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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2412.11927 | Shane Storks | Shane Storks, Itamar Bar-Yossef, Yayuan Li, Zheyuan Zhang, Jason J. Corso, Joyce Chai | Transparent and Coherent Procedural Mistake Detection | null | null | null | null | cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Procedural mistake detection (PMD) is a challenging problem of classifying whether a human user (observed through egocentric video) has successfully executed a task (specified by a procedural text). Despite significant recent efforts, machine performance in the wild remains nonviable, and the reasoning processes underlying this performance are opaque. As such, we extend PMD to require generating visual self-dialog rationales to inform decisions. Given the impressive, mature image understanding capabilities observed in recent vision-and-language models (VLMs), we curate a suitable benchmark dataset for PMD based on individual frames. As our reformulation enables unprecedented transparency, we leverage a natural language inference (NLI) model to formulate two automated metrics for the coherence of generated rationales. We establish baselines for this reframed task, showing that while VLMs struggle off-the-shelf, their accuracy, coherence, and efficiency can be improved by incorporating these metrics into common inference and fine-tuning methods- though not without tradeoff. Lastly, our multi-faceted metrics visualize common outcomes, highlighting areas for further improvement. | [
{
"version": "v1",
"created": "Mon, 16 Dec 2024 16:13:55 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:59:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Storks",
"Shane",
""
],
[
"Bar-Yossef",
"Itamar",
""
],
[
"Li",
"Yayuan",
""
],
[
"Zhang",
"Zheyuan",
""
],
[
"Corso",
"Jason J.",
""
],
[
"Chai",
"Joyce",
""
]
] |
2412.12040 | Anthony Hughes | Anthony Hughes, Ning Ma, Nikolaos Aletras | How Private are Language Models in Abstractive Summarization? | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In sensitive domains such as medical and legal, protecting sensitive information is critical, with protective laws strictly prohibiting the disclosure of personal data. This poses challenges for sharing valuable data such as medical reports and legal cases summaries. While language models (LMs) have shown strong performance in text summarization, it is still an open question to what extent they can provide privacy-preserving summaries from non-private source documents. In this paper, we perform a comprehensive study of privacy risks in LM-based summarization across two closed- and four open-weight models of different sizes and families. We experiment with both prompting and fine-tuning strategies for privacy-preservation across a range of summarization datasets including medical and legal domains. Our quantitative and qualitative analysis, including human evaluation, shows that LMs frequently leak personally identifiable information in their summaries, in contrast to human-generated privacy-preserving summaries, which demonstrate significantly higher privacy protection levels. These findings highlight a substantial gap between current LM capabilities and expert human expert performance in privacy-sensitive summarization tasks. | [
{
"version": "v1",
"created": "Mon, 16 Dec 2024 18:08:22 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:23:24 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Hughes",
"Anthony",
""
],
[
"Ma",
"Ning",
""
],
[
"Aletras",
"Nikolaos",
""
]
] |
2412.12243 | Edoardo Lauria El | Dario Benedetti, Edoardo Lauria, Dalimil Maz\'a\v{c}, Philine van Vliet | 1d Ising model with $1/r^{1.99}$ interaction | 5+1 pages. v2: published version | Phys. Rev. Lett. 134, 201602 (2025) | 10.1103/PhysRevLett.134.201602 | null | hep-th cond-mat.stat-mech math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We study the 1d Ising model with long-range interactions decaying as $1/r^{1+s}$. The critical model corresponds to a family of 1d conformal field theories (CFTs) whose data depends nontrivially on $s$ in the range $1/2\leq s\leq 1$. The model is known to be described by a generalized free field with quartic interaction, which is weakly coupled near $s=1/2$ but strongly coupled near the short-range crossover at $s=1$. We propose a dual description which becomes weakly coupled at $s=1$. At $s=1$, our model becomes an exactly solvable conformal boundary condition for the 2d free scalar. We perform a number of consistency checks of our proposal and calculate the perturbative CFT data around $s=1$ analytically using both 1) our proposed field theory and 2) the analytic conformal bootstrap. Our results show complete agreement between the two methods. | [
{
"version": "v1",
"created": "Mon, 16 Dec 2024 19:00:00 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 10:29:17 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Benedetti",
"Dario",
""
],
[
"Lauria",
"Edoardo",
""
],
[
"Mazáč",
"Dalimil",
""
],
[
"van Vliet",
"Philine",
""
]
] |
2412.12451 | Qi Zhang | Qi Zhang, Xiaoyang Wang, Rong Shi, Xinguo Ren, Han Wang, Lei Wang | Neural Canonical Transformations for Quantum Anharmonic Solids of Lithium | 7+7 pages, 3+4 figures | null | null | null | cond-mat.mtrl-sci physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Lithium is a typical quantum solid, characterized by cubic structures at ambient pressure. As the pressure increases, it forms more complex structures and undergoes a metal-to-semiconductor transformation, complicating theoretical and experimental analyses. We employ the neural canonical transformation approach, an \textit{ab initio} variational method based on probabilistic generative models, to investigate the quantum anharmonic effects in lithium solids at finite temperatures. This approach combines a normalizing flow for phonon excited-state wave functions with a probabilistic model for the occupation of energy levels, optimized jointly to minimize the free energy. Our results indicate that quantum anharmonicity lowers the \textit{bcc}-\textit{fcc} transition temperature compared to classical molecular dynamics predictions. At high pressures, the predicted fractional coordinates of lithium atoms in the \textit{cI16} structure show good quantitative agreement with experimental observations. Finally, contrary to previous beliefs, we find that the poor metallic \textit{oC88} structure is stabilized by the potential energy surface obtained via high-accuracy electronic structure calculations, rather than thermal or quantum nuclear effects. | [
{
"version": "v1",
"created": "Tue, 17 Dec 2024 01:33:59 GMT"
},
{
"version": "v2",
"created": "Thu, 26 Dec 2024 05:26:20 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 08:06:07 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhang",
"Qi",
""
],
[
"Wang",
"Xiaoyang",
""
],
[
"Shi",
"Rong",
""
],
[
"Ren",
"Xinguo",
""
],
[
"Wang",
"Han",
""
],
[
"Wang",
"Lei",
""
]
] |
2412.12472 | Yang Deng | Moxin Li, Yong Zhao, Wenxuan Zhang, Shuaiyi Li, Wenya Xie, See-Kiong Ng, Tat-Seng Chua, Yang Deng | Knowledge Boundary of Large Language Models: A Survey | Accepted to ACL 2025 (main) | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Although large language models (LLMs) store vast amount of knowledge in their parameters, they still have limitations in the memorization and utilization of certain knowledge, leading to undesired behaviors such as generating untruthful and inaccurate responses. This highlights the critical need to understand the knowledge boundary of LLMs, a concept that remains inadequately defined in existing research. In this survey, we propose a comprehensive definition of the LLM knowledge boundary and introduce a formalized taxonomy categorizing knowledge into four distinct types. Using this foundation, we systematically review the field through three key lenses: the motivation for studying LLM knowledge boundaries, methods for identifying these boundaries, and strategies for mitigating the challenges they present. Finally, we discuss open challenges and potential research directions in this area. We aim for this survey to offer the community a comprehensive overview, facilitate access to key issues, and inspire further advancements in LLM knowledge research. | [
{
"version": "v1",
"created": "Tue, 17 Dec 2024 02:14:02 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 03:21:26 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Li",
"Moxin",
""
],
[
"Zhao",
"Yong",
""
],
[
"Zhang",
"Wenxuan",
""
],
[
"Li",
"Shuaiyi",
""
],
[
"Xie",
"Wenya",
""
],
[
"Ng",
"See-Kiong",
""
],
[
"Chua",
"Tat-Seng",
""
],
[
"Deng",
"Yang",
""
]
] |
2412.12633 | Muchen Ju | Muchen Ju, Junjie Ni, Kaixin Wang, Yihan Xiao | Arborescences of Random Covering Graphs | 10 pages,4 figures | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | A rooted arborescence of a directed graph is a spanning tree directed towards a particular vertex. A recent work of Chepuri et al. showed that the arborescences of a covering graph of a directed graph G are closely related to the arborescences of G. In this paper, we study the weighted sum of arborescences of a random covering graph and give a formula for the expected value, resolving a conjecture of Chepuri et al. | [
{
"version": "v1",
"created": "Tue, 17 Dec 2024 07:56:10 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Dec 2024 10:05:11 GMT"
},
{
"version": "v3",
"created": "Mon, 26 May 2025 18:23:40 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Ju",
"Muchen",
""
],
[
"Ni",
"Junjie",
""
],
[
"Wang",
"Kaixin",
""
],
[
"Xiao",
"Yihan",
""
]
] |
2412.12810 | Angel Paredes | Angel Paredes, Jose Guerra-Carmenate, Jose R. Salgueiro, Daniele Tommasini, Humberto Michinel | Traveling Bubbles and Vortex Pairs within Symmetric 2D Quantum Droplets | 12 pages, 13 figures. In v2, title changed, figures added, references added, discussion improved, data availability statement added, typos corrected | Phys. Rev. E 111, 054217 (2025) | 10.1103/PhysRevE.111.054217 | null | nlin.PS cond-mat.quant-gas | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We disclose a class of stable nonlinear traveling waves moving at specific constant velocities within symmetric two-dimensional quantum droplets. We present a comprehensive analysis of these traveling bubbles and identify three qualitatively distinct regions within the one-parameter family of solutions, classified by velocity: (i) well-separated phase singularities at low velocity, (ii) singularities within the same density dip at intermediate velocity, and (iii) rarefaction pulses without singularities at higher (subsonic) velocities. Then, we generalize the discussion to unstable cases, incorporating higher order vortex-antivortex pairs and arrays of vortices that move cohesively with a common velocity within the fluid. In all cases, we provide analytic approximations that aid the understanding of the results in different regimes. | [
{
"version": "v1",
"created": "Tue, 17 Dec 2024 11:26:57 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 12:07:20 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Paredes",
"Angel",
""
],
[
"Guerra-Carmenate",
"Jose",
""
],
[
"Salgueiro",
"Jose R.",
""
],
[
"Tommasini",
"Daniele",
""
],
[
"Michinel",
"Humberto",
""
]
] |
2412.13407 | David Schaich | Raghav G. Jha, Anosh Joseph, David Schaich | Finite-temperature phase diagram of the BMN matrix model on the lattice | null | Phys. Rev. D 111, 094516 (2025) | 10.1103/PhysRevD.111.094516 | null | hep-lat hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the thermal phase structure of the Berenstein--Maldacena--Nastase (BMN) matrix model using non-perturbative lattice Monte Carlo calculations. Our main analyses span three orders of magnitude in the coupling, involving systems with sizes up to $N_{\tau} = 24$ lattice sites and SU($N$) gauge groups with $8 \leq N \leq 16$. In addition, we carry out extended checks of discretization artifacts for $N_{\tau} \leq 128$ and gauge group SU(4). We find results for the deconfinement temperature that interpolate between the perturbative prediction at weak coupling and the large-$N$ dual supergravity calculation at strong coupling. While we confirm that the phase transition is first order for strong coupling, it appears to be continuous for weaker couplings. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 00:46:59 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 15:33:25 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Jha",
"Raghav G.",
""
],
[
"Joseph",
"Anosh",
""
],
[
"Schaich",
"David",
""
]
] |
2412.13523 | Yike Wang | Yike Wang, Yusha Chen | Strictly monotone mean-variance preferences with applications to portfolio selection | 47 pages | null | null | null | q-fin.MF | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper extends the monotone mean-variance (MMV) preference to a broader class of strictly monotone mean-variance (SMMV) preferences, and demonstrates its applications to portfolio selection problems. For the single-period portfolio problem under the SMMV preference, we derive the gradient condition for the optimal strategy, and investigate its association with the optimal mean-variance (MV) static strategy. A novel contribution of this work is the reduction of the problem to solving a set of linear equations by analyzing the saddle point of some minimax problem. Building on this advancement, we conduct numerical experiments and compare our results with those of Maccheroni, et al. (Math. Finance 19(3): 487-521, 2009). The findings indicate that our SMMV preferences provide a more rational basis for assessing given prospects. For the continuous-time portfolio problem with the SMMV preference, we consider continuous price processes with random coefficients. We establish the condition under which the optimal dynamic strategies for SMMV and MV preferences coincide, and characterize the optimal solution using the dynamic programming principle and the martingale convex duality method, respectively. Consequently, the problem is reduced to solving a stochastic Hamilton-Jacobi-Bellman-Isaacs equation, or a multi-stage linear-quadratic optimization problem with the embedding technique. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 05:47:06 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 08:43:14 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Wang",
"Yike",
""
],
[
"Chen",
"Yusha",
""
]
] |
2412.13690 | Xiwen Geng | Xiwen Geng, Suyun Zhao, Yixin Yu, Borui Peng, Pan Du, Hong Chen, Cuiping Li, Mengdie Wang | Personalized Clustering via Targeted Representation Learning | Accepted to AAAI 2025 main conference | Geng, X., Zhao, S., Yu, Y., Peng, B., Du, P., Chen, H., Li, C., & Wang, M. (2025). Personalized Clustering via Targeted Representation Learning. Proceedings of the AAAI Conference on Artificial Intelligence, 39(16), 16790-16798 | 10.1609/aaai.v39i16.33845 | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Clustering traditionally aims to reveal a natural grouping structure within unlabeled data. However, this structure may not always align with users' preferences. In this paper, we propose a personalized clustering method that explicitly performs targeted representation learning by interacting with users via modicum task information (e.g., $\textit{must-link}$ or $\textit{cannot-link}$ pairs) to guide the clustering direction. We query users with the most informative pairs, i.e., those pairs most hard to cluster and those most easy to miscluster, to facilitate the representation learning in terms of the clustering preference. Moreover, by exploiting attention mechanism, the targeted representation is learned and augmented. By leveraging the targeted representation and constrained contrastive loss as well, personalized clustering is obtained. Theoretically, we verify that the risk of personalized clustering is tightly bounded, guaranteeing that active queries to users do mitigate the clustering risk. Experimentally, extensive results show that our method performs well across different clustering tasks and datasets, even when only a limited number of queries are available. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 10:28:51 GMT"
},
{
"version": "v2",
"created": "Fri, 20 Dec 2024 12:08:30 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 09:51:25 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Geng",
"Xiwen",
""
],
[
"Zhao",
"Suyun",
""
],
[
"Yu",
"Yixin",
""
],
[
"Peng",
"Borui",
""
],
[
"Du",
"Pan",
""
],
[
"Chen",
"Hong",
""
],
[
"Li",
"Cuiping",
""
],
[
"Wang",
"Mengdie",
""
]
] |
2412.13877 | Kun Wu | Kun Wu, Chengkai Hou, Jiaming Liu, Zhengping Che, Xiaozhu Ju, Zhuqin Yang, Meng Li, Yinuo Zhao, Zhiyuan Xu, Guang Yang, Shichao Fan, Xinhua Wang, Fei Liao, Zhen Zhao, Guangyu Li, Zhao Jin, Lecheng Wang, Jilei Mao, Ning Liu, Pei Ren, Qiang Zhang, Yaoxu Lyu, Mengzhen Liu, Jingyang He, Yulin Luo, Zeyu Gao, Chenxuan Li, Chenyang Gu, Yankai Fu, Di Wu, Xingyu Wang, Sixiang Chen, Zhenyu Wang, Pengju An, Siyuan Qian, Shanghang Zhang, Jian Tang | RoboMIND: Benchmark on Multi-embodiment Intelligence Normative Data for Robot Manipulation | 21 pages, 17 figures, Robotics: Science and Systems 2025 | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce RoboMIND (Multi-embodiment Intelligence Normative Data for Robot Manipulation), a dataset containing 107k demonstration trajectories across 479 diverse tasks involving 96 object classes. RoboMIND is collected through human teleoperation and encompasses comprehensive robotic-related information, including multi-view observations, proprioceptive robot state information, and linguistic task descriptions. To ensure data consistency and reliability for imitation learning, RoboMIND is built on a unified data collection platform and a standardized protocol, covering four distinct robotic embodiments: the Franka Emika Panda, the UR5e, the AgileX dual-arm robot, and a humanoid robot with dual dexterous hands. Our dataset also includes 5k real-world failure demonstrations, each accompanied by detailed causes, enabling failure reflection and correction during policy learning. Additionally, we created a digital twin environment in the Isaac Sim simulator, replicating the real-world tasks and assets, which facilitates the low-cost collection of additional training data and enables efficient evaluation. To demonstrate the quality and diversity of our dataset, we conducted extensive experiments using various imitation learning methods for single-task settings and state-of-the-art Vision-Language-Action (VLA) models for multi-task scenarios. By leveraging RoboMIND, the VLA models achieved high manipulation success rates and demonstrated strong generalization capabilities. To the best of our knowledge, RoboMIND is the largest multi-embodiment teleoperation dataset collected on a unified platform, providing large-scale and high-quality robotic training data. Our project is at https://x-humanoid-robomind.github.io/. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 14:17:16 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Feb 2025 14:32:16 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 01:46:53 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Wu",
"Kun",
""
],
[
"Hou",
"Chengkai",
""
],
[
"Liu",
"Jiaming",
""
],
[
"Che",
"Zhengping",
""
],
[
"Ju",
"Xiaozhu",
""
],
[
"Yang",
"Zhuqin",
""
],
[
"Li",
"Meng",
""
],
[
"Zhao",
"Yinuo",
""
],
[
"Xu",
"Zhiyuan",
""
],
[
"Yang",
"Guang",
""
],
[
"Fan",
"Shichao",
""
],
[
"Wang",
"Xinhua",
""
],
[
"Liao",
"Fei",
""
],
[
"Zhao",
"Zhen",
""
],
[
"Li",
"Guangyu",
""
],
[
"Jin",
"Zhao",
""
],
[
"Wang",
"Lecheng",
""
],
[
"Mao",
"Jilei",
""
],
[
"Liu",
"Ning",
""
],
[
"Ren",
"Pei",
""
],
[
"Zhang",
"Qiang",
""
],
[
"Lyu",
"Yaoxu",
""
],
[
"Liu",
"Mengzhen",
""
],
[
"He",
"Jingyang",
""
],
[
"Luo",
"Yulin",
""
],
[
"Gao",
"Zeyu",
""
],
[
"Li",
"Chenxuan",
""
],
[
"Gu",
"Chenyang",
""
],
[
"Fu",
"Yankai",
""
],
[
"Wu",
"Di",
""
],
[
"Wang",
"Xingyu",
""
],
[
"Chen",
"Sixiang",
""
],
[
"Wang",
"Zhenyu",
""
],
[
"An",
"Pengju",
""
],
[
"Qian",
"Siyuan",
""
],
[
"Zhang",
"Shanghang",
""
],
[
"Tang",
"Jian",
""
]
] |
2412.13904 | Mateo Galdeano | Xenia de la Ossa, Mateo Galdeano, Enrico Marchetto | $\mathcal{SW}$-algebras and strings with torsion | 46 pages plus appendices, 1 figure, 6 tables; v2: typos fixed, matches published version | JHEP 04 (2025) 157 | 10.1007/JHEP04(2025)157 | DESY-24-197 | hep-th math.QA | http://creativecommons.org/licenses/by/4.0/ | We explore the connection between super $\mathcal{W}$-algebras ($\mathcal{SW}$-algebras) and $\mathrm{G}$-structures with torsion. The former are realised as symmetry algebras of strings with $\mathcal{N}=(1,0)$ supersymmetry on the worldsheet, while the latter are associated with generic string backgrounds with non-trivial Neveu-Schwarz flux $H$. In particular, we focus on manifolds featuring $\mathrm{Spin}(7)$, $\mathrm{G}_2$, $\mathrm{SU}(2)$, and $\mathrm{SU}(3)$-structures. We compare the full quantum algebras with their classical limits, obtained by studying the commutators of superconformal and $\mathcal{W}$-symmetry transformations, which preserve the action of the $(1,0)$ non-linear $\sigma$-model. We show that, at first order in the string length scale $\ell_s$, the torsion deforms some of the OPE coefficients corresponding to special holonomy through a scalar torsion class. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 14:44:41 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 10:10:10 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"de la Ossa",
"Xenia",
""
],
[
"Galdeano",
"Mateo",
""
],
[
"Marchetto",
"Enrico",
""
]
] |
2412.13931 | Sebastian Chenery | Sebastian Chenery and Stephen Theriault | Gyration Stability for Projective Planes | 39 pages, comments welcome. Lightly revised version accepted by Topology and its Applications, with thanks to the anonymous referee | Topology and its Applications, Volume 369, Paper No. 109420 (2025) | 10.1016/j.topol.2025.109420 | null | math.AT math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Gyrations are operations on manifolds that arise in geometric topology, where a manifold $M$ may exhibit distinct gyrations depending on the chosen twisting. For a given $M$, we ask a natural question: do all gyrations of $M$ share the same homotopy type regardless of the twisting? A manifold with this property is said to have gyration stability. Inspired by recent work by Duan, which demonstrated that the quaternionic projective plane is not gyration stable with respect to diffeomorphism, we explore this question for projective planes in general. We obtain a complete description of gyration stability for the complex, quaternionic, and octonionic projective planes up to homotopy. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 15:17:01 GMT"
},
{
"version": "v2",
"created": "Mon, 12 May 2025 13:44:23 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Chenery",
"Sebastian",
""
],
[
"Theriault",
"Stephen",
""
]
] |
