id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.02999 | Nathan Broomhead | Nathan Broomhead, Mariam Pirashvili | An isometry theorem for persistent homology of circle-valued functions | 33 pages | null | null | null | math.AT math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper explores persistence modules for circle-valued functions, presenting a new extension of the interleaving and bottleneck distances in this setting. We propose a natural generalisation of barcodes in terms of arcs on a geometric model associated to the derived category of quiver representations. The main result is an isometry theorem that establishes an equivalence between the interleaving distance and the bottleneck distance for circle-valued persistence modules. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:35:42 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Broomhead",
"Nathan",
""
],
[
"Pirashvili",
"Mariam",
""
]
] |
2506.03000 | Giovanni Gaudino | Giovanni Gaudino | $\mathcal{B}(B\rightarrow\tau\nu)$ measurement with the hadronic FEI at Belle~II | Contribution to the 2025 Electroweak session of the 59th Rencontres de Moriond | null | null | null | hep-ex | http://creativecommons.org/publicdomain/zero/1.0/ | We presented at Moriond EW 2025 a measurement of the branching fraction of $B\to\tau\nu$ decays using $(387\pm 6) \times 10^6$ $\Upsilon (4S)$ collected between 2019 and 2022 with the Belle~II detector at the SuperKEKB $e^+e^-$ collider. We reconstruct the accompanying $B^-$ meson using the hadronic tagging method, while $B\to\tau\nu$ candidates are identified in the recoil. We find evidence for $B\to\tau\nu$ decays at 3.0 standard deviations, including systematic uncertainties. The measured branching fraction is $\mathcal{B}(B\rightarrow\tau\nu) = [1.24 \pm 0.41~\text{(stat.)} \pm 0.19~\text{(syst.)}] \times 10^{-4}$ | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:37:52 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Gaudino",
"Giovanni",
""
]
] |
2506.03001 | Yury Yanovich | Irina Lebedeva, Dmitrii Umnov, Yury Yanovich, Ignat Melnikov, George Ovchinnikov | Dynamic Fee for Reducing Impermanent Loss in Decentralized Exchanges | To be presented on IEEE ICBC 2025 | null | null | null | cs.GT cs.DC | http://creativecommons.org/publicdomain/zero/1.0/ | Decentralized exchanges (DEXs) are crucial to decentralized finance (DeFi) as they enable trading without intermediaries. However, they face challenges like impermanent loss (IL), where liquidity providers (LPs) see their assets' value change unfavorably within a liquidity pool compared to outside it. To tackle these issues, we propose dynamic fee mechanisms over traditional fixed-fee structures used in automated market makers (AMM). Our solution includes asymmetric fees via block-adaptive, deal-adaptive, and the "ideal but unattainable" oracle-based fee algorithm, utilizing all data available to arbitrageurs to mitigate IL. We developed a simulation-based framework to compare these fee algorithms systematically. This framework replicates trading on a DEX, considering both informed and uninformed users and a psychological relative loss factor. Results show that adaptive algorithms outperform fixed-fee baselines in reducing IL while maintaining trading activity among uninformed users. Additionally, insights from oracle-based performance underscore the potential of dynamic fee strategies to lower IL, boost LP profitability, and enhance overall market efficiency. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:41:30 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lebedeva",
"Irina",
""
],
[
"Umnov",
"Dmitrii",
""
],
[
"Yanovich",
"Yury",
""
],
[
"Melnikov",
"Ignat",
""
],
[
"Ovchinnikov",
"George",
""
]
] |
2506.03002 | Alex NieMiera | Alex NieMiera, William Good, Huey-Wen Lin | Kaon Gluon Parton Distribution and Momentum Fraction from 2+1+1 Lattice-QCD with High Statistics | null | null | null | null | hep-lat | http://creativecommons.org/licenses/by/4.0/ | We present a high-statistics lattice-QCD determination of the kaon gluon parton distribution function and gluon momentum fraction. We use clover valence fermion action to take 1,296,640 kaon-correlator measurements on a HISQ ensemble with $a \approx 0.12$~fm and 310-MeV pion mass, generated by the MILC collaboration. A detailed investigation into the impact of gauge-link smearing on the gluonic matrix elements indicates that five steps of hypercubic smearing offer an effective balance between signal quality and preservation of long-distance physics. We report a nonperturbatively renormalized kaon gluon momentum fraction of $\langle x \rangle_g^{\overline{\text{MS}}, K} = 0.557(18)_\text{stat}(24)_\text{NPR}(56)_\text{mixing}$ at $\mu = 2$ GeV in in the $\overline{\text{MS}}$ scheme. Using reduced pseudo-ITD matrix elements and pseudo-PDF matching, we extract the kaon gluon PDF and compare with the prediction from the Dyson-Schwinger equation and with the pion PDF obtained from the same ensemble. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:42:21 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"NieMiera",
"Alex",
""
],
[
"Good",
"William",
""
],
[
"Lin",
"Huey-Wen",
""
]
] |
2506.03003 | Sheehan Olver | Sheehan Olver | Newtonian potentials of Legendre polynomials on rectangles have displacement structure | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | Particular solutions of the Poisson equation can be constructed via Newtonian potentials, integrals involving the corresponding Green's function which in two-dimensions has a logarithmic singularity. The singularity represents a significant challenge for computing the integrals, which is typically overcome via specially designed quadrature methods involving a large number of evaluations of the function and kernel. We present an attractive alternative: we show that Newtonian potentials (and their gradient) applied to (tensor products of) Legendre polynomials can be expressed in terms of complex integrals which satisfy simple and explicit recurrences that can be utilised to exactly compute singular integrals, i.e., singular integral quadrature is completely avoided. The inhomogeneous part of the recurrence has low rank structure (its rank is at most three for the Newtonian potential) and hence these recurrences have displacement structure. Using the recurrence directly is a fast approach for evaluation on or near the integration domain that remains accurate for low degree polynomial approximations, while high-precision arithmetic allows accurate use of the approach for moderate degree polynomials. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:42:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Olver",
"Sheehan",
""
]
] |
2506.03004 | Junyu Liu | Junyu Liu, R. Kenny Jones, Daniel Ritchie | PartComposer: Learning and Composing Part-Level Concepts from Single-Image Examples | null | null | null | null | cs.GR cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present PartComposer: a framework for part-level concept learning from single-image examples that enables text-to-image diffusion models to compose novel objects from meaningful components. Existing methods either struggle with effectively learning fine-grained concepts or require a large dataset as input. We propose a dynamic data synthesis pipeline generating diverse part compositions to address one-shot data scarcity. Most importantly, we propose to maximize the mutual information between denoised latents and structured concept codes via a concept predictor, enabling direct regulation on concept disentanglement and re-composition supervision. Our method achieves strong disentanglement and controllable composition, outperforming subject and part-level baselines when mixing concepts from the same, or different, object categories. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:43:28 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Liu",
"Junyu",
""
],
[
"Jones",
"R. Kenny",
""
],
[
"Ritchie",
"Daniel",
""
]
] |
2506.03005 | Mahmoud Melkemi Dr | Mahmoud Melkemi | Multi-scale representation of integer sets: application to prime numbers | null | null | null | null | math.RA | http://creativecommons.org/licenses/by/4.0/ | We propose a multi-scale analysis method for studying arithmetic properties of integer sets, such as primality. Our approach organizes information through a hierarchy of nested sequences, where each level enables a hierarchical expression of the studied property by examining patterns at varying levels of granularity. To illustrate the method, we apply it to three cases: the distribution of prime numbers, twin primes, and Mersenne numbers. While this does not claim any new breakthroughs on these classical problems, the approach allows for efficient analysis of the studied property across large integer sequences and reveals new characteristics observable at different scales. This framework provides an alternative perspective for investigating the fundamental arithmetic properties. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:44:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Melkemi",
"Mahmoud",
""
]
] |
2506.03006 | Zhiyuan Peng | Zhiyuan Peng, Xin Yin, Chenhao Ying, Chao Ni, Yuan Luo | A Preference-Driven Methodology for High-Quality Solidity Code Generation | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | While Large Language Models (LLMs) have demonstrated remarkable progress in generating functionally correct Solidity code, they continue to face critical challenges in producing gas-efficient and secure code, which are critical requirements for real-world smart contract deployment. Although recent advances leverage Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) for code preference alignment, existing approaches treat functional correctness, gas optimization, and security as independent objectives, resulting in contracts that may achieve operational soundness but suffer from prohibitive execution costs or dangerous vulnerabilities. To address these limitations, we propose PrefGen, a novel framework that extends standard DPO beyond human preferences to incorporate quantifiable blockchain-specific metrics, enabling holistic multi-objective optimization specifically tailored for smart contract generation. Our framework introduces a comprehensive evaluation methodology with four complementary metrics: Pass@k (functional correctness), Compile@k (syntactic correctness), Gas@k (gas efficiency), and Secure@k (security assessment), providing rigorous multi-dimensional contract evaluation. Through extensive experimentation, we demonstrate that PrefGen significantly outperforms existing approaches across all critical dimensions, achieving 66.7% Pass@5, 58.9% Gas@5, and 62.5% Secure@5, while generating production-ready smart contracts that are functionally correct, cost-efficient, and secure. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:45:31 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Peng",
"Zhiyuan",
""
],
[
"Yin",
"Xin",
""
],
[
"Ying",
"Chenhao",
""
],
[
"Ni",
"Chao",
""
],
[
"Luo",
"Yuan",
""
]
] |
2506.03007 | JiaRui Wang | Jiarui Wang, Huiyu Duan, Juntong Wang, Ziheng Jia, Woo Yi Yang, Xiaorong Zhu, Yu Zhao, Jiaying Qian, Yuke Xing, Guangtao Zhai, Xiongkuo Min | DFBench: Benchmarking Deepfake Image Detection Capability of Large Multimodal Models | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | With the rapid advancement of generative models, the realism of AI-generated images has significantly improved, posing critical challenges for verifying digital content authenticity. Current deepfake detection methods often depend on datasets with limited generation models and content diversity that fail to keep pace with the evolving complexity and increasing realism of the AI-generated content. Large multimodal models (LMMs), widely adopted in various vision tasks, have demonstrated strong zero-shot capabilities, yet their potential in deepfake detection remains largely unexplored. To bridge this gap, we present \textbf{DFBench}, a large-scale DeepFake Benchmark featuring (i) broad diversity, including 540,000 images across real, AI-edited, and AI-generated content, (ii) latest model, the fake images are generated by 12 state-of-the-art generation models, and (iii) bidirectional benchmarking and evaluating for both the detection accuracy of deepfake detectors and the evasion capability of generative models. Based on DFBench, we propose \textbf{MoA-DF}, Mixture of Agents for DeepFake detection, leveraging a combined probability strategy from multiple LMMs. MoA-DF achieves state-of-the-art performance, further proving the effectiveness of leveraging LMMs for deepfake detection. Database and codes are publicly available at https://github.com/IntMeGroup/DFBench. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:45:41 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wang",
"Jiarui",
""
],
[
"Duan",
"Huiyu",
""
],
[
"Wang",
"Juntong",
""
],
[
"Jia",
"Ziheng",
""
],
[
"Yang",
"Woo Yi",
""
],
[
"Zhu",
"Xiaorong",
""
],
[
"Zhao",
"Yu",
""
],
[
"Qian",
"Jiaying",
""
],
[
"Xing",
"Yuke",
""
],
[
"Zhai",
"Guangtao",
""
],
[
"Min",
"Xiongkuo",
""
]
] |
2506.03008 | Ziad Sakr | Euclid Collaboration: I. S. Albuquerque (1), N. Frusciante (2), Z. Sakr (3 and 4 and 5), S. Srinivasan (6), L. Atayde (1), B. Bose (7), V. F. Cardone (8 and 9), S. Casas (10 and 11), M. Martinelli (8 and 9), J. Noller (12 and 11), E. M. Teixeira (13), D. B. Thomas (14), I. Tutusaus (4), M. Cataneo (15 and 16), K. Koyama (11), L. Lombriser (17), F. Pace (18 and 19 and 20), A. Silvestri (21), N. Aghanim (22), A. Amara (23), S. Andreon (24), N. Auricchio (25), C. Baccigalupi (26 and 27 and 28 and 29), M. Baldi (30 and 25 and 31), S. Bardelli (25), A. Biviano (27 and 26), D. Bonino (20), E. Branchini (32 and 33 and 24), M. Brescia (34 and 35 and 36), J. Brinchmann (37 and 38), S. Camera (18 and 19 and 20), G. Ca\~nas-Herrera (39 and 21), V. Capobianco (20), C. Carbone (40), J. Carretero (41 and 42), M. Castellano (8), G. Castignani (25), S. Cavuoti (35 and 36), K. C. Chambers (43), A. Cimatti (44), C. Colodro-Conde (45), G. Congedo (7), C. J. Conselice (14), L. Conversi (46 and 47), Y. Copin (48), L. Corcione (20), F. Courbin (49 and 50 and 51), H. M. Courtois (52), A. Da Silva (53 and 1), H. Degaudenzi (54), S. de la Torre (55), G. De Lucia (27), A. M. Di Giorgio (56), H. Dole (22), F. Dubath (54), C. A. J. Duncan (14), X. Dupac (47), S. Dusini (57), A. Ealet (48), S. Escoffier (58), M. Farina (56), S. Farrens (59), F. Faustini (60 and 8), S. Ferriol (48), F. Finelli (25 and 61), P. Fosalba (62 and 63), S. Fotopoulou (64), M. Frailis (27), E. Franceschi (25), M. Fumana (40), S. Galeotta (27), B. Gillis (7), C. Giocoli (25 and 31), J. Gracia-Carpio (65), A. Grazian (66), F. Grupp (65 and 6), L. Guzzo (67 and 24 and 68), S. V. H. Haugan (69), W. Holmes (70), F. Hormuth (71), A. Hornstrup (72 and 73), P. Hudelot (74), S. Ili\'c (75 and 4), K. Jahnke (76), M. Jhabvala (77), B. Joachimi (12), E. Keih\"anen (78), S. Kermiche (58), A. Kiessling (70), M. Kilbinger (59), B. Kubik (48), M. Kunz (17), H. Kurki-Suonio (79 and 80), A. M. C. Le Brun (81), S. Ligori (20), P. B. Lilje (69), V. Lindholm (79 and 80), I. Lloro (82), G. Mainetti (83), D. Maino (67 and 40 and 68), E. Maiorano (25), O. Mansutti (27), O. Marggraf (16), K. Markovic (70), N. Martinet (55), F. Marulli (84 and 25 and 31), R. Massey (85), E. Medinaceli (25), S. Mei (86), Y. Mellier (87 and 74), M. Meneghetti (25 and 31), E. Merlin (8), G. Meylan (49), A. Mora (88), M. Moresco (84 and 25), L. Moscardini (84 and 25 and 31), R. C. Nichol (23), S.-M. Niemi (39), J. W. Nightingale (89), C. Padilla (90), S. Paltani (54), F. Pasian (27), W. J. Percival (91 and 92 and 93), V. Pettorino (39), S. Pires (59), G. Polenta (60), M. Poncet (94), L. A. Popa (95), L. Pozzetti (25), F. Raison (65), R. Rebolo (45 and 96 and 97), A. Renzi (98 and 57), J. Rhodes (70), G. Riccio (35), E. Romelli (27), M. Roncarelli (25), R. Saglia (6 and 65), D. Sapone (99), B. Sartoris (6 and 27), J. A. Schewtschenko (7), M. Schirmer (76), T. Schrabback (100), A. Secroun (58), E. Sefusatti (27 and 26 and 28), G. Seidel (76), S. Serrano (62 and 101 and 63), C. Sirignano (98 and 57), G. Sirri (31), L. Stanco (57), J. Steinwagner (65), P. Tallada-Cresp\'i (41 and 42), A. N. Taylor (7), I. Tereno (53 and 102), N. Tessore (12), S. Toft (103 and 104), R. Toledo-Moreo (105), F. Torradeflot (42 and 41), E. A. Valentijn (106), L. Valenziano (25 and 61), J. Valiviita (79 and 80), T. Vassallo (6 and 27), G. Verdoes Kleijn (106), A. Veropalumbo (24 and 33 and 107), Y. Wang (108), J. Weller (6 and 65), G. Zamorani (25), E. Zucca (25), E. Bozzo (54), C. Burigana (109 and 61), M. Calabrese (110 and 40), D. Di Ferdinando (31), J. A. Escartin Vigo (65), S. Matthew (7), N. Mauri (44 and 31), A. Pezzotta (65), M. P\"ontinen (79), C. Porciani (16), V. Scottez (87 and 111), M. Tenti (31), M. Viel (26 and 27 and 29 and 28 and 112), M. Wiesmann (69), Y. Akrami (113 and 114), V. Allevato (35), S. Alvi (115), S. Anselmi (57 and 98 and 81), M. Archidiacono (67 and 68), F. Atrio-Barandela (116), A. Balaguera-Antolinez (45 and 96), M. Ballardini (115 and 25 and 117), D. Bertacca (98 and 66 and 57), A. Blanchard (4), L. Blot (118 and 81), S. Borgani (119 and 26 and 27 and 28 and 112), M. L. Brown (14), S. Bruton (120), R. Cabanac (4), A. Calabro (8), B. Camacho Quevedo (62 and 63), A. Cappi (25 and 121), F. Caro (8), C. S. Carvalho (102), T. Castro (27 and 28 and 26 and 112), F. Cogato (84 and 25), A. R. Cooray (122), S. Davini (33), G. Desprez (123), A. D\'iaz-S\'anchez (124), S. Di Domizio (32 and 33), A. G. Ferrari (31), I. Ferrero (69), A. Finoguenov (79), A. Fontana (8), F. Fornari (61), L. Gabarra (125), K. Ganga (86), J. Garc\'ia-Bellido (113), T. Gasparetto (27), V. Gautard (126), E. Gaztanaga (63 and 62 and 11), F. Giacomini (31), F. Gianotti (25), G. Gozaliasl (127), A. Gregorio (119 and 27 and 28), M. Guidi (30 and 25), C. M. Gutierrez (128), S. Hemmati (129), H. Hildebrandt (15), J. Hjorth (130), A. Jimenez Mu\~noz (131), J. J. E. Kajava (132 and 133), Y. Kang (54), V. Kansal (134 and 135), D. Karagiannis (136 and 137), C. C. Kirkpatrick (78), S. Kruk (47), M. Lattanzi (117), S. Lee (70), J. Le Graet (58), L. Legrand (138 and 139), M. Lembo (115 and 117), J. Lesgourgues (10), T. I. Liaudat (140), A. Loureiro (141 and 142), J. Macias-Perez (131), M. Magliocchetti (56), F. Mannucci (143), R. Maoli (144 and 8), J. Mart\'in-Fleitas (88), C. J. A. P. Martins (145 and 37), L. Maurin (22), R. B. Metcalf (84 and 25), M. Migliaccio (146 and 147), M. Miluzio (47 and 148), P. Monaco (119 and 27 and 28 and 26), C. Moretti (29 and 112 and 27 and 26 and 28), G. Morgante (25), C. Murray (86), S. Nadathur (11), K. Naidoo (12), Nicholas A. Walton (149), K. Paterson (76), L. Patrizii (31), A. Pisani (58 and 150), V. Popa (95), D. Potter (151), I. Risso (152), P.-F. Rocci (22), M. Sahl\'en (153), E. Sarpa (29 and 112 and 28), A. Schneider (151), M. Schultheis (121), D. Sciotti (8 and 9), E. Sellentin (154 and 155), M. Sereno (25 and 31), L. C. Smith (149), K. Tanidis (125), C. Tao (58), G. Testera (33), R. Teyssier (150), S. Tosi (32 and 33), A. Troja (98 and 57), M. Tucci (54), C. Valieri (31), D. Vergani (25), F. Vernizzi (156), G. Verza (157 and 158), P. Vielzeuf (58) | Euclid preparation. Constraining parameterised models of modifications of gravity with the spectroscopic and photometric primary probes | 21 pages, 9 figures | null | null | null | astro-ph.CO gr-qc | http://creativecommons.org/licenses/by/4.0/ | The Euclid mission has the potential to understand the fundamental physical nature of late-time cosmic acceleration and, as such, of deviations from the standard cosmological model, LCDM. In this paper, we focus on model-independent methods to modify the evolution of scalar perturbations at linear scales. We consider two approaches: the first is based on the two phenomenological modified gravity (PMG) parameters, $\mu_{\rm mg}$ and $\Sigma_{\rm mg}$, which are phenomenologically connected to the clustering of matter and weak lensing, respectively; and the second is the effective field theory (EFT) of dark energy and modified gravity, which we use to parameterise the braiding function, $\alpha_{\rm B}$, which defines the mixing between the metric and the dark energy field. We discuss the predictions from spectroscopic and photometric primary probes by Euclid on the cosmological parameters and a given set of additional parameters featuring the PMG and EFT models. We use the Fisher matrix method applied to spectroscopic galaxy clustering (GCsp), weak lensing (WL), photometric galaxy clustering (GCph), and cross-correlation (XC) between GCph and WL. For the modelling of photometric predictions on nonlinear scales, we use the halo model to cover two limits for the screening mechanism: the unscreened (US) case, for which the screening mechanism is not present; and the super-screened (SS) case, which assumes strong screening. We also assume scale cuts to account for our uncertainties in the modelling of nonlinear perturbation evolution. We choose a time-dependent form for $\{\mu_{\rm mg},\Sigma_{\rm mg}\}$, with two fiducial sets of values for the corresponding model parameters at the present time, $\{\bar{\mu}_0,\bar{\Sigma}_0\}$, and two forms for $\alpha_{\rm B}$, with one fiducial set of values for each of the model parameters, $\alpha_{\rm B,0}$ and $\{\alpha_{\rm B,0},m\}$. (Abridged) | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:48:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Euclid Collaboration",
"",
""
],
[
"Albuquerque",
"I. S.",
"",
"3 and 4 and 5"
],
[
"Frusciante",
"N.",
"",
"3 and 4 and 5"
],
[
"Sakr",
"Z.",
"",
"3 and 4 and 5"
],
[
"Srinivasan",
"S.",
"",
"8 and 9"
],
[
"Atayde",
"L.",
"",
"8 and 9"
],
[
"Bose",
"B.",
"",
"8 and 9"
],
[
"Cardone",
"V. F.",
"",
"8 and 9"
],
[
"Casas",
"S.",
"",
"10 and 11"
],
[
"Martinelli",
"M.",
"",
"8 and 9"
],
[
"Noller",
"J.",
"",
"12 and 11"
],
[
"Teixeira",
"E. M.",
"",
"15 and 16"
],
[
"Thomas",
"D. B.",
"",
"15 and 16"
],
[
"Tutusaus",
"I.",
"",
"15 and 16"
],
[
"Cataneo",
"M.",
"",
"15 and 16"
],
[
"Koyama",
"K.",
"",
"18 and 19 and 20"
],
[
"Lombriser",
"L.",
"",
"18 and 19 and 20"
],
[
"Pace",
"F.",
"",
"18 and 19 and 20"
],
[
"Silvestri",
"A.",
"",
"26 and 27 and 28 and 29"
],
[
"Aghanim",
"N.",
"",
"26 and 27 and 28 and 29"
],
[
"Amara",
"A.",
"",
"26 and 27 and 28 and 29"
],
[
"Andreon",
"S.",
"",
"26 and 27 and 28 and 29"
],
[
"Auricchio",
"N.",
"",
"26 and 27 and 28 and 29"
],
[
"Baccigalupi",
"C.",
"",
"26 and 27 and 28 and 29"
],
[
"Baldi",
"M.",
"",
"30 and 25 and 31"
],
[
"Bardelli",
"S.",
"",
"27 and 26"
],
[
"Biviano",
"A.",
"",
"27 and 26"
],
[
"Bonino",
"D.",
"",
"32 and 33 and 24"
],
[
"Branchini",
"E.",
"",
"32 and 33 and 24"
],
[
"Brescia",
"M.",
"",
"34 and 35 and 36"
],
[
"Brinchmann",
"J.",
"",
"37 and 38"
],
[
"Camera",
"S.",
"",
"18 and 19 and 20"
],
[
"Cañas-Herrera",
"G.",
"",
"39 and 21"
],
[
"Capobianco",
"V.",
"",
"41 and 42"
],
[
"Carbone",
"C.",
"",
"41 and 42"
],
[
"Carretero",
"J.",
"",
"41 and 42"
],
[
"Castellano",
"M.",
"",
"35 and 36"
],
[
"Castignani",
"G.",
"",
"35 and 36"
],
[
"Cavuoti",
"S.",
"",
"35 and 36"
],
[
"Chambers",
"K. C.",
"",
"46 and 47"
],
[
"Cimatti",
"A.",
"",
"46 and 47"
],
[
"Colodro-Conde",
"C.",
"",
"46 and 47"
],
[
"Congedo",
"G.",
"",
"46 and 47"
],
[
"Conselice",
"C. J.",
"",
"46 and 47"
],
[
"Conversi",
"L.",
"",
"46 and 47"
],
[
"Copin",
"Y.",
"",
"49 and 50 and 51"
],
[
"Corcione",
"L.",
"",
"49 and 50 and 51"
],
[
"Courbin",
"F.",
"",
"49 and 50 and 51"
],
[
"Courtois",
"H. M.",
"",
"53 and 1"
],
[
"Da Silva",
"A.",
"",
"53 and 1"
],
[
"Degaudenzi",
"H.",
"",
"60 and 8"
],
[
"de la Torre",
"S.",
"",
"60 and 8"
],
[
"De Lucia",
"G.",
"",
"60 and 8"
],
[
"Di Giorgio",
"A. M.",
"",
"60 and 8"
],
[
"Dole",
"H.",
"",
"60 and 8"
],
[
"Dubath",
"F.",
"",
"60 and 8"
],
[
"Duncan",
"C. A. J.",
"",
"60 and 8"
],
[
"Dupac",
"X.",
"",
"60 and 8"
],
[
"Dusini",
"S.",
"",
"60 and 8"
],
[
"Ealet",
"A.",
"",
"60 and 8"
],
[
"Escoffier",
"S.",
"",
"60 and 8"
],
[
"Farina",
"M.",
"",
"60 and 8"
],
[
"Farrens",
"S.",
"",
"60 and 8"
],
[
"Faustini",
"F.",
"",
"60 and 8"
],
[
"Ferriol",
"S.",
"",
"25 and 61"
],
[
"Finelli",
"F.",
"",
"25 and 61"
],
[
"Fosalba",
"P.",
"",
"62 and 63"
],
[
"Fotopoulou",
"S.",
"",
"25 and 31"
],
[
"Frailis",
"M.",
"",
"25 and 31"
],
[
"Franceschi",
"E.",
"",
"25 and 31"
],
[
"Fumana",
"M.",
"",
"25 and 31"
],
[
"Galeotta",
"S.",
"",
"25 and 31"
],
[
"Gillis",
"B.",
"",
"25 and 31"
],
[