2412.14069 | Amalio Fernandez-Pacheco | John Fullerton, Na\"emi Leo, Jakub Jurczyk, Claire Donnelly, D\'edalo Sanz-Hern\'andez, Luka Skoric, Nicolas Mille, Stefan Stanescu, Donald A. MacLaren, Rachid Belkhou, Aurelio Hierro-Rodriguez, Amalio Fern\'andez-Pacheco | Fractional Skyrmion Tubes in Chiral-Interfaced Three-Dimensional Magnetic Nanowires | null | null | 10.1002/adfm.202501615 | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Magnetic skyrmions are chiral spin textures with rich physics and great potential for unconventional computing. Typically, skyrmions form in bulk crystals with reduced symmetry or ultrathin film multilayers involving heavy metals. Here, we demonstrate the formation of fractional Bloch skyrmion tubes at room temperature by 3D printing ferromagnetic double-helix nanowires with two regions of opposite chirality. Using X-ray microscopy and micromagnetic simulations, we show that the coexistence of vortex and anti-parallel spin states induces the formation of fractional skyrmion tubes at zero magnetic fields, minimising the energy cost of breaking the coupling between geometric and magnetic chirality. We also demonstrate control over zero-field states, including pure vortex, or mixed skyrmion-vortex states, highlighting the magnetic reconfigurability of these 3D nanowires. This work shows how interfacing chiral geometries at the nanoscale can enable advanced forms of topological spintronics. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 17:13:21 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Fullerton",
"John",
""
],
[
"Leo",
"Naëmi",
""
],
[
"Jurczyk",
"Jakub",
""
],
[
"Donnelly",
"Claire",
""
],
[
"Sanz-Hernández",
"Dédalo",
""
],
[
"Skoric",
"Luka",
""
],
[
"Mille",
"Nicolas",
""
],
[
"Stanescu",
"Stefan",
""
],
[
"MacLaren",
"Donald A.",
""
],
[
"Belkhou",
"Rachid",
""
],
[
"Hierro-Rodriguez",
"Aurelio",
""
],
[
"Fernández-Pacheco",
"Amalio",
""
]
] |
2412.14125 | Vladimir Rovenski | Vladimir Rovenski | $\eta$-Ricci solitons and $\eta$-Einstein metrics on weak $\beta$-Kenmotsu $f$-manifolds | 14 pages | null | null | null | math.DG | http://creativecommons.org/licenses/by/4.0/ | Recent interest among geometers in $f$-structures of K. Yano is due to the study of topology and dynamics of contact foliations, which generalize the flow of the Reeb vector field on contact manifolds to higher dimensions. Weak metric structures introduced by V. Rovenski and R. Wolak as a generalization of Hermitian and K\"{a}hler structures, as well as $f$-structures, allow a fresh look at the classical theory. In this paper, we study a new $f$-structure of this kind, called the weak $\beta$-Kenmotsu $f$-structure, as a generalization of K. Kenmotsu's concept. We prove that a weak $\beta$-Kenmotsu $f$-manifold is locally a twisted product of the Euclidean space and a weak K\"{a}hler manifold. Our main results show that such manifolds with $\beta=const$ and equipped with an $\eta$-Ricci soliton structure whose potential vector field satisfies certain conditions are $\eta$-Einstein manifolds of constant scalar curvature. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 18:14:32 GMT"
},
{
"version": "v2",
"created": "Mon, 31 Mar 2025 14:31:14 GMT"
},
{
"version": "v3",
"created": "Thu, 01 May 2025 08:59:23 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 17:35:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Rovenski",
"Vladimir",
""
]
] |
2412.14300 | Nima Moshayedi | Alberto S. Cattaneo, Nima Moshayedi, Alberto Smailovic Funcasta | 3D Supergravity in the Batalin-Vilkovisky Formalism | 25 pages. Added references. Improved Introduction | Ann. Henri Poincar\'e (2025) | 10.1007/s00023-025-01588-6 | null | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Three-dimensional supergravity in the Batalin-Vilkovisky formalism is constructed by showing that the theory including the Rarita-Schwinger term is equivalent to an AKSZ theory. | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 20:08:55 GMT"
},
{
"version": "v2",
"created": "Sun, 09 Feb 2025 20:41:39 GMT"
},
{
"version": "v3",
"created": "Thu, 27 Mar 2025 09:58:29 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Cattaneo",
"Alberto S.",
""
],
[
"Moshayedi",
"Nima",
""
],
[
"Funcasta",
"Alberto Smailovic",
""
]
] |
2412.14838 | Wenbin Wang | Xiabin Zhou, Wenbin Wang, Minyan Zeng, Jiaxian Guo, Xuebo Liu, Li Shen, Min Zhang, Liang Ding | DynamicKV: Task-Aware Adaptive KV Cache Compression for Long Context LLMs | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficient KV cache management in LLMs is crucial for long-context tasks like RAG and summarization. Existing KV cache compression methods enforce a fixed pattern, neglecting task-specific characteristics and reducing the retention of essential information. However, we observe distinct activation patterns across layers in various tasks, highlighting the need for adaptive strategies tailored to each task's unique demands. Based on this insight, we propose DynamicKV, a method that dynamically optimizes token retention by adjusting the number of tokens retained at each layer to adapt to the specific task. DynamicKV establishes global and per-layer maximum KV cache budgets, temporarily retaining the maximum budget for the current layer, and periodically updating the KV cache sizes of all preceding layers during inference. Our method retains only 1.7% of the KV cache size while achieving ~85% of the Full KV cache performance on LongBench. Notably, even under extreme compression (0.9%), DynamicKV surpasses state-of-the-art (SOTA) methods by 11% in the Needle-in-a-Haystack test using Mistral-7B-Instruct-v0.2. The code will be released. | [
{
"version": "v1",
"created": "Thu, 19 Dec 2024 13:28:42 GMT"
},
{
"version": "v2",
"created": "Mon, 17 Feb 2025 14:34:58 GMT"
},
{
"version": "v3",
"created": "Mon, 26 May 2025 08:31:13 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 03:08:57 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhou",
"Xiabin",
""
],
[
"Wang",
"Wenbin",
""
],
[
"Zeng",
"Minyan",
""
],
[
"Guo",
"Jiaxian",
""
],
[
"Liu",
"Xuebo",
""
],
[
"Shen",
"Li",
""
],
[
"Zhang",
"Min",
""
],
[
"Ding",
"Liang",
""
]
] |
2412.15088 | Tom Hepworth | Tom R. Hepworth, Daniel K. Ruttley, Fritz von Gierke, Philip D. Gregory, Alexander Guttridge, Simon L. Cornish | Long-lived multilevel coherences and spin-1 dynamics encoded in the rotational states of ultracold molecules | 13 pages, 5 figures | null | null | null | physics.atom-ph cond-mat.quant-gas quant-ph | http://creativecommons.org/licenses/by/4.0/ | Rotational states of ultracold polar molecules possess long radiative lifetimes, microwave-domain coupling, and tunable dipolar interactions. The availability of numerous rotational states has inspired many proposed applications, including simulations of quantum magnetism, encodings of information in high-dimensional qudits, and synthetic dimensions with many synthetic lattice sites. Many of these applications are yet to be realised, primarily because engineering long-lived coherent superpositions of multiple rotational states is highly challenging. Here, we investigate how multilevel coherences between rotational states can be engineered by using optical tweezer traps operating close to a magic wavelength for a given pair of states. By performing precision Ramsey spectroscopy we find the exact magic wavelengths and sensitivities to detuning errors for multiple rotational state superpositions. We find that, for a trap polarised parallel to the quantisation axis, the magic wavelengths are closely clustered enabling long-lived coherence across multiple rotational states simultaneously. As an example, we demonstrate simultaneous second-scale coherence between three rotational states. Utilising this extended coherence, we perform multiparameter estimation using a generalised Ramsey sequence and demonstrate coherent spin-1 dynamics encoded in the rotational states. With modest experimental improvements, we predict that second-scale coherent dynamics of ten rotational states should be readily achievable. | [
{
"version": "v1",
"created": "Thu, 19 Dec 2024 17:35:57 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 15:41:43 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Hepworth",
"Tom R.",
""
],
[
"Ruttley",
"Daniel K.",
""
],
[
"von Gierke",
"Fritz",
""
],
[
"Gregory",
"Philip D.",
""
],
[
"Guttridge",
"Alexander",
""
],
[
"Cornish",
"Simon L.",
""
]
] |
2412.15335 | Ryan Rizaldy | Ryan Rizaldy, Tian Zhou, Sougato Bose, Anupam Mazumdar | Rotational stability in nanorotor and spin contrast in one-loop interferometry in the Stern-Gerlach setup | 29 pages, 12 figures | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rotation of a nanoparticle in a quantum system has many applications, from theory to experiments. This paper will treat nanoparticle rotational dynamics for spin-embedded nanorotors. We will model it as a rigid body that properly treats the rotation in the co-frame of the nanorotor in the presence of external fields. Besides rotation, we will further investigate how to create large spatial superpositions in the inhomogeneous external magnetic field, such as in the case of the Stern-Gerlach apparatus. The spin-embedded nanorotors play a crucial role in creating matter-wave interferometers through their spin and external magnetic field interaction Hamiltonian. We aim to provide a holistic interpretation of the dynamics of three Euler angles, their quantum evolution, and the nanorotor's spatial motion in a Stern-Gerlach-type setup where we will consider one-full-loop interferometry. We will then study how the quantum evolution of all the Euler angles leads to a spin coherence loss upon interference and what manifests the Einstein-de Haas effect in an external magnetic field. In particular, we show that by imparting rotation along the direction of the magnetic field, we can stabilise the nanorotor's libration mode. We will also extend our analysis to a case where the initial state of the libration mode is thermal and discuss the contrast loss due to interference of the nanorotor upon one-loop completion. | [
{
"version": "v1",
"created": "Thu, 19 Dec 2024 19:04:46 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 09:28:08 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Rizaldy",
"Ryan",
""
],
[
"Zhou",
"Tian",
""
],
[
"Bose",
"Sougato",
""
],
[
"Mazumdar",
"Anupam",
""
]
] |
2412.15576 | Xinyang Tong | Xinyang Tong, Pengxiang Ding, Yiguo Fan, Donglin Wang, Wenjie Zhang, Can Cui, Mingyang Sun, Han Zhao, Hongyin Zhang, Yonghao Dang, Siteng Huang, Shangke Lyu | QUART-Online: Latency-Free Large Multimodal Language Model for Quadruped Robot Learning | Accepted to ICRA 2025; Github page: https://quart-online.github.io | null | null | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper addresses the inherent inference latency challenges associated with deploying multimodal large language models (MLLM) in quadruped vision-language-action (QUAR-VLA) tasks. Our investigation reveals that conventional parameter reduction techniques ultimately impair the performance of the language foundation model during the action instruction tuning phase, making them unsuitable for this purpose. We introduce a novel latency-free quadruped MLLM model, dubbed QUART-Online, designed to enhance inference efficiency without degrading the performance of the language foundation model. By incorporating Action Chunk Discretization (ACD), we compress the original action representation space, mapping continuous action values onto a smaller set of discrete representative vectors while preserving critical information. Subsequently, we fine-tune the MLLM to integrate vision, language, and compressed actions into a unified semantic space. Experimental results demonstrate that QUART-Online operates in tandem with the existing MLLM system, achieving real-time inference in sync with the underlying controller frequency, significantly boosting the success rate across various tasks by 65%. Our project page is https://quart-online.github.io. | [
{
"version": "v1",
"created": "Fri, 20 Dec 2024 05:17:06 GMT"
},
{
"version": "v2",
"created": "Mon, 23 Dec 2024 06:06:17 GMT"
},
{
"version": "v3",
"created": "Tue, 11 Mar 2025 14:09:50 GMT"
},
{
"version": "v4",
"created": "Thu, 24 Apr 2025 08:00:36 GMT"
},
{
"version": "v5",
"created": "Tue, 27 May 2025 07:05:44 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Tong",
"Xinyang",
""
],
[
"Ding",
"Pengxiang",
""
],
[
"Fan",
"Yiguo",
""
],
[
"Wang",
"Donglin",
""
],
[
"Zhang",
"Wenjie",
""
],
[
"Cui",
"Can",
""
],
[
"Sun",
"Mingyang",
""
],
[
"Zhao",
"Han",
""
],
[
"Zhang",
"Hongyin",
""
],
[
"Dang",
"Yonghao",
""
],
[
"Huang",
"Siteng",
""
],
[
"Lyu",
"Shangke",
""
]
] |
2412.15949 | Shi Yin | Wei-jie Fu, Jan M. Pawlowski, Robert D. Pisarski, Fabian Rennecke, Rui Wen, Shi Yin | The QCD moat regime and its real-time properties | 20 pages, 16 figures | Phys. Rev. D 111, 094026 (2025) | 10.1103/PhysRevD.111.094026 | null | hep-ph hep-th nucl-th | http://creativecommons.org/licenses/by/4.0/ | Dense QCD matter may exhibit crystalline phases. Their existence is reflected in a moat regime, where mesonic correlations feature spatial modulations. We study the realtime properties of pions at finite temperature and density in QCD in order to elucidate the nature of this regime. We show that the moat regime arises from particle-hole-like fluctuations near the Fermi surface. This gives rise to a characteristic peak in the spectral function of the pion at nonzero \emph{spacelike} momentum. This peak can be interpreted as a new quasi particle, the moaton. In addition, our framework also allows us to directly test the stability of the homogeneous chiral phase against the formation of an inhomogeneous condensate in QCD. We find that the formation of such a phase is highly unlikely for baryon chemical potentials $\mu_B \leq 630$\,MeV. | [
{
"version": "v1",
"created": "Fri, 20 Dec 2024 14:44:24 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Fu",
"Wei-jie",
""
],
[
"Pawlowski",
"Jan M.",
""
],
[
"Pisarski",
"Robert D.",
""
],
[
"Rennecke",
"Fabian",
""
],
[
"Wen",
"Rui",
""
],
[
"Yin",
"Shi",
""
]
] |
2412.16023 | Ricard Ravell Rodr\'iguez | Ricard Ravell Rodr\'iguez and Simon Morelli | Knowledge-dependent optimal Gaussian strategies for phase estimation | We corrected some typos, improved the readability and added some explanations | SciPost Phys. 18, 167 (2025) | 10.21468/SciPostPhys.18.5.167 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | When estimating an unknown phase rotation of a continuous-variable system with homodyne detection, the optimal probe state strongly depends on the value of the estimated parameter. In this article, we identify the optimal pure single-mode Gaussian probe states depending on the knowledge of the estimated phase parameter before the measurement. We find that for a large prior uncertainty, the optimal probe states are close to coherent states, a result in line with findings from noisy parameter estimation. But with increasingly precise estimates of the parameter it becomes beneficial to put more of the available energy into the squeezing of the probe state. Surprisingly, there is a clear jump, where the optimal probe state changes abruptly to a squeezed vacuum state, which maximizes the Fisher information for this estimation task. We use our results to study repeated measurements and compare different methods to adapt the probe state based on the changing knowledge of the parameter according to the previous findings. | [
{
"version": "v1",
"created": "Fri, 20 Dec 2024 16:16:33 GMT"
},
{
"version": "v2",
"created": "Thu, 23 Jan 2025 08:20:42 GMT"
},
{
"version": "v3",
"created": "Tue, 15 Apr 2025 14:15:06 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Rodríguez",
"Ricard Ravell",
""
],
[
"Morelli",
"Simon",
""
]
] |
2412.16216 | Yue Yu | Ting Bai, Yue Yu, Le Huang, Zenan Xu, Zhe Zhao, Chuan Shi | GMoE: Empowering LLMs Fine-Tuning via MoE Graph Collaboration | 9 pages, 25 figures | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The sparse Mixture-of-Experts (MoE) architecture of large language models (LLMs) confronts an inherent issue of load imbalance arising from the simplistic linear router strategy, which ultimately causes the instability and inefficient learning of LLMs. To address this challenge, we introduce a novel MoE graph-based framework $\textbf{GMoE}$, aimed at enhancing the collaboration among multiple experts. In GMoE, a graph router function is designed to capture the collaboration signals among experts. This enables all experts to dynamically allocate information derived from input data by sharing information with their neighboring experts. Moreover, we put forward two coordination strategies in GMoE: the $\textit{Poisson distribution-based distinction strategy}$ and the $\textit{Normal distribution-based balance strategy}$, to further release the capacity of each expert and increase the model stability in the fine-tuning of LLMs. Specifically, we leverage a parameter-efficient fine-tuning technique, i.e., Low-Rank Adaptation (LoRA), to implement the graph MoE architecture. Extensive experiments on four real-world benchmark datasets demonstrate the effectiveness of GMoE, showing the benefits of facilitating collaborations of multiple experts in LLM fine-tuning. The code of experimental implementation is available at https://github.com/BAI-LAB/GMoE | [
{
"version": "v1",
"created": "Wed, 18 Dec 2024 02:18:57 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 16:31:22 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 02:34:28 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Bai",
"Ting",
""
],
[
"Yu",
"Yue",
""
],
[
"Huang",
"Le",
""
],
[
"Xu",
"Zenan",
""
],
[
"Zhao",
"Zhe",
""
],
[
"Shi",
"Chuan",
""
]
] |
2412.17143 | Giovambattista Ianni | Francesco Calimeri, Giovambattista Ianni, Francesco Pacenza, Simona Perri, Jessica Zangari | ASP-based Multi-shot Reasoning via DLV2 with Incremental Grounding | Under consideration in Theory and Practice of Logic Programming (TPLP) | null | 10.1017/S1471068425000067 | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | DLV2 is an AI tool for Knowledge Representation and Reasoning which supports Answer Set Programming (ASP) - a logic-based declarative formalism, successfully used in both academic and industrial applications. Given a logic program modelling a computational problem, an execution of DLV2 produces the so-called answer sets that correspond one-to-one to the solutions to the problem at hand. The computational process of DLV2 relies on the typical Ground & Solve approach where the grounding step transforms the input program into a new, equivalent ground program, and the subsequent solving step applies propositional algorithms to search for the answer sets. Recently, emerging applications in contexts such as stream reasoning and event processing created a demand for multi-shot reasoning: here, the system is expected to be reactive while repeatedly executed over rapidly changing data. In this work, we present a new incremental reasoner obtained from the evolution of DLV2 towards iterated reasoning. Rather than restarting the computation from scratch, the system remains alive across repeated shots, and it incrementally handles the internal grounding process. At each shot, the system reuses previous computations for building and maintaining a large, more general ground program, from which a smaller yet equivalent portion is determined and used for computing answer sets. Notably, the incremental process is performed in a completely transparent fashion for the user. We describe the system, its usage, its applicability and performance in some practically relevant domains. Under consideration in Theory and Practice of Logic Programming (TPLP). | [
{
"version": "v1",
"created": "Sun, 22 Dec 2024 19:46:49 GMT"
},
{
"version": "v2",
"created": "Tue, 24 Dec 2024 13:37:03 GMT"
},
{
"version": "v3",
"created": "Mon, 10 Mar 2025 23:03:10 GMT"
},
{
"version": "v4",
"created": "Tue, 01 Apr 2025 16:43:40 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Calimeri",
"Francesco",
""
],
[
"Ianni",
"Giovambattista",
""
],
[
"Pacenza",
"Francesco",
""
],
[
"Perri",
"Simona",
""
],
[
"Zangari",
"Jessica",
""
]
] |
2412.17564 | Narutaka Ozawa | Narutaka Ozawa | Contractibility of the automorphism group of a von Neumann algebra | 10 pages; typos corrected and a remark added (v2); proofs of lemmas 2 and 4 overhauled (v3) | null | null | null | math.OA | http://creativecommons.org/licenses/by/4.0/ | We prove that the approximately inner automorphism group of a separable strongly stable von Neumann algebra is contractible in the u-topology. Thus the automorphism group of the hyperfinite type III_1 factor is contractible. | [
{
"version": "v1",
"created": "Mon, 23 Dec 2024 13:35:16 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Feb 2025 07:52:27 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 07:24:02 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Ozawa",
"Narutaka",
""
]
] |
2412.17629 | Kaichen Ouyang | Kaichen Ouyang, Shengwei Fu, Zong Ke, Renxiang Guan, Ke Liang, Dayu Hu | Learn from Global Correlations: Enhancing Evolutionary Algorithm via Spectral GNN | 22 pages, 9 figures | null | null | null | cs.NE cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Evolutionary algorithms (EAs) simulate natural selection but have two main limitations: (1) they rarely update individuals based on global correlations, limiting comprehensive learning; (2) they struggle with balancing exploration and exploitation, where excessive exploitation causes premature convergence, and excessive exploration slows down the search. Moreover, EAs often depend on manual parameter settings, which can disrupt the exploration-exploitation balance.To address these issues, we propose Graph Neural Evolution (GNE), a novel EA framework. GNE represents the population as a graph, where nodes represent individuals, and edges capture their relationships, enabling global information usage. GNE utilizes spectral graph neural networks (GNNs) to decompose evolutionary signals into frequency components, applying a filtering function to fuse these components. High-frequency components capture diverse global information, while low-frequency ones capture more consistent information. This explicit frequency filtering strategy directly controls global-scale features through frequency components, overcoming the limitations of manual parameter settings and making the exploration-exploitation control more interpretable and effective.Extensive tests on nine benchmark functions (e.g., Sphere, Rastrigin, Rosenbrock) show that GNE outperforms classical (GA, DE, CMA-ES) and advanced algorithms (SDAES, RL-SHADE) under various conditions, including noise-corrupted and optimal solution deviation scenarios. GNE achieves solutions several orders of magnitude better (e.g., 3.07e-20 mean on Sphere vs. 1.51e-07). | [
{
"version": "v1",
"created": "Mon, 23 Dec 2024 15:06:37 GMT"
},
{
"version": "v2",
"created": "Tue, 24 Dec 2024 13:27:44 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 11:23:38 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Ouyang",
"Kaichen",
""
],
[
"Fu",
"Shengwei",
""
],
[
"Ke",
"Zong",
""
],
[
"Guan",
"Renxiang",
""
],
[
"Liang",
"Ke",
""
],
[
"Hu",
"Dayu",
""
]
] |