"Giocoli",
"C.",
"",
"25 and 31"
],
[
"Gracia-Carpio",
"J.",
"",
"65 and 6"
],
[
"Grazian",
"A.",
"",
"65 and 6"
],
[
"Grupp",
"F.",
"",
"65 and 6"
],
[
"Guzzo",
"L.",
"",
"67 and 24 and 68"
],
[
"Haugan",
"S. V. H.",
"",
"72 and 73"
],
[
"Holmes",
"W.",
"",
"72 and 73"
],
[
"Hormuth",
"F.",
"",
"72 and 73"
],
[
"Hornstrup",
"A.",
"",
"72 and 73"
],
[
"Hudelot",
"P.",
"",
"75 and 4"
],
[
"Ilić",
"S.",
"",
"75 and 4"
],
[
"Jahnke",
"K.",
"",
"79 and 80"
],
[
"Jhabvala",
"M.",
"",
"79 and 80"
],
[
"Joachimi",
"B.",
"",
"79 and 80"
],
[
"Keihänen",
"E.",
"",
"79 and 80"
],
[
"Kermiche",
"S.",
"",
"79 and 80"
],
[
"Kiessling",
"A.",
"",
"79 and 80"
],
[
"Kilbinger",
"M.",
"",
"79 and 80"
],
[
"Kubik",
"B.",
"",
"79 and 80"
],
[
"Kunz",
"M.",
"",
"79 and 80"
],
[
"Kurki-Suonio",
"H.",
"",
"79 and 80"
],
[
"Brun",
"A. M. C. Le",
"",
"79 and 80"
],
[
"Ligori",
"S.",
"",
"79 and 80"
],
[
"Lilje",
"P. B.",
"",
"79 and 80"
],
[
"Lindholm",
"V.",
"",
"79 and 80"
],
[
"Lloro",
"I.",
"",
"67 and 40 and 68"
],
[
"Mainetti",
"G.",
"",
"67 and 40 and 68"
],
[
"Maino",
"D.",
"",
"67 and 40 and 68"
],
[
"Maiorano",
"E.",
"",
"84 and 25 and 31"
],
[
"Mansutti",
"O.",
"",
"84 and 25 and 31"
],
[
"Marggraf",
"O.",
"",
"84 and 25 and 31"
],
[
"Markovic",
"K.",
"",
"84 and 25 and 31"
],
[
"Martinet",
"N.",
"",
"84 and 25 and 31"
],
[
"Marulli",
"F.",
"",
"84 and 25 and 31"
],
[
"Massey",
"R.",
"",
"87 and 74"
],
[
"Medinaceli",
"E.",
"",
"87 and 74"
],
[
"Mei",
"S.",
"",
"87 and 74"
],
[
"Mellier",
"Y.",
"",
"87 and 74"
],
[
"Meneghetti",
"M.",
"",
"25 and 31"
],
[
"Merlin",
"E.",
"",
"84 and 25"
],
[
"Meylan",
"G.",
"",
"84 and 25"
],
[
"Mora",
"A.",
"",
"84 and 25"
],
[
"Moresco",
"M.",
"",
"84 and 25"
],
[
"Moscardini",
"L.",
"",
"84 and 25 and 31"
],
[
"Nichol",
"R. C.",
"",
"91 and 92 and 93"
],
[
"Niemi",
"S. -M.",
"",
"91 and 92 and 93"
],
[
"Nightingale",
"J. W.",
"",
"91 and 92 and 93"
],
[
"Padilla",
"C.",
"",
"91 and 92 and 93"
],
[
"Paltani",
"S.",
"",
"91 and 92 and 93"
],
[
"Pasian",
"F.",
"",
"91 and 92 and 93"
],
[
"Percival",
"W. J.",
"",
"91 and 92 and 93"
],
[
"Pettorino",
"V.",
"",
"45 and 96 and 97"
],
[
"Pires",
"S.",
"",
"45 and 96 and 97"
],
[
"Polenta",
"G.",
"",
"45 and 96 and 97"
],
[
"Poncet",
"M.",
"",
"45 and 96 and 97"
],
[
"Popa",
"L. A.",
"",
"45 and 96 and 97"
],
[
"Pozzetti",
"L.",
"",
"45 and 96 and 97"
],
[
"Raison",
"F.",
"",
"45 and 96 and 97"
],
[
"Rebolo",
"R.",
"",
"45 and 96 and 97"
],
[
"Renzi",
"A.",
"",
"98 and 57"
],
[
"Rhodes",
"J.",
"",
"6 and 65"
],
[
"Riccio",
"G.",
"",
"6 and 65"
],
[
"Romelli",
"E.",
"",
"6 and 65"
],
[
"Roncarelli",
"M.",
"",
"6 and 65"
],
[
"Saglia",
"R.",
"",
"6 and 65"
],
[
"Sapone",
"D.",
"",
"6 and 27"
],
[
"Sartoris",
"B.",
"",
"6 and 27"
],
[
"Schewtschenko",
"J. A.",
"",
"27 and 26 and 28"
],
[
"Schirmer",
"M.",
"",
"27 and 26 and 28"
],
[
"Schrabback",
"T.",
"",
"27 and 26 and 28"
],
[
"Secroun",
"A.",
"",
"27 and 26 and 28"
],
[
"Sefusatti",
"E.",
"",
"27 and 26 and 28"
],
[
"Seidel",
"G.",
"",
"62 and 101 and 63"
],
[
"Serrano",
"S.",
"",
"62 and 101 and 63"
],
[
"Sirignano",
"C.",
"",
"98 and 57"
],
[
"Sirri",
"G.",
"",
"41 and 42"
],
[
"Stanco",
"L.",
"",
"41 and 42"
],
[
"Steinwagner",
"J.",
"",
"41 and 42"
],
[
"Tallada-Crespí",
"P.",
"",
"41 and 42"
],
[
"Taylor",
"A. N.",
"",
"53 and 102"
],
[
"Tereno",
"I.",
"",
"53 and 102"
],
[
"Tessore",
"N.",
"",
"103 and 104"
],
[
"Toft",
"S.",
"",
"103 and 104"
],
[
"Toledo-Moreo",
"R.",
"",
"42 and 41"
],
[
"Torradeflot",
"F.",
"",
"42 and 41"
],
[
"Valentijn",
"E. A.",
"",
"25 and 61"
],
[
"Valenziano",
"L.",
"",
"25 and 61"
],
[
"Valiviita",
"J.",
"",
"79 and 80"
],
[
"Vassallo",
"T.",
"",
"6 and 27"
],
[
"Kleijn",
"G. Verdoes",
"",
"24 and 33 and 107"
],
[
"Veropalumbo",
"A.",
"",
"24 and 33 and 107"
],
[
"Wang",
"Y.",
"",
"6 and 65"
],
[
"Weller",
"J.",
"",
"6 and 65"
],
[
"Zamorani",
"G.",
"",
"109 and 61"
],
[
"Zucca",
"E.",
"",
"109 and 61"
],
[
"Bozzo",
"E.",
"",
"109 and 61"
],
[
"Burigana",
"C.",
"",
"109 and 61"
],
[
"Calabrese",
"M.",
"",
"110 and 40"
],
[
"Di Ferdinando",
"D.",
"",
"44 and 31"
],
[
"Vigo",
"J. A. Escartin",
"",
"44 and 31"
],
[
"Matthew",
"S.",
"",
"44 and 31"
],
[
"Mauri",
"N.",
"",
"44 and 31"
],
[
"Pezzotta",
"A.",
"",
"87 and 111"
],
[
"Pöntinen",
"M.",
"",
"87 and 111"
],
[
"Porciani",
"C.",
"",
"87 and 111"
],
[
"Scottez",
"V.",
"",
"87 and 111"
],
[
"Tenti",
"M.",
"",
"26 and 27 and 29 and 28 and 112"
],
[
"Viel",
"M.",
"",
"26 and 27 and 29 and 28 and 112"
],
[
"Wiesmann",
"M.",
"",
"113 and 114"
],
[
"Akrami",
"Y.",
"",
"113 and 114"
],
[
"Allevato",
"V.",
"",
"57 and 98 and 81"
],
[
"Alvi",
"S.",
"",
"57 and 98 and 81"
],
[
"Anselmi",
"S.",
"",
"57 and 98 and 81"
],
[
"Archidiacono",
"M.",
"",
"67 and 68"
],
[
"Atrio-Barandela",
"F.",
"",
"45 and 96"
],
[
"Balaguera-Antolinez",
"A.",
"",
"45 and 96"
],
[
"Ballardini",
"M.",
"",
"115 and 25 and 117"
],
[
"Bertacca",
"D.",
"",
"98 and 66 and 57"
],
[
"Blanchard",
"A.",
"",
"118 and 81"
],
[
"Blot",
"L.",
"",
"118 and 81"
],
[
"Borgani",
"S.",
"",
"119 and 26 and 27 and 28 and 112"
],
[
"Brown",
"M. L.",
"",
"62 and 63"
],
[
"Bruton",
"S.",
"",
"62 and 63"
],
[
"Cabanac",
"R.",
"",
"62 and 63"
],
[
"Calabro",
"A.",
"",
"62 and 63"
],
[
"Quevedo",
"B. Camacho",
"",
"62 and 63"
],
[
"Cappi",
"A.",
"",
"25 and 121"
],
[
"Caro",
"F.",
"",
"27 and 28 and 26 and 112"
],
[
"Carvalho",
"C. S.",
"",
"27 and 28 and 26 and 112"
],
[
"Castro",
"T.",
"",
"27 and 28 and 26 and 112"
],
[
"Cogato",
"F.",
"",
"84 and 25"
],
[
"Cooray",
"A. R.",
"",
"32 and 33"
],
[
"Davini",
"S.",
"",
"32 and 33"
],
[
"Desprez",
"G.",
"",
"32 and 33"
],
[
"Díaz-Sánchez",
"A.",
"",
"32 and 33"
],
[
"Di Domizio",
"S.",
"",
"32 and 33"
],
[
"Ferrari",
"A. G.",
"",
"63 and 62 and 11"
],
[
"Ferrero",
"I.",
"",
"63 and 62 and 11"
],
[
"Finoguenov",
"A.",
"",
"63 and 62 and 11"
],
[
"Fontana",
"A.",
"",
"63 and 62 and 11"
],
[
"Fornari",
"F.",
"",
"63 and 62 and 11"
],
[
"Gabarra",
"L.",
"",
"63 and 62 and 11"
],
[
"Ganga",
"K.",
"",
"63 and 62 and 11"
],
[
"García-Bellido",
"J.",
"",
"63 and 62 and 11"
],
[
"Gasparetto",
"T.",
"",
"63 and 62 and 11"
],
[
"Gautard",
"V.",
"",
"63 and 62 and 11"
],
[
"Gaztanaga",
"E.",
"",
"63 and 62 and 11"
],
[
"Giacomini",
"F.",
"",
"119 and 27 and 28"
],
[
"Gianotti",
"F.",
"",
"119 and 27 and 28"
],
[
"Gozaliasl",
"G.",
"",
"119 and 27 and 28"
],
[
"Gregorio",
"A.",
"",
"119 and 27 and 28"
],
[
"Guidi",
"M.",
"",
"30 and 25"
],
[
"Gutierrez",
"C. M.",
"",
"132 and 133"
],
[
"Hemmati",
"S.",
"",
"132 and 133"
],
[
"Hildebrandt",
"H.",
"",
"132 and 133"
],
[
"Hjorth",
"J.",
"",
"132 and 133"
],
[
"Muñoz",
"A. Jimenez",
"",
"132 and 133"
],
[
"Kajava",
"J. J. E.",
"",
"132 and 133"
],
[
"Kang",
"Y.",
"",
"134 and 135"
],
[
"Kansal",
"V.",
"",
"134 and 135"
],
[
"Karagiannis",
"D.",
"",
"136 and 137"
],
[
"Kirkpatrick",
"C. C.",
"",
"138 and 139"
],
[
"Kruk",
"S.",
"",
"138 and 139"
],
[
"Lattanzi",
"M.",
"",
"138 and 139"
],
[
"Lee",
"S.",
"",
"138 and 139"
],
[
"Graet",
"J. Le",
"",
"138 and 139"
],
[
"Legrand",
"L.",
"",
"138 and 139"
],
[
"Lembo",
"M.",
"",
"115 and 117"
],
[
"Lesgourgues",
"J.",
"",
"141 and 142"
],
[
"Liaudat",
"T. I.",
"",
"141 and 142"
],
[
"Loureiro",
"A.",
"",
"141 and 142"
],
[
"Macias-Perez",
"J.",
"",
"144 and 8"
],
[
"Magliocchetti",
"M.",
"",
"144 and 8"
],
[
"Mannucci",
"F.",
"",
"144 and 8"
],
[
"Maoli",
"R.",
"",
"144 and 8"
],
[
"Martín-Fleitas",
"J.",
"",
"145 and 37"
],
[
"Martins",
"C. J. A. P.",
"",
"145 and 37"
],
[
"Maurin",
"L.",
"",
"84 and 25"
],
[
"Metcalf",
"R. B.",
"",
"84 and 25"
],
[
"Migliaccio",
"M.",
"",
"146 and 147"
],
[
"Miluzio",
"M.",
"",
"47 and 148"
],
[
"Monaco",
"P.",
"",
"119 and 27 and 28 and 26"
],
[
"Moretti",
"C.",
"",
"29 and 112 and 27 and 26 and 28"
],
[
"Morgante",
"G.",
"",
"58 and 150"
],
[
"Murray",
"C.",
"",
"58 and 150"
],
[
"Nadathur",
"S.",
"",
"58 and 150"
],
[
"Naidoo",
"K.",
"",
"58 and 150"
],
[
"Walton",
"Nicholas A.",
"",
"58 and 150"
],
[
"Paterson",
"K.",
"",
"58 and 150"
],
[
"Patrizii",
"L.",
"",
"58 and 150"
],
[
"Pisani",
"A.",
"",
"58 and 150"
],
[
"Popa",
"V.",
"",
"29 and 112 and 28"
],
[
"Potter",
"D.",
"",
"29 and 112 and 28"
],
[
"Risso",
"I.",
"",
"29 and 112 and 28"
],
[
"Rocci",
"P. -F.",
"",
"29 and 112 and 28"
],
[
"Sahlén",
"M.",
"",
"29 and 112 and 28"
],
[
"Sarpa",
"E.",
"",
"29 and 112 and 28"
],
[
"Schneider",
"A.",
"",
"8 and 9"
],
[
"Schultheis",
"M.",
"",
"8 and 9"
],
[
"Sciotti",
"D.",
"",
"8 and 9"
],
[
"Sellentin",
"E.",
"",
"154 and 155"
],
[
"Sereno",
"M.",
"",
"25 and 31"
],
[
"Smith",
"L. C.",
"",
"32 and 33"
],
[
"Tanidis",
"K.",
"",
"32 and 33"
],
[
"Tao",
"C.",
"",
"32 and 33"
],
[
"Testera",
"G.",
"",
"32 and 33"
],
[
"Teyssier",
"R.",
"",
"32 and 33"
],
[
"Tosi",
"S.",
"",
"32 and 33"
],
[
"Troja",
"A.",
"",
"98 and 57"
],
[
"Tucci",
"M.",
"",
"157 and 158"
],
[
"Valieri",
"C.",
"",
"157 and 158"
],
[
"Vergani",
"D.",
"",
"157 and 158"
],
[
"Vernizzi",
"F.",
"",
"157 and 158"
],
[
"Verza",
"G.",
"",
"157 and 158"
],
[
"Vielzeuf",
"P.",
""
]
] |
2506.03009 | Frederike Zufall | Florian Ludwig, Torsten Zesch, Frederike Zufall | Conditioning Large Language Models on Legal Systems? Detecting Punishable Hate Speech | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The assessment of legal problems requires the consideration of a specific legal system and its levels of abstraction, from constitutional law to statutory law to case law. The extent to which Large Language Models (LLMs) internalize such legal systems is unknown. In this paper, we propose and investigate different approaches to condition LLMs at different levels of abstraction in legal systems. This paper examines different approaches to conditioning LLMs at multiple levels of abstraction in legal systems to detect potentially punishable hate speech. We focus on the task of classifying whether a specific social media posts falls under the criminal offense of incitement to hatred as prescribed by the German Criminal Code. The results show that there is still a significant performance gap between models and legal experts in the legal assessment of hate speech, regardless of the level of abstraction with which the models were conditioned. Our analysis revealed, that models conditioned on abstract legal knowledge lacked deep task understanding, often contradicting themselves and hallucinating answers, while models using concrete legal knowledge performed reasonably well in identifying relevant target groups, but struggled with classifying target conducts. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:50:27 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Ludwig",
"Florian",
""
],
[
"Zesch",
"Torsten",
""
],
[
"Zufall",
"Frederike",
""
]
] |
2506.03010 | Hao Wu | Hao Wu, Clayton Z. C. Ho, Grant D. Mitts, Joshua A. Rabinowitz and Eric R. Hudson | Floquet-engineered decoherence-resilient protocols for wideband sensing beyond the linear standard quantum limit | 13 pages; 3 figures in the main text, 5 figures in the SI | null | null | null | physics.atom-ph quant-ph | http://creativecommons.org/licenses/by/4.0/ | A key advantage of quantum metrology is the ability to surpass the standard quantum limit for measurement precision through the use of non-classical states. However, there is typically little to no improvement in precision with the use of non-classical states for measurements whose duration exceeds the decoherence time of the underlying quantum states. Measurements aimed at the ultimate possible precision are thus performed almost exclusively with classical states and, therefore, are constrained by the standard quantum limit. Here, we demonstrate that by using the phenomenon of subharmonic excitation, in combination with a recently demonstrated technique of Raman excitation of a harmonic oscillator, the frequency of an electric field can be measured at a resolution below the standard quantum limit of the corresponding linear measurement. As the input states can be classical, this metrological gain persists to long timescales and improves the ultimate possible precision. While we demonstrate this technique using motional Raman subharmonic excitation of a single 40ca^+ ion through engineered Floquet states, this technique is expected to be extendable to other platforms, such as NV centers, solid-state qubits, and neutral atoms, where it can provide metrological gain for sensing across the radio frequency, microwave, and optical domains. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:50:34 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wu",
"Hao",
""
],
[
"Ho",
"Clayton Z. C.",
""
],
[
"Mitts",
"Grant D.",
""
],
[
"Rabinowitz",
"Joshua A.",
""
],
[
"Hudson",
"Eric R.",
""
]
] |
2506.03011 | Aditya Bharat Soni | Aditya Bharat Soni, Boxuan Li, Xingyao Wang, Valerie Chen, Graham Neubig | Coding Agents with Multimodal Browsing are Generalist Problem Solvers | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Modern human labor is characterized by specialization; we train for years and develop particular tools that allow us to perform well across a variety of tasks. In addition, AI agents have been specialized for domains such as software engineering, web navigation, and workflow automation. However, this results in agents that are good for one thing but fail to generalize beyond their intended scope. One reason for this is that agent developers provide a highly specialized set of tools or make architectural decisions optimized for a specific use case or benchmark. In this work, we ask the question: what is the minimal set of general tools that can be used to achieve high performance across a diverse set of tasks? Our answer is OpenHands-Versa, a generalist agent built with a modest number of general tools: code editing and execution, web search, as well as multimodal web browsing and file access. Importantly, OpenHands-Versa demonstrates superior or competitive performance over leading specialized agents across three diverse and challenging benchmarks: SWE-Bench Multimodal, GAIA, and The Agent Company, outperforming the best-performing previously published results with absolute improvements in success rate of 9.1, 1.3, and 9.1 points respectively. Further, we show how existing state-of-the-art multi-agent systems fail to generalize beyond their target domains. These results demonstrate the feasibility of developing a generalist agent to solve diverse tasks and establish OpenHands-Versa as a strong baseline for future research. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:50:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Soni",
"Aditya Bharat",
""
],
[
"Li",
"Boxuan",
""
],
[
"Wang",
"Xingyao",
""
],
[
"Chen",
"Valerie",
""
],
[
"Neubig",
"Graham",
""
]
] |
2506.03012 | Nasir Eisty | Nasir U. Eisty, Jeffrey C. Carver, Johanna Cohoon, Ian A. Cosden, Carole Goble, Samuel Grayson | Ten Simple Rules for Catalyzing Collaborations and Building Bridges between Research Software Engineers and Software Engineering Researchers | Published in IEEE Computing in Science & Engineering (CiSE) Magazine | null | 10.1109/MCSE.2025.3569786 | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | In the evolving landscape of scientific and scholarly research, effective collaboration between Research Software Engineers (RSEs) and Software Engineering Researchers (SERs) is pivotal for advancing innovation and ensuring the integrity of computational methodologies. This paper presents ten strategic guidelines aimed at fostering productive partnerships between these two distinct yet complementary communities. The guidelines emphasize the importance of recognizing and respecting the cultural and operational differences between RSEs and SERs, proactively initiating and nurturing collaborations, and engaging within each other's professional environments. They advocate for identifying shared challenges, maintaining openness to emerging problems, ensuring mutual benefits, and serving as advocates for one another. Additionally, the guidelines highlight the necessity of vigilance in monitoring collaboration dynamics, securing institutional support, and defining clear, shared objectives. By adhering to these principles, RSEs and SERs can build synergistic relationships that enhance the quality and impact of research outcomes. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:51:17 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Eisty",
"Nasir U.",
""
],
[
"Carver",
"Jeffrey C.",
""
],
[
"Cohoon",
"Johanna",
""
],
[
"Cosden",
"Ian A.",
""
],
[
"Goble",
"Carole",
""
],
[
"Grayson",
"Samuel",
""
]
] |
2506.03013 | Alexandra Gonz\'alez | Alexandra Gonz\'alez, Xavier Franch, David Lo, Silverio Mart\'inez-Fern\'andez | How do Pre-Trained Models Support Software Engineering? An Empirical Study in Hugging Face | null | null | null | null | cs.SE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Open-Source Pre-Trained Models (PTMs) provide extensive resources for various Machine Learning (ML) tasks, yet these resources lack a classification tailored to Software Engineering (SE) needs. To address this gap, we derive a taxonomy encompassing 147 SE tasks and apply an SE-oriented classification to PTMs in a popular open-source ML repository, Hugging Face (HF). Our repository mining study began with a systematically gathered database of PTMs from the HF API, considering their model card descriptions and metadata, and the abstract of the associated arXiv papers. We confirmed SE relevance through multiple filtering steps: detecting outliers, identifying near-identical PTMs, and the use of Gemini 2.0 Flash, which was validated with five pilot studies involving three human annotators. This approach uncovered 2,205 SE PTMs. We find that code generation is the most common SE task among PTMs, primarily focusing on software implementation, while requirements engineering and software design activities receive limited attention. In terms of ML tasks, text generation dominates within SE PTMs. Notably, the number of SE PTMs has increased markedly since 2023 Q2. Our classification provides a solid foundation for future automated SE scenarios, such as the sampling and selection of suitable PTMs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:51:17 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"González",
"Alexandra",
""
],
[
"Franch",
"Xavier",
""
],
[
"Lo",
"David",
""
],
[
"Martínez-Fernández",
"Silverio",
""
]
] |
2506.03014 | Tobias Hartung | Tobias Hartung, Karl Jansen | Convergence and efficiency proof of quantum imaginary time evolution for bounded order systems | 15 pages | null | null | null | quant-ph cs.CC physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | Many current and near-future applications of quantum computing utilise parametric families of quantum circuits and variational methods to find optimal values for these parameters. Solving a quantum computational problem with such variational methods relies on minimising some cost function, e.g., the energy of a physical system. As such, this is similar to the training process in machine learning and variational quantum simulations can therefore suffer from similar problems encountered in machine learning training. This includes non-convergence to the global minimum due to local minima as well as critical slowing down. In this article, we analyse the imaginary time evolution as a means of compiling parametric quantum circuits and finding optimal parameters, and show that it guarantees convergence to the global minimum without critical slowing down. We also show that the compilation process, including the task of finding optimal parameters, can be performed efficiently up to an arbitrary error threshold if the underlying physical system is of bounded order. This includes many relevant computational problems, e.g., local physical theories and combinatorial optimisation problems such as the flight-to-gate assignment problem. In particular, we show a priori estimates on the success probability for these combinatorial optimisation problems. There seem to be no known classical methods with similar efficiency and convergence guarantees. Meanwhile the imaginary time evolution method can be implemented on current quantum computers. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:52:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Hartung",
"Tobias",
""
],
[
"Jansen",
"Karl",
""
]
] |
2506.03015 | Rui Lan Jun | Rui Lan Jun (UTokyo), Tom Theuns (ICC, Durham), Kana Moriwaki (UTokyo), Sownak Bose (ICC, Durham) | Scatter in the star formation rate-halo mass relation: secondary bias and its impact on line-intensity mapping | 25 pages, 21 figures. We welcome comments | null | null | null | astro-ph.GA astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | We use the IllustrisTNG cosmological hydrodynamical simulations to study the impact of secondary bias -- specifically, the correlation between star formation rate (SFR) and halo bias at fixed halo mass -- on the line-intensity mapping (LIM) power spectrum. In LIM, the galaxy contributions are flux-weighted, and for many emission lines (e.g., H$\alpha$), flux scales with SFR. We show that the (ensemble-averaged) large-scale two-halo term of the power spectrum depends only on the mean luminosity-halo mass relation if the scatter is uncorrelated with halo bias. However, when luminosity correlates with halo bias at fixed mass, this assumption breaks down. In IllustrisTNG, this secondary bias increases the two-halo term by 5 per cent at $z \sim 1.5$ compared to a model with random scatter. We also find that SFRs of central and satellite galaxies are correlated, enhancing the one-halo term -- sensitive to intra-halo SFR distribution -- by 10 per cent relative to random pairings. To mitigate secondary bias in the two-halo term, we identify halo concentration (for haloes with mass $\log M_h \lesssim 12$) and satellite mass (for $\log M_h \gtrsim 12$) as effective secondary parameters. These results highlight the need to account for secondary bias when building mock catalogues and interpreting LIM observations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:53:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jun",
"Rui Lan",
"",
"UTokyo"
],
[
"Theuns",
"Tom",
"",
"ICC, Durham"
],
[
"Moriwaki",
"Kana",
"",
"UTokyo"
],
[
"Bose",
"Sownak",
"",
"ICC, Durham"
]
] |
2506.03016 | Heming Bai | Heming Bai and Xin Bian | Hybrid deep learning and iterative methods for accelerated solutions of viscous incompressible flow | null | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The pressure Poisson equation, central to the fractional step method in incompressible flow simulations, incurs high computational costs due to the iterative solution of large-scale linear systems. To address this challenge, we introduce HyDEA, a novel framework that synergizes deep learning with classical iterative solvers. It leverages the complementary strengths of a DeepONet - capable of capturing large-scale features of the solution - and the CG or a PCG method, which efficiently resolves fine-scale errors. Specifically, within the framework of line-search methods, the DeepONet predicts search directions to accelerate convergence in solving sparse, symmetric-positive-definite linear systems, while the CG/ PCG method ensures robustness through iterative refinement. The framework seamlessly extends to flows over solid structures via the decoupled immersed boundary projection method. Crucially, the DeepONet is trained on fabricated linear systems rather than flow specific data, endowing it with inherent generalization across geometric complexities and Reynolds numbers without retraining. Benchmarks demonstrate superior efficiency and accuracy of HyDEA over the CG/PCG methods for flows with no obstacles, single or multiple stationary obstacles, and one moving obstacle - using fixed network weights. Remarkably, HyDEA also exhibits super-resolution capability: although the DeepONet is trained on a 128*128 grid for Re=1000, the hybrid solver delivers accurate solutions on a 512*512 grid for Re=10000 via interpolation, despite discretizations mismatch. In contrast, a purely data-driven DeepONet fails for complex flows, underscoring the necessity of hybridizing deep learning with iterative methods. Robustness, efficiency, and generalization across geometries, resolutions, and Reynolds numbers of HyDEA highlight its potential as a transformative solver for real world fluid dynamics problems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:54:17 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bai",
"Heming",
""
],
[
"Bian",
"Xin",
""
]
] |