2412.17738 | Ioannis Gialamas | Ioannis D. Gialamas, Antonio Racioppi | Symmetry-breaking inflation in non-minimal metric-affine gravity | 19 pages, 10 figures, matches published version | JCAP 05, 072 (2025) | null | null | gr-qc astro-ph.CO hep-ph | http://creativecommons.org/licenses/by/4.0/ | We study symmetry-breaking inflation within the framework of metric-affine gravity. By introducing a non-minimal coupling, $\beta(\phi)\tilde{\cal R}$, between the Holst invariant and the inflaton, both small-field and large-field inflationary predictions can be brought into agreement with the latest observational constraints. Remarkably, even for sub-Planckian vacuum expectation values, appropriately chosen values of $\beta(\phi)$ enable viable inflation, a scenario previously considered unattainable. | [
{
"version": "v1",
"created": "Mon, 23 Dec 2024 17:41:54 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 22:01:24 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Gialamas",
"Ioannis D.",
""
],
[
"Racioppi",
"Antonio",
""
]
] |
2412.17820 | Martin Guillemaud | Martin Guillemaud, Alice Longhena, Louis Cousyn, Valerio Frazzini, Bertrand Mathon, Vincent Navarro, Mario Chavez | Hyperbolic embedding of brain networks can predict the surgery outcome in temporal lobe epilepsy | null | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Epilepsy surgery, particularly for temporal lobe epilepsy (TLE), remains a vital treatment option for patients with drug-resistant seizures. However, accurately predicting surgical outcomes remains a significant challenge. This study introduces a novel biomarker derived from brain connectivity, analyzed using non-Euclidean network geometry, to predict the surgery outcome in TLE. Using structural and diffusion magnetic resonance imaging (MRI) data from 51 patients, we examined differences in structural connectivity networks associated to surgical outcomes. Our approach uniquely utilized hyperbolic embeddings of pre- and post-surgery brain networks, successfully distinguishing patients with favorable outcomes from those with poor outcomes. Notably, the method identified regions in the contralateral hemisphere relative to the epileptogenic zone, whose connectivity patterns emerged as a potential biomarker for favorable surgical outcomes. The prediction model achieves an area under the curve (AUC) of 0.87 and a balanced accuracy of 0.81. These results underscore the predictive capability of our model and its effectiveness in individual outcome forecasting based on structural network changes. Our findings highlight the value of non-Euclidean representation of brain networks in gaining deeper insights into connectivity alterations in epilepsy, and advancing personalized prediction of surgical outcomes in TLE. | [
{
"version": "v1",
"created": "Mon, 09 Dec 2024 11:45:13 GMT"
},
{
"version": "v2",
"created": "Sat, 08 Feb 2025 14:00:23 GMT"
},
{
"version": "v3",
"created": "Tue, 04 Mar 2025 14:39:24 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 16:20:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Guillemaud",
"Martin",
""
],
[
"Longhena",
"Alice",
""
],
[
"Cousyn",
"Louis",
""
],
[
"Frazzini",
"Valerio",
""
],
[
"Mathon",
"Bertrand",
""
],
[
"Navarro",
"Vincent",
""
],
[
"Chavez",
"Mario",
""
]
] |
2412.17920 | Haohong Lin | Haohong Lin, Xin Huang, Tung Phan-Minh, David S. Hayden, Huan Zhang, Ding Zhao, Siddhartha Srinivasa, Eric M. Wolff, Hongge Chen | Causal Composition Diffusion Model for Closed-loop Traffic Generation | null | null | null | null | cs.AI cs.LG cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Simulation is critical for safety evaluation in autonomous driving, particularly in capturing complex interactive behaviors. However, generating realistic and controllable traffic scenarios in long-tail situations remains a significant challenge. Existing generative models suffer from the conflicting objective between user-defined controllability and realism constraints, which is amplified in safety-critical contexts. In this work, we introduce the Causal Compositional Diffusion Model (CCDiff), a structure-guided diffusion framework to address these challenges. We first formulate the learning of controllable and realistic closed-loop simulation as a constrained optimization problem. Then, CCDiff maximizes controllability while adhering to realism by automatically identifying and injecting causal structures directly into the diffusion process, providing structured guidance to enhance both realism and controllability. Through rigorous evaluations on benchmark datasets and in a closed-loop simulator, CCDiff demonstrates substantial gains over state-of-the-art approaches in generating realistic and user-preferred trajectories. Our results show CCDiff's effectiveness in extracting and leveraging causal structures, showing improved closed-loop performance based on key metrics such as collision rate, off-road rate, FDE, and comfort. | [
{
"version": "v1",
"created": "Mon, 23 Dec 2024 19:20:29 GMT"
},
{
"version": "v2",
"created": "Wed, 05 Feb 2025 16:08:12 GMT"
},
{
"version": "v3",
"created": "Mon, 26 May 2025 22:43:40 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Lin",
"Haohong",
""
],
[
"Huang",
"Xin",
""
],
[
"Phan-Minh",
"Tung",
""
],
[
"Hayden",
"David S.",
""
],
[
"Zhang",
"Huan",
""
],
[
"Zhao",
"Ding",
""
],
[
"Srinivasa",
"Siddhartha",
""
],
[
"Wolff",
"Eric M.",
""
],
[
"Chen",
"Hongge",
""
]
] |
2412.18038 | Mirko Zaffaroni | Mirko Zaffaroni, Federico Signoretta, Marco Grangetto, Attilio Fiandrotti | AA-SGAN: Adversarially Augmented Social GAN with Synthetic Data | null | null | 10.1007/978-3-031-91767-7_2 | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Accurately predicting pedestrian trajectories is crucial in applications such as autonomous driving or service robotics, to name a few. Deep generative models achieve top performance in this task, assuming enough labelled trajectories are available for training. To this end, large amounts of synthetically generated, labelled trajectories exist (e.g., generated by video games). However, such trajectories are not meant to represent pedestrian motion realistically and are ineffective at training a predictive model. We propose a method and an architecture to augment synthetic trajectories at training time and with an adversarial approach. We show that trajectory augmentation at training time unleashes significant gains when a state-of-the-art generative model is evaluated over real-world trajectories. | [
{
"version": "v1",
"created": "Mon, 23 Dec 2024 23:17:44 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zaffaroni",
"Mirko",
""
],
[
"Signoretta",
"Federico",
""
],
[
"Grangetto",
"Marco",
""
],
[
"Fiandrotti",
"Attilio",
""
]
] |
2412.18872 | Jie Feng | Pengwei Zhao, Jianqi Yan, Alex P. Leung and Jie Feng | Proton Flux Measurement from Neutron Monitor Data Using Neural Networks | 15 pages, 8 figures | null | null | null | astro-ph.SR astro-ph.HE astro-ph.IM hep-ex physics.space-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate measurements of cosmic ray proton flux are essential for studying the modulation processes of cosmic rays during the solar activity cycle. A proton flux measurement method, based on ground-based neutron monitor (NM) data and deep learning techniques, is presented. After the necessary pre-processing of ground-based NM data using a convolutional neural network (CNN) model, we simulate the relationship between NM observations and proton flux measured by the Alpha Magnetic Spectrometer (AMS). The daily proton flux data, ranging from 1 GV to 100 GV, are obtained for the period from 2011 to 2024, showing strong agreement with the observed values. In addition, daily proton flux measurements are provided for periods when AMS data were unavailable due to operational reasons. For the first time, hourly proton fluxes as a function of rigidity are calculated dedicated to the short-time solar activity studies. | [
{
"version": "v1",
"created": "Wed, 25 Dec 2024 11:14:13 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 10:18:17 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhao",
"Pengwei",
""
],
[
"Yan",
"Jianqi",
""
],
[
"Leung",
"Alex P.",
""
],
[
"Feng",
"Jie",
""
]
] |
2412.19032 | Nobuyoshi Komatsu | Nobuyoshi Komatsu | Cosmological scenario based on the first and second laws of thermodynamics: Thermodynamic constraints on a generalized cosmological model | Final version accepted for publication in EPJC. Several paragraphs and references are revised. [11 pages, 1 figure] | null | null | null | gr-qc astro-ph.CO hep-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The first and second laws of thermodynamics should lead to a consistent scenario for discussing the cosmological constant problem. In the present study, to establish such a thermodynamic scenario, cosmological equations in a flat Friedmann-Lema\^{i}tre-Robertson-Walker universe were derived from the first law, using an arbitrary entropy $S_{H}$ on a cosmological horizon. Then, the cosmological equations were formulated based on a general formulation that includes two extra driving terms, $f_{\Lambda}(t)$ and $h_{\textrm{B}}(t)$, which are usually used for, e.g., time-varying $\Lambda (t)$ cosmology and bulk viscous cosmology, respectively. In addition, thermodynamic constraints on the two terms are examined using the second law of thermodynamics, extending a previous analysis [Phys. Rev. D 99, 043523 (2019) (arXiv:1810.11138)]. It is found that a deviation $S_{\Delta}$ of $S_{H}$ from the Bekenstein-Hawking entropy plays important roles in the two terms. The second law should constrain the upper limits of $f_{\Lambda}(t)$ and $h_{\textrm{B}}(t)$ in our late Universe. The orders of the two terms are likely consistent with the order of the cosmological constant $\Lambda_{\textrm{obs}}$ measured by observations. In particular, when the deviation $S_{\Delta}$ is close to zero, $h_{\textrm{B}}(t)$ and $f_{\Lambda}(t)$ should reduce to zero and a constant value (consistent with the order of $\Lambda_{\textrm{obs}}$), respectively, as if a consistent and viable scenario could be obtained from thermodynamics. | [
{
"version": "v1",
"created": "Thu, 26 Dec 2024 03:04:19 GMT"
},
{
"version": "v2",
"created": "Wed, 05 Mar 2025 06:07:24 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 02:38:35 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Komatsu",
"Nobuyoshi",
""
]
] |
2412.19590 | Hiroki Kuji | Hiroki Kuji, Yuta Shingu, Tetsuro Nikuni, Takashi Imoto, Kenji Sugisaki, Yuichiro Matsuzaki | Robust phase estimation of the ground-state energy without controlled time evolution on a quantum device | null | Phys. Rev. A 111, 042618 (2025) | 10.1103/PhysRevA.111.042618 | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Estimating the ground-state energy of Hamiltonians in quantum systems is an important task. In this work, we demonstrate that the ground-state energy can be accurately estimated without controlled time evolution by using adiabatic state preparation (ASP) and Ramsey-type measurement. By considering the symmetry of the Hamiltonian governing the time evolution during ASP, we can prepare a superposition of the ground state and reference state whose eigenvalue is known. This enables the estimation of the ground-state energy via Ramsey-type measurement. Furthermore, our method is robust against non-adiabatic transitions, making it suitable for use with early fault-tolerant quantum computers and quantum annealing. | [
{
"version": "v1",
"created": "Fri, 27 Dec 2024 11:20:45 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:10:06 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Kuji",
"Hiroki",
""
],
[
"Shingu",
"Yuta",
""
],
[
"Nikuni",
"Tetsuro",
""
],
[
"Imoto",
"Takashi",
""
],
[
"Sugisaki",
"Kenji",
""
],
[
"Matsuzaki",
"Yuichiro",
""
]
] |
2412.19748 | Jianping Yao | Jianping Yao, Zeyu Yang, Zai Yang, Jie Xu, and Tony Q. S. Quek | UAV-Enabled Secure ISAC Against Dual Eavesdropping Threats: Joint Beamforming and Trajectory Design | 8 pages, 6 figures, submitted for possible publication. It overlaps with the former version (arXiv:2412.19748) | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we study an unmanned aerial vehicle (UAV)-enabled secure integrated sensing and communication (ISAC) system, where a UAV serves as an aerial base station (BS) to simultaneously perform communication with a user and detect a target on the ground, while a dual-functional eavesdropper attempts to intercept the signals for both sensing and communication. Facing the dual eavesdropping threats, we aim to enhance the average achievable secrecy rate for the communication user by jointly designing the UAV trajectory together with the transmit information and sensing beamforming, while satisfying the requirements on sensing performance and sensing security, as well as the UAV power and flight constraints. To address the non-convex nature of the optimization problem, we employ the alternating optimization (AO) strategy, jointly with the successive convex approximation (SCA) and semidefinite relaxation (SDR) methods. Numerical results validate the proposed approach, demonstrating its ability to achieve a high secrecy rate while meeting the required sensing and security constraints. | [
{
"version": "v1",
"created": "Fri, 27 Dec 2024 17:15:20 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 06:28:19 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Yao",
"Jianping",
""
],
[
"Yang",
"Zeyu",
""
],
[
"Yang",
"Zai",
""
],
[
"Xu",
"Jie",
""
],
[
"Quek",
"Tony Q. S.",
""
]
] |
2412.20000 | Thomas Bouetou Bouetou | Marius Landry Foka, Michel Bertrand Ngaha Djiadeu, Thomas Bouetou Bouetou | Schouten like metrics on five dimensional nilpotents Lie groups | null | null | null | null | math.DG | http://creativecommons.org/licenses/by/4.0/ | The prescribed Ricci curvature problem involves finding a Riemannian metric g that satisfies the equation ric(g) = T, where T is a fixed symmetric (0, 2)-tensor field on a differential manifold M. In this paper, we introduce the concept of Schouten-like metrics as particular solutions to the prescribed Ricci curvature problem. We classify these metrics on five-dimensional nilpotent Lie groups by establishing a connection with algebraic Schouten solitons. This approach also enables us to classify five-dimensional nilsolitons, providing a comprehensive understanding of their geometric structures and properties. | [
{
"version": "v1",
"created": "Sat, 28 Dec 2024 03:55:54 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 12:39:44 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Foka",
"Marius Landry",
""
],
[
"Djiadeu",
"Michel Bertrand Ngaha",
""
],
[
"Bouetou",
"Thomas Bouetou",
""
]
] |
2412.20367 | Yangfan He | Junqiao Wang, Zeng Zhang, Yangfan He, Zihao Zhang, Yuyang Song, Tianyu Shi, Yuchen Li, Hengyuan Xu, Kunyu Wu, Xin Yi, Zhongwei Wan, Xinhang Yuan, Kuan Lu, Menghao Huo, Guangwu Qian, Keqin Li, Qiuwu Chen, Lewei He | Enhancing Code LLMs with Reinforcement Learning in Code Generation: A Survey | null | null | null | null | cs.SE cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reinforcement learning (RL) has emerged as a powerful paradigm for enhancing large language models (LLMs) in code generation and optimization. This survey systematically reviews RL-driven techniques across the code development lifecycle, from compiler-level optimizations and resource allocation strategies to end-to-end code synthesis frameworks. We first examine classical and modern RL algorithms -- spanning policy gradients, actor-critic methods, human-feedback alignment, and preference-based optimization -- and their adaptations to the unique challenges of code generation, such as sparse and delayed rewards. Next, we analyze key benchmarks, datasets, and evaluation metrics that drive progress in RL-augmented Code LLMs. Finally, we identify open problems, including the need for richer feedback sources, support for low-level and domain-specific languages, and methods to reduce computational overhead. By consolidating current insights and outlining future directions, this work aims to guide researchers and practitioners in leveraging RL to produce more robust, efficient, and human-aligned code generation systems. | [
{
"version": "v1",
"created": "Sun, 29 Dec 2024 06:15:41 GMT"
},
{
"version": "v2",
"created": "Thu, 02 Jan 2025 09:43:43 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 05:41:19 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Wang",
"Junqiao",
""
],
[
"Zhang",
"Zeng",
""
],
[
"He",
"Yangfan",
""
],
[
"Zhang",
"Zihao",
""
],
[
"Song",
"Yuyang",
""
],
[
"Shi",
"Tianyu",
""
],
[
"Li",
"Yuchen",
""
],
[
"Xu",
"Hengyuan",
""
],
[
"Wu",
"Kunyu",
""
],
[
"Yi",
"Xin",
""
],
[
"Wan",
"Zhongwei",
""
],
[
"Yuan",
"Xinhang",
""
],
[
"Lu",
"Kuan",
""
],
[
"Huo",
"Menghao",
""
],
[
"Qian",
"Guangwu",
""
],
[
"Li",
"Keqin",
""
],
[
"Chen",
"Qiuwu",
""
],
[
"He",
"Lewei",
""
]
] |
2412.20422 | Ohad Rahamim | Ohad Rahamim, Ori Malca, Dvir Samuel, Gal Chechik | Bringing Objects to Life: training-free 4D generation from 3D objects through view consistent noise | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent advancements in generative models have enabled the creation of dynamic 4D content - 3D objects in motion - based on text prompts, which holds potential for applications in virtual worlds, media, and gaming. Existing methods provide control over the appearance of generated content, including the ability to animate 3D objects. However, their ability to generate dynamics is limited to the mesh datasets they were trained on, lacking any growth or structural development capability. In this work, we introduce a training-free method for animating 3D objects by conditioning on textual prompts to guide 4D generation, enabling custom general scenes while maintaining the original object's identity. We first convert a 3D mesh into a static 4D Neural Radiance Field (NeRF) that preserves the object's visual attributes. Then, we animate the object using an Image-to-Video diffusion model driven by text. To improve motion realism, we introduce a view-consistent noising protocol that aligns object perspectives with the noising process to promote lifelike movement, and a masked Score Distillation Sampling (SDS) loss that leverages attention maps to focus optimization on relevant regions, better preserving the original object. We evaluate our model on two different 3D object datasets for temporal coherence, prompt adherence, and visual fidelity, and find that our method outperforms the baseline based on multiview training, achieving better consistency with the textual prompt in hard scenarios. | [
{
"version": "v1",
"created": "Sun, 29 Dec 2024 10:12:01 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 17:40:59 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Rahamim",
"Ohad",
""
],
[
"Malca",
"Ori",
""
],
[
"Samuel",
"Dvir",
""
],
[
"Chechik",
"Gal",
""
]
] |
2412.20761 | Jie Jing | Jie Jing, Qing Lin, Shuangpeng Han, Lucia Schiatti, Yen-Ling Kuo, Mengmi Zhang | Unforgettable Lessons from Forgettable Images: Intra-Class Memorability Matters in Computer Vision | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce intra-class memorability, where certain images within the same class are more memorable than others despite shared category characteristics. To investigate what features make one object instance more memorable than others, we design and conduct human behavior experiments, where participants are shown a series of images, and they must identify when the current image matches the image presented a few steps back in the sequence. To quantify memorability, we propose the Intra-Class Memorability score (ICMscore), a novel metric that incorporates the temporal intervals between repeated image presentations into its calculation. Furthermore, we curate the Intra-Class Memorability Dataset (ICMD), comprising over 5,000 images across ten object classes with their ICMscores derived from 2,000 participants' responses. Subsequently, we demonstrate the usefulness of ICMD by training AI models on this dataset for various downstream tasks: memorability prediction, image recognition, continual learning, and memorability-controlled image editing. Surprisingly, high-ICMscore images impair AI performance in image recognition and continual learning tasks, while low-ICMscore images improve outcomes in these tasks. Additionally, we fine-tune a state-of-the-art image diffusion model on ICMD image pairs with and without masked semantic objects. The diffusion model can successfully manipulate image elements to enhance or reduce memorability. Our contributions open new pathways in understanding intra-class memorability by scrutinizing fine-grained visual features behind the most and least memorable images and laying the groundwork for real-world applications in computer vision. We will release all code, data, and models publicly. | [
{
"version": "v1",
"created": "Mon, 30 Dec 2024 07:09:28 GMT"
},
{
"version": "v2",
"created": "Mon, 03 Feb 2025 04:31:57 GMT"
},
{
"version": "v3",
"created": "Tue, 20 May 2025 06:01:29 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 09:03:03 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Jing",
"Jie",
""
],
[
"Lin",
"Qing",
""
],
[
"Han",
"Shuangpeng",
""
],
[
"Schiatti",
"Lucia",
""
],
[
"Kuo",
"Yen-Ling",
""
],
[
"Zhang",
"Mengmi",
""
]
] |
2412.20982 | Fengxing Zhu | Fengxing Zhu | Bootstrap percolation on a generalized Hamming cube | null | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | We consider the $r$-neighbor bootstrap percolation process on the graph with vertex set $V=\{0,1\}^n$ and edges connecting the pairs at Hamming distance $1,2,\dots,k$, where $k\ge 2$. We find asymptotics of the critical probability of percolation for $r=2,3$. In the deterministic setting, we obtain several results for the size of the smallest percolating set for $k\ge 2$. | [
{
"version": "v1",
"created": "Mon, 30 Dec 2024 14:44:54 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 15:32:51 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 15:23:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhu",
"Fengxing",
""
]
] |
2412.20993 | Yichao Fu | Yichao Fu, Junda Chen, Siqi Zhu, Zheyu Fu, Zhongdongming Dai, Yonghao Zhuang, Yian Ma, Aurick Qiao, Tajana Rosing, Ion Stoica, Hao Zhang | Efficiently Scaling LLM Reasoning with Certaindex | null | null | null | null | cs.LG cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Test-time reasoning algorithms such as chain-of-thought, self-consistency, and MCTS enhance LLM problem-solving but can wastefully generate many tokens without improving accuracy. At the same time, we observe that these algorithms exhibit answer stabilization: their intermediate solutions often cease to change after a certain point, and further investment of compute does not change their final answer. To quantify this phenomenon, we introduce Certaindex, an algorithm-agnostic metric measuring this evolving stability, signaling when further computation is unlikely to alter the final result. Certaindex is lightweight, can accelerate reasoning program inference via early exit, and further enables dynamic token allocation, gang scheduling, and many opportunities when integrated with real-world LLM serving systems. To quantify real-world benefits, we built Certaindex as a scheduler into Dynasor, our reasoning-aware LLM serving system, and demonstrate up to 50% compute savings and 3.3x higher throughput in real workloads with no accuracy drop. Our code is available at https://github.com/hao-ai-lab/Dynasor.git | [
{
"version": "v1",
"created": "Mon, 30 Dec 2024 14:57:53 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:41:53 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Fu",