2506.03017 | Botao Lin | Botao Lin, Tinghua Zhang, Sishen Yuan, Tiantian Wang, Jiaole Wang, Wu Yuan, Hongliang Ren | Adjusting Tissue Puncture Omnidirectionally In Situ with Pneumatic Rotatable Biopsy Mechanism and Hierarchical Airflow Management in Tortuous Luminal Pathways | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In situ tissue biopsy with an endoluminal catheter is an efficient approach for disease diagnosis, featuring low invasiveness and few complications. However, the endoluminal catheter struggles to adjust the biopsy direction by distal endoscope bending or proximal twisting for tissue sampling within the tortuous luminal organs, due to friction-induced hysteresis and narrow spaces. Here, we propose a pneumatically-driven robotic catheter enabling the adjustment of the sampling direction without twisting the catheter for an accurate in situ omnidirectional biopsy. The distal end of the robotic catheter consists of a pneumatic bending actuator for the catheter's deployment in torturous luminal organs and a pneumatic rotatable biopsy mechanism (PRBM). By hierarchical airflow control, the PRBM can adjust the biopsy direction under low airflow and deploy the biopsy needle with higher airflow, allowing for rapid omnidirectional sampling of tissue in situ. This paper describes the design, modeling, and characterization of the proposed robotic catheter, including repeated deployment assessments of the biopsy needle, puncture force measurement, and validation via phantom tests. The PRBM prototype has six sampling directions evenly distributed across 360 degrees when actuated by a positive pressure of 0.3 MPa. The pneumatically-driven robotic catheter provides a novel biopsy strategy, potentially facilitating in situ multidirectional biopsies in tortuous luminal organs with minimum invasiveness. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:54:27 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lin",
"Botao",
""
],
[
"Zhang",
"Tinghua",
""
],
[
"Yuan",
"Sishen",
""
],
[
"Wang",
"Tiantian",
""
],
[
"Wang",
"Jiaole",
""
],
[
"Yuan",
"Wu",
""
],
[
"Ren",
"Hongliang",
""
]
] |
2506.03018 | Ricky Elwell | Ricky Elwell, James E. S. Terhune, Christian Schneider, Harry W. T. Morgan, Hoang Bao Tran Tan, Udeshika C. Perera, Daniel A. Rehn, Marisa C. Alfonso, Lars von der Wense, Benedict Seiferle, Kevin Scharl, Peter G. Thirolf, Andrei Derevianko, and Eric R. Hudson | $^{229}$Th Nuclear Spectroscopy in an Opaque Material: Laser-Based Conversion Electron M\"ossbauer Spectroscopy of $^{229}$ThO$_2$ | 18 pages, 7 figures | null | null | null | physics.atom-ph nucl-ex physics.optics quant-ph | http://creativecommons.org/licenses/by/4.0/ | Here, we report the first demonstration of laser-induced conversion electron M\"{o}ssbauer spectroscopy of the $^{229}$Th nuclear isomeric state, which provides the ability to probe the nuclear transition in a material that is opaque to light resonant with the nuclear transition. Specifically, we excite the nuclear transition in a thin ThO$_2$ sample whose band gap ($\sim$ 6 eV) is considerably smaller than the nuclear isomeric state energy (8.4 eV). As a result, the excited nucleus can quickly decay by internal conversion, resulting in the ejection of electrons from the surface. By collecting these conversion electrons, nuclear spectroscopy can be recorded. Unlike fluorescence spectroscopy, this technique is compatible with materials whose work function is less than the nuclear transition energy, opening a wider class of systems to study. Further, because ThO$_2$ can be made from spinless isotopes and the internal conversion decay process reduces the isomeric state lifetime to only $\sim$10 $\mu$s, allowing $\sim$10$^8$ relative reduction in clock interrogation time, a conversion-electron-based nuclear clock could lead to a $\sim$10$^4$ reduction in clock instability. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:54:46 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Elwell",
"Ricky",
""
],
[
"Terhune",
"James E. S.",
""
],
[
"Schneider",
"Christian",
""
],
[
"Morgan",
"Harry W. T.",
""
],
[
"Tan",
"Hoang Bao Tran",
""
],
[
"Perera",
"Udeshika C.",
""
],
[
"Rehn",
"Daniel A.",
""
],
[
"Alfonso",
"Marisa C.",
""
],
[
"von der Wense",
"Lars",
""
],
[
"Seiferle",
"Benedict",
""
],
[
"Scharl",
"Kevin",
""
],
[
"Thirolf",
"Peter G.",
""
],
[
"Derevianko",
"Andrei",
""
],
[
"Hudson",
"Eric R.",
""
]
] |
2506.03019 | Giuseppe Gaetano Luciano Dr | Giuseppe Gaetano Luciano, Andronikos Paliathanasis and Emmanuel N. Saridakis | Constraints on Barrow and Tsallis Holographic Dark Energy from DESI DR2 BAO data | 7 pages, 1 figure, 2 Tabs | null | null | null | gr-qc hep-th | http://creativecommons.org/licenses/by/4.0/ | Barrow and Tsallis Holographic Dark Energy (HDE) are two recently proposed extensions of the standard HDE framework, incorporating generalized corrections to horizon entropy through the use of Barrow and Tsallis entropies. Tsallis entropy arises from non-extensive statistical phenomena which account for long-range correlations and deviations from additivity, while Barrow entropy emerges from quantum-gravitational effects on the horizon geometry, associated with fractal modifications and deformations. At the cosmological level, both scenarios lead to the same equations, nevertheless the involved parameters obey different theoretical bounds. In this work, we use observational data from Supernova Type Ia (SNIa), Cosmic Chronometers (CC) and Baryonic acoustic oscillations (BAO), including the recently released DESI DR2 dataset, to place constraints on both scenaria. We show that both can be in agreement with observations, although they cannot alleviate the $H_0$ tension. However, applying information criteria we deduce that both of them are not favoured comparing to $\Lambda$CDM concordance cosmological paradigm. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:56:26 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Luciano",
"Giuseppe Gaetano",
""
],
[
"Paliathanasis",
"Andronikos",
""
],
[
"Saridakis",
"Emmanuel N.",
""
]
] |
2506.03020 | Chaeyoung Jung | Chaeyoung Jung, Hojoon Ki, Ji-Hoon Kim, Junmo Kim, Joon Son Chung | InfiniteAudio: Infinite-Length Audio Generation with Consistency | null | null | null | null | eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents InfiniteAudio, a simple yet effective strategy for generating infinite-length audio using diffusion-based text-to-audio methods. Current approaches face memory constraints because the output size increases with input length, making long duration generation challenging. A common workaround is to concatenate short audio segments, but this often leads to inconsistencies due to the lack of shared temporal context. To address this, InfiniteAudio integrates seamlessly into existing pipelines without additional training. It introduces two key techniques: FIFO sampling, a first-in, first-out inference strategy with fixed-size inputs, and curved denoising, which selectively prioritizes key diffusion steps for efficiency. Experiments show that InfiniteAudio achieves comparable or superior performance across all metrics. Audio samples are available on our project page. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:57:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Jung",
"Chaeyoung",
""
],
[
"Ki",
"Hojoon",
""
],
[
"Kim",
"Ji-Hoon",
""
],
[
"Kim",
"Junmo",
""
],
[
"Chung",
"Joon Son",
""
]
] |
2506.03021 | Akshay Patil | Akshay Patil and Clara Garc\'ia-S\'anchez | Should we care about the spatial heterogeneity in coral reefs under unidirectional turbulent flows? | null | null | null | null | physics.flu-dyn physics.geo-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work, we systematically investigate the similarities and differences observed between a hydraulically rough wall comprised of an array of cylinders, massive corals, and branching corals arranged in a staggered manner, along with a stochastically generated coral bed using a scale-resolving computational framework. Our data suggests that for all the flow parameters of interest, there is a substantial difference observed between the stochastic coral bed and the regularly arranged coral bed. By analysing the double-averaged statistics and time-averaged spatial heterogeneity in the hydrodynamic response, we explain the differences observed between the four cases that bring out significant local effects. These observations have important consequences for modelling coral-like roughness in numerical and experimental settings to better understand the mean flow statistics and the spatial heterogeneity induced as a consequence of the underlying coral geometry. Our results can help inform the coastal ocean modelling efforts to further improve the inclusion of coral heterogeneity within two-equation closure models by further investigating the impact of spatially stochastic, rough bottom boundary layers. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:58:47 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Patil",
"Akshay",
""
],
[
"García-Sánchez",
"Clara",
""
]
] |
2506.03022 | Ryan Mukherjee | David McVicar, Brian Avant, Adrian Gould, Diego Torrejon, Charles Della Porta, Ryan Mukherjee | Smartflow: Enabling Scalable Spatiotemporal Geospatial Research | null | IGARSS 2023 | 10.1109/IGARSS52108.2023.10283095 | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | BlackSky introduces Smartflow, a cloud-based framework enabling scalable spatiotemporal geospatial research built on open-source tools and technologies. Using STAC-compliant catalogs as a common input, heterogeneous geospatial data can be processed into standardized datacubes for analysis and model training. Model experimentation is managed using a combination of tools, including ClearML, Tensorboard, and Apache Superset. Underpinning Smartflow is Kubernetes, which orchestrates the provisioning and execution of workflows to support both horizontal and vertical scalability. This combination of features makes Smartflow well-suited for geospatial model development and analysis over large geographic areas, time scales, and expansive image archives.
We also present a novel neural architecture, built using Smartflow, to monitor large geographic areas for heavy construction. Qualitative results based on data from the IARPA Space-based Machine Automated Recognition Technique (SMART) program are presented that show the model is capable of detecting heavy construction throughout all major phases of development. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:58:52 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"McVicar",
"David",
""
],
[
"Avant",
"Brian",
""
],
[
"Gould",
"Adrian",
""
],
[
"Torrejon",
"Diego",
""
],
[
"Della Porta",
"Charles",
""
],
[
"Mukherjee",
"Ryan",
""
]
] |
2506.03024 | Madhusudan Srinivasan | Madhusudan Srinivasan, Jubril Abdel | GenFair: Systematic Test Generation for Fairness Fault Detection in Large Language Models | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) are increasingly deployed in critical domains, yet they often exhibit biases inherited from training data, leading to fairness concerns. This work focuses on the problem of effectively detecting fairness violations, especially intersectional biases that are often missed by existing template-based and grammar-based testing methods. Previous approaches, such as CheckList and ASTRAEA, provide structured or grammar-driven test generation but struggle with low test diversity and limited sensitivity to complex demographic interactions. To address these limitations, we propose GenFair, a metamorphic fairness testing framework that systematically generates source test cases using equivalence partitioning, mutation operators, and boundary value analysis. GenFair improves fairness testing by generating linguistically diverse, realistic, and intersectional test cases. It applies metamorphic relations (MR) to derive follow-up cases and detects fairness violations via tone-based comparisons between source and follow-up responses. In experiments with GPT-4.0 and LLaMA-3.0, GenFair outperformed two baseline methods. It achieved a fault detection rate (FDR) of 0.73 (GPT-4.0) and 0.69 (LLaMA-3.0), compared to 0.54/0.51 for template-based and 0.39/0.36 for ASTRAEA. GenFair also showed the highest test case diversity (syntactic:10.06, semantic: 76.68) and strong coherence (syntactic: 291.32, semantic: 0.7043), outperforming both baselines. These results demonstrate the effectiveness of GenFair in uncovering nuanced fairness violations. The proposed method offers a scalable and automated solution for fairness testing and contributes to building more equitable LLMs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:00:30 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Srinivasan",
"Madhusudan",
""
],
[
"Abdel",
"Jubril",
""
]
] |
2506.03025 | Ludovico Bruni Bruno | Ludovico Bruni Bruno, Francesco Dell'Accio, Wolfgang Erb, Federico Nudo | Bivariate polynomial histopolation techniques on Padua, Fekete and Leja triangles | 24 pages, 15 figures | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | This paper explores the reconstruction of a real-valued function $f$ defined over a domain $\Omega \subset \mathbb{R}^2$ using bivariate polynomials that satisfy triangular histopolation conditions. More precisely, we assume that only the averages of $f$ over a given triangulation $\mathcal{T}_N$ of $\Omega$ are available and seek a bivariate polynomial that approximates $f$ using a histopolation approach, potentially flanked by an additional regression technique. This methodology relies on the selection of a subset of triangles $\mathcal{T}_M \subset \mathcal{T}_N$ for histopolation, ensuring both the solvability and the well-conditioning of the problem. The remaining triangles can potentially be used to enhance the accuracy of the polynomial approximation through a simultaneous regression. We will introduce histopolation and combined histopolation-regression methods using the Padua points, discrete Leja sequences, and approximate Fekete nodes. The proposed algorithms are implemented and evaluated through numerical experiments that demonstrate their effectiveness in function approximation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:00:34 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bruno",
"Ludovico Bruni",
""
],
[
"Dell'Accio",
"Francesco",
""
],
[
"Erb",
"Wolfgang",
""
],
[
"Nudo",
"Federico",
""
]
] |
2506.03026 | Sridhar Venkatesh | Hyunsuk Kim, Sridhar Venkatesh | Lefschetz morphisms on singular cohomology and local cohomological dimension of toric varieties | 24 pages, comments welcome! | null | null | null | math.AG math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given a proper toric variety and a line bundle on it, we describe the morphism on singular cohomology given by the cup product with the Chern class of that line bundle in terms of the data of the associated fan. Using that, we relate the local cohomological dimension of an affine toric variety with the Lefschetz morphism on the singular cohomology of a projective toric variety of one dimension lower. As a corollary, we show that the local cohomological defect is not a combinatorial invariant. We also produce numerous examples of toric varieties in every dimension with any possible local cohomological defect, by showing that the local cohomological defect remains unchanged under taking a pyramid. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:00:54 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Kim",
"Hyunsuk",
""
],
[
"Venkatesh",
"Sridhar",
""
]
] |
2506.03027 | Konduri Aditya | Aswin Kumar Arumugam, Shubham Kumar Goswami, Nagabhushana Rao Vadlamani, Konduri Aditya | Accuracy and scalability of asynchronous compressible flow solver for transitional flows | null | null | null | null | physics.comp-ph physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | To overcome the communication bottlenecks observed in state-of-the-art parallel time-dependent flow solvers at extreme scales, an asynchronous computing approach that relaxes communication and synchronization at a mathematical level was previously developed. This approach preserves high-order accuracy of computations near processing element boundaries using asynchrony-tolerant (AT) schemes while significantly improving the scalability. The numerical properties of the AT schemes were studied based on simple linear and nonlinear partial differential equations (PDEs) in previous works. Allowing asynchrony in numerical schemes can minimize communication overheads in a parallel setting in two ways: one that avoids communication over a few predetermined time steps, and the other that initiates communications without enforcing synchronization. In this study, the asynchronous algorithms are incorporated into the high-order compressible flow solver COMP-SQUARE, which solves practically relevant flow problems in complex geometries in a multi-block framework. The numerical efficacy and scalability of the two asynchronous algorithms are demonstrated for three test cases: isentropic advection of a vortex, the Taylor-Green vortex, and a much more sensitive case of the flow transitioning on a NACA0012 airfoil. Speed-ups of up to $4\times$ with respect to the baseline synchronous algorithm are observed in the scaling experiments performed on up to 18,432 cores. The results of this study demonstrate the applicability of AT schemes on established CFD solvers to improve scalability at extreme scales as the scientific computing environment moves to the exascale era. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:02:05 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Arumugam",
"Aswin Kumar",
""
],
[
"Goswami",
"Shubham Kumar",
""
],
[
"Vadlamani",
"Nagabhushana Rao",
""
],
[
"Aditya",
"Konduri",
""
]
] |
2506.03028 | Junde Xu | Junde Xu, Zijun Gao, Xinyi Zhou, Jie Hu, Xingyi Cheng, Le Song, Guangyong Chen, Pheng-Ann Heng, Jiezhong Qiu | Protein Inverse Folding From Structure Feedback | null | null | null | null | cs.LG q-bio.BM | http://creativecommons.org/licenses/by/4.0/ | The inverse folding problem, aiming to design amino acid sequences that fold into desired three-dimensional structures, is pivotal for various biotechnological applications. Here, we introduce a novel approach leveraging Direct Preference Optimization (DPO) to fine-tune an inverse folding model using feedback from a protein folding model. Given a target protein structure, we begin by sampling candidate sequences from the inverse-folding model, then predict the three-dimensional structure of each sequence with the folding model to generate pairwise structural-preference labels. These labels are used to fine-tune the inverse-folding model under the DPO objective. Our results on the CATH 4.2 test set demonstrate that DPO fine-tuning not only improves sequence recovery of baseline models but also leads to a significant improvement in average TM-Score from 0.77 to 0.81, indicating enhanced structure similarity. Furthermore, iterative application of our DPO-based method on challenging protein structures yields substantial gains, with an average TM-Score increase of 79.5\% with regard to the baseline model. This work establishes a promising direction for enhancing protein sequence design ability from structure feedback by effectively utilizing preference optimization. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:02:12 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Xu",
"Junde",
""
],
[
"Gao",
"Zijun",
""
],
[
"Zhou",
"Xinyi",
""
],
[
"Hu",
"Jie",
""
],
[
"Cheng",
"Xingyi",
""
],
[
"Song",
"Le",
""
],
[
"Chen",
"Guangyong",
""
],
[
"Heng",
"Pheng-Ann",
""
],
[
"Qiu",
"Jiezhong",
""
]
] |
2506.03029 | Meret Preu{\ss} | Meret Preu{\ss} | Simplifying higher-order perturbation theory for ring-shaped Bose-Hubbard systems | 16 pages, 3 figures. Phys. Scr. 2025 | null | 10.1088/1402-4896/addfb0 | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | In this paper, higher-order perturbation theory is applied and tailored to one-dimensional ring-shaped Bose-Hubbard systems. Spectral and geometrical properties are used to structurally simplify the contributions and reduce computational effort without sacrificing accuracy. For this, a guide for the computation of the individual perturbational orders up to order nine is provided, alongside a both system-specific and parametrization-dependent convergence criterion. The simplification scheme described is found to be applicable to a wider class of Bose-Hubbard systems with different lattice geometries. An exemplary validation of these findings is included in the form of explicit calculations of ground state energies of the three-site Bose-Hubbard system with repulsive on-site interactions. These calculations are successfully checked against numerical computations of exact diagonalization results. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:03:09 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Preuß",
"Meret",
""
]
] |
2506.03030 | Petra Hozzov\'a | Filip B\'artek, Ahmed Bhayat, Robin Coutelier, M\'arton Hajdu, Matthias Hetzenberger, Petra Hozzov\'a, Laura Kov\'acs, Jakob Rath, Michael Rawson, Giles Reger, Martin Suda, Johannes Schoisswohl, Andrei Voronkov | The Vampire Diary | 15 pages, 3 figures | null | null | null | cs.LO | http://creativecommons.org/licenses/by/4.0/ | During the past decade of continuous development, the theorem prover Vampire has become an automated solver for the combined theories of commonly-used data structures. Vampire now supports arithmetic, induction, and higher-order logic. These advances have been made to meet the demands of software verification, enabling Vampire to effectively complement SAT/SMT solvers and aid proof assistants. We explain how best to use Vampire in practice and review the main changes Vampire has undergone since its last tool presentation, focusing on the engineering principles and design choices we made during this process. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:03:37 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bártek",
"Filip",
""
],
[
"Bhayat",
"Ahmed",
""
],
[
"Coutelier",
"Robin",
""
],
[
"Hajdu",
"Márton",
""
],
[
"Hetzenberger",
"Matthias",
""
],
[
"Hozzová",
"Petra",
""
],
[
"Kovács",
"Laura",
""
],
[
"Rath",
"Jakob",
""
],
[
"Rawson",
"Michael",
""
],
[
"Reger",
"Giles",
""
],
[
"Suda",
"Martin",
""
],
[
"Schoisswohl",
"Johannes",
""
],
[
"Voronkov",
"Andrei",
""
]
] |
2506.03031 | Yang Huang | Yang Huang, Shih-Han Wang, Shuyi Cao, Luke E. K. Achenie and Hongliang Xin | Understanding Stability Mechanisms in Single-Atom Alloys with Theory-infused Deep Learning | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present an interpretable deep learning model that enhances the prediction of cohesive energy in transition metal alloys (TMAs) by incorporating cohesion theory into a graph neural network (GNN) framework. The model not only predicts the total cohesive energy-an indicator of crystal stability-but also disentangles its various contributing factors and underlying physical parameters. The physics insights extracted from the model clarify the stability trends of transition metal surfaces across the periodic table. Furthermore, by applying the model to single-atom alloys (SAAs), a class of catalytically significant next-generation TMAs, we analyze and explain the relative stability of monomer/dimer (in-plane symmetry breaking) and top-/sub-layer (out-of-plane symmetry breaking) configurations. These two types of symmetry breaking lead to distinct thermodynamic preferences in SAAs, governed by localized effects (e.g. d-orbital coupling) and delocalized effects (e.g. wavefunction renormalization). The model is thus positioned as a powerful tool for understanding and strategically designing TMAs, enabling the tailored development of materials with improved stability for advanced applications in catalysis and materials science. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:07:23 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Huang",
"Yang",
""
],
[
"Wang",
"Shih-Han",
""
],
[
"Cao",
"Shuyi",
""
],
[
"Achenie",
"Luke E. K.",
""
],
[
"Xin",
"Hongliang",
""
]
] |
2506.03032 | Junhao Yu | Junhao Yu, Yan Zhuang, YuXuan Sun, Weibo Gao, Qi Liu, Mingyue Cheng, Zhenya Huang, Enhong Chen | TestAgent: An Adaptive and Intelligent Expert for Human Assessment | 24 pages,10 figures | null | null | null | cs.AI cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Accurately assessing internal human states is key to understanding preferences, offering personalized services, and identifying challenges in real-world applications. Originating from psychometrics, adaptive testing has become the mainstream method for human measurement and has now been widely applied in education, healthcare, sports, and sociology. It customizes assessments by selecting the fewest test questions . However, current adaptive testing methods face several challenges. The mechanized nature of most algorithms leads to guessing behavior and difficulties with open-ended questions. Additionally, subjective assessments suffer from noisy response data and coarse-grained test outputs, further limiting their effectiveness. To move closer to an ideal adaptive testing process, we propose TestAgent, a large language model (LLM)-powered agent designed to enhance adaptive testing through interactive engagement. This is the first application of LLMs in adaptive testing. TestAgent supports personalized question selection, captures test-takers' responses and anomalies, and provides precise outcomes through dynamic, conversational interactions. Experiments on psychological, educational, and lifestyle assessments show our approach achieves more accurate results with 20% fewer questions than state-of-the-art baselines, and testers preferred it in speed, smoothness, and other dimensions. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:07:54 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yu",
"Junhao",
""
],
[
"Zhuang",
"Yan",
""
],
[
"Sun",
"YuXuan",
""
],
[
"Gao",
"Weibo",
""
],
[
"Liu",
"Qi",
""
],
[
"Cheng",
"Mingyue",
""
],
[
"Huang",
"Zhenya",
""
],
[
"Chen",
"Enhong",
""
]
] |
2506.03033 | Yves Kini | Yves Kini, Anna L. Watts, Tuomo Salmi, Anna Bilous, Serena Vinciguerra, Sebastien Guillot, David R. Ballantyne, Erik Kuulkers, Slavko Bogdanov, Valery Suleimanov | Impact of Accretion Assumptions on Pulse Profile Modelling of Superburst Oscillations in 4U 1636-536 | Accepted for publication in MNRAS. The Zenodo link will be made public after the paper is published | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Modelling the coherent pulsations observed during thermonuclear bursts offers a valuable method to probe the poorly understood equation of state of dense and cold matter. Here we apply the pulse profile modelling technique to the pulsations observed with RXTE during the 2001 superburst of 4U 1636$-$536. By employing a single, uniform-temperature hot spot model with varying size and temperature, along with various assumptions for background/accretion contribution, we find that each assumption leads to different inferred mass, radius, and compactness constraints. This highlights the critical need to better understand the mass accretion rate enhancement/reduction during thermonuclear bursts to accurately model burst oscillation sources using pulse profile modelling. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:16:59 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Kini",
"Yves",
""
],
[
"Watts",
"Anna L.",
""
],
[
"Salmi",
"Tuomo",
""
],
[
"Bilous",
"Anna",
""
],
[
"Vinciguerra",
"Serena",
""
],
[
"Guillot",
"Sebastien",
""
],
[
"Ballantyne",
"David R.",
""
],
[
"Kuulkers",
"Erik",
""
],
[
"Bogdanov",
"Slavko",
""
],
[
"Suleimanov",
"Valery",
""
]
] |
2506.03034 | Xinyuan You | Xinyuan You, Andy C. Y. Li, Tanay Roy, Shaojiang Zhu, Alexander Romanenko, Anna Grassellino, Yao Lu, and Srivatsan Chakram | Floquet-Engineered Fast SNAP gates in weakly coupled cQED systems | 18 pages, 9 figures | null | null | FERMILAB-PUB-25-0374-SQMS | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Superconducting cavities with high quality factors, coupled to a fixed-frequency transmon, provide a state-of-the-art platform for quantum information storage and manipulation. The commonly used selective number-dependent arbitrary phase (SNAP) gate faces significant challenges in ultra-high-coherence cavities, where the weak dispersive shifts necessary for preserving high coherence typically result in prolonged gate times. Here, we propose a protocol to achieve high-fidelity SNAP gates that are orders of magnitude faster than the standard implementation, surpassing the speed limit set by the bare dispersive shift. We achieve this enhancement by dynamically amplifying the dispersive coupling via sideband interactions, followed by quantum optimal control on the Floquet-engineered system. We also present a unified perturbation theory that explains both the gate acceleration and the associated benign drive-induced decoherence, corroborated by Floquet-Markov simulations. These results pave the way for the experimental realization of high-fidelity, selective control of weakly coupled, high-coherence cavities, and expanding the scope of optimal control techniques to a broader class of Floquet quantum systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:18:43 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"You",
"Xinyuan",
""
],
[
"Li",
"Andy C. Y.",
""
],
[
"Roy",
"Tanay",
""
],
[
"Zhu",
"Shaojiang",
""
],
[
"Romanenko",
"Alexander",
""
],
[
"Grassellino",
"Anna",
""
],
[
"Lu",
"Yao",
""
],
[
"Chakram",
"Srivatsan",
""
]
] |
2506.03035 | Christophe Servan | Pierre Lepagnol and Sahar Ghannay and Thomas Gerald and Christophe Servan and Sophie Rosset | Leveraging Information Retrieval to Enhance Spoken Language Understanding Prompts in Few-Shot Learning | Conference paper accepted to INTERSPEECH 2025 | INTERSPEECH 2025 | null | null | cs.CL cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | Understanding user queries is fundamental in many applications, such as home assistants, booking systems, or recommendations. Accordingly, it is crucial to develop accurate Spoken Language Understanding (SLU) approaches to ensure the reliability of the considered system. Current State-of-the-Art SLU techniques rely on large amounts of training data; however, only limited annotated examples are available for specific tasks or languages.
In the meantime, instruction-tuned large language models (LLMs) have shown exceptional performance on unseen tasks in a few-shot setting when provided with adequate prompts. In this work, we propose to explore example selection by leveraging Information retrieval (IR) approaches to build an enhanced prompt that is applied to an SLU task. We evaluate the effectiveness of the proposed method on several SLU benchmarks. Experimental results show that lexical IR methods significantly enhance performance without increasing prompt length. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:18:45 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lepagnol",
"Pierre",
""
],
[
"Ghannay",
"Sahar",
""
],
[
"Gerald",
"Thomas",
""
],
[
"Servan",
"Christophe",
""
],
[
"Rosset",
"Sophie",
""
]
] |
2506.03036 | Anton Klimek | Anton Klimek, Benjamin A. Dalton, Roland R. Netz | Subdiffusion from competition between multi-exponential friction memory and energy barriers | 14 pages main text including 6 figures and 11 pages supplementary information (SI) including 4 additional figures | null | null | null | cond-mat.soft cond-mat.stat-mech physics.bio-ph | http://creativecommons.org/licenses/by/4.0/ | Subdiffusion is a hallmark of complex systems, ranging from protein folding to transport in viscoelastic media. However, despite its pervasiveness, the mechanistic origins of subdiffusion remain contested. Here, we analyze both Markovian and non-Markovian dynamics, in the presence and absence of energy barriers, in order to disentangle the distinct contributions of memory-dependent friction and energy barriers to the emergence of subdiffusive behavior. Focusing on the mean squared displacement (MSD), we develop an analytical framework that connects subdiffusion to multiscale memory effects in the generalized Langevin equation (GLE), and derive the subdiffusive scaling behavior of the MSD for systems governed by multi-exponential memory kernels. We identify persistence and relaxation timescales that delineate dynamical regimes in which subdiffusion arises from either memory or energy barrier effects. By comparing analytical predictions with simulations, we confirm that memory dominates the overdamped dynamics for barrier heights up to approximately $2\,k_BT$, a regime recently shown to be relevant for protein folding. Overall, our results advance the theoretical understanding of anomalous diffusion and provide practical tools that are broadly applicable to fields as diverse as molecular biophysics, polymer physics, and active matter systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:19:09 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Klimek",
"Anton",
""
],
[
"Dalton",
"Benjamin A.",
""
],
[
"Netz",
"Roland R.",
""
]
] |
2506.03037 | Shubhendu Trivedi | Shubhendu Trivedi, Brian D. Nord | On the Need to Align Intent and Implementation in Uncertainty Quantification for Machine Learning | null | null | null | null | cs.LG stat.ME stat.ML | http://creativecommons.org/licenses/by/4.0/ | Quantifying uncertainties for machine learning (ML) models is a foundational challenge in modern data analysis. This challenge is compounded by at least two key aspects of the field: (a) inconsistent terminology surrounding uncertainty and estimation across disciplines, and (b) the varying technical requirements for establishing trustworthy uncertainties in diverse problem contexts. In this position paper, we aim to clarify the depth of these challenges by identifying these inconsistencies and articulating how different contexts impose distinct epistemic demands. We examine the current landscape of estimation targets (e.g., prediction, inference, simulation-based inference), uncertainty constructs (e.g., frequentist, Bayesian, fiducial), and the approaches used to map between them. Drawing on the literature, we highlight and explain examples of problematic mappings. To help address these issues, we advocate for standards that promote alignment between the \textit{intent} and \textit{implementation} of uncertainty quantification (UQ) approaches. We discuss several axes of trustworthiness that are necessary (if not sufficient) for reliable UQ in ML models, and show how these axes can inform the design and evaluation of uncertainty-aware ML systems. Our practical recommendations focus on scientific ML, offering illustrative cases and use scenarios, particularly in the context of simulation-based inference (SBI). | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:19:59 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Trivedi",
"Shubhendu",
""
],
[
"Nord",
"Brian D.",
""
]
] |
2506.03038 | Jintian Shao | Jintian Shao, Yiming Cheng | Towards Analyzing and Understanding the Limitations of VAPO: A Theoretical Perspective | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) enhances large language models (LLMs) in complex, long-chain-of-thought (long-CoT) reasoning. The advanced VAPO framework, despite sophisticated mechanisms like Decoupled GAE, theoretically faces fundamental limitations in comprehensively modeling and leveraging deep, long-term value for fine-grained, step-by-step policy guidance in extended reasoning chains. We argue these limitations stem from inherent difficulties in credit assignment, value function representational capacity with temporally abstracted goals, and translating global value signals into local policy improvements, especially with sparse rewards. Our theoretical analysis examines these aspects to illuminate VAPO's boundaries in long-term value modeling, aiming to deepen understanding of current RL for advanced reasoning and suggest future research for more robust LLM agents. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:20:47 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Shao",
"Jintian",
""
],
[
"Cheng",
"Yiming",
""
]
] |
2506.03039 | Yohance Osborne | Yohance A. P. Osborne and Iain Smears | Rates of convergence of finite element approximations of second-order mean field games with nondifferentiable Hamiltonians | 31 pages | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | We prove a rate of convergence for finite element approximations of stationary, second-order mean field games with nondifferentiable Hamiltonians posed in general bounded polytopal Lipschitz domains with strongly monotone running costs. In particular, we obtain a rate of convergence in the $H^1$-norm for the value function approximations and in the $L^2$-norm for the approximations of the density. We also establish a rate of convergence for the error between the exact solution of the MFG system with a nondifferentiable Hamiltonian and the finite element discretizations of the corresponding MFG system with a regularized Hamiltonian. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:21:30 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Osborne",
"Yohance A. P.",
""
],
[
"Smears",
"Iain",
""
]
] |
2506.03040 | Hari Padma | Hari Padma, Filippo Glerean, Sophia F. R. TenHuisen, Zecheng Shen, Haoxin Wang, Luogen Xu, Joshua D. Elliott, Christopher C. Homes, Elizabeth Skoropata, Hiroki Ueda, Biaolong Liu, Eugenio Paris, Arnau Romaguera, Byungjune Lee, Wei He, Yu Wang, Seng Huat Lee, Hyeongi Choi, Sang-Youn Park, Zhiqiang Mao, Matteo Calandra, Hoyoung Jang, Elia Razzoli, Mark P. M. Dean, Yao Wang, Matteo Mitrano | Symmetry-protected electronic metastability in an optically driven cuprate ladder | Submitted version. Main text: 27 pages, 6 figures. SI: 36 pages, 19 figures | Nature Materials (2025), advanced online publication (URL: https://www.nature.com/articles/s41563-025-02254-2) | 10.1038/s41563-025-02254-2 | null | cond-mat.str-el cond-mat.mtrl-sci cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | Optically excited quantum materials exhibit nonequilibrium states with remarkable emergent properties, but these phenomena are usually transient, decaying on picosecond timescales and limiting practical applications. Advancing the design and control of nonequilibrium phases requires the development of targeted strategies to achieve long-lived, metastable phases. Here, we report the discovery of symmetry-protected electronic metastability in the model cuprate ladder Sr$_{14}$Cu$_{24}$O$_{41}$. Using femtosecond resonant x-ray scattering and spectroscopy, we show that this metastability is driven by a transfer of holes from chain-like charge reservoirs into the ladders. This ultrafast charge redistribution arises from the optical dressing and activation of a hopping pathway that is forbidden by symmetry at equilibrium. Relaxation back to the ground state is hence suppressed after the pump coherence dissipates. Our findings highlight how dressing materials with electromagnetic fields can dynamically activate terms in the electronic Hamiltonian, and provide a rational design strategy for nonequilibrium phases of matter. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:21:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Padma",
"Hari",
""
],
[
"Glerean",
"Filippo",
""
],
[
"TenHuisen",
"Sophia F. R.",
""
],
[
"Shen",
"Zecheng",
""
],
[
"Wang",
"Haoxin",
""
],
[
"Xu",
"Luogen",
""
],
[
"Elliott",
"Joshua D.",
""
],
[
"Homes",
"Christopher C.",
""
],
[
"Skoropata",
"Elizabeth",
""
],
[
"Ueda",
"Hiroki",
""
],
[
"Liu",
"Biaolong",
""
],
[
"Paris",
"Eugenio",
""
],
[
"Romaguera",
"Arnau",
""
],
[
"Lee",
"Byungjune",
""
],
[
"He",
"Wei",
""
],
[
"Wang",
"Yu",
""
],
[
"Lee",
"Seng Huat",
""
],
[
"Choi",
"Hyeongi",
""
],
[
"Park",
"Sang-Youn",
""
],
[
"Mao",
"Zhiqiang",
""
],
[
"Calandra",
"Matteo",
""
],
[
"Jang",
"Hoyoung",
""
],
[
"Razzoli",
"Elia",
""
],
[
"Dean",
"Mark P. M.",
""
],
[
"Wang",
"Yao",
""
],
[
"Mitrano",
"Matteo",
""
]
] |
2506.03041 | Sabab Al Farabi | Sabab Al Farabi (Department of Industrial Engineering, Lamar University, Beaumont, Texas, USA) | AI-Augmented OTDR Fault Localization Framework for Resilient Rural Fiber Networks in the United States | null | null | null | null | cs.NI eess.SP | http://creativecommons.org/licenses/by/4.0/ | This research presents a novel framework that combines traditional Optical Time-Domain Reflectometer (OTDR) signal analysis with machine learning to localize and classify fiber optic faults in rural broadband infrastructures. The proposed system addresses a critical need in the expansion of middle-mile and last-mile networks, particularly in regions targeted by the U.S. Broadband Equity, Access, and Deployment (BEAD) Program. By enhancing fault diagnosis through a predictive, AI-based model, this work enables proactive network maintenance in low-resource environments. Experimental evaluations using a controlled fiber testbed and synthetic datasets simulating rural network conditions demonstrate that the proposed method significantly improves detection accuracy and reduces false positives compared to conventional thresholding techniques. The solution offers a scalable, field-deployable tool for technicians and ISPs engaged in rural broadband deployment. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:23:12 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Farabi",
"Sabab Al",
"",
"Department of Industrial Engineering, Lamar University, Beaumont, Texas, USA"
]
] |
2506.03042 | Damilya Saduakhas | Damilya Saduakhas, David Bolin, Xiaotian Jin, Alexandre B. Simas, Jonas Wallin | Incorporating Correlated Nugget Effects in Multivariate Spatial Models: An Application to Argo Ocean Data | null | null | null | null | stat.ME stat.AP | http://creativecommons.org/licenses/by/4.0/ | Accurate analysis of global oceanographic data, such as temperature and salinity profiles from the Argo program, requires geostatistical models capable of capturing complex spatial dependencies. This study introduces Gaussian and non-Gaussian hierarchical multivariate Mat\'ern-SPDE models with correlated nugget effects to account for small-scale variability and measurement error correlations. Using simulations and Argo data, we demonstrate that incorporating correlated nugget effects significantly improves the accuracy of parameter estimation and spatial prediction in both Gaussian and non-Gaussian multivariate spatial processes. When applied to global ocean temperature and salinity data, our model yields lower correlation estimates between fields compared to models that assume independent noise. This suggests that traditional models may overestimate the underlying field correlation. By separating these effects, our approach captures fine-scale oceanic patterns more effectively. These findings show the importance of relaxing the assumption of independent measurement errors in multivariate hierarchical models. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:24:05 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Saduakhas",
"Damilya",
""
],
[
"Bolin",
"David",
""
],
[
"Jin",
"Xiaotian",
""
],
[
"Simas",
"Alexandre B.",
""
],
[
"Wallin",
"Jonas",
""
]
] |
2506.03043 | Nikita Puchkin | Nikita Puchkin, Iurii Pustovalov, Yuri Sapronov, Denis Suchkov, Alexey Naumov, and Denis Belomestny | Sample complexity of Schr\"odinger potential estimation | 60 pages | null | null | null | cs.LG math.ST stat.ML stat.TH | http://creativecommons.org/licenses/by/4.0/ | We address the problem of Schr\"odinger potential estimation, which plays a crucial role in modern generative modelling approaches based on Schr\"odinger bridges and stochastic optimal control for SDEs. Given a simple prior diffusion process, these methods search for a path between two given distributions $\rho_0$ and $\rho_T^*$ requiring minimal efforts. The optimal drift in this case can be expressed through a Schr\"odinger potential. In the present paper, we study generalization ability of an empirical Kullback-Leibler (KL) risk minimizer over a class of admissible log-potentials aimed at fitting the marginal distribution at time $T$. Under reasonable assumptions on the target distribution $\rho_T^*$ and the prior process, we derive a non-asymptotic high-probability upper bound on the KL-divergence between $\rho_T^*$ and the terminal density corresponding to the estimated log-potential. In particular, we show that the excess KL-risk may decrease as fast as $O(\log^2 n / n)$ when the sample size $n$ tends to infinity even if both $\rho_0$ and $\rho_T^*$ have unbounded supports. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:26:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Puchkin",
"Nikita",
""
],
[
"Pustovalov",
"Iurii",
""
],
[
"Sapronov",
"Yuri",
""
],
[
"Suchkov",
"Denis",
""
],
[
"Naumov",
"Alexey",
""
],
[
"Belomestny",
"Denis",
""
]
] |
2506.03044 | Laurentiu Marchis | Laurentiu Andrei Marchis, Po-Ling Loh | On the Benefits of Accelerated Optimization in Robust and Private Estimation | 91 pages, 8 figures | null | null | null | math.ST cs.LG stat.ML stat.TH | http://creativecommons.org/licenses/by/4.0/ | We study the advantages of accelerated gradient methods, specifically based on the Frank-Wolfe method and projected gradient descent, for privacy and heavy-tailed robustness. Our approaches are as follows: For the Frank-Wolfe method, our technique is based on a tailored learning rate and a uniform lower bound on the gradient of the $\ell_2$-norm over the constraint set. For accelerating projected gradient descent, we use the popular variant based on Nesterov's momentum, and we optimize our objective over $\mathbb{R}^p$. These accelerations reduce iteration complexity, translating into stronger statistical guarantees for empirical and population risk minimization. Our analysis covers three settings: non-random data, random model-free data, and parametric models (linear regression and generalized linear models). Methodologically, we approach both privacy and robustness based on noisy gradients. We ensure differential privacy via the Gaussian mechanism and advanced composition, and we achieve heavy-tailed robustness using a geometric median-of-means estimator, which also sharpens the dependency on the dimension of the covariates. Finally, we compare our rates to existing bounds and identify scenarios where our methods attain optimal convergence. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:26:30 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Marchis",
"Laurentiu Andrei",
""
],
[
"Loh",
"Po-Ling",
""
]
] |
2506.03046 | Mikolaj Walczak | Mikolaj Walczak, Romina Aalishah, Wyatt Mackey, Brittany Story, David L. Boothe Jr., Nicholas Waytowich, Xiaomin Lin, Tinoosh Mohsenin | EDEN: Entorhinal Driven Egocentric Navigation Toward Robotic Deployment | null | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep reinforcement learning agents are often fragile while humans remain adaptive and flexible to varying scenarios. To bridge this gap, we present EDEN, a biologically inspired navigation framework that integrates learned entorhinal-like grid cell representations and reinforcement learning to enable autonomous navigation. Inspired by the mammalian entorhinal-hippocampal system, EDEN allows agents to perform path integration and vector-based navigation using visual and motion sensor data. At the core of EDEN is a grid cell encoder that transforms egocentric motion into periodic spatial codes, producing low-dimensional, interpretable embeddings of position. To generate these activations from raw sensory input, we combine fiducial marker detections in the lightweight MiniWorld simulator and DINO-based visual features in the high-fidelity Gazebo simulator. These spatial representations serve as input to a policy trained with Proximal Policy Optimization (PPO), enabling dynamic, goal-directed navigation. We evaluate EDEN in both MiniWorld, for rapid prototyping, and Gazebo, which offers realistic physics and perception noise. Compared to baseline agents using raw state inputs (e.g., position, velocity) or standard convolutional image encoders, EDEN achieves a 99% success rate, within the simple scenarios, and >94% within complex floorplans with occluded paths with more efficient and reliable step-wise navigation. In addition, as a replacement of ground truth activations, we present a trainable Grid Cell encoder enabling the development of periodic grid-like patterns from vision and motion sensor data, emulating the development of such patterns within biological mammals. This work represents a step toward biologically grounded spatial intelligence in robotics, bridging neural navigation principles with reinforcement learning for scalable deployment. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:28:33 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Walczak",
"Mikolaj",
""
],
[
"Aalishah",
"Romina",
""
],
[
"Mackey",
"Wyatt",
""
],
[
"Story",
"Brittany",
""
],
[
"Boothe",
"David L.",
"Jr."