"Yichao",
""
],
[
"Chen",
"Junda",
""
],
[
"Zhu",
"Siqi",
""
],
[
"Fu",
"Zheyu",
""
],
[
"Dai",
"Zhongdongming",
""
],
[
"Zhuang",
"Yonghao",
""
],
[
"Ma",
"Yian",
""
],
[
"Qiao",
"Aurick",
""
],
[
"Rosing",
"Tajana",
""
],
[
"Stoica",
"Ion",
""
],
[
"Zhang",
"Hao",
""
]
] |
2501.00649 | Andrzej Derdzinski | Andrzej Derdzinski, Yunhee Euh, Sinhwi Kim, JeongHyeong Park | On weakly Einstein K\"ahler surfaces | Many minor corrections and additions | null | null | null | math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Riemannian four-manifolds in which the triple contraction of the curvature tensor against itself yields a functional multiple of the metric are called weakly Einstein. We focus on weakly Einstein K\"ahler surfaces. We provide several conditions characterizing those K\"ahler surfaces which are weakly Einstein, classify weakly Einstein K\"ahler surfaces having some specific additional properties, and construct new examples. | [
{
"version": "v1",
"created": "Tue, 31 Dec 2024 21:10:29 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:28:27 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Derdzinski",
"Andrzej",
""
],
[
"Euh",
"Yunhee",
""
],
[
"Kim",
"Sinhwi",
""
],
[
"Park",
"JeongHyeong",
""
]
] |
2501.00701 | Yuanchao Xu | Yuanchao Xu, Kaidi Shao, Nikos Logothetis and Zhongwei Shen | ResKoopNet: Learning Koopman Representations for Complex Dynamics with Spectral Residuals | null | null | null | null | cs.LG math.DS | http://creativecommons.org/licenses/by/4.0/ | Analyzing the long-term behavior of high-dimensional nonlinear dynamical systems remains a significant challenge. While the Koopman operator framework provides a powerful global linearization tool, current methods for approximating its spectral components often face theoretical limitations and depend on predefined dictionaries. Residual Dynamic Mode Decomposition (ResDMD) advanced the field by introducing the \emph{spectral residual} to assess Koopman operator approximation accuracy; however, its approach of only filtering precomputed spectra prevents the discovery of the operator's complete spectral information, a limitation known as the `spectral inclusion' problem. We introduce ResKoopNet (Residual-based Koopman-learning Network), a novel method that directly addresses this by explicitly minimizing the \emph{spectral residual} to compute Koopman eigenpairs. This enables the identification of a more precise and complete Koopman operator spectrum. Using neural networks, our approach provides theoretical guarantees while maintaining computational adaptability. Experiments on a variety of physical and biological systems show that ResKoopNet achieves more accurate spectral approximations than existing methods, particularly for high-dimensional systems and those with continuous spectra, which demonstrates its effectiveness as a tool for analyzing complex dynamical systems. | [
{
"version": "v1",
"created": "Wed, 01 Jan 2025 02:19:42 GMT"
},
{
"version": "v2",
"created": "Fri, 31 Jan 2025 16:40:16 GMT"
},
{
"version": "v3",
"created": "Tue, 01 Apr 2025 19:41:00 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 14:24:52 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Xu",
"Yuanchao",
""
],
[
"Shao",
"Kaidi",
""
],
[
"Logothetis",
"Nikos",
""
],
[
"Shen",
"Zhongwei",
""
]
] |
2501.01264 | Xiaoshuai Song | Xiaoshuai Song, Yanan Wu, Weixun Wang, Jiaheng Liu, Wenbo Su, Bo Zheng | ProgCo: Program Helps Self-Correction of Large Language Models | Accpeted at ACL2025 Main | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-Correction aims to enable large language models (LLMs) to self-verify and self-refine their initial responses without external feedback. However, LLMs often fail to effectively self-verify and generate correct feedback, further misleading refinement and leading to the failure of self-correction, especially in complex reasoning tasks. In this paper, we propose Program-driven Self-Correction (ProgCo). First, program-driven verification (ProgVe) achieves complex verification logic and extensive validation through self-generated, self-executing verification pseudo-programs. Then, program-driven refinement (ProgRe) receives feedback from ProgVe, conducts dual reflection and refinement on both responses and verification programs to mitigate misleading of incorrect feedback in complex reasoning tasks. Experiments on three instruction-following and mathematical benchmarks indicate that ProgCo achieves effective self-correction, and can be further enhance performance when combined with real program tools. We release our code at https://github.com/songxiaoshuai/progco. | [
{
"version": "v1",
"created": "Thu, 02 Jan 2025 13:59:20 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 05:04:23 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Song",
"Xiaoshuai",
""
],
[
"Wu",
"Yanan",
""
],
[
"Wang",
"Weixun",
""
],
[
"Liu",
"Jiaheng",
""
],
[
"Su",
"Wenbo",
""
],
[
"Zheng",
"Bo",
""
]
] |
2501.01473 | Hadi Askari | Hadi Askari, Shivanshu Gupta, Terry Tong, Fei Wang, Anshuman Chhabra, Muhao Chen | Unraveling Indirect In-Context Learning Using Influence Functions | Under Review | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | In this work, we introduce a novel paradigm for generalized In-Context Learning (ICL), termed Indirect In-Context Learning. In Indirect ICL, we explore demonstration selection strategies tailored for two distinct real-world scenarios: Mixture of Tasks and Noisy ICL. We systematically evaluate the effectiveness of Influence Functions (IFs) as a selection tool for these settings, highlighting the potential of IFs to better capture the informativeness of examples within the demonstration pool. For the Mixture of Tasks setting, demonstrations are drawn from 28 diverse tasks, including MMLU, BigBench, StrategyQA, and CommonsenseQA. We demonstrate that combining BertScore-Recall (BSR) with an IF surrogate model can further improve performance, leading to average absolute accuracy gains of 0.37\% and 1.45\% for 3-shot and 5-shot setups when compared to traditional ICL metrics. In the Noisy ICL setting, we examine scenarios where demonstrations might be mislabeled or have adversarial noise. Our experiments show that reweighting traditional ICL selectors (BSR and Cosine Similarity) with IF-based selectors boosts accuracy by an average of 2.90\% for Cosine Similarity and 2.94\% for BSR on noisy GLUE benchmarks. For the adversarial sub-setting, we show the utility of using IFs for task-agnostic demonstration selection for backdoor attack mitigation. Showing a 32.89\% reduction in Attack Success Rate compared to task-aware methods. In sum, we propose a robust framework for demonstration selection that generalizes beyond traditional ICL, offering valuable insights into the role of IFs for Indirect ICL. | [
{
"version": "v1",
"created": "Wed, 01 Jan 2025 12:37:12 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:17:49 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Askari",
"Hadi",
""
],
[
"Gupta",
"Shivanshu",
""
],
[
"Tong",
"Terry",
""
],
[
"Wang",
"Fei",
""
],
[
"Chhabra",
"Anshuman",
""
],
[
"Chen",
"Muhao",
""
]
] |
2501.02052 | Cristina Mondino | Lorenzo Mirasola, Cristina Mondino, Francesco Amicucci, Nils Siemonsen, Cristiano Palomba, Sabrina D'Antonio, Paola Leaci, Luca D'Onofrio, Pia Astone, Daniel Egana-Ugrinovic, Junwu Huang, Masha Baryakhtar, William E. East | Search for continuous gravitational wave signals from luminous dark photon superradiance clouds with LVK O3 observations | 26 pages, 8 figures; v2: published version, minor comments added | Phys. Rev. D 111 (2025), 084032 | 10.1103/PhysRevD.111.084032 | null | gr-qc astro-ph.HE hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Superradiance clouds of kinetically-mixed dark photons around spinning black holes can produce observable multi-messenger electromagnetic and gravitational wave signals. The cloud generates electric fields of up to a Teravolt-per-meter, which lead to a cascade production of charged particles, yielding a turbulent quasi-equilibrium plasma around the black hole, and resulting in electromagnetic fluxes ranging from supernova to pulsar-like luminosities. For stellar mass black holes, such systems resemble millisecond pulsars and are expected to emit pulsating radio waves and continuous gravitational waves (CWs) within the LIGO-Virgo-KAGRA (LVK) sensitivity band. We select 44 sources with approximately coincident frequencies or positive frequency drifts from existing pulsar catalogs as potential candidates of long-lasting superradiance clouds around old galactic black holes. For a subset of 34 sources that are well measured and have not been previously targeted, we perform the first search for CW emission in LVK data from the third observing run. We find no evidence of a CW signal and place 95% confidence level upper limits on the emitted strain amplitude. We interpret these results, together with limits from previous searches, in terms of the underlying dark photon theory by performing an analysis of the expected signals from superradiance clouds from galactic black holes. We find that, even for moderately spinning black holes, the absence of an observed CW signal disfavors a discrete set of dark photon masses between about $10^{-13}$ $\rm{eV}/c^2$ and $10^{-12}$ $\rm{eV}/c^2$ and kinetic mixing couplings in the range of $10^{-9}$-$10^{-7}$, subject to assumptions about the properties of the black hole population and the cloud's electromagnetic emission. | [
{
"version": "v1",
"created": "Fri, 03 Jan 2025 19:00:01 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 14:38:53 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Mirasola",
"Lorenzo",
""
],
[
"Mondino",
"Cristina",
""
],
[
"Amicucci",
"Francesco",
""
],
[
"Siemonsen",
"Nils",
""
],
[
"Palomba",
"Cristiano",
""
],
[
"D'Antonio",
"Sabrina",
""
],
[
"Leaci",
"Paola",
""
],
[
"D'Onofrio",
"Luca",
""
],
[
"Astone",
"Pia",
""
],
[
"Egana-Ugrinovic",
"Daniel",
""
],
[
"Huang",
"Junwu",
""
],
[
"Baryakhtar",
"Masha",
""
],
[
"East",
"William E.",
""
]
] |
2501.02825 | Kavi Gupta | Kavi Gupta, Kate Sanders, Armando Solar-Lezama | Randomly Sampled Language Reasoning Problems Explain Limits of LLMs | 10 pages, 4 figures, 2 tables | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | While LLMs have revolutionized the field of machine learning due to their high performance across a range of tasks, they are known to perform poorly in planning, hallucinate false answers, have degraded performance on less canonical versions of the same task, and answer incorrectly on a variety of specific prompts. There are several emerging theories of LLM performance with some predictive power, among them that LLMs lack world modeling ability, that they have an undesirable bias towards an autoregressive prior, and that they perform less well on more novel problems. The existing literature on novelty has focused on tasks of relatively high complexity, studying perturbations of canonical but complex problems. In this paper, we attempt to isolate novelty as a factor in LLM underperformance. To this end, we consider an extremely simple domain: next token prediction on simple language tasks. The twist is that these language tasks are unseen, as they are randomly drawn from a large, parsimoniously defined set of languages arising from simple grammar rules. This allows us to isolate the effect of task novelty and see if it is sufficient to explain low performance. We find that LLMs uniformly underperform n-gram models (which do not have the capacity for world modeling) on these tasks, both when used as next token predictors and as reasoners. | [
{
"version": "v1",
"created": "Mon, 06 Jan 2025 07:57:51 GMT"
},
{
"version": "v2",
"created": "Tue, 07 Jan 2025 21:51:30 GMT"
},
{
"version": "v3",
"created": "Fri, 24 Jan 2025 23:02:29 GMT"
},
{
"version": "v4",
"created": "Mon, 03 Mar 2025 20:16:13 GMT"
},
{
"version": "v5",
"created": "Tue, 27 May 2025 00:11:34 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Gupta",
"Kavi",
""
],
[
"Sanders",
"Kate",
""
],
[
"Solar-Lezama",
"Armando",
""
]
] |
2501.03484 | Stoffel Janssens D. | Stoffel D. Janssens, Francisco S. Forte Neto, David V\'azquez-Cort\'es, Fernando P. Duda, and Eliot Fried | Probing growth precursor diffusion lengths by inter-surface diffusion | null | null | 10.1016/j.cpc.2025.109671 | null | cond-mat.mtrl-sci physics.chem-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Understanding and optimizing thin-film synthesis requires measuring the diffusion length $d_\alpha$ of adsorbed growth precursors. Despite technological advances, in-situ measurements of $d_\alpha$ are often unachievable due to harsh deposition conditions, such as high temperatures or reactive environments. In this paper, we propose a fitting approach to determine $d_\alpha$ from experimental data by leveraging inter-surface diffusion between a substrate and a strip obtained by, for example, processing a film. The substrate serves as a source or sink of precursors, influencing the growth dynamics and shaping the profile of the strip. By fitting simulated profiles to given profiles, we demonstrate that $d_\alpha$ can be determined. To achieve this, we develop a theoretical growth model, a simulation strategy, and a fitting procedure. The growth model incorporates inter-surface diffusion, adsorption, and desorption of growth precursors, with growth being proportional to the concentration of adsorbed precursors. In our simulations, a chain of nodes represents a profile, and growth is captured by the displacement of those nodes, while keeping the node density approximately constant. For strips significantly wider than $d_\alpha$, a scaled precursor concentration and $d_\alpha$ are the fitting parameters that are determined by minimizing a suitably defined measure of the distance between simulated and given profiles. We evaluate the robustness of our procedure by analyzing the effect of profile resolution and noise on the fitted parameters. Our approach can offer valuable insights into thin-film growth processes, such as those occurring during plasma-enhanced chemical vapor deposition. | [
{
"version": "v1",
"created": "Tue, 07 Jan 2025 03:07:08 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Janssens",
"Stoffel D.",
""
],
[
"Neto",
"Francisco S. Forte",
""
],
[
"Vázquez-Cortés",
"David",
""
],
[
"Duda",
"Fernando P.",
""
],
[
"Fried",
"Eliot",
""
]
] |
2501.04070 | Xiaoqing Zhang | Xiaoqing Zhang, Ang Lv, Yuhan Liu, Flood Sung, Wei Liu, Jian Luan, Shuo Shang, Xiuying Chen, Rui Yan | More is not always better? Enhancing Many-Shot In-Context Learning with Differentiated and Reweighting Objectives | 14 pages, 8 figures, 11 tables | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Large language models (LLMs) excel at few-shot in-context learning (ICL) without requiring parameter updates. However, as ICL demonstrations increase from a few to many, performance tends to plateau and eventually decline. We identify two primary causes for this trend: the suboptimal negative log-likelihood (NLL) optimization objective and the incremental data noise. To address these issues, we introduce \textit{DrICL}, a novel optimization method that enhances model performance through \textit{Differentiated} and \textit{Reweighting} objectives. Globally, DrICL utilizes differentiated learning to optimize the NLL objective, ensuring that many-shot performance surpasses zero-shot levels. Locally, it dynamically adjusts the weighting of many-shot demonstrations by leveraging cumulative advantages inspired by reinforcement learning, thereby mitigating the impact of noisy data. Recognizing the lack of multi-task datasets with diverse many-shot distributions, we develop the \textit{Many-Shot ICL Benchmark} (ICL-50)-a large-scale benchmark of 50 tasks that cover shot numbers from 1 to 350 within sequences of up to 8,000 tokens-for both fine-tuning and evaluation purposes. Experimental results demonstrate that LLMs enhanced with DrICL achieve significant improvements in many-shot setups across various tasks, including both in-domain and out-of-domain scenarios. We release the code and dataset hoping to facilitate further research in many-shot ICL\footnote{https://github.com/xiaoqzhwhu/DrICL}. | [
{
"version": "v1",
"created": "Tue, 07 Jan 2025 14:57:08 GMT"
},
{
"version": "v2",
"created": "Thu, 09 Jan 2025 02:20:13 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 09:48:24 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhang",
"Xiaoqing",
""
],
[
"Lv",
"Ang",
""
],
[
"Liu",
"Yuhan",
""
],
[
"Sung",
"Flood",
""
],
[
"Liu",
"Wei",
""
],
[
"Luan",
"Jian",
""
],
[
"Shang",
"Shuo",
""
],
[
"Chen",
"Xiuying",
""
],
[
"Yan",
"Rui",
""
]
] |
2501.04962 | Wenqian Cui Mr. | Wenqian Cui and Xiaoqi Jiao and Ziqiao Meng and Irwin King | VoxEval: Benchmarking the Knowledge Understanding Capabilities of End-to-End Spoken Language Models | The Version of Record of this contribution is accepted to ACL 2025 main conference | null | null | null | cs.CL cs.SD eess.AS | http://creativecommons.org/licenses/by/4.0/ | With the rising need for speech-based interaction models, end-to-end Spoken Language Models (SLMs) have emerged as a promising solution. While these models require comprehensive world knowledge for meaningful and reliable human interactions, existing question-answering (QA) benchmarks fall short in evaluating SLMs' knowledge understanding due to their inability to support end-to-end speech evaluation and account for varied input audio conditions. To address these limitations, we present VoxEval, a novel SpeechQA benchmark that assesses SLMs' knowledge understanding through pure speech interactions. Our benchmark 1) uniquely maintains speech format for both inputs and outputs, 2) evaluates model robustness across diverse input audio conditions, and 3) pioneers the assessment of complex tasks like mathematical reasoning in spoken format. Systematic evaluation demonstrates that VoxEval presents significant challenges to current SLMs, revealing their sensitivity to varying audio conditions and highlighting the need to enhance reasoning capabilities in future development. We hope this benchmark could guide the advancement of more sophisticated and reliable SLMs. VoxEval dataset is available at: https://github.com/dreamtheater123/VoxEval | [
{
"version": "v1",
"created": "Thu, 09 Jan 2025 04:30:12 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Jan 2025 10:38:05 GMT"
},
{
"version": "v3",
"created": "Tue, 18 Feb 2025 07:07:24 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 16:14:30 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Cui",
"Wenqian",
""
],
[
"Jiao",
"Xiaoqi",
""
],
[
"Meng",
"Ziqiao",
""
],
[
"King",
"Irwin",
""
]
] |
2501.05188 | Guixiang Xu | Guixiang Xu, Pengxuan Yang, and Zhuohui You | Global well-posedness of the defocusing nonlinear wave equation outside of a ball with radial data for $3<p<5$ | 15 pages, 0 figure. Any comment is welcome. arXiv admin note: text overlap with arXiv:2406.05614 | null | null | null | math.AP | http://creativecommons.org/licenses/by/4.0/ | We continue the study of the Dirichlet boundary value problem of nonlinear wave equation with radial data in the exterior $\Omega = \mathbb{R}^3\backslash \bar{B}(0,1)$. We combine the distorted Fourier truncation method in \cite{Bourgain98:FTM}, the global-in-time (endpoint) Strichartz estimates in \cite{XuYang:NLW} with the energy method in \cite{GallPlan03:NLW} to prove the global well-posedness of the radial solution to the defocusing, energy-subcriticial nonlinear wave equation outside of a ball in $\left(\dot H^{s}_{D}(\Omega) \cap L^{p+1}(\Omega) \right)\times \dot H^{s-1}_{D}(\Omega)$ with $1-\frac{(p+3)(1-s_c)}{4(2p-3)}<s<1$, $s_c=\frac{3}{2}-\frac{2}{p-1} $, which extends the result for the cubic nonlinearity in \cite{XuYang:NLW} to the case $3<p<5$. Except from the argument in \cite{XuYang:NLW}, another new ingredient is that we need make use of the radial Sobolev inequality to deal with the super-conformal nonlinearity in addition to the Sobolev inequality. | [
{
"version": "v1",
"created": "Thu, 09 Jan 2025 12:22:40 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 00:28:41 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Xu",
"Guixiang",
""
],
[
"Yang",
"Pengxuan",
""
],
[
"You",
"Zhuohui",
""
]
] |
2501.05359 | Junha Park | Junha Park, Jaehui Hwang, Ian Ryu, Hyungkeun Park, Jiyoon Kim, Jong-Seok Lee | SC-Pro: Training-Free Framework for Defending Unsafe Image Synthesis Attack | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With advances in diffusion models, image generation has shown significant performance improvements. This raises concerns about the potential abuse of image generation, such as the creation of explicit or violent images, commonly referred to as Not Safe For Work (NSFW) content. To address this, the Stable Diffusion model includes several safety checkers to censor initial text prompts and final output images generated from the model. However, recent research has shown that these safety checkers have vulnerabilities against adversarial attacks, allowing them to generate NSFW images. In this paper, we find that these adversarial attacks are not robust to small changes in text prompts or input latents. Based on this, we propose SC-Pro (Spherical or Circular Probing), a training-free framework that easily defends against adversarial attacks generating NSFW images. Moreover, we develop an approach that utilizes one-step diffusion models for efficient NSFW detection (SC-Pro-o), further reducing computational resources. We demonstrate the superiority of our method in terms of performance and applicability. | [
{
"version": "v1",
"created": "Thu, 09 Jan 2025 16:43:21 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 06:10:33 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Park",
"Junha",
""
],
[
"Hwang",
"Jaehui",
""
],
[
"Ryu",
"Ian",
""
],
[
"Park",
"Hyungkeun",
""
],
[
"Kim",
"Jiyoon",
""
],
[
"Lee",
"Jong-Seok",
""
]
] |
2501.06158 | Seul Lee | Seul Lee, Karsten Kreis, Srimukh Prasad Veccham, Meng Liu, Danny Reidenbach, Yuxing Peng, Saee Paliwal, Weili Nie, Arash Vahdat | GenMol: A Drug Discovery Generalist with Discrete Diffusion | ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Drug discovery is a complex process that involves multiple stages and tasks. However, existing molecular generative models can only tackle some of these tasks. We present Generalist Molecular generative model (GenMol), a versatile framework that uses only a single discrete diffusion model to handle diverse drug discovery scenarios. GenMol generates Sequential Attachment-based Fragment Embedding (SAFE) sequences through non-autoregressive bidirectional parallel decoding, thereby allowing the utilization of a molecular context that does not rely on the specific token ordering while having better sampling efficiency. GenMol uses fragments as basic building blocks for molecules and introduces fragment remasking, a strategy that optimizes molecules by regenerating masked fragments, enabling effective exploration of chemical space. We further propose molecular context guidance (MCG), a guidance method tailored for masked discrete diffusion of GenMol. GenMol significantly outperforms the previous GPT-based model in de novo generation and fragment-constrained generation, and achieves state-of-the-art performance in goal-directed hit generation and lead optimization. These results demonstrate that GenMol can tackle a wide range of drug discovery tasks, providing a unified and versatile approach for molecular design. | [
{
"version": "v1",
"created": "Fri, 10 Jan 2025 18:30:05 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 23:51:15 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Lee",
"Seul",
""
],
[
"Kreis",
"Karsten",
""
],
[
"Veccham",
"Srimukh Prasad",
""
],
[
"Liu",
"Meng",
""
],
[
"Reidenbach",
"Danny",
""
],
[
"Peng",
"Yuxing",
""
],
[
"Paliwal",
"Saee",
""
],
[
"Nie",
"Weili",
""
],
[
"Vahdat",
"Arash",
""
]
] |
2501.06234 | Gernot Heiser | Gernot Heiser, Ivan Velickovic, Peter Chubb, Alwin Joshy, Anuraag Ganesh, Bill Nguyen, Cheng Li, Courtney Darville, Guangtao Zhu, James Archer, Jingyao Zhou, Krishnan Winter, Lucy Parker, Szymon Duchniewicz, Tianyi Bai | Fast, Secure, Adaptable: LionsOS Design, Implementation and Performance | 14 pages, 13 figures | null | null | null | cs.OS cs.CR | http://creativecommons.org/licenses/by-sa/4.0/ | We present LionsOS, an operating system for security- and safety-critical embedded systems. LionsOS is based on the formally verified seL4 microkernel and designed with verification in mind. It uses a static architecture and features a highly modular design driven by strict separa- tion of concerns and a focus on simplicity. We demonstrate that LionsOS achieves excellent performance on system-call intensive workloads. | [
{
"version": "v1",
"created": "Wed, 08 Jan 2025 05:01:24 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 09:14:00 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Heiser",
"Gernot",
""
],
[
"Velickovic",
"Ivan",
""
],
[
"Chubb",
"Peter",
""
],
[
"Joshy",
"Alwin",
""
],
[
"Ganesh",
"Anuraag",
""
],
[
"Nguyen",
"Bill",
""
],
[
"Li",
"Cheng",
""
],
[
"Darville",
"Courtney",
""
],
[
"Zhu",
"Guangtao",
""
],
[
"Archer",
"James",
""
],
[
"Zhou",
"Jingyao",
""
],
[
"Winter",
"Krishnan",
""
],
[
"Parker",
"Lucy",
""
],
[
"Duchniewicz",
"Szymon",
""
],
[
"Bai",
"Tianyi",
""
]
] |
2501.06598 | Xuanle Zhao | Xuanle Zhao, Xianzhen Luo, Qi Shi, Chi Chen, Shuo Wang, Zhiyuan Liu, Maosong Sun | ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generation | Accepted by ACL 2025 Main, Camera Ready | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in chart understanding tasks. However, interpreting charts with textual descriptions often leads to information loss, as it fails to fully capture the dense information embedded in charts. In contrast, parsing charts into code provides lossless representations that can effectively contain all critical details. Although existing open-source MLLMs have achieved success in chart understanding tasks, they still face two major challenges when applied to chart-to-code tasks: (1) Low executability and poor restoration of chart details in the generated code and (2) Lack of large-scale and diverse training data. To address these challenges, we propose \textbf{ChartCoder}, the first dedicated chart-to-code MLLM, which leverages Code LLMs as the language backbone to enhance the executability of the generated code. Furthermore, we introduce \textbf{Chart2Code-160k}, the first large-scale and diverse dataset for chart-to-code generation, and propose the \textbf{Snippet-of-Thought (SoT)} method, which transforms direct chart-to-code generation data into step-by-step generation. Experiments demonstrate that ChartCoder, with only 7B parameters, surpasses existing open-source MLLMs on chart-to-code benchmarks, achieving superior chart restoration and code excitability. Our code is available at https://github.com/thunlp/ChartCoder. | [
{
"version": "v1",
"created": "Sat, 11 Jan 2025 17:52:22 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 05:08:29 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhao",
"Xuanle",
""
],
[
"Luo",
"Xianzhen",
""
],
[
"Shi",
"Qi",
""
],
[
"Chen",
"Chi",
""
],
[
"Wang",
"Shuo",
""
],
[
"Liu",
"Zhiyuan",
""
],
[
"Sun",
"Maosong",
""
]
] |
2501.06961 | Andreas Tersenov | Andreas Tersenov, Lucie Baumont, Jean-Luc Starck, Martin Kilbinger | Impact of weak-lensing mass-mapping algorithms on cosmology inference | null | A&A 698, A25 (2025) | 10.1051/0004-6361/202553707 | null | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Weak-lensing mass-mapping algorithms, which reconstruct the convergence field from galaxy shear measurements, are crucial for extracting higher-order statistics to constrain cosmological parameters. However, only limited research has explored whether the choice of mass-mapping algorithm affects the inference of cosmological parameters from weak-lensing higher-order statistics. This study aims to evaluate the impact of different mass-mapping algorithms on the inference of cosmological parameters measured with weak-lensing peak counts. We employ Kaiser-Squires, inpainting Kaiser-Squires, and MCALens mass-mapping algorithms to reconstruct the convergence field from simulated weak-lensing data. Using these maps, we compute the peak counts and wavelet peak counts as data vectors and perform Bayesian analysis with MCMC sampling to estimate posterior distributions of cosmological parameters. Our results indicate that the choice of mass-mapping algorithm significantly affects the constraints on cosmological parameters, with the MCALens method improving constraints by up to 157$\%$ compared to the standard Kaiser-Squires method. This improvement arises from MCALens' ability to better capture small-scale structures. In contrast, inpainting Kaiser-Squires yields constraints similar to Kaiser-Squires, indicating a limited benefit from inpainting for cosmological parameter estimation with peaks. The accuracy of mass-mapping algorithms is thus critical for cosmological inference from weak-lensing data. Advanced algorithms like MCALens, which offer superior reconstruction of the convergence field, can substantially enhance the precision of cosmological parameter estimates. These findings underscore the importance of selecting appropriate mass-mapping techniques in weak-lensing studies to fully exploit the potential of higher-order statistics for cosmological research. | [
{
"version": "v1",
"created": "Sun, 12 Jan 2025 22:47:40 GMT"
},
{
"version": "v2",
"created": "Mon, 21 Apr 2025 10:49:13 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Tersenov",
"Andreas",
""
],
[
"Baumont",
"Lucie",
""
],
[
"Starck",
"Jean-Luc",
""
],
[
"Kilbinger",
"Martin",
""
]
] |
2501.07001 | Abineet Parichha | Abineet Parichha, Stephane Colombi, Shohei Saga, Atsushi Taruya | Dark matter halo dynamics in 2D Vlasov Simulations: a self-similar approach | 15 pages, 13 figures, 2 tables | A&A 697, A218 (2025) | 10.1051/0004-6361/202553734 | YITP-25-03 | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Understanding dark matter halo dynamics can be pivotal in unravelling the nature of dark matter particles. Analytical treatment of the multistream flows inside the turnaround region of a collapsed cold dark matter (CDM) halo using various self-similar approaches already exist. In this work, we aim to determine the extent of self-similarity in 2D halo dynamics and the factors leading to deviations from it by studying numerical simulations of monolithically growing CDM halos. We have adapted the Fillmore and Goldreich (FG) self-similar solutions assuming cylindrical symmetry to data from 2D Vlasov-Poisson (ColDICE package) simulations of CDM halos seeded by sine wave initial conditions. We measured trajectories in position and phase-space, mass and density profiles and compared these to predictions from the FG model. We find that after turn-around and subsequent shell crossing, particles undergo a period of relaxation, typically about 1-2 oscillations about the center before they start to trace the self-similar fits and continue to do so as long as their orbits are predominantly radial. Overplotting the trajectories from different snapshots in scale-free position-time and phase spaces shows strikingly good superposition, a defining feature of self-similarity. The radial density profiles measured from simulations: $\rho \propto r^{-\alpha}, \alpha = 0.9 - 1.0$ are consistent with FG's prediction $\alpha=1$ for 2D halos. Deviations from the model, on the other hand, are evidently linked to relaxation, inhibited motion due to periodic boundaries, transverse motion in the halo interior, and deficit of infalling mass in limited simulation volume. It could not be conclusively established if the halos tend to grow circular over time. Extension of this work to actual 3D CDM cosmologies necessitates further detailed study of self-similar solutions with ellipsoidal collapse and transverse motion. | [
{
"version": "v1",
"created": "Mon, 13 Jan 2025 01:29:31 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Parichha",
"Abineet",
""
],
[
"Colombi",
"Stephane",
""
],
[
"Saga",
"Shohei",
""
],
[
"Taruya",
"Atsushi",
""
]
] |
2501.07229 | Wenjing Zhang | Wenjing Zhang, Yu Chen, and Yixian Gao | Limiting absorption principle of Helmholtz equation with sign changing coefficients under periodic structure | null | null | null | null | math.AP math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Negative refractive index materials have attracted significant research attention due to their unique electromagnetic response characteristics. In this paper, we employ the complementing boundary condition to establish rigorous a priori estimates for the Helmholtz equation, from which the limiting absorption principle is analytically derived. Within this mathematical framework, we conclusively establish the well-posedness of the electromagnetic transmission problem at the interface between conventional materials and negative refractive index materials in two-dimensional periodic structures. | [
{
"version": "v1",
"created": "Mon, 13 Jan 2025 11:29:53 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 06:19:05 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhang",
"Wenjing",
""
],
[
"Chen",
"Yu",
""
],
[
"Gao",
"Yixian",
""
]
] |
2501.07438 | Elena Bogliolo | Elena Bogliolo | Bounded cohomology and scl of verbal wreath products | 22 pages | null | null | null | math.GR math.AT math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the bounded cohomology and the stable commutator length of verbal wreath products $\Gamma \wr^{_W}A$, where $A$ has trivial bounded cohomology for a sufficiently large class of coefficients.\\ We prove that the stable commutator length always vanishes, and that the bounded cohomology vanishes in positive degrees for some such verbal wreath products; including the standard restricted wreath products (extending a recent result by Monod for lamplighters groups), as well as verbal wreath products arising from n-solvable, $n$-nilpotent, and $k$-Burnside $(k = 2, 3, 4, 6)$ verbal products.\ As an application, we show that every group of type $F_p$ isometrically embeds into a group of type $F_p$ with vanishing bounded cohomology in positive degrees for a large class of coefficients. | [
{
"version": "v1",
"created": "Mon, 13 Jan 2025 16:05:45 GMT"
},
{
"version": "v2",
"created": "Sun, 02 Feb 2025 21:44:40 GMT"
},
{
"version": "v3",
"created": "Mon, 26 May 2025 14:56:02 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 11:58:10 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Bogliolo",
"Elena",
""
]
] |
2501.08056 | Jan Kunes | A. Hariki, K. Sakurai, T. Okauchi, and J. Kune\v{s} | Magnetic Dichroism in Rutile NiF$_2$: Separating Altermagnetic and Ferromagnetic Effects | 6 pages, 6 figures, supplemental material | npj Quantum Mater. 10, 49 (2025) | 10.1038/s41535-025-00753-8 | null | cond-mat.mtrl-sci cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | We present numerical simulations of x-ray magnetic circular dichroism (XMCD) at the L$_{2,3}$ edge of Ni in the weakly ferromagnetic altermagnet NiF$_2$. Our results predict a significant XMCD signal for light propagating perpendicular to the magnetic moments, which are approximately aligned along the [100] easy-axis direction. The analysis shows that the altermagnetic and ferromagnetic contributions to the XMCD signal can be uniquely distinguished by their dependence on an applied magnetic field. By varying the angle of the field relative to the easy axis, the in-plane orientation of both the N\'eel vector and the net magnetization can be systematically controlled. We further demonstrate that the XMCD signal, even under fields as strong as 40 T and for any in-plane orientation, can be accurately described as a linear combination of two spectral components, with geometrical prefactors determined by the field magnitude and direction. This insight enables experimental validation of the distinctive relationship between the N\'eel vector orientation and the x-ray Hall vector in the rutile structure. Quantitative simulations supporting these findings are provided. | [
{
"version": "v1",
"created": "Tue, 14 Jan 2025 12:11:36 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 10:40:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Hariki",
"A.",
""
],
[
"Sakurai",
"K.",
""
],
[
"Okauchi",
"T.",
""
],
[
"Kuneš",
"J.",
""
]
] |
2501.10120 | Yichen He | Yichen He, Guanhua Huang, Peiyuan Feng, Yuan Lin, Yuchen Zhang, Hang Li, Weinan E | PaSa: An LLM Agent for Comprehensive Academic Paper Search | null | null | null | null | cs.IR cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We introduce PaSa, an advanced Paper Search agent powered by large language models. PaSa can autonomously make a series of decisions, including invoking search tools, reading papers, and selecting relevant references, to ultimately obtain comprehensive and accurate results for complex scholar queries. We optimize PaSa using reinforcement learning with a synthetic dataset, AutoScholarQuery, which includes 35k fine-grained academic queries and corresponding papers sourced from top-tier AI conference publications. Additionally, we develop RealScholarQuery, a benchmark collecting real-world academic queries to assess PaSa performance in more realistic scenarios. Despite being trained on synthetic data, PaSa significantly outperforms existing baselines on RealScholarQuery, including Google, Google Scholar, Google with GPT-4o for paraphrased queries, ChatGPT (search-enabled GPT-4o), GPT-o1, and PaSa-GPT-4o (PaSa implemented by prompting GPT-4o). Notably, PaSa-7B surpasses the best Google-based baseline, Google with GPT-4o, by 37.78% in recall@20 and 39.90% in recall@50, and exceeds PaSa-GPT-4o by 30.36% in recall and 4.25% in precision. Model, datasets, and code are available at https://github.com/bytedance/pasa. | [
{
"version": "v1",
"created": "Fri, 17 Jan 2025 11:12:28 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:01:29 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"He",
"Yichen",
""
],
[
"Huang",
"Guanhua",
""
],
[
"Feng",
"Peiyuan",
""
],
[
"Lin",
"Yuan",
""
],
[
"Zhang",
"Yuchen",
""
],
[
"Li",
"Hang",
""
],
[
"E",
"Weinan",
""
]
] |
2501.11091 | Bin Chen | Bin Chen | Bitcoin: A Non-Continuous Time System | null | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes a novel framework for understanding Bitcoin as a non-continuous time system shaped by the probabilistic block generation, the emergence of forks, and the non-linear confirmation of transactions. It introduces the concept of entropy collapse to describe how each block represents the irreversible resolution of uncertainty through Proof-of-Work computation, compressing a space of possible futures into a single validated past. In contrast to traditional time systems governed by synchronized clocks, Bitcoin constructs time internally through entropy dissipation and consensus-driven finality. This entropic structure reframes Bitcoin not merely as a ledger but as a decentralized clock that encodes a fundamentally asymmetric view of time in which the past is unique and irreversible while the future remains probabilistic until resolved. | [
{
"version": "v1",
"created": "Sun, 19 Jan 2025 16:04:03 GMT"
},
{
"version": "v2",
"created": "Thu, 03 Apr 2025 13:18:19 GMT"
},
{
"version": "v3",
"created": "Sat, 05 Apr 2025 08:02:08 GMT"
},
{
"version": "v4",
"created": "Fri, 25 Apr 2025 09:06:20 GMT"
},
{
"version": "v5",
"created": "Wed, 21 May 2025 15:22:29 GMT"
},
{
"version": "v6",
"created": "Tue, 27 May 2025 14:43:50 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Chen",
"Bin",
""
]
] |
2501.11153 | Huu Phong Nguyen | Huu Phong Nguyen, Shekhar Madhav Khairnar, Sofia Garces Palacios, Amr Al-Abbas, Melissa E. Hogg, Amer H. Zureikat, Patricio M. Polanco, Herbert Zeh III, Ganesh Sankaranarayanan | Efficient Frame Extraction: A Novel Approach Through Frame Similarity and Surgical Tool Tracking for Video Segmentation | 18 | null | 10.1109/ACCESS.2025.3573264 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The interest in leveraging Artificial Intelligence (AI) for surgical procedures to automate analysis has witnessed a significant surge in recent years. One of the primary tools for recording surgical procedures and conducting subsequent analyses, such as performance assessment, is through videos. However, these operative videos tend to be notably lengthy compared to other fields, spanning from thirty minutes to several hours, which poses a challenge for AI models to effectively learn from them. Despite this challenge, the foreseeable increase in the volume of such videos in the near future necessitates the development and implementation of innovative techniques to tackle this issue effectively. In this article, we propose a novel technique called Kinematics Adaptive Frame Recognition (KAFR) that can efficiently eliminate redundant frames to reduce dataset size and computation time while retaining useful frames to improve accuracy. Specifically, we compute the similarity between consecutive frames by tracking the movement of surgical tools. Our approach follows these steps: $i)$ Tracking phase: a YOLOv8 model is utilized to detect tools presented in the scene, $ii)$ Similarity phase: Similarities between consecutive frames are computed by estimating variation in the spatial positions and velocities of the tools, $iii$) Classification phase: An X3D CNN is trained to classify segmentation. We evaluate the effectiveness of our approach by analyzing datasets obtained through retrospective reviews of cases at two referral centers. The newly annotated Gastrojejunostomy (GJ) dataset covers procedures performed between 2017 and 2021, while the previously annotated Pancreaticojejunostomy (PJ) dataset spans from 2011 to 2022 at the same centers. | [
{
"version": "v1",
"created": "Sun, 19 Jan 2025 19:36:09 GMT"
},
{
"version": "v2",
"created": "Sun, 20 Apr 2025 07:28:55 GMT"
},
{
"version": "v3",
"created": "Mon, 28 Apr 2025 19:02:47 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Nguyen",
"Huu Phong",
""
],
[
"Khairnar",
"Shekhar Madhav",
""
],
[
"Palacios",
"Sofia Garces",
""
],
[
"Al-Abbas",
"Amr",
""
],
[
"Hogg",
"Melissa E.",
""
],
[
"Zureikat",
"Amer H.",
""
],
[
"Polanco",
"Patricio M.",
""
],
[
"Zeh",
"Herbert",
"III"
],
[
"Sankaranarayanan",
"Ganesh",
""
]
] |
2501.11472 | David Loeffler | David Loeffler and Sarah Livia Zerbes | Poles of p-adic Asai L-functions and distinguished representations | 10 pages; revised version, including appendix giving details of zeta-integral computation | null | null | null | math.NT math.RT | http://creativecommons.org/licenses/by/4.0/ | We give a criterion in terms of p-adic Asai L-functions for a cuspidal automorphic representation of GL(2) over a real quadratic field to be a distinguished representation, providing a p-adic counterpart of a well-known theorem of Flicker for the complex Asai L-function. | [
{
"version": "v1",
"created": "Mon, 20 Jan 2025 13:00:34 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:42:51 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Loeffler",
"David",
""
],
[
"Zerbes",
"Sarah Livia",
""
]
] |
2501.12692 | Lucien Saviot | Lucien Saviot and Vincent Laude | Strong shape-dependent intensity of inelastic light scattering by gold nanocrystals | 9 pages, 7 figures | Phys. Rev. B 111, 205431 (2025) | 10.1103/PhysRevB.111.205431 | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | We present a numerical approach to calculate inelastic light scattering spectra from gold nanocrystals, based on the finite element method. This approach is validated by comparison with previous analytic calculations for spherically symmetric scatterers. Superellipsoid nanocrystals are considered in order to smoothly vary the shape from octahedra to cubes via spheres, while preserving cubic symmetry. Spectra are calculated and discussed taking into account the irreducible representation of the involved vibration modes. A strong increase in the inelastically scattered light intensity is observed for small variations of the shape around the sphere. This increase is related to variations of the electric field inside the nanocrystals, which are very small for small nanospheres but increase quickly for non-spherical nanocrystals. This strong dependence with shape must be taken into account when interpreting experimental spectra acquired from inhomogeneous ensembles of nanocrystals whose shape dispersion are usually neglected. The overall changes in the spectra when varying the shape of the nanocrystals provide additional insight into previously published results. Preliminary calculations for chiral shapes further show a significant difference between spectra obtained with right or left circularly polarized light. | [
{
"version": "v1",
"created": "Wed, 22 Jan 2025 08:02:54 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 09:45:27 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Saviot",
"Lucien",
""
],
[
"Laude",
"Vincent",
""
]
] |
2501.12740 | Yuchi Jiang | Yuchi Jiang, Zhijie Li, Yunpeng Wang, Huiyu Yang, Jianchun Wang | An Implicit Adaptive Fourier Neural Operator for Long-term Predictions of Three-dimensional Turbulence | null | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Long-term prediction of three-dimensional (3D) turbulent flows is one of the most challenging problems for machine learning approaches. Although some existing machine learning approaches such as implicit U-net enhanced Fourier neural operator (IUFNO) have been proven to be capable of achieving stable long-term predictions for turbulent flows, their computational costs are usually high. In this paper, we use the adaptive Fourier neural operator (AFNO) as the backbone to construct a model that can predict 3D turbulence. Furthermore, we employ the implicit iteration to our constructed AFNO and propose the implicit adaptive Fourier neural operator (IAFNO). IAFNO is systematically tested in three types of 3D turbulence, including forced homogeneous isotropic turbulence (HIT), temporally evolving turbulent mixing layer and turbulent channel flow. The numerical results demonstrate that IAFNO is more accurate and stable than IUFNO and the traditional large-eddy simulation using dynamic Smagorinsky model (DSM). Meanwhile, the AFNO model exhibits instability in numerical simulations. Moreover, the training efficiency of IAFNO is 4 times higher than that of IUFNO, and the number of parameters and GPU memory occupation of IAFNO are only 1/80 and 1/3 of IUFNO, respectively in HIT. In other tests, the improvements are slightly lower but still considerable. These improvements mainly come from patching and self-attention in 3D space. Besides, the well-trained IAFNO is significantly faster than the DSM. | [