],
[
"Waytowich",
"Nicholas",
""
],
[
"Lin",
"Xiaomin",
""
],
[
"Mohsenin",
"Tinoosh",
""
]
] |
2506.03047 | Zhiyi Chi | Zhiyi Chi | Complexity of exact sampling of the first passage of a stable subordinator | null | null | null | null | stat.CO | http://creativecommons.org/licenses/by/4.0/ | We consider the exact sampling of the first passage of a stable subordinator across a non-increasing regular barrier. First, the sampling is reduced to one from a bivariate distribution parameterized by the index $\alpha$ of the subordinator and a scalar $z$ independent of the barrier. Then three algorithms are devised for different regions of $(\alpha, z)$, using the acceptance-rejection method without numerical inversion or integration. When combined, the algorithms allow the exact sampling of the first passage to be done with complexity $O(1+|\ln(1-\alpha)|)$. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:28:42 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chi",
"Zhiyi",
""
]
] |
2506.03048 | Yang Zhang | Yang Zhang, Mingyu Xu, Jie Li, Suk Hyun Sung, Sang-Wook Cheong, Weiwei Xie, and Ismail El Baggari | Mesoscopic Helices of Polar Domains in a Quadruple Perovskite | null | null | null | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | A significant effort in condensed matter physics is dedicated to the search for exotic arrangements of electric dipoles. Beyond collinear alignments of electric dipoles, more complex arrangements analogous to magnetic spin textures, such as polar vortices and skyrmions, have been realized. Here we report atomic-scale visualizations of a candidate for helical texture of electric dipoles, the lightly doped quadruple perovskite BiCu$_{x}$Mn$_{7-x}$O$_{12}$. Rather than forming a continuous incommensurate helical structure, this material exhibits a mesoscale ordering of ferroelastic and polar domains. Across domains, bismuth's polar displacement rotates discretely following a consistent handedness, giving rise to a mesoscale helical pattern. We observe both right-handed and left-handed chiralities associated with helical ordering. The mesoscopic helices of polar order observed in our work would provide an opportunity to realize emergent chiral-optical activity in ferroelectric crystals akin to supramolecular assemblies. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:28:53 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhang",
"Yang",
""
],
[
"Xu",
"Mingyu",
""
],
[
"Li",
"Jie",
""
],
[
"Sung",
"Suk Hyun",
""
],
[
"Cheong",
"Sang-Wook",
""
],
[
"Xie",
"Weiwei",
""
],
[
"Baggari",
"Ismail El",
""
]
] |
2506.03049 | Maria Walch | Maria Walch | Torsion in Persistent Homology and Neural Networks | null | null | null | null | math.AT cs.LG | http://creativecommons.org/licenses/by/4.0/ | We explore the role of torsion in hybrid deep learning models that incorporate topological data analysis, focusing on autoencoders. While most TDA tools use field coefficients, this conceals torsional features present in integer homology. We show that torsion can be lost during encoding, altered in the latent space, and in many cases, not reconstructed by standard decoders. Using both synthetic and high-dimensional data, we evaluate torsion sensitivity to perturbations and assess its recoverability across several autoencoder architectures. Our findings reveal key limitations of field-based approaches and underline the need for architectures or loss terms that preserve torsional information for robust data representation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:29:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Walch",
"Maria",
""
]
] |
2506.03050 | Ying Cui | Ying Cui, Bo Huang, Gaohong Dong, Ryuji Uozumi and Lu Tian | An IPCW Adjusted Win Statistics Approach in Clinical Trials Incorporating Equivalence Margins to Define Ties | null | null | null | null | stat.ME stat.AP | http://creativecommons.org/licenses/by/4.0/ | In clinical trials, multiple outcomes of different priorities commonly occur as the patient's response may not be adequately characterized by a single outcome. Win statistics are appealing summary measures for between-group difference at more than one endpoint. When defining the result of pairwise comparisons of a time-to-event endpoint, it is desirable to allow ties to account for incomplete follow-up and not clinically meaningful difference in endpoints of interest. In this paper, we propose a class of win statistics for time-to-event endpoints with a user-specified equivalence margin. These win statistics are identifiable in the presence of right-censoring and do not depend on the censoring distribution. We then develop estimation and inference procedures for the proposed win statistics based on inverse-probability-of-censoring {weighting} (IPCW) adjustment to handle right-censoring. We conduct extensive simulations to investigate the operational characteristics of the proposed procedure in the finite sample setting. A real oncology trial is used to illustrate the proposed approach. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:30:45 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Cui",
"Ying",
""
],
[
"Huang",
"Bo",
""
],
[
"Dong",
"Gaohong",
""
],
[
"Uozumi",
"Ryuji",
""
],
[
"Tian",
"Lu",
""
]
] |
2506.03051 | Yuval Kansal | Yuval Kansal, Shmuel Berman, Lydia Liu | Facts Do Care About Your Language: Assessing Answer Quality of Multilingual LLMs | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Factuality is a necessary precursor to useful educational tools. As adoption of Large Language Models (LLMs) in education continues of grow, ensuring correctness in all settings is paramount. Despite their strong English capabilities, LLM performance in other languages is largely untested. In this work, we evaluate the correctness of the Llama3.1 family of models in answering factual questions appropriate for middle and high school students. We demonstrate that LLMs not only provide extraneous and less truthful information, but also exacerbate existing biases against rare languages. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:31:52 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Kansal",
"Yuval",
""
],
[
"Berman",
"Shmuel",
""
],
[
"Liu",
"Lydia",
""
]
] |
2506.03052 | Hannah Nguyen | Hannah Vy Nguyen, Yu-Chun Grace Yen, Omar Shakir, Hang Huynh, Sebastian Gutierrez, June A. Smith, Sheila Jimenez, Salma Abdelgelil, Stephen MacNeil | Feedstack: Layering Structured Representations over Unstructured Feedback to Scaffold Human AI Conversation | CUI '25, Proceedings of the 7th ACM Conference on Conversational User Interfaces, July 8--10, 2025, Waterloo, ON, Canada | null | 10.1145/3719160.3737636 | null | cs.HC | http://creativecommons.org/licenses/by/4.0/ | Many conversational user interfaces facilitate linear conversations with turn-based dialogue, similar to face-to-face conversations between people. However, digital conversations can afford more than simple back-and-forth; they can be layered with interaction techniques and structured representations that scaffold exploration, reflection, and shared understanding between users and AI systems. We introduce Feedstack, a speculative interface that augments feedback conversations with layered affordances for organizing, navigating, and externalizing feedback. These layered structures serve as a shared representation of the conversation that can surface user intent and reveal underlying design principles. This work represents an early exploration of this vision using a research-through-design approach. We describe system features and design rationale, and present insights from two formative (n=8, n=8) studies to examine how novice designers engage with these layered supports. Rather than presenting a conclusive evaluation, we reflect on Feedstack as a design probe that opens up new directions for conversational feedback systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:32:10 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Nguyen",
"Hannah Vy",
""
],
[
"Yen",
"Yu-Chun Grace",
""
],
[
"Shakir",
"Omar",
""
],
[
"Huynh",
"Hang",
""
],
[
"Gutierrez",
"Sebastian",
""
],
[
"Smith",
"June A.",
""
],
[
"Jimenez",
"Sheila",
""
],
[
"Abdelgelil",
"Salma",
""
],
[
"MacNeil",
"Stephen",
""
]
] |
2506.03053 | Ram Potham | Sinem Erisken (Independent Researcher), Timothy Gothard (Independent Researcher), Martin Leitgab (Independent Researcher), Ram Potham (Independent Researcher) | MAEBE: Multi-Agent Emergent Behavior Framework | Preprint. This work has been submitted to the Multi-Agent Systems Workshop at ICML 2025 for review | null | null | null | cs.MA cs.AI cs.CL cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Traditional AI safety evaluations on isolated LLMs are insufficient as multi-agent AI ensembles become prevalent, introducing novel emergent risks. This paper introduces the Multi-Agent Emergent Behavior Evaluation (MAEBE) framework to systematically assess such risks. Using MAEBE with the Greatest Good Benchmark (and a novel double-inversion question technique), we demonstrate that: (1) LLM moral preferences, particularly for Instrumental Harm, are surprisingly brittle and shift significantly with question framing, both in single agents and ensembles. (2) The moral reasoning of LLM ensembles is not directly predictable from isolated agent behavior due to emergent group dynamics. (3) Specifically, ensembles exhibit phenomena like peer pressure influencing convergence, even when guided by a supervisor, highlighting distinct safety and alignment challenges. Our findings underscore the necessity of evaluating AI systems in their interactive, multi-agent contexts. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:33:47 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Erisken",
"Sinem",
"",
"Independent Researcher"
],
[
"Gothard",
"Timothy",
"",
"Independent Researcher"
],
[
"Leitgab",
"Martin",
"",
"Independent Researcher"
],
[
"Potham",
"Ram",
"",
"Independent Researcher"
]
] |
2506.03054 | John Dziak | John J. Dziak and Inbal Nahum-Shani | Constructing Evidence-Based Tailoring Variables for Adaptive Interventions | 51 pages, 6 figures. Submitted to Annals of Behavioral Medicine | null | null | null | stat.ME stat.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Background: An adaptive intervention (ADI) uses individual information in order to select treatment, to improve effectiveness while reducing cost and burden. ADIs require tailoring variables: person- and potentially time-specific information used to decide whether and how to deliver treatment. Specifying a tailoring variable for an intervention requires specifying what to measure, when to measure it, when to make the resulting decisions, and what cutoffs should be used in making those decisions. This involves tradeoffs between specificity versus sensitivity, and between waiting for sufficient information versus intervening quickly. These questions are causal and prescriptive (what should be done and when), not merely predictive (what would happen if current conditions persist).
Purpose: There is little specific guidance in the literature on how to empirically choose tailoring variables, including cutoffs, measurement times, and decision times. Methods: We review possible approaches for comparing potential tailoring variables and propose a framework for systematically developing tailoring variables.
Results: Although secondary observational data can be used to select tailoring variables, additional assumptions are needed. A specifically designed randomized experiment for optimization purposes (an optimization randomized clinical trial or ORCT), in the form of a multi-arm randomized trial, sequential multiple assignment randomized trial, a factorial experiment, or hybrid among them, may provide a more direct way to answer these questions.
Conclusions: Using randomization directly to inform tailoring variables would provide the most direct causal evidence, but designing a trial to compare both tailoring variables and treatments adds complexity; further methodological research is warranted. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:35:25 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Dziak",
"John J.",
""
],
[
"Nahum-Shani",
"Inbal",
""
]
] |
2506.03055 | Jacek Gatlik | Tomasz Dobrowolski, Jacek Gatlik, Zofia Bry{\l}owska, Panayotis G. Kevrekidis | Kink Dynamics in a Non-Autonomous Sine-Gordon Model | 14 pages, 8 figures | null | null | null | nlin.PS | http://creativecommons.org/licenses/by/4.0/ | The sine-Gordon model with space- and time-dependent parameters is considered. A highly accurate effective model with two degrees of freedom is constructed, allowing the description of the kink movement in this model even for extremely long times and nontrivial trajectories of the coherent structure. As a stringent test of the reduced order model, the case of a temporal drive leading to extremely complex kink motion is studied. The two-degree-of-freedom approximation is found to faithfully reproduce the behavior of the full field-theoretic model paving the way for both deeper understanding and improved design of soliton-based devices. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:35:37 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Dobrowolski",
"Tomasz",
""
],
[
"Gatlik",
"Jacek",
""
],
[
"Bryłowska",
"Zofia",
""
],
[
"Kevrekidis",
"Panayotis G.",
""
]
] |
2506.03056 | Ram Potham | Ram Potham (Independent Researcher), Max Harms (Machine Intelligence Research Institute) | Corrigibility as a Singular Target: A Vision for Inherently Reliable Foundation Models | Preprint. This work has been submitted to the Reliable and Responsible Foundation Models Workshop at ICML 2025 for review | null | null | null | cs.AI cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Foundation models (FMs) face a critical safety challenge: as capabilities scale, instrumental convergence drives default trajectories toward loss of human control, potentially culminating in existential catastrophe. Current alignment approaches struggle with value specification complexity and fail to address emergent power-seeking behaviors. We propose "Corrigibility as a Singular Target" (CAST)-designing FMs whose overriding objective is empowering designated human principals to guide, correct, and control them. This paradigm shift from static value-loading to dynamic human empowerment transforms instrumental drives: self-preservation serves only to maintain the principal's control; goal modification becomes facilitating principal guidance. We present a comprehensive empirical research agenda spanning training methodologies (RLAIF, SFT, synthetic data generation), scalability testing across model sizes, and demonstrations of controlled instructability. Our vision: FMs that become increasingly responsive to human guidance as capabilities grow, offering a path to beneficial AI that remains as tool-like as possible, rather than supplanting human judgment. This addresses the core alignment problem at its source, preventing the default trajectory toward misaligned instrumental convergence. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:36:03 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Potham",
"Ram",
"",
"Independent Researcher"
],
[
"Harms",
"Max",
"",
"Machine Intelligence Research Institute"
]
] |
2506.03057 | Andrey Skripnikov | Andrey Skripnikov, Sujit Sivadanam | Partially Regularized Ordinal Regression to Adjust Teams' Scoring for Strength of Schedule and Complementary Unit Performance in American Football | 39 pages,6 figures | null | null | null | stat.AP | http://creativecommons.org/licenses/by/4.0/ | American football is unique in that offensive and defensive units typically consist of separate players who don't share the field simultaneously, which tempts one to evaluate them independently. However, a team's offensive and defensive performances often complement each other. For instance, turnovers forced by the defense can create easier scoring opportunities for the offense. Using drive-by-drive data from 2014-2020 Division-I college football (Football Bowl Subdivision, FBS) and 2009-2017 National Football League (NFL) seasons, we identify complementary football features that impact scoring the most. We employ regularized ordinal regression with an elastic penalty, enabling variable selection and partially relaxing the proportional odds assumption. Moreover, given the importance of accounting for strength of the opposition, we incorporate unpenalized components to ensure full adjustment for strength of schedule. For residual diagnostics of our ordinal regression models we apply the surrogate approach, creatively extending its use to non-proportional odds models. We then adjust each team's offensive (defensive) performance to project it onto a league-average complementary unit, showcasing the effects of these adjustments on team scoring. Lastly, we evaluate the out-of-sample prediction performance of our selected model, highlighting improvements gained from incorporating complementary football features alongside strength-of-schedule adjustments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:36:09 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Skripnikov",
"Andrey",
""
],
[
"Sivadanam",
"Sujit",
""
]
] |
2506.03058 | Zhi Wang | Zhi Wang | Microscopic origin of the entropy of de Sitter spacetime | 19 pages, 7 figures | null | null | null | hep-th | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We construct an infinite family of semiclassical de Sitter (dS) microstates, realized as backreacted geometries of dS spacetime with a constant tension thin-shell brane located outside the dS event horizon. The contribution of wormholes in semiclassical gravitational path integral computations of overlaps between these microstates implies that they are not fully orthogonal. We count the dimension of the Hilbert space spanned by these dS microstates and find that it equals the exponential of the Gibbons-Hawking entropy of the de Sitter spacetime. Our construction provides a gravitational states counting derivation of Gibbons-Hawking formula for dS entropy. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:36:35 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wang",
"Zhi",
""
]
] |
2506.03059 | Salah Eddine Choutri | Salah Eddine Choutri, Boualem Djehiche, Prajwal Chauhan and Saif Eddin Jabari | Backpressure-based Mean-field Type Game for Scheduling in Multi-Hop Wireless Sensor Networks | Accepted to the 33rd European Signal Processing Conference (EUSIPCO 2025) | null | null | null | eess.SY cs.SY math.OC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We propose a Mean-Field Type Game (MFTG) framework for effective scheduling in multi-hop wireless sensor networks (WSNs) using backpressure as a performance criterion. Traditional backpressure algorithms leverage queue differentials to regulate data flow and maintain network stability. In this work, we extend the backpressure framework by incorporating a mean-field term into the cost functional, capturing the global behavior of the system alongside local dynamics. The resulting model utilizes the strengths of non-cooperative mean-field type games, enabling nodes to make decentralized decisions based on both individual queue states and system mean-field effects while accounting for stochastic network interactions. By leveraging the interplay between backpressure dynamics and mean-field coupling, the approach balances local optimization with global efficiency. Numerical simulations demonstrate the efficacy of the proposed method in handling congestion and scheduling in large-scale WSNs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:37:51 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Choutri",
"Salah Eddine",
""
],
[
"Djehiche",
"Boualem",
""
],
[
"Chauhan",
"Prajwal",
""
],
[
"Jabari",
"Saif Eddin",
""
]
] |
2506.03060 | Kun Fang | Kun Fang, Hamza Fawzi and Omar Fawzi | Adversarial quantum channel discrimination | Contains the sections "Application 2: adversarial quantum channel discrimination" and "Application 3: a relative entropy accumulation theorem" from arXiv:2411.04035v2. arXiv admin note: text overlap with arXiv:2411.04035 | null | null | null | quant-ph cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a new framework for quantum channel discrimination in an adversarial setting, where the tester plays against an adversary who accesses the environmental system and possesses internal quantum memory to perform adaptive strategies. We show that in asymmetric hypothesis testing, the optimal type-II error exponent is precisely characterized by the minimum output channel divergence, a new notion of quantum channel divergence in the worst-case scenario. This serves as a direct analog of the quantum Stein's lemma in the adversarial channel discrimination. Notably, the optimal error exponent can be achieved via simple non-adaptive strategies by the adversary, and its value can be efficiently computed despite its regularization. The strong converse property for quantum channel discrimination also holds in general. This adversarial quantum Stein's lemma is proved by new chain rules for measured and sandwiched relative entropies. Moreover, we derive a generalized version of the entropy accumulation theorem between two arbitrary sequences of quantum channels, extending the existing results from entropy to divergence and providing a solution to the dual formulation of the open problem presented in [IEEE FOCS, pp. 844-850 (2022)]. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:39:01 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Fang",
"Kun",
""
],
[
"Fawzi",
"Hamza",
""
],
[
"Fawzi",
"Omar",
""
]
] |
2506.03061 | Tathagata Saha | T. Saha, M. Krumpe, A. Markowitz, M. Powell, G. Leung, F. Combes, R. E. McElroy, J. S. Elford, M. Gaspari, N. Winkel, A. L. Coil, T. Urrutia | The Close AGN Reference Survey (CARS). Long-term spectral variability study of the changing look AGN Mrk 1018 | Accepted for publication in Astronomy and Astrophysics | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Changing-look AGNs (CLAGN) are accreting supermassive black hole systems that undergo variations in optical spectral type, driven by major changes in accretion rate. Mrk 1018 has undergone two transitions, a brightening event in the 1980s and a transition back to a faint state over the course of 2-3 years in the early 2010s. We characterize the evolving physical properties of the source's inner accretion flow, particularly during the bright-to-faint transition, as well as the morphological properties of its parsec-scale circumnuclear gas. We model archival X-ray spectra from XMM-Newton, Chandra, Suzaku, and Swift, using physically-motivated models to characterize X-ray spectral variations and track Fe Kalpha line flux. We also quantify Mrk 1018's long-term multi-wavelength spectral variability from optical/UV to the X-rays. Over the duration of the bright-to-faint transition, the UV and hard X-ray flux fell by differing factors, roughly 24 and 8, respectively. The soft X-ray excess faded, and was not detected by 2021. In the faint state, when the Eddington ratio drops to log Lbol/LEdd < -1.7, the hot X-ray corona photon index shows a 'softer-when-fainter' trend, similar to that seen in some black hole X-ray binaries and samples of low-luminosity AGNs. Finally, the Fe Kalpha line flux has dropped by only half the factor of the drop in the X-ray continuum. The transition from the bright state to the faint state is consistent with the inner accretion flow transitioning from a geometrically-thin disk to an ADAF-dominated state, with the warm corona disintegrating or becoming energetically negligible, while the X-ray-emitting hot flow becoming energetically dominant. Meanwhile, narrow Fe Kalpha emission has not yet fully responded to the drop in its driving continuum, likely because its emitter extends up to roughly 10 pc. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:40:15 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Saha",
"T.",
""
],
[
"Krumpe",
"M.",
""
],
[
"Markowitz",
"A.",
""
],
[
"Powell",
"M.",
""
],
[
"Leung",
"G.",
""
],
[
"Combes",
"F.",
""
],
[
"McElroy",
"R. E.",
""
],
[
"Elford",
"J. S.",
""
],
[
"Gaspari",
"M.",
""
],
[
"Winkel",
"N.",
""
],
[
"Coil",
"A. L.",
""
],
[
"Urrutia",
"T.",
""
]
] |
2506.03062 | Qining Zhang | Qining Zhang, Tanner Fiez, Yi Liu, Wenyang Liu | Multi-Metric Adaptive Experimental Design under Fixed Budget with Validation | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Standard A/B tests in online experiments face statistical power challenges when testing multiple candidates simultaneously, while adaptive experimental designs (AED) alone fall short in inferring experiment statistics such as the average treatment effect, especially with many metrics (e.g., revenue, safety) and heterogeneous variances. This paper proposes a fixed-budget multi-metric AED framework with a two-phase structure: an adaptive exploration phase to identify the best treatment, and a validation phase with an A/B test to verify the treatment's quality and infer statistics. We propose SHRVar, which generalizes sequential halving (SH) (Karnin et al., 2013) with a novel relative-variance-based sampling and an elimination strategy built on reward z-values. It achieves a provable error probability that decreases exponentially, where the exponent generalizes the complexity measure for SH (Karnin et al., 2013) and SHVar (Lalitha et al., 2023) with homogeneous and heterogeneous variances, respectively. Numerical experiments verify our analysis and demonstrate the superior performance of this new framework. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:41:11 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhang",
"Qining",
""
],
[
"Fiez",
"Tanner",
""
],
[
"Liu",
"Yi",
""
],
[
"Liu",
"Wenyang",
""
]
] |
2506.03063 | Deqiao Gan | Deqiao Gan, Xiaoxia Xu, Jiakuo Zuo, Xiaohu Ge and Yuanwei Liu | Joint Beamforming for NOMA Assisted Pinching Antenna Systems (PASS) | null | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pinching antenna system (PASS) configures the positions of pinching antennas (PAs) along dielectric waveguides to change both large-scale fading and small-scale scattering, which is known as pinching beamforming. A novel non-orthogonal multiple access (NOMA) assisted PASS framework is proposed for downlink multi-user multiple-input multiple-output (MIMO) communications. The transmit power minimization problem is formulated to jointly optimize the transmit beamforming, pinching beamforming, and power allocation. To solve this highly nonconvex problem, both gradient-based and swarm-based optimization methods are developed. 1) For gradient-based method, a majorization-minimization and penalty dual decomposition (MM-PDD) algorithm is developed. The Lipschitz gradient surrogate function is constructed based on MM to tackle the nonconvex terms of this problem. Then, the joint optimization problem is decomposed into subproblems that are alternatively optimized based on PDD to obtain stationary closed-form solutions. 2) For swarm-based method, a fast-convergent particle swarm optimization and zero forcing (PSO-ZF) algorithm is proposed. Specifically, the PA position-seeking particles are constructed to explore high-quality pinching beamforming solutions. Moreover, ZF-based transmit beamforming is utilized by each particle for fast fitness function evaluation. Simulation results demonstrate that: i) The proposed NOMA assisted PASS and algorithms outperforms the conventional NOMA assisted massive antenna system. The proposed framework reduces over 95.22% transmit power compared to conventional massive MIMO-NOMA systems. ii) Swarm-based optimization outperforms gradient-based optimization by searching effective solution subspace to avoid stuck in undesirable local optima. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:41:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Gan",
"Deqiao",
""
],
[
"Xu",
"Xiaoxia",
""
],
[
"Zuo",
"Jiakuo",
""
],
[
"Ge",
"Xiaohu",
""
],
[
"Liu",
"Yuanwei",
""
]
] |
2506.03064 | Bal\'azs N\'emeth | Bal\'azs N\'emeth and Ronojoy Adhikari | Paramagnetism in spherically confined charged active matter | 6 pages, 3 figures. Accepted for publication as a Letter in Physical Review Research | null | 10.1103/ccv9-c322 | null | cond-mat.soft cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | The celebrated theorem of Bohr and van Leeuwen guarantees that a classical charged system cannot have a magnetization in thermal equilibrium. Quantum mechanically, however, a diamagnetic response is obtained. In contrast, we show here that a classical charged active system, consisting of a motile particle confined to the surface of a sphere, has a nonzero magnetization and a paramagnetic response. We numerically sample Langevin trajectories of this system and compare with limiting analytical solutions of the Fokker-Planck equation, at small and large temperatures, to find excellent agreement in the magnetic response. Our Letter suggests experimental routes to controlling and extracting work from charged active matter. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:42:00 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Németh",