{
"version": "v1",
"created": "Wed, 22 Jan 2025 09:19:52 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 14:44:46 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Jiang",
"Yuchi",
""
],
[
"Li",
"Zhijie",
""
],
[
"Wang",
"Yunpeng",
""
],
[
"Yang",
"Huiyu",
""
],
[
"Wang",
"Jianchun",
""
]
] |
2501.12962 | Kristof Meding | Kristof Meding | It's complicated. The relationship of algorithmic fairness and non-discrimination regulations for high-risk systems in the EU AI Act | null | null | null | null | cs.LG cs.AI cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | What constitutes a fair decision? This question is not only difficult for humans but becomes more challenging when Artificial Intelligence (AI) models are used. In light of discriminatory algorithmic behaviors, the EU has recently passed the AI Act, which mandates specific rules for high-risk systems, incorporating both traditional legal non-discrimination regulations and machine learning based algorithmic fairness concepts. This paper aims to bridge these two different concepts in the AI Act through: First, a necessary high-level introduction of both concepts targeting legal and computer science-oriented scholars, and second, an in-depth analysis of the AI Act's relationship between legal non-discrimination regulations and algorithmic fairness. Our analysis reveals three key findings: (1.) Most non-discrimination regulations target only high-risk AI systems. (2.) The regulation of high-risk systems encompasses both data input requirements and output monitoring, though these regulations are partly inconsistent and raise questions of computational feasibility. (3.) Finally, we consider the possible (future) interaction of classical EU non-discrimination law and the AI Act regulations. We recommend developing more specific auditing and testing methodologies for AI systems. This paper aims to serve as a foundation for future interdisciplinary collaboration between legal scholars and computer science-oriented machine learning researchers studying discrimination in AI systems. | [
{
"version": "v1",
"created": "Wed, 22 Jan 2025 15:38:09 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Mar 2025 15:05:09 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 15:14:55 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Meding",
"Kristof",
""
]
] |
2501.13228 | Jin Kim | Jin Kim, Shane Schweitzer, Christoph Riedl, and David De Cremer | The AI Penalization Effect: People Reduce Compensation for Workers Who Use AI | null | null | null | null | econ.GN q-fin.EC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We investigate whether and why people might adjust compensation for workers who use AI tools. Across 11 studies (N = 3,846), participants consistently lowered compensation for AI-assisted workers compared to those who were unassisted. This "AI Penalization" effect was robust across (1) different types of work (e.g., specific tasks or general work scenarios) and worker statuses (e.g., full-time, part-time, or freelance), (2) different forms of compensation (e.g., required payments or optional bonuses) and their timing, (3) various methods of eliciting compensation (e.g., slider scale, multiple choice, and numeric entry), and (4) conditions where workers' output quality was held constant, subject to varying inferences, or statistically controlled. Moreover, the effect emerged not only in hypothetical compensation scenarios (Studies 1-9) but also with real gig workers and real monetary compensation (Studies 10 and 11). People reduced compensation for workers using AI because they believed these workers deserved less credit than those who did not use AI (Studies 7 and 8). This mediated effect attenuated when it was less permissible to reduce worker compensation, such as when employment contracts provide stricter constraints (Study 8). Our findings suggest that adoption of AI tools in the workplace may exacerbate inequality among workers, as those protected by structured contracts are less vulnerable to compensation reductions, while those without such protections are at greater risk of financial penalties for using AI. | [
{
"version": "v1",
"created": "Wed, 22 Jan 2025 21:27:28 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 00:39:48 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Kim",
"Jin",
""
],
[
"Schweitzer",
"Shane",
""
],
[
"Riedl",
"Christoph",
""
],
[
"De Cremer",
"David",
""
]
] |
2501.13229 | Daniel Gratzer | Daniel Gratzer, Jonathan Weinberger, Ulrik Buchholtz | The Yoneda embedding in simplicial type theory | null | null | null | null | cs.LO math.AT math.CT | http://creativecommons.org/licenses/by/4.0/ | Riehl and Shulman introduced simplicial type theory (STT), a variant of homotopy type theory which aimed to study not just homotopy theory, but its fusion with category theory: $(\infty,1)$-category theory. While notoriously technical, manipulating $\infty$-categories in simplicial type theory is often easier than working with ordinary categories, with the type theory handling infinite stacks of coherences in the background. We capitalize on recent work by Gratzer et al. defining the $(\infty,1)$-category of $\infty$-groupoids in STT to define presheaf categories within STT and systematically develop their theory. In particular, we construct the Yoneda embedding, prove the universal property of presheaf categories, refine the theory of adjunctions in STT, introduce the theory of Kan extensions, and prove Quillen's Theorem A. | [
{
"version": "v1",
"created": "Wed, 22 Jan 2025 21:28:14 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 15:01:27 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Gratzer",
"Daniel",
""
],
[
"Weinberger",
"Jonathan",
""
],
[
"Buchholtz",
"Ulrik",
""
]
] |
2501.13710 | I\~naki Erregue | I\~naki Erregue, Kamal Nasrollahi, Sergio Escalera | YOLO11-JDE: Fast and Accurate Multi-Object Tracking with Self-Supervised Re-ID | This paper has been accepted to the 5th Workshop on Real-World Surveillance: Applications and Challenges (WACV 2025) | null | 10.1109/WACVW65960.2025.00092 | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We introduce YOLO11-JDE, a fast and accurate multi-object tracking (MOT) solution that combines real-time object detection with self-supervised Re-Identification (Re-ID). By incorporating a dedicated Re-ID branch into YOLO11s, our model performs Joint Detection and Embedding (JDE), generating appearance features for each detection. The Re-ID branch is trained in a fully self-supervised setting while simultaneously training for detection, eliminating the need for costly identity-labeled datasets. The triplet loss, with hard positive and semi-hard negative mining strategies, is used for learning discriminative embeddings. Data association is enhanced with a custom tracking implementation that successfully integrates motion, appearance, and location cues. YOLO11-JDE achieves competitive results on MOT17 and MOT20 benchmarks, surpassing existing JDE methods in terms of FPS and using up to ten times fewer parameters. Thus, making our method a highly attractive solution for real-world applications. | [
{
"version": "v1",
"created": "Thu, 23 Jan 2025 14:38:40 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Erregue",
"Iñaki",
""
],
[
"Nasrollahi",
"Kamal",
""
],
[
"Escalera",
"Sergio",
""
]
] |
2501.13886 | Taha El Bakkali El Kadi | Taha El Bakkali El Kadi, Omar Saadi | On the Almost Sure Convergence of the Stochastic Three Points Algorithm | null | null | null | null | math.OC | http://creativecommons.org/licenses/by/4.0/ | The stochastic three points (STP) algorithm is a derivative-free optimization technique designed for unconstrained optimization problems in $\mathbb{R}^d$. In this paper, we analyze this algorithm for three classes of functions: smooth functions that may lack convexity, smooth convex functions, and smooth functions that are strongly convex. Our work provides the first almost sure convergence results of the STP algorithm, alongside some convergence results in expectation. For the class of smooth functions, we establish that the best gradient iterate of the STP algorithm converges almost surely to zero at a rate of $o(1/{T^{\frac{1}{2}-\epsilon}})$ for any $\epsilon\in (0,\frac{1}{2})$, where $T$ is the number of iterations. Furthermore, within the same class of functions, we establish both almost sure convergence and convergence in expectation of the final gradient iterate towards zero. For the class of smooth convex functions, we establish that $f(\theta^T)$ converges to $\inf_{\theta \in \mathbb{R}^d} f(\theta)$ almost surely at a rate of $o(1/{T^{1-\epsilon}})$ for any $\epsilon\in (0,1)$, and in expectation at a rate of $O(\frac{d}{T})$ where $d$ is the dimension of the space. Finally, for the class of smooth functions that are strongly convex, we establish that when step sizes are obtained by approximating the directional derivatives of the function, $f(\theta^T)$ converges to $\inf_{\theta \in \mathbb{R}^d} f(\theta)$ in expectation at a rate of $O((1-\frac{\mu}{2\pi dL})^T)$, and almost surely at a rate of $o((1-s\frac{\mu}{2\pi dL})^T)$ for any $s\in (0,1)$, where $\mu$ and $L$ are the strong convexity and smoothness parameters of the function. | [
{
"version": "v1",
"created": "Thu, 23 Jan 2025 17:58:42 GMT"
},
{
"version": "v2",
"created": "Fri, 24 Jan 2025 02:50:11 GMT"
},
{
"version": "v3",
"created": "Sat, 22 Mar 2025 20:05:29 GMT"
},
{
"version": "v4",
"created": "Mon, 26 May 2025 20:31:44 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Kadi",
"Taha El Bakkali El",
""
],
[
"Saadi",
"Omar",
""
]
] |
2501.14261 | Anastasiia Zalogina Dr | Anastasiia Zalogina, Chi Li, Ivan Zhigulin, Nathan Coste, Hossein Alijani, Otto Cranwell Schaeper, Hugo Charlton, Joseph Ward, Haoran Ren, Igor Aharonovich | An Inverse Design Wavelength Demultiplexer for On-Chip Photoluminescence Sorting in TMDC Heterostructures | null | null | null | null | physics.optics physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Emerging two-dimensional transition metal dichalcogenides (TMDCs) offer a promising platform for on-chip integrated photonics because of their unique optical and electronic properties. Their naturally passivated surfaces make them highly tolerant to lattice mismatch, enabling seamless heterogeneous integration by stacking different van der Waals materials, a crucial step in the development of advanced photonic devices. Here, we demonstrate the use of an inverse design wavelength demultiplexing waveguides for on-chip sorting and routing of distinct photoluminescence from the heterojunction formed by WS2 and WSe2 monolayers. The integrated nanophotonic chip splits and sorts excitonic emission into individual waveguides at room temperature. Our demonstration opens up new perspectives for integrating light sources in van der Waals materials with functional integrated photonics, offering a versatile platform for both fundamental research and practical applications. | [
{
"version": "v1",
"created": "Fri, 24 Jan 2025 06:00:19 GMT"
},
{
"version": "v2",
"created": "Wed, 16 Apr 2025 06:18:47 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 06:07:28 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zalogina",
"Anastasiia",
""
],
[
"Li",
"Chi",
""
],
[
"Zhigulin",
"Ivan",
""
],
[
"Coste",
"Nathan",
""
],
[
"Alijani",
"Hossein",
""
],
[
"Schaeper",
"Otto Cranwell",
""
],
[
"Charlton",
"Hugo",
""
],
[
"Ward",
"Joseph",
""
],
[
"Ren",
"Haoran",
""
],
[
"Aharonovich",
"Igor",
""
]
] |
2501.14315 | Chao-Chung Wu | Chao-Chung Wu, Zhi Rui Tam, Chieh-Yen Lin, Yun-Nung Chen, Shao-Hua Sun, Hung-yi Lee | Mitigating Forgetting in LLM Fine-Tuning via Low-Perplexity Token Learning | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Maintaining consistent model performance across domains is a fundamental challenge in machine learning. While recent work has explored using LLM-generated data for fine-tuning, its impact on cross-domain generalization remains poorly understood. This paper presents a systematic analysis revealing that fine-tuning with LLM-generated data not only improves target task performance but also reduces non-target task degradation compared to fine-tuning with ground truth data. Through analyzing the data sequence in tasks of various domains, we demonstrate that this enhancement of non-target task robustness stems from the reduction of high perplexity tokens found in LLM-generated sequences. Following our findings, we showed that masking high perplexity tokens in ground truth training data achieves similar non-target task performance preservation, comparable to using LLM-generated data. Extensive experiments across different model families and scales, including Gemma 2 IT 2B, Llama 3 8B Instruct, and 3 additional models, agree with our findings. To the best of our knowledge, this is the first work to provide an empirical explanation based on token perplexity reduction to mitigate catastrophic forgetting in LLMs after fine-tuning, offering valuable insights for developing more robust fine-tuning strategies. | [
{
"version": "v1",
"created": "Fri, 24 Jan 2025 08:18:56 GMT"
},
{
"version": "v2",
"created": "Tue, 20 May 2025 14:10:24 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 08:28:53 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Wu",
"Chao-Chung",
""
],
[
"Tam",
"Zhi Rui",
""
],
[
"Lin",
"Chieh-Yen",
""
],
[
"Chen",
"Yun-Nung",
""
],
[
"Sun",
"Shao-Hua",
""
],
[
"Lee",
"Hung-yi",
""
]
] |
2501.14878 | Alessandro Traspadini | Alexander Bonora, Alessandro Traspadini, Marco Giordani, and Michele Zorzi | Performance Evaluation of Satellite-Based Data Offloading on Starlink Constellations | 7 pages, 6 figures, 1 table. This paper has been accepted for the 2025 IEEE Wireless Communications and Networking Conference (WCNC). \c{opyright} 2025 IEEE. Cite as: A. Bonora, A. Traspadini, M. Giordani, and M. Zorzi, "Performance Evaluation of Satellite-Based Data Offloading on Starlink Constellations," in Proc. IEEE Wireless Communications and Networking Conference (WCNC), 2025 | null | 10.1109/WCNC61545.2025.10978515 | null | cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vehicular Edge Computing (VEC) is a key research area in autonomous driving. As Intelligent Transportation Systems (ITSs) continue to expand, ground vehicles (GVs) face the challenge of handling huge amounts of sensor data to drive safely. Specifically, due to energy and capacity limitations, GVs will need to offload resource-hungry tasks to external (cloud) computing units for faster processing. In 6th generation (6G) wireless systems, the research community is exploring the concept of Non-Terrestrial Networks (NTNs), where satellites can serve as space edge computing nodes to aggregate, store, and process data from GVs. In this paper we propose new data offloading strategies between a cluster of GVs and satellites in the Low Earth Orbits (LEOs), to optimize the trade-off between coverage and end-to-end delay. For the accuracy of the simulations, we consider real data and orbits from the Starlink constellation, one of the most representative and popular examples of commercial satellite deployments for communication. Our results demonstrate that Starlink satellites can support real-time offloading under certain conditions that depend on the onboard computational capacity of the satellites, the frame rate of the sensors, and the number of GVs. | [
{
"version": "v1",
"created": "Fri, 24 Jan 2025 19:04:17 GMT"
},
{
"version": "v2",
"created": "Tue, 11 Mar 2025 22:59:55 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Bonora",
"Alexander",
""
],
[
"Traspadini",
"Alessandro",
""
],
[
"Giordani",
"Marco",
""
],
[
"Zorzi",
"Michele",
""
]
] |
2501.15222 | Francesco Di Clemente | Francesco Di Clemente, Matteo Scialpi, Micha{\l} Bejger | Explainable autoencoder for neutron star dense matter parameter estimation | null | null | 10.1088/2632-2153/add3bd | null | physics.comp-ph astro-ph.HE astro-ph.IM nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a physics-informed autoencoder designed to encode the equation of state of neutron stars into an interpretable latent space. In particular the input will be encoded in the mass, radius, and tidal deformability values of a neutron star. Unlike traditional black-box models, our approach incorporates additional loss functions to enforce explainability in the encoded representations. This method enhances the transparency of machine learning models in physics, providing a robust proof-of-concept tool to study compact stars data. Our results demonstrate that the proposed autoencoder not only accurately estimates the EoS parameters and central density/pressure but also offers insights into the physical connection between equation of state and observable physical quantities. This framework conceptualizes the physical differential equations themselves as the ``encoders", allowing interpretability of the latent space. | [
{
"version": "v1",
"created": "Sat, 25 Jan 2025 13:59:58 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Di Clemente",
"Francesco",
""
],
[
"Scialpi",
"Matteo",
""
],
[
"Bejger",
"Michał",
""
]
] |
2501.15419 | Mitsuaki Obara | Mitsuaki Obara, Takayuki Okuno, Akiko Takeda | A primal-dual interior point trust region method for second-order stationary points of Riemannian inequality-constrained optimization problems | 44 pages, 4 figures. An earlier version of this work presented both global and local convergence analyses. The local analysis has since been extended to cover both inequality- and equality-constrained settings, and is now reported separately in arXiv:2505.19724 | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider Riemannian inequality-constrained optimization problems. Such problems inherit the benefits of Riemannian approach developed in the unconstrained setting and naturally arise from applications in control, machine learning, and other fields. We propose a Riemannian primal-dual interior point trust region method (RIPTRM) for solving them. We prove its global convergence to an approximate Karush-Kuhn-Tucker point and a second-order stationary point. To the best of our knowledge, this is the first algorithm that incorporates the trust region strategy for constrained optimization on Riemannian manifolds, and has the second-order convergence property for optimization problems on Riemannian manifolds with nonlinear inequality constraints. We conduct numerical experiments in which we introduce a truncated conjugate gradient method and an eigenvalue-based subsolver for RIPTRM to approximately and exactly solve the trust region subproblems, respectively. Empirical results show that RIPTRMs find solutions with higher accuracy compared to an existing Riemannian interior point method and other algorithms. Additionally, we observe that RIPTRM with the exact search direction shows promising performance in an instance where the Hessian of the Lagrangian has a large negative eigenvalue. | [
{
"version": "v1",
"created": "Sun, 26 Jan 2025 06:46:56 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 15:21:47 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Obara",
"Mitsuaki",
""
],
[
"Okuno",
"Takayuki",
""
],
[
"Takeda",
"Akiko",
""
]
] |
2501.15502 | Xiaoyu Lai | Xinyi Zhong and Xiaoyu Lai | The role of equation of state on the spin-up of millisecond pulsars | 8 pages, 5 figures. Submitted | null | null | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Millisecond pulsars (MSPs) are recycled pulsars which have been spun-up due to mass accretion during the phase of mass exchange in binaries. Although the interactions with companion stars play important roles on the spin-up process, the global properties of pulsars determined by the equation of state (EoS), such as mass, radius and the moment of inertia, should also play a role. We investigate the spin-up of MSPs in neutron star (NS) and strangeon star (SS) models, both of which have passed the tests by the existence of high-mass pulsars and the tidal deformability of GW170817. Combining the spin-up condition and the transferred angular momentum, we can constrain the accreted mass and the magnetic field strength. The magnetic field strength will be important in impeding the spin-up of NSs, especially to the ones with low masses, but the impeding effect would not be important to SSs. The results indicate that it is easier for an SS to form a fully recycled low-mass MPS, whose spin-period is below 10 ms and mass is below about $1.5 M_\odot$, than that for an NS. Finding more low-mass and fully recycled MSPs, with accurate mass-measurement and better constraints on magnetic field strength, could help to put more strict constraints on the EoS of pulsars. | [
{
"version": "v1",
"created": "Sun, 26 Jan 2025 12:22:14 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 08:25:05 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhong",
"Xinyi",
""
],
[
"Lai",
"Xiaoyu",
""
]
] |
2501.16243 | Yang Xu | Yang Xu and Vaneet Aggarwal | Accelerating Quantum Reinforcement Learning with a Quantum Natural Policy Gradient Based Approach | null | null | null | null | quant-ph cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We address the problem of quantum reinforcement learning (QRL) under model-free settings with quantum oracle access to the Markov Decision Process (MDP). This paper introduces a Quantum Natural Policy Gradient (QNPG) algorithm, which replaces the random sampling used in classical Natural Policy Gradient (NPG) estimators with a deterministic gradient estimation approach, enabling seamless integration into quantum systems. While this modification introduces a bounded bias in the estimator, the bias decays exponentially with increasing truncation levels. This paper demonstrates that the proposed QNPG algorithm achieves a sample complexity of $\tilde{\mathcal{O}}(\epsilon^{-1.5})$ for queries to the quantum oracle, significantly improving the classical lower bound of $\tilde{\mathcal{O}}(\epsilon^{-2})$ for queries to the MDP. | [
{
"version": "v1",
"created": "Mon, 27 Jan 2025 17:38:30 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 21:50:51 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Xu",
"Yang",
""
],
[
"Aggarwal",
"Vaneet",
""
]
] |
2501.16511 | Jakob Nordin | Jakob Nordin, Valery Brinnel, Jakob van Santen, Simeon Reusch, Marek Kowalski | AMPEL workflows for LSST: Modular and reproducible real-time photometric classification | Submitted to Astronomy & Astrophysics | A&A 698, A13 (2025) | 10.1051/0004-6361/202452481 | null | astro-ph.IM astro-ph.CO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Modern time-domain astronomical surveys produce high throughput data streams which require tools for processing and analysis. This will be critical for programs making full use of the alert stream from the Vera Rubin Observatory (VRO), where spectroscopic labels will only be available for a small subset of all transients. In this context, the AMPEL toolset can work as a code-to-data platform for the development of efficient, reproducible and flexible workflows for real-time astronomical application.