"Balázs",
""
],
[
"Adhikari",
"Ronojoy",
""
]
] |
2506.03065 | Pengtao Chen | Pengtao Chen and Xianfang Zeng and Maosen Zhao and Peng Ye and Mingzhu Shen and Wei Cheng and Gang Yu and Tao Chen | Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | While Diffusion Transformers (DiTs) have achieved breakthroughs in video generation, this long sequence generation task remains constrained by the quadratic complexity of attention mechanisms, resulting in significant inference latency. Through detailed analysis of attention maps in Video Diffusion Transformer (vDiT), we identify three recurring sparsity patterns: diagonal, multi-diagonal, and vertical-stripe structures. And even 3-6\% attention heads can be skipped. Crucially, these patterns exhibit strong layer-depth and head-position correlations but show limited dependence on the input content. Leveraging these findings, we propose Sparse-vDiT, a sparsity acceleration framework for vDiT comprising: 1) Pattern-optimized sparse kernels that replace dense attention with computationally efficient implementations for each identified sparsity pattern. 2) An offline sparse diffusion search algorithm that selects the optimal sparse computation strategy per layer and head via hardware-aware cost modeling. After determining the optimal configuration, we fuse heads within the same layer that share the same attention strategy, enhancing inference efficiency. Integrated into state-of-the-art vDiT models (CogVideoX1.5, HunyuanVideo, and Wan2.1), Sparse-vDiT achieves 2.09$\times$, 2.38$\times$, and 1.67$\times$ theoretical FLOP reduction, and actual inference speedups of 1.76$\times$, 1.85$\times$, and 1.58$\times$, respectively, while maintaining high visual fidelity, with PSNR values reaching 24.13, 27.09, and 22.59. Our work demonstrates that latent structural sparsity in vDiTs can be systematically exploited for long video synthesis. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:42:37 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Pengtao",
""
],
[
"Zeng",
"Xianfang",
""
],
[
"Zhao",
"Maosen",
""
],
[
"Ye",
"Peng",
""
],
[
"Shen",
"Mingzhu",
""
],
[
"Cheng",
"Wei",
""
],
[
"Yu",
"Gang",
""
],
[
"Chen",
"Tao",
""
]
] |
2506.03066 | Qining Zhang | Qining Zhang, Lei Ying | Provable Reinforcement Learning from Human Feedback with an Unknown Link Function | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Link functions, which characterize how human preferences are generated from the value function of an RL problem, are a crucial component in designing RLHF algorithms. Almost all RLHF algorithms, including state-of-the-art ones in empirical studies such as DPO and PPO, assume the link function is known to the agent (e.g., a logistic function according to the Bradley-Terry model), which is arguably unrealistic considering the complex nature of human preferences. To avoid link function mis-specification, this paper studies general RLHF problems with unknown link functions. We propose a novel policy optimization algorithm called ZSPO based on a new zeroth-order policy optimization method, where the key is to use human preference to construct a parameter update direction that is positively correlated with the true policy gradient direction. ZSPO achieves it by estimating the sign of the value function difference instead of estimating the gradient from the value function difference, so it does not require knowing the link function. Under mild conditions, ZSPO converges to a stationary policy with a polynomial convergence rate depending on the number of policy iterations and trajectories per iteration. Numerical results also show the superiority of ZSPO under link function mismatch. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:42:39 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Zhang",
"Qining",
""
],
[
"Ying",
"Lei",
""
]
] |
2506.03067 | Mingzhe Li | Mingzhe Li, Gehao Zhang, Zhenting Wang, Shiqing Ma, Siqi Pan, Richard Cartwright, Juan Zhai | EDITOR: Effective and Interpretable Prompt Inversion for Text-to-Image Diffusion Models | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Text-to-image generation models~(e.g., Stable Diffusion) have achieved significant advancements, enabling the creation of high-quality and realistic images based on textual descriptions. Prompt inversion, the task of identifying the textual prompt used to generate a specific artifact, holds significant potential for applications including data attribution, model provenance, and watermarking validation. Recent studies introduced a delayed projection scheme to optimize for prompts representative of the vocabulary space, though challenges in semantic fluency and efficiency remain. Advanced image captioning models or visual large language models can generate highly interpretable prompts, but they often lack in image similarity. In this paper, we propose a prompt inversion technique called \sys for text-to-image diffusion models, which includes initializing embeddings using a pre-trained image captioning model, refining them through reverse-engineering in the latent space, and converting them to texts using an embedding-to-text model. Our experiments on the widely-used datasets, such as MS COCO, LAION, and Flickr, show that our method outperforms existing methods in terms of image similarity, textual alignment, prompt interpretability and generalizability. We further illustrate the application of our generated prompts in tasks such as cross-concept image synthesis, concept manipulation, evolutionary multi-concept generation and unsupervised segmentation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:44:15 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Li",
"Mingzhe",
""
],
[
"Zhang",
"Gehao",
""
],
[
"Wang",
"Zhenting",
""
],
[
"Ma",
"Shiqing",
""
],
[
"Pan",
"Siqi",
""
],
[
"Cartwright",
"Richard",
""
],
[
"Zhai",
"Juan",
""
]
] |
2506.03068 | Manar Samad | Yina Hou, Shourav B. Rabbani, Liang Hong, Norou Diawara, Manar D. Samad | Causal Explainability of Machine Learning in Heart Failure Prediction from Electronic Health Records | 4 figures | null | null | null | stat.ML cs.CY cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | The importance of clinical variables in the prognosis of the disease is explained using statistical correlation or machine learning (ML). However, the predictive importance of these variables may not represent their causal relationships with diseases. This paper uses clinical variables from a heart failure (HF) patient cohort to investigate the causal explainability of important variables obtained in statistical and ML contexts. Due to inherent regression modeling, popular causal discovery methods strictly assume that the cause and effect variables are numerical and continuous. This paper proposes a new computational framework to enable causal structure discovery (CSD) and score the causal strength of mixed-type (categorical, numerical, binary) clinical variables for binary disease outcomes. In HF classification, we investigate the association between the importance rank order of three feature types: correlated features, features important for ML predictions, and causal features. Our results demonstrate that CSD modeling for nonlinear causal relationships is more meaningful than its linear counterparts. Feature importance obtained from nonlinear classifiers (e.g., gradient-boosting trees) strongly correlates with the causal strength of variables without differentiating cause and effect variables. Correlated variables can be causal for HF, but they are rarely identified as effect variables. These results can be used to add the causal explanation of variables important for ML-based prediction modeling. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:46:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Hou",
"Yina",
""
],
[
"Rabbani",
"Shourav B.",
""
],
[
"Hong",
"Liang",
""
],
[
"Diawara",
"Norou",
""
],
[
"Samad",
"Manar D.",
""
]
] |
2506.03069 | Simon Corrodi | The Muon $g-2$ Collaboration: D. P. Aguillard (33), T. Albahri (30), D. Allspach (7), J. Annala (7), K. Badgley (7), S. Bae{\ss}ler (35), I. Bailey (16), L. Bailey (27), E. Barlas-Yucel (28), T. Barrett (6), E. Barzi (7), F. Bedeschi (10), M. Berz (17), M. Bhattacharya (7), H. P. Binney (36), P. Bloom (18), J. Bono (7), E. Bottalico (30), T. Bowcock (30), S. Braun (36), M. Bressler (32), G. Cantatore (12), R. M. Carey (2), B. C. K. Casey (7), D. Cauz (26), R. Chakraborty (29), A. Chapelain (6), S. Chappa (7), S. Charity (30), C. Chen (21 and 22), M. Cheng (28), R. Chislett (27), Z. Chu (21), T. E. Chupp (33), C. Claessens (36), F. Confortini (9), M. E. Convery (7), S. Corrodi (1), L. Cotrozzi (30), J. D. Crnkovic (7), S. Dabagov (8), P. T. Debevec (28), S. Di Falco (10), G. Di Sciascio (11), S. Donati (10), B. Drendel (7), A. Driutti (10 and 29), M. Eads (19), A. Edmonds (2 and 37), J. Esquivel (7), M. Farooq (33), R. Fatemi (29), K. Ferraby (30), C. Ferrari (10), M. Fertl (14), A. T. Fienberg (36), A. Fioretti (10), D. Flay (32), S. B. Foster (2 and 29), H. Friedsam (7), N. S. Froemming (19), C. Gabbanini (10), I. Gaines (7), S. Ganguly (7), J. George (32), L. K. Gibbons (6), A. Gioiosa (25), K. L. Giovanetti (13), P. Girotti (10), W. Gohn (29), L. Goodenough (7), T. Gorringe (29), J. Grange (33), S. Grant (1 and 27), F. Gray (20), S. Haciomeroglu (5), T. Halewood-Leagas (30), D. Hampai (8), F. Han (29), J. Hempstead (36), D. W. Hertzog (36), G. Hesketh (27), E. Hess (10), A. Hibbert (30), Z. Hodge (36), S. Y. Hoh (21 and 22), K. W. Hong (35), R. Hong (1 and 29), T. Hu (21 and 22), Y. Hu (21), M. Iacovacci (9), M. Incagli (10), S. Israel (2 and 32), P. Kammel (36), M. Kargiantoulakis (7), M. Karuza (12), J. Kaspar (36), D. Kawall (32), L. Kelton (23 and 29), A. Keshavarzi (31), D. S. Kessler (32), K. S. Khaw (21 and 22), Z. Khechadoorian (6), B. Kiburg (7), M. Kiburg (7 and 18), O. Kim (34), N. Kinnaird (2), E. Kraegeloh (33), J. LaBounty (36), K. R. Labe (6), M. Lancaster (31), S. Lee (5), B. Li (21), D. Li (21), L. Li (21), I. Logashenko (4), A. Lorente Campos (29), Z. Lu (21), A. Luc\`a (7), G. Lukicov (27), A. Lusiani (10), A. L. Lyon (7), B. MacCoy (36), R. Madrak (7), K. Makino (17), S. Mastroianni (9), R. McCarthy (2), J. P. Miller (2), S. Miozzi (11), B. Mitra (34), J. P. Morgan (7), W. M. Morse (3), J. Mott (7), A. Nath (9), J. K. Ng (21 and 22), H. Nguyen (7), Y. Oksuzian (1), Z. Omarov (5 and 15), W. Osar (6), R. Osofsky (36), S. Park (5), G. Pauletta (26), J. Peck (29), G. M. Piacentino (25), R. N. Pilato (30), K. T. Pitts (28 and b), B. Plaster (29), N. Pohlman (19), C. C. Polly (7), D. Po\v{c}ani\'c (35), J. Price (30), B. Quinn (34), M. U. H. Qureshi (14), G. Rakness (7), S. Ramachandran (1), E. Ramberg (7), R. Reimann (14), B. L. Roberts (2), D. L. Rubin (6), M. Sakurai (27), L. Santi (26), C. Schlesier (28), A. Schreckenberger (7), Y. K. Semertzidis (5 and 15), M. Sorbara (11), J. Stapleton (7), D. Still (7), C. Stoughton (7), D. Stratakis (7), D. St\"ockinger (24), H. E. Swanson (36), G. Sweetmore (31), D. A. Sweigart (6), M. J. Syphers (19), Y. Takeuchi (21 and 22), D. A. Tarazona (6), T. Teubner (30), A. E. Tewsley-Booth (29 and 33), V. Tishchenko (3), N. H. Tran (2), W. Turner (30), E. Valetov (17), D. Vasilkova (30), G. Venanzoni (30), T. Walton (7), A. Weisskopf (17), L. Welty-Rieger (7), P. Winter (1), Y. Wu (1), B. Yu (34), M. Yucel (7), E. Zaid (30), Y. Zeng (21 and 22), C. Zhang (30) ((1) Argonne National Laboratory, Lemont, Illinois, USA, (2) Boston University, Boston, Massachusetts, USA, (3) Brookhaven National Laboratory, Upton, New York, USA, (4) Budker Institute of Nuclear Physics, Novosibirsk, Russia, (5) Center for Axion and Precision Physics (CAPP) / Institute for Basic Science (IBS), Daejeon, Republic of Korea, (6) Cornell University, Ithaca, New York, USA, (7) Fermi National Accelerator Laboratory, Batavia, Illinois, USA, (8) INFN, Laboratori Nazionali di Frascati, Frascati, Italy, (9) INFN, Sezione di Napoli, Naples, Italy, (10) INFN, Sezione di Pisa, Pisa, Italy, (11) INFN, Sezione di Roma Tor Vergata, Rome, Italy, (12) INFN, Sezione di Trieste, Trieste, Italy, (13) Department of Physics and Astronomy, James Madison University, Harrisonburg, Virginia, USA, (14) Institute of Physics and Cluster of Excellence PRISMA+, Johannes Gutenberg University Mainz, Mainz, Germany, (15) Department of Physics, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea, (16) Lancaster University, Lancaster, United Kingdom, (17) Michigan State University, East Lansing, Michigan, USA, (18) North Central College, Naperville, Illinois, USA, (19) Northern Illinois University, DeKalb, Illinois, USA, (20) Regis University, Denver, Colorado, USA, (21) School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China, (22) Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, China, (23) Department of Physics and Astronomy, Trinity University, San Antonio, Texas, USA, (24) Institut f\"ur Kern- und Teilchenphysik, Technische Universit\"at Dresden, Dresden, Germany, (25) Universit\`a del Molise, Campobasso, Italy, (26) Universit\`a di Udine, Udine, Italy, (27) Department of Physics and Astronomy, University College London, London, United Kingdom, (28) University of Illinois at Urbana-Champaign, Urbana, Illinois, USA, (29) University of Kentucky, Lexington, Kentucky, USA, (30) University of Liverpool, Liverpool, United Kingdom, (31) Department of Physics and Astronomy, University of Manchester, Manchester, United Kingdom, (32) Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA, (33) University of Michigan, Ann Arbor, Michigan, USA, (34) University of Mississippi, University, Mississippi, USA, (35) University of Virginia, Charlottesville, Virginia, USA, (36) University of Washington, Seattle, Washington, USA, (37) City University of New York at York College, Jamaica, New York, USA) | Measurement of the Positive Muon Anomalous Magnetic Moment to 127 ppb | 11 pages, 3 figures; 2 pages of supplement material | null | null | FERMILAB-PUB-25-0364-PPD | hep-ex nucl-ex | http://creativecommons.org/licenses/by/4.0/ | A new measurement of the magnetic anomaly $a_{\mu}$ of the positive muon is presented based on data taken from 2020 to 2023 by the Muon $g-2$ Experiment at Fermi National Accelerator Laboratory (FNAL). This dataset contains over 2.5 times the total statistics of our previous results. From the ratio of the precession frequencies for muons and protons in our storage ring magnetic field, together with precisely known ratios of fundamental constants, we determine $a_{\mu} = 116\,592\,0710(162) \times 10^{-12}$ (139 ppb) for the new datasets, and $a_{\mu} = 116\,592\,0705(148) \times 10^{-12}$ (127 ppb) when combined with our previous results. The new experimental world average, dominated by the measurements at FNAL, is $a_{\mu}(\text{exp}) =116\,592\,0715(145) \times 10^{-12}$ (124 ppb). The measurements at FNAL have improved the precision on the world average by over a factor of four. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:46:28 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Collaboration",
"The Muon $g-2$",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
":",
"",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"Aguillard",
"D. P.",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"Albahri",
"T.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Allspach",
"D.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Annala",
"J.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Badgley",
"K.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Baeßler",
"S.",
"",
"University of Virginia, Charlottesville, Virginia, USA"
],
[
"Bailey",
"I.",
"",
"Lancaster University, Lancaster, United Kingdom"
],
[
"Bailey",
"L.",
"",
"Department of Physics and Astronomy, University College London, London, United Kingdom"
],
[
"Barlas-Yucel",
"E.",
"",
"University of Illinois at Urbana-Champaign, Urbana, Illinois, USA"
],
[
"Barrett",
"T.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Barzi",
"E.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Bedeschi",
"F.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Berz",
"M.",
"",
"Michigan State University, East Lansing, Michigan, USA"
],
[
"Bhattacharya",
"M.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Binney",
"H. P.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Bloom",
"P.",
"",
"North Central College, Naperville, Illinois, USA"
],
[
"Bono",
"J.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Bottalico",
"E.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Bowcock",
"T.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Braun",
"S.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Bressler",
"M.",
"",
"Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA"
],
[
"Cantatore",
"G.",
"",
"INFN, Sezione di Trieste, Trieste, Italy"
],
[
"Carey",
"R. M.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Casey",
"B. C. K.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Cauz",
"D.",
"",
"Università di Udine, Udine, Italy"
],
[
"Chakraborty",
"R.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Chapelain",
"A.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Chappa",
"S.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Charity",
"S.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Chen",
"C.",
"",
"21 and 22"
],
[
"Cheng",
"M.",
"",
"University of Illinois at Urbana-Champaign, Urbana, Illinois, USA"
],
[
"Chislett",
"R.",
"",
"Department of Physics and Astronomy, University College London, London, United Kingdom"
],
[
"Chu",
"Z.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Chupp",
"T. E.",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"Claessens",
"C.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Confortini",
"F.",
"",
"INFN, Sezione di Napoli, Naples, Italy"
],
[
"Convery",
"M. E.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Corrodi",
"S.",
"",
"Argonne National Laboratory, Lemont, Illinois, USA"
],
[
"Cotrozzi",
"L.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Crnkovic",
"J. D.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Dabagov",
"S.",
"",
"INFN, Laboratori Nazionali di Frascati, Frascati, Italy"
],
[
"Debevec",
"P. T.",
"",
"University of Illinois at Urbana-Champaign, Urbana, Illinois, USA"
],
[
"Di Falco",
"S.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Di Sciascio",
"G.",
"",
"INFN, Sezione di Roma Tor Vergata, Rome, Italy"
],
[
"Donati",
"S.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Drendel",
"B.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Driutti",
"A.",
"",
"10 and 29"
],
[
"Eads",
"M.",
"",
"Northern Illinois University, DeKalb, Illinois, USA"
],
[
"Edmonds",
"A.",
"",
"2 and 37"
],
[
"Esquivel",
"J.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Farooq",
"M.",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"Fatemi",
"R.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Ferraby",
"K.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Ferrari",
"C.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Fertl",
"M.",
"",
"Institute of Physics and Cluster of Excellence PRISMA+, Johannes Gutenberg University Mainz, Mainz, Germany"
],
[
"Fienberg",
"A. T.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Fioretti",
"A.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Flay",
"D.",
"",
"Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA"
],
[
"Foster",
"S. B.",
"",
"2 and 29"
],
[
"Friedsam",
"H.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Froemming",
"N. S.",
"",
"Northern Illinois University, DeKalb, Illinois, USA"
],
[
"Gabbanini",
"C.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Gaines",
"I.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Ganguly",
"S.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"George",
"J.",
"",
"Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA"
],
[
"Gibbons",
"L. K.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Gioiosa",
"A.",
"",
"Università del Molise, Campobasso, Italy"
],
[
"Giovanetti",
"K. L.",
"",
"Department of Physics and Astronomy, James Madison University, Harrisonburg, Virginia, USA"
],
[
"Girotti",
"P.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Gohn",
"W.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Goodenough",
"L.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Gorringe",
"T.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Grange",
"J.",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"Grant",
"S.",
"",
"1 and 27"
],
[
"Gray",
"F.",
"",
"Regis University, Denver, Colorado, USA"
],
[
"Haciomeroglu",
"S.",
"",
"Center for Axion and Precision Physics"
],
[
"Halewood-Leagas",
"T.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Hampai",
"D.",
"",
"INFN, Laboratori Nazionali di Frascati, Frascati, Italy"
],
[
"Han",
"F.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Hempstead",
"J.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Hertzog",
"D. W.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Hesketh",
"G.",
"",
"Department of Physics and Astronomy, University College London, London, United Kingdom"
],
[
"Hess",
"E.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Hibbert",
"A.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Hodge",
"Z.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Hoh",
"S. Y.",
"",
"21 and 22"
],
[
"Hong",
"K. W.",
"",
"University of Virginia, Charlottesville, Virginia, USA"
],
[
"Hong",
"R.",
"",
"1 and 29"
],
[
"Hu",
"T.",
"",
"21 and 22"
],
[
"Hu",
"Y.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Iacovacci",
"M.",
"",
"INFN, Sezione di Napoli, Naples, Italy"
],
[
"Incagli",
"M.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Israel",
"S.",
"",
"2 and 32"
],
[
"Kammel",
"P.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Kargiantoulakis",
"M.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Karuza",
"M.",
"",
"INFN, Sezione di Trieste, Trieste, Italy"
],
[
"Kaspar",
"J.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Kawall",
"D.",
"",
"Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA"
],
[
"Kelton",
"L.",
"",
"23 and 29"
],
[
"Keshavarzi",
"A.",
"",
"Department of Physics and Astronomy, University of Manchester, Manchester, United Kingdom"
],
[
"Kessler",
"D. S.",
"",
"Department of Physics, University of Massachusetts, Amherst, Massachusetts, USA"
],
[
"Khaw",
"K. S.",
"",
"21 and 22"
],
[
"Khechadoorian",
"Z.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Kiburg",
"B.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Kiburg",
"M.",
"",
"7 and 18"
],
[
"Kim",
"O.",
"",
"University of Mississippi, University, Mississippi, USA"
],
[
"Kinnaird",
"N.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Kraegeloh",
"E.",
"",
"University of Michigan, Ann Arbor, Michigan, USA"
],
[
"LaBounty",
"J.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Labe",
"K. R.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Lancaster",
"M.",
"",
"Department of Physics and Astronomy, University of Manchester, Manchester, United Kingdom"
],
[
"Lee",
"S.",
"",
"Center for Axion and Precision Physics"
],
[
"Li",
"B.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Li",
"D.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Li",
"L.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Logashenko",
"I.",
"",
"Budker Institute of Nuclear Physics, Novosibirsk, Russia"
],
[
"Campos",
"A. Lorente",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Lu",
"Z.",
"",
"School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, China"
],
[
"Lucà",
"A.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Lukicov",
"G.",
"",
"Department of Physics and Astronomy, University College London, London, United Kingdom"
],
[
"Lusiani",
"A.",
"",
"INFN, Sezione di Pisa, Pisa, Italy"
],
[
"Lyon",
"A. L.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"MacCoy",
"B.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Madrak",
"R.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Makino",
"K.",
"",
"Michigan State University, East Lansing, Michigan, USA"
],
[
"Mastroianni",
"S.",
"",
"INFN, Sezione di Napoli, Naples, Italy"
],
[
"McCarthy",
"R.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Miller",
"J. P.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Miozzi",
"S.",
"",
"INFN, Sezione di Roma Tor Vergata, Rome, Italy"
],
[
"Mitra",
"B.",
"",
"University of Mississippi, University, Mississippi, USA"
],
[
"Morgan",
"J. P.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Morse",
"W. M.",
"",
"Brookhaven National Laboratory, Upton, New York, USA"
],
[
"Mott",
"J.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Nath",
"A.",
"",
"INFN, Sezione di Napoli, Naples, Italy"
],
[
"Ng",
"J. K.",
"",
"21 and 22"
],
[
"Nguyen",
"H.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Oksuzian",
"Y.",
"",
"Argonne National Laboratory, Lemont, Illinois, USA"
],
[
"Omarov",
"Z.",
"",
"5 and 15"
],
[
"Osar",
"W.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Osofsky",
"R.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Park",
"S.",
"",
"Center for Axion and Precision Physics"
],
[
"Pauletta",
"G.",
"",
"Università di Udine, Udine, Italy"
],
[
"Peck",
"J.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Piacentino",
"G. M.",
"",
"Università del Molise, Campobasso, Italy"
],
[
"Pilato",
"R. N.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Pitts",
"K. T.",
"",
"28 and b"
],
[
"Plaster",
"B.",
"",
"University of Kentucky, Lexington, Kentucky, USA"
],
[
"Pohlman",
"N.",
"",
"Northern Illinois University, DeKalb, Illinois, USA"
],
[
"Polly",
"C. C.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Počanić",
"D.",
"",
"University of Virginia, Charlottesville, Virginia, USA"
],
[
"Price",
"J.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Quinn",
"B.",
"",
"University of Mississippi, University, Mississippi, USA"
],
[
"Qureshi",
"M. U. H.",
"",
"Institute of Physics and Cluster of Excellence PRISMA+, Johannes Gutenberg University Mainz, Mainz, Germany"
],
[
"Rakness",
"G.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Ramachandran",
"S.",
"",
"Argonne National Laboratory, Lemont, Illinois, USA"
],
[
"Ramberg",
"E.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Reimann",
"R.",
"",
"Institute of Physics and Cluster of Excellence PRISMA+, Johannes Gutenberg University Mainz, Mainz, Germany"
],
[
"Roberts",
"B. L.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Rubin",
"D. L.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Sakurai",
"M.",
"",
"Department of Physics and Astronomy, University College London, London, United Kingdom"
],
[
"Santi",
"L.",
"",
"Università di Udine, Udine, Italy"
],
[
"Schlesier",
"C.",
"",
"University of Illinois at Urbana-Champaign, Urbana, Illinois, USA"
],
[
"Schreckenberger",
"A.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Semertzidis",
"Y. K.",
"",
"5 and 15"
],
[
"Sorbara",
"M.",
"",
"INFN, Sezione di Roma Tor Vergata, Rome, Italy"
],
[
"Stapleton",
"J.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Still",
"D.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Stoughton",
"C.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Stratakis",
"D.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Stöckinger",
"D.",
"",
"Institut für Kern- und Teilchenphysik, Technische Universität Dresden, Dresden, Germany"
],
[
"Swanson",
"H. E.",
"",
"University of Washington, Seattle, Washington, USA"
],
[
"Sweetmore",
"G.",
"",
"Department of Physics and Astronomy, University of Manchester, Manchester, United Kingdom"
],
[
"Sweigart",
"D. A.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Syphers",
"M. J.",
"",
"Northern Illinois University, DeKalb, Illinois, USA"
],
[
"Takeuchi",
"Y.",
"",
"21 and 22"
],
[
"Tarazona",
"D. A.",
"",
"Cornell University, Ithaca, New York, USA"
],
[
"Teubner",
"T.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Tewsley-Booth",
"A. E.",
"",
"29 and 33"
],
[
"Tishchenko",
"V.",
"",
"Brookhaven National Laboratory, Upton, New York, USA"
],
[
"Tran",
"N. H.",
"",
"Boston University, Boston, Massachusetts, USA"
],
[
"Turner",
"W.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Valetov",
"E.",
"",
"Michigan State University, East Lansing, Michigan, USA"
],
[
"Vasilkova",
"D.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Venanzoni",
"G.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Walton",
"T.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Weisskopf",
"A.",
"",
"Michigan State University, East Lansing, Michigan, USA"
],
[
"Welty-Rieger",
"L.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Winter",
"P.",
"",
"Argonne National Laboratory, Lemont, Illinois, USA"
],
[
"Wu",
"Y.",
"",
"Argonne National Laboratory, Lemont, Illinois, USA"
],
[
"Yu",
"B.",
"",
"University of Mississippi, University, Mississippi, USA"
],
[
"Yucel",
"M.",
"",
"Fermi National Accelerator Laboratory, Batavia, Illinois, USA"
],
[
"Zaid",
"E.",
"",
"University of Liverpool, Liverpool, United Kingdom"
],
[
"Zeng",
"Y.",
"",
"21 and 22"
],
[
"Zhang",
"C.",
"",
"University of Liverpool, Liverpool, United Kingdom"
]
] |
2506.03070 | Tyler Chen | Tyler Chen, Pradeep Niroula, Archan Ray, Pragna Subrahmanya, Marco Pistoia, Niraj Kumar | GPU-Parallelizable Randomized Sketch-and-Precondition for Linear Regression using Sparse Sign Sketches | null | null | null | null | cs.DS cs.DC cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A litany of theoretical and numerical results have established the sketch-and-precondition paradigm as a powerful approach to solving large linear regression problems in standard computing environments. Perhaps surprisingly, much less work has been done on understanding how sketch-and-precondition performs on graphics processing unit (GPU) systems. We address this gap by benchmarking an implementation of sketch-and-precondition based on sparse sign-sketches on single and multi-GPU systems. In doing so, we describe a novel, easily parallelized, rejection-sampling based method for generating sparse sign sketches. Our approach, which is particularly well-suited for GPUs, is easily adapted to a variety of computing environments. Taken as a whole, our numerical experiments indicate that sketch-and-precondition with sparse sign sketches is particularly well-suited for GPUs, and may be suitable for use in black-box least-squares solvers. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:48:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Tyler",
""
],
[
"Niroula",
"Pradeep",
""
],
[
"Ray",
"Archan",
""
],
[
"Subrahmanya",
"Pragna",
""
],
[
"Pistoia",
"Marco",
""
],
[
"Kumar",
"Niraj",
""
]
] |
2506.03071 | Juan S\'anchez-Baena | Ra\'ul Bomb\'in, Jordi Boronat, Ferran Mazzanti, and Juan S\'anchez-Baena | Creating and melting a supersolid by heating a quantum dipolar system | 4 figures | null | null | null | cond-mat.quant-gas | http://creativecommons.org/licenses/by/4.0/ | Recent experiments have shown that rising the temperature of a dipolar gas under certain conditions leads to a transition to a supersolid state. Here, we employ the path integral Monte Carlo method, which exactly accounts for both thermal and correlation effects, to study that phenomenology in a system of $^{162}$Dy atoms in the canonical ensemble. Our microscopic description allows to quantitatively characterize the emergence of spatial order and superfluidity, the two ingredients that define a supersolid state. Our calculations prove that temperature on its own can promote the formation of a supersolid in a dipolar system. Furthermore, we bridge this exotic phenomenology with the more usual melting of the supersolid at a higher temperature. Our results offer insight into the interplay between thermal excitations, the dipole-dipole interaction, quantum statistics and supersolidity. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:49:27 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Bombín",
"Raúl",
""
],
[
"Boronat",
"Jordi",
""
],
[
"Mazzanti",
"Ferran",
""
],
[
"Sánchez-Baena",
"Juan",
""
]
] |
2506.03072 | Marcin Kucharczyk | LHCb collaboration | Three-pion Bose-Einstein correlations measured in proton-proton collisions | All figures and tables, along with machine-readable versions and any supplementary material and additional information, are available at https://lbfence.cern.ch/alcm/public/analysis/full-details/3759/ (LHCb public pages) | null | null | CERN-EP-2025-104, LHCb-PAPER-2025-007 | hep-ex | http://creativecommons.org/licenses/by/4.0/ | A study on the Bose-Einstein correlations for triplets of same-sign pions is presented. The analysis is performed using proton-proton collisions at a centre-of-mass energy of $\sqrt{s}$ = 7 TeV, recorded by the LHCb experiment, corresponding to an integrated luminosity of 1.0 fb$^{-1}$. For the first time, the results are interpreted in the core-halo model. The parameters of the model are determined in regions of charged-particle multiplicity. This measurement provides insight into the nature of hadronisation in terms of coherence, showing a coherent emission of pions. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:51:11 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"LHCb collaboration",
"",
""
]
] |
2506.03073 | Roman Titkov | Roman Titkov, Egor Zubkov, Dmitry Yudin, Jaafar Mahmoud, Malik Mohrat, Gennady Sidorov | LEG-SLAM: Real-Time Language-Enhanced Gaussian Splatting for SLAM | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Modern Gaussian Splatting methods have proven highly effective for real-time photorealistic rendering of 3D scenes. However, integrating semantic information into this representation remains a significant challenge, especially in maintaining real-time performance for SLAM (Simultaneous Localization and Mapping) applications. In this work, we introduce LEG-SLAM -- a novel approach that fuses an optimized Gaussian Splatting implementation with visual-language feature extraction using DINOv2 followed by a learnable feature compressor based on Principal Component Analysis, while enabling an online dense SLAM. Our method simultaneously generates high-quality photorealistic images and semantically labeled scene maps, achieving real-time scene reconstruction with more than 10 fps on the Replica dataset and 18 fps on ScanNet. Experimental results show that our approach significantly outperforms state-of-the-art methods in reconstruction speed while achieving competitive rendering quality. The proposed system eliminates the need for prior data preparation such as camera's ego motion or pre-computed static semantic maps. With its potential applications in autonomous robotics, augmented reality, and other interactive domains, LEG-SLAM represents a significant step forward in real-time semantic 3D Gaussian-based SLAM. Project page: https://titrom025.github.io/LEG-SLAM/ | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:51:59 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Titkov",
"Roman",
""
],
[
"Zubkov",
"Egor",
""
],
[
"Yudin",
"Dmitry",
""
],
[
"Mahmoud",
"Jaafar",
""
],
[
"Mohrat",
"Malik",
""
],
[
"Sidorov",
"Gennady",
""
]
] |
2506.03075 | Tom Waknine | Bogdan Chornomaz, Yonatan Koren, Shay Moran, Tom Waknine | Agnostic Learning under Targeted Poisoning: Optimal Rates and the Role of Randomness | null | null | null | null | cs.LG math.PR | http://creativecommons.org/licenses/by/4.0/ | We study the problem of learning in the presence of an adversary that can corrupt an $\eta$ fraction of the training examples with the goal of causing failure on a specific test point. In the realizable setting, prior work established that the optimal error under such instance-targeted poisoning attacks scales as $\Theta(d\eta)$, where $d$ is the VC dimension of the hypothesis class arXiv:2210.02713. In this work, we resolve the corresponding question in the agnostic setting. We show that the optimal excess error is $\tilde{\Theta}(\sqrt{d\eta})$, answering one of the main open problems left by Hanneke et al. To achieve this rate, it is necessary to use randomized learners: Hanneke et al. showed that deterministic learners can be forced to suffer error close to 1, even under small amounts of poisoning. Perhaps surprisingly, our upper bound remains valid even when the learner's random bits are fully visible to the adversary . In the other direction, our lower bound is stronger than standard PAC-style bounds: instead of tailoring a hard distribution separately for each sample size, we exhibit a single fixed distribution under which the adversary can enforce an excess error of $\Omega(\sqrt{d\eta})$ infinitely often. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:53:20 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chornomaz",
"Bogdan",
""
],
[
"Koren",
"Yonatan",
""
],
[
"Moran",
"Shay",
""
],
[
"Waknine",
"Tom",
""
]
] |
2506.03076 | Philip Tee | Paul C.W. Davies, Philip Tee | Discrete Spacetime Theories Can Explain the Muon Magnetic Moment Discrepancy | null | null | 10.1103/9jty-782x | null | hep-ph hep-th | http://creativecommons.org/licenses/by/4.0/ | An unsolved problem of particle physics is a discrepancy between the measured value of the muon anomalous magnetic moment and the theoretical prediction based on standard quantum electrodynamics. In this paper we show that if spacetime possesses a fundamental length scale, the ensuing modifications to the photon propagator can account for the discrepancy if the scale is chosen to be $10^{-22}$~m; the corresponding energy being about $30$~TeV. The possibility that spacetime possesses a graininess on a fine enough scale has a long history. One class of theories that develops this idea is Doubly Special Relativity (DSR), and we choose this as a model for our calculation. We note that the derived length scale is many orders of magnitude larger than the Planck length, but comparable to that of some higher dimensional gravitational theories. It is also within scope of experimental confirmation in the next generation of colliders. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:54:09 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Davies",
"Paul C. W.",
""
],
[
"Tee",
"Philip",
""
]
] |
2506.03077 | Qijun Luo | Qijun Luo, Mengqi Li, Lei Zhao, Xiao Li | StreamBP: Memory-Efficient Exact Backpropagation for Long Sequence Training of LLMs | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Training language models on long sequence data is a demanding requirement for enhancing the model's capability on complex tasks, e.g., long-chain reasoning. However, as the sequence length scales up, the memory cost for storing activation values becomes huge during the Backpropagation (BP) process, even with the application of gradient checkpointing technique. To tackle this challenge, we propose a memory-efficient and exact BP method called StreamBP, which performs a linear decomposition of the chain rule along the sequence dimension in a layer-wise manner, significantly reducing the memory cost of activation values and logits. The proposed method is applicable to common objectives such as SFT, GRPO, and DPO. From an implementation perspective, StreamBP achieves less computational FLOPs and faster BP speed by leveraging the causal structure of the language model. Compared to gradient checkpointing, StreamBP scales up the maximum sequence length of BP by 2.8-5.5 times larger, while using comparable or even less BP time. Note that StreamBP's sequence length scaling ability can be directly transferred to batch size scaling for accelerating training. We further develop a communication-efficient distributed StreamBP to effectively support multi-GPU training and broaden its applicability. Our code can be easily integrated into the training pipeline of any transformer models and is available at https://github.com/Ledzy/StreamBP. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:54:15 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Luo",
"Qijun",
""
],
[
"Li",
"Mengqi",
""
],
[
"Zhao",
"Lei",
""
],
[
"Li",
"Xiao",
""
]
] |
2506.03078 | Christian Peterson | Christian Peterson and David Hilditch | Strong hyperboloidal compactification for the spherical DF-GHG formulation of GR | 10 pages, 4 figures | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The use of compactified hyperboloidal coordinates for metric formulations of the Einstein field Equations introduces formally singular terms in the equations of motion whose numerical treatment requires care. In this paper we study a particular choice of constraint addition, choice of gauge and reduction fields in order to minimize the number of these terms in a spherically symmetric reduction of the Dual-Foliation Generalized Harmonic Gauge formulation of General Relativity. We proceed to the numerical implementation of a more aggressive compactification, as compared to our previous work. With the present setup there is a direct analogy with conformal compactification used in other approaches to the use of hyperboloidal coordinates. We present numerical results of constraints violating and satisfying perturbations on top of a Schwarzschild black hole. For small perturbations we recover the expected physics from linear theory, corresponding to quasi normal mode ringing and tail decay for a scalar field, both extracted directly at future null infinity from our numerical data. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:57:01 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Peterson",
"Christian",
""
],
[
"Hilditch",
"David",
""
]
] |
2506.03079 | Xiuyu Yang | Xiuyu Yang, Bohan Li, Shaocong Xu, Nan Wang, Chongjie Ye, Zhaoxi Chen, Minghan Qin, Yikang Ding, Xin Jin, Hang Zhao, Hao Zhao | ORV: 4D Occupancy-centric Robot Video Generation | Project page: https://orangesodahub.github.io/ORV/ ; Code: https://github.com/OrangeSodahub/ORV | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Acquiring real-world robotic simulation data through teleoperation is notoriously time-consuming and labor-intensive. Recently, action-driven generative models have gained widespread adoption in robot learning and simulation, as they eliminate safety concerns and reduce maintenance efforts. However, the action sequences used in these methods often result in limited control precision and poor generalization due to their globally coarse alignment. To address these limitations, we propose ORV, an Occupancy-centric Robot Video generation framework, which utilizes 4D semantic occupancy sequences as a fine-grained representation to provide more accurate semantic and geometric guidance for video generation. By leveraging occupancy-based representations, ORV enables seamless translation of simulation data into photorealistic robot videos, while ensuring high temporal consistency and precise controllability. Furthermore, our framework supports the simultaneous generation of multi-view videos of robot gripping operations - an important capability for downstream robotic learning tasks. Extensive experimental results demonstrate that ORV consistently outperforms existing baseline methods across various datasets and sub-tasks. Demo, Code and Model: https://orangesodahub.github.io/ORV | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:00:32 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yang",
"Xiuyu",
""
],
[
"Li",
"Bohan",
""
],
[
"Xu",
"Shaocong",
""
],
[
"Wang",
"Nan",
""
],
[
"Ye",
"Chongjie",
""
],
[
"Chen",
"Zhaoxi",
""
],
[
"Qin",
"Minghan",
""
],
[
"Ding",
"Yikang",
""
],
[
"Jin",
"Xin",
""
],
[
"Zhao",
"Hang",
""
],
[
"Zhao",
"Hao",
""
]
] |
2506.03080 | Bjarne Sch\"ulke | David Conlon, Bjarne Sch\"ulke | Hypergraphs accumulate infinitely often | 10 pages | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | We show that the set $\Pi^{(k)}$ of Tur\'an densities of $k$-uniform hypergraphs has infinitely many accumulation points in $[0,1)$ for every $k \geq 3$. This extends an earlier result of ours showing that $\Pi^{(k)}$ has at least one such accumulation point. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:01:33 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Conlon",
"David",
""
],
[
"Schülke",
"Bjarne",
""
]
] |
2506.03081 | Michael Dumbser | Walter Boscheri, Michael Dumbser, Raphael Loub\`ere, Pierre-Henri Maire | A structure-preserving and thermodynamically compatible cell-centered Lagrangian finite volume scheme for continuum mechanics | null | null | null | null | math.NA cs.NA physics.comp-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In this work we present a novel structure-preserving scheme for the discretization of the Godunov-Peshkov-Romenski (GPR) model of continuum mechanics written in Lagrangian form. This model admits an extra conservation law for the total energy (first principle of thermodynamics) and satisfies the entropy inequality (second principle of thermodynamics). Furthermore, in the absence of algebraic source terms, the distortion field of the continuum and the specific thermal impulse satisfy a curl-free condition, provided the initial data are curl-free. Last but not least, the determinant of the distortion field is related to the density of the medium, i.e. the system is also endowed with a nonlinear algebraic constraint.
The objective of this work is to construct and analyze a new semi-discrete thermodynamically compatible cell-centered Lagrangian finite volume scheme on moving unstructured meshes that satisfies the following structural properties of the governing PDE exactly at the discrete level: i) compatibility with the first law of thermodynamics, i.e. discrete total energy conservation; ii) compatibility with the second law of thermodynamics, i.e. discrete entropy inequality; iii) exact discrete compatibility between the density and the determinant of the distortion field; iv) exact preservation of the curl-free property of the distortion field and of the specific thermal impulse in the absence of algebraic source terms. We show that it is possible to achieve all above properties simultaneously. Unlike in existing schemes, we choose to directly discretize the entropy inequality, hence obtaining total energy conservation as a consequence of an appropriate and thermodynamically compatible discretization of all the other equations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:01:54 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Boscheri",
"Walter",
""
],
[
"Dumbser",
"Michael",
""
],
[
"Loubère",
"Raphael",
""
],
[
"Maire",
"Pierre-Henri",
""
]
] |
2506.03082 | Ssharvien Kumar Sivakumar | Ssharvien Kumar Sivakumar, Yannik Frisch, Ghazal Ghazaei, Anirban Mukhopadhyay | SG2VID: Scene Graphs Enable Fine-Grained Control for Video Synthesis | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Surgical simulation plays a pivotal role in training novice surgeons, accelerating their learning curve and reducing intra-operative errors. However, conventional simulation tools fall short in providing the necessary photorealism and the variability of human anatomy. In response, current methods are shifting towards generative model-based simulators. Yet, these approaches primarily focus on using increasingly complex conditioning for precise synthesis while neglecting the fine-grained human control aspect. To address this gap, we introduce SG2VID, the first diffusion-based video model that leverages Scene Graphs for both precise video synthesis and fine-grained human control. We demonstrate SG2VID's capabilities across three public datasets featuring cataract and cholecystectomy surgery. While SG2VID outperforms previous methods both qualitatively and quantitatively, it also enables precise synthesis, providing accurate control over tool and anatomy's size and movement, entrance of new tools, as well as the overall scene layout. We qualitatively motivate how SG2VID can be used for generative augmentation and present an experiment demonstrating its ability to improve a downstream phase detection task when the training set is extended with our synthetic videos. Finally, to showcase SG2VID's ability to retain human control, we interact with the Scene Graphs to generate new video samples depicting major yet rare intra-operative irregularities. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:02:38 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Sivakumar",
"Ssharvien Kumar",
""
],
[
"Frisch",
"Yannik",
""
],
[
"Ghazaei",
"Ghazal",
""
],
[
"Mukhopadhyay",
"Anirban",
""
]
] |
2506.03083 | Adrian de Wynter | Adrian de Wynter | Labelling Data with Unknown References | null | null | null | null | cs.DS cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | An evaluator is trustworthy when there exists some agreed-upon way to measure its performance as a labeller. The two ways to establish trustworthiness are either by testing it, or by assuming the evaluator `knows' somehow the way to label the corpus. However, if labelled references (e.g., a development set) are unavailable, neither of these approaches work: the former requires the data, and the latter is an assumption, not evidence. To address this, we introduce an algorithm (the `No-Data Algorithm') by which to establish trust in an evaluator without any existing references. Our algorithm works by successively posing challenges to said evaluator. We show that this is sufficient to establish trustworthiness w.h.p., in such a way that when the evaluator actually knows the way to label the corpus, the No-Data Algorithm accepts its output; and, conversely, flags untrustworthy evaluators when these are unable to prove it. We present formal proofs of correctness and limited experiments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:04:22 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"de Wynter",
"Adrian",
""
]
] |
2506.03084 | Yingying Sun | Zizhao Wu, Yingying Sun, Yiming Chen, Xiaoling Gu, Ruyu Liu, Jiazhou Chen | InterMamba: Efficient Human-Human Interaction Generation with Adaptive Spatio-Temporal Mamba | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Human-human interaction generation has garnered significant attention in motion synthesis due to its vital role in understanding humans as social beings. However, existing methods typically rely on transformer-based architectures, which often face challenges related to scalability and efficiency. To address these issues, we propose a novel, efficient human-human interaction generation method based on the Mamba framework, designed to meet the demands of effectively capturing long-sequence dependencies while providing real-time feedback. Specifically, we introduce an adaptive spatio-temporal Mamba framework that utilizes two parallel SSM branches with an adaptive mechanism to integrate the spatial and temporal features of motion sequences. To further enhance the model's ability to capture dependencies within individual motion sequences and the interactions between different individual sequences, we develop two key modules: the self-adaptive spatio-temporal Mamba module and the cross-adaptive spatio-temporal Mamba module, enabling efficient feature learning. Extensive experiments demonstrate that our method achieves state-of-the-art results on two interaction datasets with remarkable quality and efficiency. Compared to the baseline method InterGen, our approach not only improves accuracy but also requires a minimal parameter size of just 66M ,only 36% of InterGen's, while achieving an average inference speed of 0.57 seconds, which is 46% of InterGen's execution time. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:05:06 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Wu",
"Zizhao",
""
],
[
"Sun",
"Yingying",
""
],
[
"Chen",
"Yiming",
""
],
[
"Gu",
"Xiaoling",
""
],
[
"Liu",
"Ruyu",
""
],
[
"Chen",
"Jiazhou",
""
]
] |
2506.03085 | Thomas Chen | Thomas Chen, Tengyu Ma, Zhiyuan Li | Non-Asymptotic Length Generalization | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Length generalization is the ability of a learning algorithm to learn a hypothesis which generalizes to longer inputs than the inputs in the training set. In this paper, we provide provable guarantees of length generalization for various classes of functions in an idealized setting. First, we formalize the framework of non-asymptotic length generalization, which requires a computable upper bound for the minimum input length that guarantees length generalization, as a function of the complexity of ground-truth function under some given complexity measure. We refer to this minimum input length to length generalize as length complexity. We show the Minimum-Complexity Interpolator learning algorithm achieves optimal length complexity. We further show that whether a function class admits non-asymptotic length generalization is equivalent to the decidability of its language equivalence problem, which implies that there is no computable upper bound for the length complexity of Context-Free Grammars. On the positive side, we show that the length complexity of Deterministic Finite Automata is $2n - 2$ where $n$ is the number of states of the ground-truth automaton. Our main results are upper bounds of length complexity for a subset of a transformer-related function class called C-RASP (Yang & Chiang, 2024). We show that the length complexity of 1-layer C-RASP functions is $O(T^2)$ when the ground-truth function has precision $T$, and that the length complexity of 2-layer C-RASP functions is $O(T^{O(K)})$ when the ground-truth function has precision $T$ and $K$ heads. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:09:34 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Chen",
"Thomas",
""
],
[
"Ma",
"Tengyu",
""
],
[
"Li",
"Zhiyuan",
""
]
] |
2506.03086 | Nan Xi | Nan Miles Xi, Man Mandy Jin, Lin Wang, Xin Huang | Synergy-Informed Design of Platform Trials for Combination Therapies | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | Combination drug therapies hold significant promise for enhancing treatment efficacy, particularly in fields such as oncology, immunotherapy, and infectious diseases. However, designing clinical trials for these regimens poses unique statistical challenges due to multiple hypothesis testing, shared control groups, and overlapping treatment components that induce complex correlation structures. In this paper, we develop a novel statistical framework tailored for early-phase translational combination therapy trials, with a focus on platform trial designs. Our methodology introduces a generalized Dunnett's procedure that controls false positive rates by accounting for the correlations between treatment arms. Additionally, we propose strategies for power analysis and sample size optimization that leverage preclinical data to estimate effect sizes, synergy parameters, and inter-arm correlations. Simulation studies demonstrate that our approach not only controls various false positive metrics under diverse trial scenarios but also informs optimal allocation ratios to maximize power. A real-data application further illustrates the integration of translational preclinical insights into the clinical trial design process. An open-source R package is provided to support the application of our methods in practice. Overall, our framework offers statistically rigorous guidance for the design of early-phase combination therapy trials, aiming to enhance the efficiency of the bench-to-bedside transition. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:10:49 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Xi",
"Nan Miles",
""
],
[
"Jin",
"Man Mandy",
""
],
[
"Wang",
"Lin",
""
],
[
"Huang",
"Xin",
""
]
] |
2506.03087 | Bin Ma | Bin Ma, Yuyuan Feng, Minhua Lin, Enyan Dai | How Explanations Leak the Decision Logic: Stealing Graph Neural Networks via Explanation Alignment | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph Neural Networks (GNNs) have become essential tools for analyzing graph-structured data in domains such as drug discovery and financial analysis, leading to growing demands for model transparency. Recent advances in explainable GNNs have addressed this need by revealing important subgraphs that influence predictions, but these explanation mechanisms may inadvertently expose models to security risks. This paper investigates how such explanations potentially leak critical decision logic that can be exploited for model stealing. We propose {\method}, a novel stealing framework that integrates explanation alignment for capturing decision logic with guided data augmentation for efficient training under limited queries, enabling effective replication of both the predictive behavior and underlying reasoning patterns of target models. Experiments on molecular graph datasets demonstrate that our approach shows advantages over conventional methods in model stealing. This work highlights important security considerations for the deployment of explainable GNNs in sensitive domains and suggests the need for protective measures against explanation-based attacks. Our code is available at https://github.com/beanmah/EGSteal. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:11:05 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Ma",
"Bin",
""
],
[
"Feng",
"Yuyuan",
""
],
[
"Lin",
"Minhua",
""
],
[
"Dai",
"Enyan",
""
]
] |
2506.03088 | Lloyd Pellatt | Lloyd Pellatt, Fotios Drakopoulos, Shievanie Sabesan and Nicholas A. Lesica | Modelling the Effects of Hearing Loss on Neural Coding in the Auditory Midbrain with Variational Conditioning | 12 pages, 3 figures | null | null | null | q-bio.NC cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The mapping from sound to neural activity that underlies hearing is highly non-linear. The first few stages of this mapping in the cochlea have been modelled successfully, with biophysical models built by hand and, more recently, with DNN models trained on datasets simulated by biophysical models. Modelling the auditory brain has been a challenge because central auditory processing is too complex for models to be built by hand, and datasets for training DNN models directly have not been available. Recent work has taken advantage of large-scale high resolution neural recordings from the auditory midbrain to build a DNN model of normal hearing with great success. But this model assumes that auditory processing is the same in all brains, and therefore it cannot capture the widely varying effects of hearing loss.