We here introduce three different AMPEL channels constructed to highlight different uses of alert streams: to rapidly find infant transients (SNGuess), to provide unbiased transient samples for follow-up (FollowMe) and to deliver final transient classifications (FinalBet). These pipelines already contain placeholders for mechanisms which will be essential for the optimal usage of VRO alerts: combining different classifiers, including host galaxy information, population priors and sampling non-gaussian photometric redshift distributions. Based on the ELAsTiCC simulation, all three channels are already working at a high level: SNGuess correctly tags 99% of all young supernovae, FollowMe illustrates how an unbiased subset of alerts can be selected for spectroscopic follow-up in the context of cosmological probes and FinalBet includes priors to achieve successful classifications for >~80% of all extragalactic transients.
The fully functional workflows presented here are all public and can be used as starting points for any group wishing to optimize pipelines for their specific VRO science programs. AMPEL is designed to allow this to be done in accordance with FAIR principles: both software and results can be easily shared and results reproduced. The code-to-data environment ensures that models developed this way can be directly applied to the real-time LSST stream parsed by AMPEL. | [
{
"version": "v1",
"created": "Mon, 27 Jan 2025 21:19:23 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Nordin",
"Jakob",
""
],
[
"Brinnel",
"Valery",
""
],
[
"van Santen",
"Jakob",
""
],
[
"Reusch",
"Simeon",
""
],
[
"Kowalski",
"Marek",
""
]
] |
2501.16716 | Tongxu Zhang | Tongxu Zhang and Bei Wang | PUSSM: Point Cloud Upsampling as Implicit Statistical Shape Model | 14 pages, 4 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | This paper proposes a framework for high-fidelity reconstruction of pelvic structures by integrating medical image segmentation and point cloud upsampling. By point cloud upsampling to learn shape priors from MedShapePelvic without requiring landmarks or PCA, our method functions as an implicit statistical shape model. Evaluations on Pelvic1k show significant improvements in surface quality and anatomical accuracy. This approach is generalizable and applicable to other skeletal regions. | [
{
"version": "v1",
"created": "Tue, 28 Jan 2025 05:47:50 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Feb 2025 14:38:36 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 11:57:58 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Zhang",
"Tongxu",
""
],
[
"Wang",
"Bei",
""
]
] |
2501.16809 | Remi Carles | R\'emi Carles (IRMAR), Fangyuan Dong (Department of Applied Mathematics) | Propagation of coherent states in the logarithmic Schrodinger equation | Some typos fixed. More comments in the introduction | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the logarithmic Schr{\"o}dinger equation in a semiclassical scaling, in the presence of a smooth, at most quadratic, external potential. For initial data under the form of a single coherent state, we identify the notion of criticality as far as the nonlinear coupling constant is concerned, in the semiclassical limit. In the critical case, we prove a general error estimate, and improve it in the case of initial Gaussian profiles. In this critical case, when the initial datum is the sum of two Gaussian coherent states with different centers in phase space, we prove a nonlinear superposition principle. | [
{
"version": "v1",
"created": "Tue, 28 Jan 2025 09:25:27 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:16:54 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Carles",
"Rémi",
"",
"IRMAR"
],
[
"Dong",
"Fangyuan",
"",
"Department of Applied Mathematics"
]
] |
2501.17089 | Felix Hoops | Felix Hoops, Jonas Gebele, Florian Matthes | CRSet: Private Non-Interactive Verifiable Credential Revocation | Preprint | null | null | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | Like any digital certificate, Verifiable Credentials (VCs) require a way to revoke them in case of an error or key compromise. Existing solutions for VC revocation, most prominently Bitstring Status List, are not viable for many use cases because they may leak the issuer's activity, which in turn leaks internal business metrics. For instance, staff fluctuation through the revocation of employee IDs. We identify the protection of issuer activity as a key gap and propose formal definitions for corresponding characteristics of a revocation mechanism. Then, we introduce CRSet, a non-interactive mechanism that trades some space efficiency to reach these privacy characteristics. For that, we provide a proof sketch. Issuers periodically encode revocation data and publish it via Ethereum blob-carrying transactions, ensuring secure and private availability. Relying Parties (RPs) can download it to perform revocation checks locally. Sticking to a non-interactive design also makes adoption easier because it requires no changes to wallet agents and exchange protocols. We also implement and empirically evaluate CRSet, finding its real-world behavior to match expectations. One Ethereum blob fits revocation data for about 170k VCs. | [
{
"version": "v1",
"created": "Tue, 28 Jan 2025 17:23:45 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 21:58:55 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Hoops",
"Felix",
""
],
[
"Gebele",
"Jonas",
""
],
[
"Matthes",
"Florian",
""
]
] |
2501.17178 | Omar Swelam | David Salinas, Omar Swelam, Frank Hutter | Tuning LLM Judge Design Decisions for 1/1000 of the Cost | null | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Evaluating Large Language Models (LLMs) often requires costly human annotations. To address this, LLM-based judges have been proposed, which compare the outputs of two LLMs enabling the ranking of models without human intervention. While several approaches have been proposed, many confounding factors are present between different papers. For instance the model, the prompt and other hyperparameters are typically changed at the same time making apple-to-apple comparisons challenging. In this paper, we propose to systematically analyze and tune the hyperparameters of LLM judges. To alleviate the high cost of evaluating a judge, we propose to leverage multi-objective multi-fidelity which allows to find judges that trade accuracy for cost and also significantly reduce the cost of the search. Our method identifies judges that not only outperform existing benchmarks in accuracy and cost-efficiency but also utilize open-weight models, ensuring greater accessibility and reproducibility. The code to reproduce our experiments is available at this repository https://github.com/geoalgo/judgetuning . | [
{
"version": "v1",
"created": "Fri, 24 Jan 2025 17:01:14 GMT"
},
{
"version": "v2",
"created": "Tue, 04 Feb 2025 08:21:00 GMT"
},
{
"version": "v3",
"created": "Tue, 18 Mar 2025 09:09:29 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 07:58:46 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Salinas",
"David",
""
],
[
"Swelam",
"Omar",
""
],
[
"Hutter",
"Frank",
""
]
] |
2501.17421 | Shuhang Li | Shuhang Li, Zijun Zhao, and Francesco A. Evangelista | Equation-of-motion internally contracted multireference unitary coupled-cluster theory | null | J. Chem. Phys. 162, 154112 (2025) | 10.1063/5.0261000 | null | physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | The accurate computation of excited states remains a challenge in electronic structure theory, especially for systems with a ground state that requires a multireference treatment. In this work, we introduce a novel equation-of-motion (EOM) extension of the internally contracted multireference unitary coupled-cluster framework (ic-MRUCC), termed EOM-ic-MRUCC. EOM-ic-MRUCC follows the transform-then-diagonalize approach, in analogy to its non-unitary counterpart [Datta and Nooijen, J. Chem. Phys. 137, 204107 (2012)]. By employing a projective approach to optimize the ground state, the method retains additive separability and proper scaling with system size. We show that excitation energies are size intensive if the EOM operator satisfies the "killer" and the projective conditions. Furthermore, we propose to represent changes in reference state upon electron excitation via projected many-body operators that span active orbitals and show that the EOM equations formulated in this way are invariant with respect to active orbital rotations. We test the EOM-ic-MRUCC method truncated to single and double excitations by computing the potential energy curves for several excited states of a BeH$_2$ model system, the HF molecule, and water undergoing symmetric dissociation. Across these systems, our method delivers accurate excitation energies and potential energy curves within 5 m$E_\mathrm{h}$ (ca. 0.14 eV) from full configuration interaction. We find that truncating the Baker-Campbell-Hausdorff series to four-fold commutators contributes negligible errors (on the order of $10^{-5}$ $E_\mathrm{h}$ or less), offering a practical route to highly accurate excited-state calculations with reduced computational overhead. | [
{
"version": "v1",
"created": "Wed, 29 Jan 2025 05:23:03 GMT"
},
{
"version": "v2",
"created": "Fri, 07 Feb 2025 16:41:33 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Li",
"Shuhang",
""
],
[
"Zhao",
"Zijun",
""
],
[
"Evangelista",
"Francesco A.",
""
]
] |
2501.18282 | Yunzhen Yao | Yunzhen Yao, Lie He, Michael Gastpar | Leveraging Sparsity for Sample-Efficient Preference Learning: A Theoretical Perspective | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | This paper considers the sample-efficiency of preference learning, which models and predicts human choices based on comparative judgments. The minimax optimal estimation error rate $\Theta(d/n)$ in classical estimation theory requires that the number of samples $n$ scales linearly with the dimensionality of the feature space $d$. However, the high dimensionality of the feature space and the high cost of collecting human-annotated data challenge the efficiency of traditional estimation methods. To remedy this, we leverage sparsity in the preference model and establish sharp error rates. We show that under the sparse random utility model, where the parameter of the reward function is $k$-sparse, the minimax optimal rate can be reduced to $\Theta(k/n \log(d/k))$. Furthermore, we analyze the $\ell_{1}$-regularized estimator and show that it achieves near-optimal rate under mild assumptions on the Gram matrix. Experiments on synthetic data and LLM alignment data validate our theoretical findings, showing that sparsity-aware methods significantly reduce sample complexity and improve prediction accuracy. | [
{
"version": "v1",
"created": "Thu, 30 Jan 2025 11:41:13 GMT"
},
{
"version": "v2",
"created": "Fri, 31 Jan 2025 02:13:55 GMT"
},
{
"version": "v3",
"created": "Mon, 26 May 2025 19:24:08 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Yao",
"Yunzhen",
""
],
[
"He",
"Lie",
""
],
[
"Gastpar",
"Michael",
""
]
] |
2501.18326 | Petr Hlin\v{e}n\'y | Petr Hlin\v{e}n\'y and Jan Jedelsk\'y | Transductions of Graph Classes Admitting Product Structure | null | null | null | null | cs.LO math.CO | http://creativecommons.org/licenses/by-sa/4.0/ | In a quest to thoroughly understand the first-order transduction hierarchy of hereditary graph classes, some questions in particular stand out; such as, what properties hold for graph classes that are first-order transductions of planar graphs (and of similar classes)? When addressing this (so-far wide open) question, we turn to the concept of a product structure - being a subgraph of the strong product of a path and a graph of bounded tree-width, introduced by Dujmovic et al. [JACM 2020]. Namely, we prove that any graph class which is a first-order transduction of a class admitting such product structure, up to perturbations also meets a structural description generalizing the concept of a product structure in a dense hereditary way - the latter concept being introduced just recently by Hlineny and Jedelsky under the name of H-clique-width [MFCS 2024]. Using this characterization, we show that the class of the 3D grids, as well as a class of certain modifications of 2D grids, are not first-order transducible from classes admitting a product structure, and in particular not from the class of planar graphs. | [
{
"version": "v1",
"created": "Thu, 30 Jan 2025 13:13:28 GMT"
},
{
"version": "v2",
"created": "Fri, 21 Mar 2025 10:45:25 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 08:13:59 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Hliněný",
"Petr",
""
],
[
"Jedelský",
"Jan",
""
]
] |
2501.18470 | Alistair Carson | Alistair Carson, Vesa V\"alim\"aki, Alec Wright, Stefan Bilbao | Resampling Filter Design for Multirate Neural Audio Effect Processing | Accepted for publication in IEEE Transactions on Audio, Speech and Language Processing | null | null | null | eess.AS cs.LG cs.SD eess.SP | http://creativecommons.org/licenses/by/4.0/ | Neural networks have become ubiquitous in audio effects modelling, especially for guitar amplifiers and distortion pedals. One limitation of such models is that the sample rate of the training data is implicitly encoded in the model weights and therefore not readily adjustable at inference. Recent work explored modifications to recurrent neural network architecture to approximate a sample rate independent system, enabling audio processing at a rate that differs from the original training rate. This method works well for integer oversampling and can reduce aliasing caused by nonlinear activation functions. For small fractional changes in sample rate, fractional delay filters can be used to approximate sample rate independence, but in some cases this method fails entirely. Here, we explore the use of real-time signal resampling at the input and output of the neural network as an alternative solution. We investigate several resampling filter designs and show that a two-stage design consisting of a half-band IIR filter cascaded with a Kaiser window FIR filter can give similar or better results to the previously proposed model adjustment method with many fewer filtering operations per sample and less than one millisecond of latency at typical audio rates. Furthermore, we investigate interpolation and decimation filters for the task of integer oversampling and show that cascaded half-band IIR and FIR designs can be used in conjunction with the model adjustment method to reduce aliasing in a range of distortion effect models. | [
{
"version": "v1",
"created": "Thu, 30 Jan 2025 16:44:49 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 09:56:21 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Carson",
"Alistair",
""
],
[
"Välimäki",
"Vesa",
""
],
[
"Wright",
"Alec",
""
],
[
"Bilbao",
"Stefan",
""
]
] |
2501.18486 | Lajos Di\'osi | Lajos Di\'osi | A healthier stochastic semiclassical gravity: world without Schr\"{o}dinger cats | 10pp, based on the author's talk at the Lema\^itre 2024 conference on Black Holes, Gravitational Waves and Space-Time Singularities (June 17-21, 2024, Castel Gandolfo, Specola Vaticana) | Gen. Rel. Grav. 57, 62 (2025) | 10.1007/s10714-025-03396-z | null | gr-qc quant-ph | http://creativecommons.org/licenses/by/4.0/ | Semiclassical gravity couples classical gravity to the quantized matter in meanfield approximation. The meanfield coupling is problematic for two reasons. First, it ignores the quantum fluctuation of matter distribution. Second, it violates the linearity of the quantum dynamics. The first problem can be mitigated by allowing stochastic fluctuations of the geometry but the second problem lies deep in quantum foundations. Restoration of quantum linearity requires a conceptual approach to hybrid classical-quantum coupling. Studies of the measurement problem and the quantum-classical transition point to the solution. It is based on a postulated mechanism of spontaneous quantum monitoring plus feedback. This approach eliminates Schr\"{o}dinger cat states, takes quantum fluctuations into account, and restores the linearity of quantum dynamics. Such a captivating conceptionally `healthier' semiclassical theory exists in the Newtonian limit, but its relativistic covariance hits a wall. Here we will briefly recapitulate the concept and its realization in the nonrelativistic limit. We emphasize that the long-known obstacles to the relativistic extension lie in quantum foundations. | [
{
"version": "v1",
"created": "Thu, 30 Jan 2025 17:03:38 GMT"
},
{
"version": "v2",
"created": "Wed, 02 Apr 2025 21:24:10 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Diósi",
"Lajos",
""
]
] |
2501.19124 | Rahul Marathe | Monojit Chatterjee, Viktor Holubec and Rahul Marathe | Optimal constrained control for generally damped Brownian heat engines | 18 pages, 7 figures, 1 table | null | null | null | cond-mat.stat-mech cond-mat.soft | http://creativecommons.org/licenses/by/4.0/ | Optimization of cyclic stochastic heat engines, a topic spanning decades of research, commonly assumes fixed control or response parameters at discrete points in the cycle-a limitation that often leads to experimentally impractical protocols. We overcome this with a general algorithm, adapted from optimal control theory, that optimizes full-cycle dynamics under realistic constraints, such as stiffness and temperature bounds, across diverse systems. Unlike geometric or mass transport methods, which rely on fixed endpoints and are unsuitable for unconstrained cycles, our approach simultaneously tunes both cycle time and control variations. Applied to a generally damped Brownian particle in a harmonic potential-an experimentally relevant case-our method is validated in the overdamped regime and extended to arbitrary damping rates. As damping decreases, maximum power vanishes and cycle time diverges; at fixed cycle times, efficiency follows a similar trend, with optimal protocols exhibiting non-monotonic complexity. Notably, optimizing temperature profiles-often overlooked-significantly enhances efficiency in intermediate damping regimes. Our work establishes the first systematic framework for optimizing cyclic stochastic processes under experimental constraints, broadening the scope of power and efficiency optimization in nonequilibrium thermodynamics. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 13:28:25 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 06:19:33 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Chatterjee",
"Monojit",
""
],
[
"Holubec",
"Viktor",
""
],
[
"Marathe",
"Rahul",
""
]
] |
2501.19254 | Xinyu Liu | Xinyu Liu, Zixuan Xie, Shangtong Zhang | Linear $Q$-Learning Does Not Diverge in $L^2$: Convergence Rates to a Bounded Set | null | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | $Q$-learning is one of the most fundamental reinforcement learning algorithms. It is widely believed that $Q$-learning with linear function approximation (i.e., linear $Q$-learning) suffers from possible divergence until the recent work Meyn (2024) which establishes the ultimate almost sure boundedness of the iterates of linear $Q$-learning. Building on this success, this paper further establishes the first $L^2$ convergence rate of linear $Q$-learning iterates (to a bounded set). Similar to Meyn (2024), we do not make any modification to the original linear $Q$-learning algorithm, do not make any Bellman completeness assumption, and do not make any near-optimality assumption on the behavior policy. All we need is an $\epsilon$-softmax behavior policy with an adaptive temperature. The key to our analysis is the general result of stochastic approximations under Markovian noise with fast-changing transition functions. As a side product, we also use this general result to establish the $L^2$ convergence rate of tabular $Q$-learning with an $\epsilon$-softmax behavior policy, for which we rely on a novel pseudo-contraction property of the weighted Bellman optimality operator. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 16:10:50 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Feb 2025 19:26:06 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Feb 2025 16:39:27 GMT"
},
{
"version": "v4",
"created": "Tue, 27 May 2025 16:10:41 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Liu",
"Xinyu",
""
],
[
"Xie",
"Zixuan",
""
],
[
"Zhang",
"Shangtong",
""
]
] |