We propose a novel variational-conditional model to learn to encode the space of hearing loss directly from recordings of neural activity in the auditory midbrain of healthy and noise exposed animals. With hearing loss parametrised by only 6 free parameters per animal, our model accurately predicts 62\% of the explainable variance in neural responses from normal hearing animals and 68% for hearing impaired animals, within a few percentage points of state of the art animal specific models. We demonstrate that the model can be used to simulate realistic activity from out of sample animals by fitting only the learned conditioning parameters with Bayesian optimisation, achieving crossentropy loss within 2% of the optimum in 15-30 iterations. Including more animals in the training data slightly improved the performance on unseen animals. This model will enable future development of parametrised hearing loss compensation models trained to directly restore normal neural coding in hearing impaired brains, which can be quickly fitted for a new user by human in the loop optimisation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:12:21 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Pellatt",
"Lloyd",
""
],
[
"Drakopoulos",
"Fotios",
""
],
[
"Sabesan",
"Shievanie",
""
],
[
"Lesica",
"Nicholas A.",
""
]
] |
2506.03089 | Lucas Piper | Lucas Piper, Arlindo L. Oliveira, Tiago Marques | Explicitly Modeling Subcortical Vision with a Neuro-Inspired Front-End Improves CNN Robustness | null | null | null | null | cs.CV q-bio.NC | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Convolutional neural networks (CNNs) trained on object recognition achieve high task performance but continue to exhibit vulnerability under a range of visual perturbations and out-of-domain images, when compared with biological vision. Prior work has demonstrated that coupling a standard CNN with a front-end block (VOneBlock) that mimics the primate primary visual cortex (V1) can improve overall model robustness. Expanding on this, we introduce Early Vision Networks (EVNets), a new class of hybrid CNNs that combine the VOneBlock with a novel SubcorticalBlock, whose architecture draws from computational models in neuroscience and is parameterized to maximize alignment with subcortical responses reported across multiple experimental studies. Without being optimized to do so, the assembly of the SubcorticalBlock with the VOneBlock improved V1 alignment across most standard V1 benchmarks, and better modeled extra-classical receptive field phenomena. In addition, EVNets exhibit stronger emergent shape bias and overperform the base CNN architecture by 8.5% on an aggregate benchmark of robustness evaluations, including adversarial perturbations, common corruptions, and domain shifts. Finally, we show that EVNets can be further improved when paired with a state-of-the-art data augmentation technique, surpassing the performance of the isolated data augmentation approach by 7.3% on our robustness benchmark. This result reveals complementary benefits between changes in architecture to better mimic biology and training-based machine learning approaches. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:13:51 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Piper",
"Lucas",
""
],
[
"Oliveira",
"Arlindo L.",
""
],
[
"Marques",
"Tiago",
""
]
] |
2506.03090 | Katherine Thai | Katherine Thai and Mohit Iyyer | Literary Evidence Retrieval via Long-Context Language Models | ACL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | How well do modern long-context language models understand literary fiction? We explore this question via the task of literary evidence retrieval, repurposing the RELiC dataset of That et al. (2022) to construct a benchmark where the entire text of a primary source (e.g., The Great Gatsby) is provided to an LLM alongside literary criticism with a missing quotation from that work. This setting, in which the model must generate the missing quotation, mirrors the human process of literary analysis by requiring models to perform both global narrative reasoning and close textual examination. We curate a high-quality subset of 292 examples through extensive filtering and human verification. Our experiments show that recent reasoning models, such as Gemini Pro 2.5 can exceed human expert performance (62.5% vs. 50% accuracy). In contrast, the best open-weight model achieves only 29.1% accuracy, highlighting a wide gap in interpretive reasoning between open and closed-weight models. Despite their speed and apparent accuracy, even the strongest models struggle with nuanced literary signals and overgeneration, signaling open challenges for applying LLMs to literary analysis. We release our dataset and evaluation code to encourage future work in this direction. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:19:45 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Thai",
"Katherine",
""
],
[
"Iyyer",
"Mohit",
""
]
] |
2506.03091 | Manuel Meyer | Jacob Egge and Manuel Meyer | Sensitivity of a Gigahertz Fabry-P\'erot Resonator for Axion Dark Matter Detection | 10 pages, 8 figures, 1 table. Submitted | null | null | null | physics.ins-det astro-ph.IM hep-ex hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Axions are hypothetical pseudo-Nambu Goldstone bosons that could explain the observed cold dark matter density and solve the strong CP problem of quantum chromodynamics (QCD). Haloscope experiments commonly employ resonant cavities to search for a conversion of axion dark matter into photons in external magnetic fields. As the expected signal power degrades with increasing frequency, this approach becomes challenging at frequencies beyond tens of Gigahertz. Here, we propose a novel haloscope design based on an open Fabry-P\'erot resonator. Operating a small-scale resonator at cryogenic temperatures and at modest magnetic fields should already lead to an unparalleled sensitivity for photon-axion couplings $g_{a\gamma} \gtrsim 3\times10^{-12}\,\mathrm{GeV}^{-1}$ at 35GHz. We demonstrate how this sensitivity could be further improved using graded-phase mirrors and sketch possibilities to probe benchmark models of the QCD axion. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:20:57 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Egge",
"Jacob",
""
],
[
"Meyer",
"Manuel",
""
]
] |
2506.03092 | D. N\'obrega-Siverio | D. N\'obrega-Siverio, R. Joshi, E. Sola-Viladesau, D. Berghmans, D. Lim | Thin coronal jets and plasmoid-mediated reconnection: Insights from Solar Orbiter observations and Bifrost simulations | Submitted to A&A, 15 pages, 9 figures, movies available upon request | null | null | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Coronal jets are ubiquitous, collimated million-degree ejections that contribute to the energy and mass supply of the upper solar atmosphere and the solar wind. Solar Orbiter provides an unprecedented opportunity to observe fine-scale jets from a unique vantage point close to the Sun. We aim to (1) uncover thin jets originating from Coronal Bright Points (CBPs), revealing previously unresolved contributions to coronal upflows; and (2) improve our understanding of plasmoid-mediated reconnection and its observable signatures. We analyze eleven datasets from the High Resolution Imager 174 \r{A} of the Extreme Ultraviolet Imager (HRIEUV) onboard Solar Orbiter, focusing on narrow jets from CBPs and signatures of magnetic reconnection within current sheets and outflow regions. To support the observations, we compare with CBP simulations performed with the Bifrost code. We have identified thin coronal jets originating from CBPs with widths ranging from 253 km to 706 km: scales that could not be resolved with previous EUV imaging instruments. Remarkably, these jets are 30-85% brighter than their surroundings and can extend up to 22 Mm while maintaining their narrow form. In one of the datasets, we directly identify plasmoid-mediated reconnection through the development within the current sheet of a small-scale plasmoid that reaches a size of 332 km and propagates at 40 km/s. In another dataset, we infer plasmoid signatures through the intermittent boomerang-like pattern that appears in the outflow region. Both direct and indirect plasmoid-mediated reconnection signatures are supported by comparisons with the synthetic HRIEUV emission from the simulations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:24:08 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Nóbrega-Siverio",
"D.",
""
],
[
"Joshi",
"R.",
""
],
[
"Sola-Viladesau",
"E.",
""
],
[
"Berghmans",
"D.",
""
],
[
"Lim",
"D.",
""
]
] |
2506.03093 | Val\'erie Costa | Val\'erie Costa, Thomas Fel, Ekdeep Singh Lubana, Bahareh Tolooshams, Demba Ba | From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuit | Preprint | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motivated by the hypothesis that neural network representations encode abstract, interpretable features as linearly accessible, approximately orthogonal directions, sparse autoencoders (SAEs) have become a popular tool in interpretability. However, recent work has demonstrated phenomenology of model representations that lies outside the scope of this hypothesis, showing signatures of hierarchical, nonlinear, and multi-dimensional features. This raises the question: do SAEs represent features that possess structure at odds with their motivating hypothesis? If not, does avoiding this mismatch help identify said features and gain further insights into neural network representations? To answer these questions, we take a construction-based approach and re-contextualize the popular matching pursuits (MP) algorithm from sparse coding to design MP-SAE -- an SAE that unrolls its encoder into a sequence of residual-guided steps, allowing it to capture hierarchical and nonlinearly accessible features. Comparing this architecture with existing SAEs on a mixture of synthetic and natural data settings, we show: (i) hierarchical concepts induce conditionally orthogonal features, which existing SAEs are unable to faithfully capture, and (ii) the nonlinear encoding step of MP-SAE recovers highly meaningful features, helping us unravel shared structure in the seemingly dichotomous representation spaces of different modalities in a vision-language model, hence demonstrating the assumption that useful features are solely linearly accessible is insufficient. We also show that the sequential encoder principle of MP-SAE affords an additional benefit of adaptive sparsity at inference time, which may be of independent interest. Overall, we argue our results provide credence to the idea that interpretability should begin with the phenomenology of representations, with methods emerging from assumptions that fit it. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:24:55 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Costa",
"Valérie",
""
],
[
"Fel",
"Thomas",
""
],
[
"Lubana",
"Ekdeep Singh",
""
],
[
"Tolooshams",
"Bahareh",
""
],
[
"Ba",
"Demba",
""
]
] |
2506.03094 | Michael Beverland E | Theodore J. Yoder, Eddie Schoute, Patrick Rall, Emily Pritchett, Jay M. Gambetta, Andrew W. Cross, Malcolm Carroll, Michael E. Beverland | Tour de gross: A modular quantum computer based on bivariate bicycle codes | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present the bicycle architecture, a modular quantum computing framework based on high-rate, low-overhead quantum LDPC codes identified in prior work. For two specific bivariate bicycle codes with distances 12 and 18, we construct explicit fault-tolerant logical instruction sets and estimate the logical error rate of the instructions under circuit noise. We develop a compilation strategy adapted to the constraints of the bicycle architecture, enabling large-scale universal quantum circuit execution. Integrating these components, we perform end-to-end resource estimates demonstrating that an order of magnitude larger logical circuits can be implemented with a given number of physical qubits on the bicycle architecture than on surface code architectures. We anticipate further improvements through advances in code constructions, circuit designs, and compilation techniques. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:25:13 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Yoder",
"Theodore J.",
""
],
[
"Schoute",
"Eddie",
""
],
[
"Rall",
"Patrick",
""
],
[
"Pritchett",
"Emily",
""
],
[
"Gambetta",
"Jay M.",
""
],
[
"Cross",
"Andrew W.",
""
],
[
"Carroll",
"Malcolm",
""
],
[
"Beverland",
"Michael E.",
""
]
] |
2506.03095 | Man Luo | Man Luo, David Cobbley, Xin Su, Shachar Rosenman, Vasudev Lal, Shao-Yen Tseng, Phillip Howard | DPO Learning with LLMs-Judge Signal for Computer Use Agents | null | null | null | null | cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | Computer use agents (CUA) are systems that automatically interact with graphical user interfaces (GUIs) to complete tasks. CUA have made significant progress with the advent of large vision-language models (VLMs). However, these agents typically rely on cloud-based inference with substantial compute demands, raising critical privacy and scalability concerns, especially when operating on personal devices. In this work, we take a step toward privacy-preserving and resource-efficient agents by developing a lightweight vision-language model that runs entirely on local machines. To train this compact agent, we introduce an LLM-as-Judge framework that automatically evaluates and filters synthetic interaction trajectories, producing high-quality data for reinforcement learning without human annotation. Experiments on the OS-World benchmark demonstrate that our fine-tuned local model outperforms existing baselines, highlighting a promising path toward private, efficient, and generalizable GUI agents. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:27:04 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Luo",
"Man",
""
],
[
"Cobbley",
"David",
""
],
[
"Su",
"Xin",
""
],
[
"Rosenman",
"Shachar",
""
],
[
"Lal",
"Vasudev",
""
],
[
"Tseng",
"Shao-Yen",
""
],
[
"Howard",
"Phillip",
""
]
] |
2506.03096 | Christian Schlarmann | Christian Schlarmann, Francesco Croce, Nicolas Flammarion, Matthias Hein | FuseLIP: Multimodal Embeddings via Early Fusion of Discrete Tokens | Code and models available at https://github.com/chs20/fuselip | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Contrastive language-image pre-training aligns the features of text-image pairs in a common latent space via distinct encoders for each modality. While this approach achieves impressive performance in several zero-shot tasks, it cannot natively handle multimodal inputs, i.e., encoding image and text into a single feature vector. As a remedy, it is common practice to use additional modules to merge the features extracted by the unimodal encoders. In this work, we present FuseLIP, an alternative architecture for multimodal embedding. Leveraging recent progress in discrete image tokenizers, we propose to use a single transformer model which operates on an extended vocabulary of text and image tokens. This early fusion approach allows the different modalities to interact at each depth of encoding and obtain richer representations compared to common late fusion. We collect new datasets for multimodal pre-training and evaluation, designing challenging tasks for multimodal encoder models. We show that FuseLIP outperforms other approaches in multimodal embedding tasks such as VQA and text-guided image transformation retrieval, while being comparable to baselines on unimodal tasks. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:27:12 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Schlarmann",
"Christian",
""
],
[
"Croce",
"Francesco",
""
],
[
"Flammarion",
"Nicolas",
""
],
[
"Hein",
"Matthias",
""
]
] |
2506.03097 | Shrey Pandit | Ashwin Vinod, Shrey Pandit, Aditya Vavre, Linshen Liu | EgoVLM: Policy Optimization for Egocentric Video Understanding | Our Code can be found at https://github.com/adityavavre/VidEgoVLM | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Emerging embodied AI applications, such as wearable cameras and autonomous agents, have underscored the need for robust reasoning from first person video streams. We introduce EgoVLM, a vision-language model specifically designed to integrate visual comprehension and spatial-temporal reasoning within egocentric video contexts. EgoVLM is fine-tuned via Group Relative Policy Optimization (GRPO), a reinforcement learning method adapted to align model outputs with human-like reasoning steps. Following DeepSeek R1-Zero's approach, we directly tune using RL without any supervised fine-tuning phase on chain-of-thought (CoT) data. We evaluate EgoVLM on egocentric video question answering benchmarks and show that domain-specific training substantially improves performance over general-purpose VLMs. Our EgoVLM-3B, trained exclusively on non-CoT egocentric data, outperforms the base Qwen2.5-VL 3B and 7B models by 14.33 and 13.87 accuracy points on the EgoSchema benchmark, respectively. By explicitly generating reasoning traces, EgoVLM enhances interpretability, making it well-suited for downstream applications. Furthermore, we introduce a novel keyframe-based reward that incorporates salient frame selection to guide reinforcement learning optimization. This reward formulation opens a promising avenue for future exploration in temporally grounded egocentric reasoning. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:28:00 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Vinod",
"Ashwin",
""
],
[
"Pandit",
"Shrey",
""
],
[
"Vavre",
"Aditya",
""
],
[
"Liu",
"Linshen",
""
]
] |
2506.03098 | Danilo Triggiani | Francesco Di Lena, Fabrizio Sgobba, Danilo Triggiani, Andrea Andrisani, Cosmo Lupo, Piergiorgio Daniele, Gennaro Fratta, Giulia Acconcia, Ivan Rech, Luigi Santamaria Amato | High-precision measurement of time delay with frequency-resolved Hong-Ou-Mandel interference of weak coherent states | 12 pages, 6 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We demonstrate a scheme for high-precision measurements of time delay based on frequency-resolved Hong-Ou-Mandel (HOM) interference. Our approach is applied to weak coherent states and exploits an array of single-photon avalanche diodes (SPADs). Unlike conventional HOM experiments, our setup enables high-precision measurements even for photons separated by delays much greater than their coherence time. This result confirms our newly developed theoretical predictions that consider, differently from previous theoretical results, a finite frequency resolution in the detection. We compare the performance of this scheme against the conventional non-resolved case. Experimental data align well with the predictions of quantum estimation theory, demonstrating a significant reduction in the uncertainty. Due to the physics of the frequency-resolved HOM effect, the gain in precision is particularly high when the estimated time delay is much longer than the coherence time. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:28:10 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Di Lena",
"Francesco",
""
],
[
"Sgobba",
"Fabrizio",
""
],
[
"Triggiani",
"Danilo",
""
],
[
"Andrisani",
"Andrea",
""
],
[
"Lupo",
"Cosmo",
""
],
[
"Daniele",
"Piergiorgio",
""
],
[
"Fratta",
"Gennaro",
""
],
[
"Acconcia",
"Giulia",
""
],
[
"Rech",
"Ivan",
""
],
[
"Amato",
"Luigi Santamaria",
""
]
] |
2506.03099 | Weimin Wang | Chetwin Low, Weimin Wang | TalkingMachines: Real-Time Audio-Driven FaceTime-Style Video via Autoregressive Diffusion Models | null | null | null | null | cs.SD cs.AI cs.GR | http://creativecommons.org/licenses/by/4.0/ | In this paper, we present TalkingMachines -- an efficient framework that transforms pretrained video generation models into real-time, audio-driven character animators. TalkingMachines enables natural conversational experiences by integrating an audio large language model (LLM) with our video generation foundation model. Our primary contributions include: (1) We adapt a pretrained SOTA image-to-video DiT into an audio-driven avatar generation model of 18 billion parameters; (2) We enable infinite video streaming without error accumulation through asymmetric knowledge distillation from a bidirectional teacher model into a sparse causal, autoregressive student model; (3) We design a high-throughput, low-latency inference pipeline incorporating several key engineering optimizations such as: (a) disaggregation of the DiT and VAE decoder across separate devices, (b) efficient overlap of inter-device communication and computation using CUDA streams, (c) elimination of redundant recomputations to maximize frame-generation throughput. Please see demo videos here - https://aaxwaz.github.io/TalkingMachines/ | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:29:28 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Low",
"Chetwin",
""
],
[
"Wang",
"Weimin",
""
]
] |
2506.03101 | Jonas F. Lotz | Jonas F. Lotz, Ant\'onio V. Lopes, Stephan Peitz, Hendra Setiawan, Leonardo Emili | Beyond Text Compression: Evaluating Tokenizers Across Scales | ACL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | The choice of tokenizer can profoundly impact language model performance, yet accessible and reliable evaluations of tokenizer quality remain an open challenge. Inspired by scaling consistency, we show that smaller models can accurately predict significant differences in tokenizer impact on larger models at a fraction of the compute cost. By systematically evaluating both English-centric and multilingual tokenizers, we find that tokenizer choice has negligible effects on tasks in English but results in consistent performance differences in multilingual settings. We propose new intrinsic tokenizer metrics inspired by Zipf's law that correlate more strongly with downstream performance than text compression when modeling unseen languages. By combining several metrics to capture multiple aspects of tokenizer behavior, we develop a reliable framework for intrinsic tokenizer evaluations. Our work offers a more efficient path to informed tokenizer selection in future language model development. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:35:56 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Lotz",
"Jonas F.",
""
],
[
"Lopes",
"António V.",
""
],
[
"Peitz",
"Stephan",
""
],
[
"Setiawan",
"Hendra",
""
],
[
"Emili",
"Leonardo",
""
]
] |
2506.03102 | Sophie Greenwood | Sophie Greenwood, Karen Levy, Solon Barocas, Hoda Heidari, and Jon Kleinberg | Designing Algorithmic Delegates: The Role of Indistinguishability in Human-AI Handoff | Accepted at the Twenty-Sixth ACM Conference on Economics and Computation (EC'25) | null | null | null | cs.GT cs.AI cs.CY | http://creativecommons.org/licenses/by/4.0/ | As AI technologies improve, people are increasingly willing to delegate tasks to AI agents. In many cases, the human decision-maker chooses whether to delegate to an AI agent based on properties of the specific instance of the decision-making problem they are facing. Since humans typically lack full awareness of all the factors relevant to this choice for a given decision-making instance, they perform a kind of categorization by treating indistinguishable instances -- those that have the same observable features -- as the same. In this paper, we define the problem of designing the optimal algorithmic delegate in the presence of categories. This is an important dimension in the design of algorithms to work with humans, since we show that the optimal delegate can be an arbitrarily better teammate than the optimal standalone algorithmic agent. The solution to this optimal delegation problem is not obvious: we discover that this problem is fundamentally combinatorial, and illustrate the complex relationship between the optimal design and the properties of the decision-making task even in simple settings. Indeed, we show that finding the optimal delegate is computationally hard in general. However, we are able to find efficient algorithms for producing the optimal delegate in several broad cases of the problem, including when the optimal action may be decomposed into functions of features observed by the human and the algorithm. Finally, we run computational experiments to simulate a designer updating an algorithmic delegate over time to be optimized for when it is actually adopted by users, and show that while this process does not recover the optimal delegate in general, the resulting delegate often performs quite well. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:36:20 GMT"
}
] | 2025-06-04T00:00:00 | [
[
"Greenwood",
"Sophie",
""
],
[
"Levy",
"Karen",
""
],
[
"Barocas",
"Solon",
""
],
[
"Heidari",
"Hoda",
""
],
[
"Kleinberg",
"Jon",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.