2501.19388 | Antoine Scheid | Antoine Scheid, Etienne Boursier, Alain Durmus, Eric Moulines, Michael Jordan | Online Decision-Making in Tree-Like Multi-Agent Games with Transfers | null | null | null | null | cs.GT | http://creativecommons.org/licenses/by/4.0/ | The widespread deployment of Machine Learning systems everywhere raises challenges, such as dealing with interactions or competition between multiple learners. In that goal, we study multi-agent sequential decision-making by considering principal-agent interactions in a tree structure. In this problem, the reward of a player is influenced by the actions of her children, who are all self-interested and non-cooperative, hence the complexity of making good decisions. Our main finding is that it is possible to steer all the players towards the globally optimal set of actions by simply allowing single-step transfers between them. A transfer is established between a principal and one of her agents: the principal actually offers the proposed payment if the agent picks the recommended action. The analysis poses specific challenges due to the intricate interactions between the nodes of the tree and the propagation of the regret within this tree. Considering a bandit setup, we propose algorithmic solutions for the players to end up being no-regret with respect to the optimal pair of actions and incentives. In the long run, allowing transfers between players makes them act as if they were collaborating together, although they remain self-interested non-cooperative: transfers restore efficiency. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 18:43:29 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 14:43:52 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Scheid",
"Antoine",
""
],
[
"Boursier",
"Etienne",
""
],
[
"Durmus",
"Alain",
""
],
[
"Moulines",
"Eric",
""
],
[
"Jordan",
"Michael",
""
]
] |
2501.19406 | Matthew Chen | Matthew Chen, Joshua Engels, Max Tegmark | Low-Rank Adapting Models for Sparse Autoencoders | Code available at https://github.com/matchten/LoRA-Models-for-SAEs | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Sparse autoencoders (SAEs) decompose language model representations into a sparse set of linear latent vectors. Recent works have improved SAEs using language model gradients, but these techniques require many expensive backward passes during training and still cause a significant increase in cross entropy loss when SAE reconstructions are inserted into the model. In this work, we improve on these limitations by taking a fundamentally different approach: we use low-rank adaptation (LoRA) to finetune the \textit{language model itself} around a previously trained SAE. We analyze our method across SAE sparsity, SAE width, language model size, LoRA rank, and model layer on the Gemma Scope family of SAEs. In these settings, our method reduces the cross entropy loss gap by 30\% to 55\% when SAEs are inserted during the forward pass. We also find that compared to end-to-end (e2e) SAEs, our approach achieves the same downstream cross entropy loss 3$\times$ to 20$\times$ faster on \gemma and 2$\times$ to 10$\times$ faster on \llama. We further show that our technique improves downstream metrics and can adapt multiple SAEs at once without harming general language model capabilities. Our results demonstrate that improving model interpretability is not limited to post-hoc SAE training; Pareto improvements can also be achieved by directly optimizing the model itself. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 18:59:16 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 06:39:54 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Chen",
"Matthew",
""
],
[
"Engels",
"Joshua",
""
],
[
"Tegmark",
"Max",
""
]
] |
2502.00131 | Soumik Dey | Soumik Dey, Wei Zhang, Hansi Wu, Bingfeng Dong, Binbin Li | Middleman Bias in Advertising: Aligning Relevance of Keyphrase Recommendations with Search | null | WWW '25: Companion Proceedings of the ACM on Web Conference 2025 | 10.1145/3701716.3717858 | null | cs.IR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | E-commerce sellers are recommended keyphrases based on their inventory on which they advertise to increase buyer engagement (clicks/sales). Keyphrases must be pertinent to items; otherwise, it can result in seller dissatisfaction and poor targeting -- towards that end relevance filters are employed. In this work, we describe the shortcomings of training relevance filter models on biased click/sales signals. We re-conceptualize advertiser keyphrase relevance as interaction between two dynamical systems -- Advertising which produces the keyphrases and Search which acts as a middleman to reach buyers. We discuss the bias of search relevance systems (middleman bias) and the need to align advertiser keyphrases with search relevance signals. We also compare the performance of cross encoders and bi-encoders in modeling this alignment and the scalability of such a solution for sellers at eBay. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 19:28:26 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Feb 2025 05:05:02 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Dey",
"Soumik",
""
],
[
"Zhang",
"Wei",
""
],
[
"Wu",
"Hansi",
""
],
[
"Dong",
"Bingfeng",
""
],
[
"Li",
"Binbin",
""
]
] |
2502.00191 | Armando Casta\~neda | Armando Casta\~neda, Gilde Valeria Rodr\'iguez | Asynchronous Fault-Tolerant Language Decidability for Runtime Verification of Distributed Systems | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | Implementing correct distributed systems is an error-prone task. Runtime Verification (RV) offers a lightweight formal method to improve reliability by monitoring system executions against correctness properties. However, applying RV in distributed settings - where no process has global knowledge - poses fundamental challenges, particularly under full asynchrony and fault tolerance. This paper addresses the Distributed Runtime Verification (DRV) problem under such conditions. In our model, each process in a distributed monitor receives a fragment of the input word describing system behavior and must decide whether this word belongs to the language representing the correctness property being verified. Hence, the goal is to decide languages in a distributed fault-tolerant manner. We propose several decidability definitions, study the relations among them, and prove possibility and impossibility results. One of our main results is a characterization of the correctness properties that can be decided asynchronously. Remarkably, it applies to any language decidability definition. Intuitively, the characterization is that only properties with no real-time order constraints can be decided in asynchronous fault-tolerant settings. These results expose the expressive limits of DRV in realistic systems, as several properties of practical interest rely on reasoning about real-time order of events in executions. To overcome these limitations, we introduce a weaker model where the system under inspection is verified indirectly. Under this weaker model we define predictive decidability, a decidability definition that turn some real-time sensitive correctness properties verifiable. Our framework unifies and extends existing DRV theory and sharpens the boundary of runtime monitorability under different assumptions. | [
{
"version": "v1",
"created": "Fri, 31 Jan 2025 22:17:12 GMT"
},
{
"version": "v2",
"created": "Mon, 26 May 2025 20:01:37 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Castañeda",
"Armando",
""
],
[
"Rodríguez",
"Gilde Valeria",
""
]
] |
2502.00267 | Jonathan R. Owens | Jonathan R. Owens and Bojun Feng and Jie Liu and David Moore | A combined statistical mechanical and ab initio approach to understanding H2O/CO2 co-adsorption in mmen-Mg2(dobpdc) | 9 pages, 4 figures | null | null | null | cond-mat.mtrl-sci physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | We study the effects of H2O on CO2 adsorption in an amine-appended variant of the metal-organic framework Mg2(dobpdc), which is known to exhibit chaining behavior that presents in a step-shaped adsorption isotherm. We first show how the presence of different levels of local H2O affects this chaining behavior and the energetics of CO2 adsorption, based on a series of ab initio calculations, giving insight into the atomic-scale environment. In particular, we predict a novel adsorbed configuration, in which H2O and CO2 intertwine to make a braided chain down the MOF pore. We then show how an existing lattice model can be adapted to incorporate the effect of water, and predict the CO2 isotherms for the various water levels, observing a sharp shift the uptake at low partial pressures. The manifestation of this braided chain in the lattice model points to the potential emergence of a shift from cooperative capture to that of a phase transition. In addition to the physical insights, this work may serve as a launching off point for further work on this and related materials. | [
{
"version": "v1",
"created": "Sat, 01 Feb 2025 01:47:01 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 11:21:50 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Owens",
"Jonathan R.",
""
],
[
"Feng",
"Bojun",
""
],
[
"Liu",
"Jie",
""
],
[
"Moore",
"David",
""
]
] |
2502.00431 | Thomas Christopoulos | Thomas Christopoulos, Johanne Hizanidis, Georgios Nousios, Emmanouil E. Kriezis, Odysseas Tsilipakos | Can 2D materials enable passively $Q$-switched lasers in the nanoscale? | null | null | null | null | physics.optics physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Achieving compact on-chip pulsed lasers with attractive performance metrics and compatibility with the silicon photonics platform is an important, yet elusive, goal in contemporary nanophotonics. Here, the fundamental question of whether 2D materials can be utilized as both gain and saturable absorption media to enable compact integrated passively Q-switched nanophotonic lasers is posed and addressed by examining a broad range of 2D material families. The study is conducted by developing a temporal coupled-mode theory framework involving semi-classical rate equations that is capable of rigorously handling gain and saturable absorption by 2D materials, allowing to perform stability and bifurcation analysis covering broad parameter spaces. The range of pulse-train metrics (repetition rate, pulse width, peak power) that can be obtained via different 2D materials is thoroughly assessed. Our work illustrates that nanophotonic cavities enhanced with 2D materials can enable passive Q-switching with repetition rates ranging up to 50~GHz, short pulse duration down to few picoseconds, and peak power exceeding several milliwatts. Such attractive metrics, along with the ultrathin nature of 2D materials and the ability to electrically tune their properties, demonstrate the potential of the proposed platform for compact and flexible integrated laser sources. | [
{
"version": "v1",
"created": "Sat, 01 Feb 2025 13:39:52 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 13:12:58 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Christopoulos",
"Thomas",
""
],
[
"Hizanidis",
"Johanne",
""
],
[
"Nousios",
"Georgios",
""
],
[
"Kriezis",
"Emmanouil E.",
""
],
[
"Tsilipakos",
"Odysseas",
""
]
] |
2502.01512 | Thibault De Surrel | Thibault de Surrel, Fabien Lotte, Sylvain Chevallier, Florian Yger | Wrapped Gaussian on the manifold of Symmetric Positive Definite Matrices | null | null | null | null | stat.ME cs.LG math.ST stat.ML stat.TH | http://creativecommons.org/licenses/by/4.0/ | Circular and non-flat data distributions are prevalent across diverse domains of data science, yet their specific geometric structures often remain underutilized in machine learning frameworks. A principled approach to accounting for the underlying geometry of such data is pivotal, particularly when extending statistical models, like the pervasive Gaussian distribution. In this work, we tackle those issue by focusing on the manifold of symmetric positive definite (SPD) matrices, a key focus in information geometry. We introduce a non-isotropic wrapped Gaussian by leveraging the exponential map, we derive theoretical properties of this distribution and propose a maximum likelihood framework for parameter estimation. Furthermore, we reinterpret established classifiers on SPD through a probabilistic lens and introduce new classifiers based on the wrapped Gaussian model. Experiments on synthetic and real-world datasets demonstrate the robustness and flexibility of this geometry-aware distribution, underscoring its potential to advance manifold-based data analysis. This work lays the groundwork for extending classical machine learning and statistical methods to more complex and structured data. | [
{
"version": "v1",
"created": "Mon, 03 Feb 2025 16:46:46 GMT"
},
{
"version": "v2",
"created": "Wed, 12 Feb 2025 18:11:46 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 15:27:04 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"de Surrel",
"Thibault",
""
],
[
"Lotte",
"Fabien",
""
],
[
"Chevallier",
"Sylvain",
""
],
[
"Yger",
"Florian",
""
]
] |
2502.01607 | Tilemachos Bolioudakis | Tilemachos Bolioudakis (1), Theodoros Diamantopoulos (1), Peter J. Diamessis (1), Ren-Chieh Lien (2), Kevin G. Lamb (3), Gustavo Rivera-Rosario (1) and Greg N. Thomsen (4) ((1) School of Civil and Environmental Engineering, Cornell University, Ithaca, New York, (2) Applied Physics Laboratory, University of Washington, Seattle, Washington, (3) Department of Applied Mathematics, University of Waterloo, Waterloo, Canada, (4) Wandering Wakhs Research, Austin, Texas) | Formation and evolution of turbulence in convectively unstable internal solitary waves of depression shoaling over gentle slopes in the South China Sea | 63 pages, 16 figures, submitted to Journal of Physical Oceanography | null | null | null | physics.ao-ph physics.flu-dyn | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The shoaling of high-amplitude Internal Solitary Waves (ISWs) of depression in the South China Sea (SCS) is examined through large-scale parallel turbulence-resolving high-accuracy/resolution simulations. A select, near-isobath-normal, bathymetric transect of the gentle SCS continental slope is employed together with stratification and current profiles obtained by in-situ measurements. Three simulations of separate ISWs with initial deep-water amplitudes in the range [136m, 150m] leverage a novel wave-tracking capability for a propagation distance of 80km and accurately reproduce key features of in-situ-observed phenomena with significantly higher spatiotemporal resolution. The interplay between convective and shear instability and the associated turbulence formation and evolution, as a function of deep-water ISW amplitude are further studied in-part revealing processes previously not observed in the field. Across all three waves, the convective instability develops in a similar fashion. Heavier water entrained from the wave rear plunges into its interior, giving rise to transient, yet distinct, subsurface vortical structures. Ultimately, a gravity current is triggered which horizontally advances through the wave interior and mixes it down to pycnocline's base. Although the waveform remains distinctly symmetric, Kelvin-Helmholtz billows emerge near the well-mixed ISW trough, disturb the wave's trailing edge and give rise to an active wake. The evolution of the kinetic energy associated with finer-scale perturbations to the ISW-induced velocity field shows two different growth regimes, each dominated by either convective or shear instability. The wake's perturbation kinetic energy is nonlinearly dependent on deep-water wave amplitude and can become a sizable fraction of the kinetic energy of the deep-water ISW. | [
{
"version": "v1",
"created": "Mon, 03 Feb 2025 18:42:17 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 16:03:15 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Bolioudakis",
"Tilemachos",
"",
"School of Civil and Environmental Engineering, Cornell University, Ithaca, New York"
],
[
"Diamantopoulos",
"Theodoros",
"",
"School of Civil and Environmental Engineering, Cornell University, Ithaca, New York"
],
[
"Diamessis",
"Peter J.",
"",
"School of Civil and Environmental Engineering, Cornell University, Ithaca, New York"
],
[
"Lien",
"Ren-Chieh",
"",
"Applied Physics Laboratory, University of Washington, Seattle, Washington"
],
[
"Lamb",
"Kevin G.",
"",
"Department of Applied Mathematics, University of Waterloo, Waterloo, Canada"
],
[
"Rivera-Rosario",
"Gustavo",
"",
"School of Civil and Environmental Engineering, Cornell University, Ithaca, New York"
],
[
"Thomsen",
"Greg N.",
"",
"Wandering Wakhs Research, Austin, Texas"
]
] |
2502.01734 | Suzanne Bintanja | Alexandre Belin, Suzanne Bintanja, Alejandra Castro, Waltraut Knop | Symmetric Product Orbifold Universality and the Mirage of an Emergent Spacetime | 51 pages, v2: minor changes, v3: minor changes, as published in JHEP | J. High Energ. Phys. 2025, 190 (2025) | 10.1007/JHEP05(2025)190 | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study thermal two-point functions and four-point functions involving two heavy twisted operators and two light probes in symmetric product orbifolds. We identify cases where they are universal at large $N$, that is, they are only sensitive to the orbifold structure. Surprisingly, such observables mimic correlators obtained from the BTZ background, even though symmetric product orbifolds are not dual to semi-classical gravity. We discuss the interpretation of these results in light of the criteria for emergence of spacetime via Von Neumann algebras. Our analysis implies that a condition on the infinite $N$ thermal two-point functions cannot be stringent enough to define an emergent spacetime and the concept of a sharp horizon. | [
{
"version": "v1",
"created": "Mon, 03 Feb 2025 19:00:01 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Feb 2025 19:38:12 GMT"
},
{
"version": "v3",
"created": "Tue, 27 May 2025 00:24:50 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Belin",
"Alexandre",
""
],
[
"Bintanja",
"Suzanne",
""
],
[
"Castro",
"Alejandra",
""
],
[
"Knop",
"Waltraut",
""
]
] |
2502.01740 | Roberto Soria | Roberto Soria, Thomas D. Russell, Eli Wiston, Siying Cheng, Raffaella Margutti, Kovi Rose, Stuart Ryder, Giacomo Terreran | The radio re-brightening of the Type IIb SN 2001ig | 14 pages, 3.3 MB. Accepted by PASA on 2025 Jan 29 | Publ. Astron. Soc. Aust. 42 (2025) e050 | 10.1017/pasa.2025.38 | null | astro-ph.HE astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We study the late-time evolution of the compact Type IIb SN 2001ig in the spiral galaxy NGC 7424, with new and unpublished archival data from the Australia Telescope Compact Array and the Australian Square Kilometre Array Pathfinder. More than two decades after the SN explosion, its radio luminosity is showing a substantial re-brightening: it is now two orders of magnitude brighter than expected from the standard model of a shock expanding into a uniform circumstellar wind (i.e., with a density scaling as R^{-2}). This suggests that the SN ejecta have reached a denser shell, perhaps compressed by the fast wind of the Wolf-Rayet progenitor or expelled centuries before the final stellar collapse. We model the system parameters (circumstellar density profile, shock velocity, mass loss rate), finding that the denser layer was encountered when the shock reached a distance of ~0.1 pc; the mass-loss rate of the progenitor immediately before the explosion was Mdot/v_w ~ 10^{-7} Msun/yr/(km/s). We compare SN 2001ig with other SNe that have shown late-time re-brightenings, and highlight the opposite behaviour of some extended Type IIb SNe which show instead a late-time flux cut-off. | [
{
"version": "v1",
"created": "Mon, 03 Feb 2025 19:00:02 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Soria",
"Roberto",
""
],
[
"Russell",
"Thomas D.",
""
],
[
"Wiston",
"Eli",
""
],
[
"Cheng",
"Siying",
""
],
[
"Margutti",
"Raffaella",
""
],
[
"Rose",
"Kovi",
""
],
[
"Ryder",
"Stuart",
""
],
[
"Terreran",
"Giacomo",
""
]
] |
2502.01792 | Haruka Kiyohara | Haruka Kiyohara, Fan Yao, Sarah Dean | Policy Design for Two-sided Platforms with Participation Dynamics | ICML2025 | null | null | null | cs.GT cs.IR cs.LG cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In two-sided platforms (e.g., video streaming or e-commerce), viewers and providers engage in interactive dynamics: viewers benefit from increases in provider populations, while providers benefit from increases in viewer population. Despite the importance of such "population effects" on long-term platform health, recommendation policies do not generally take the participation dynamics into account. This paper thus studies the dynamics and recommender policy design on two-sided platforms under the population effects for the first time. Our control- and game-theoretic findings warn against the use of the standard "myopic-greedy" policy and shed light on the importance of provider-side considerations (i.e., effectively distributing exposure among provider groups) to improve social welfare via population growth. We also present a simple algorithm to optimize long-term social welfare by taking the population effects into account, and demonstrate its effectiveness in synthetic and real-data experiments. Our experiment code is available at https://github.com/sdean-group/dynamics-two-sided-market. | [
{
"version": "v1",
"created": "Mon, 03 Feb 2025 20:12:01 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 02:26:09 GMT"
}
] | 2025-05-28T00:00:00 | [
[
"Kiyohara",
"Haruka",
""
],
[
"Yao",
"Fan",
""
],
[
"Dean",
"Sarah",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.