text
stringlengths 1
3.65k
| source
stringlengths 15
79
|
|---|---|
compressing neural network architectures is important to allow the deployment of models to embedded or mobile devices, and pruning and quantization are the major approaches to compress neural networks nowadays. both methods benefit when compression parameters are selected specifically for each layer. finding good combinations of compression parameters, so - called compression policies, is hard as the problem spans an exponentially large search space. effective compression policies consider the influence of the specific hardware architecture on the used compression methods. we propose an algorithmic framework called galen to search such policies using reinforcement learning utilizing pruning and quantization, thus providing automatic compression for neural networks. contrary to other approaches we use inference latency measured on the target hardware device as an optimization goal. with that, the framework supports the compression of models specific to a given hardware target. we validate our approach using three different reinforcement learning agents for pruning, quantization and joint pruning and quantization. besides proving the functionality of our approach we were able to compress a resnet18 for cifar - 10, on an embedded arm processor, to 20 % of the original inference latency without significant loss of accuracy. moreover, we can demonstrate that a joint search and compression using pruning and quantization is superior to an individual search for policies using a single compression method.
|
arxiv:2212.07818
|
we present synthetic line observations of a simulated molecular cloud, utilising a self - consistent treatment of the dynamics and time - dependent chemical evolution. we investigate line emission from the three most common co isotopologues ( $ ^ { 12 } $ co, $ ^ { 13 } $ co, c $ ^ { 18 } $ o ) and six supposed tracers of dense gas ( nh $ _ 3 $, hcn, n $ _ 2 $ h $ ^ + $, hco $ ^ + $, cs, hnc ). our simulation produces a range of line intensities consistent with that observed in real molecular clouds. the hcn - to - co intensity ratio is relatively invariant with column density, making hcn ( and chemically - similar species such as cs ) a poor tracer of high - density material in the cloud. the ratio of n $ _ 2 $ h $ ^ + $ to hcn or co, on the other hand, is highly selective of regions with densities above $ 10 ^ { 22 } \, { \ rm cm ^ { - 2 } } $, and the n $ _ 2 $ h $ ^ + $ line is a very good tracer of the dynamics of high volume density ( $ > 10 ^ 4 \, { \ rm cm ^ { - 3 } } $ ) material. focusing on cores formed within the simulated cloud, we find good agreement with the line intensities of an observational sample of prestellar cores, including reproducing observed cs line intensities with an undepleted elemental abundance of sulphur. however, agreement between cores formed in the simulation, and models of isolated cores which have otherwise - comparable properties, is poor. the formation from and interaction with the large - scale environment has a significant impact on the line emission properties of the cores, making isolated models unsuitable for interpreting observational data.
|
arxiv:2406.06702
|
" neither sherlock holmes nor babylon : a reassessment of plimpton 322 ". historia math. 28 ( 3 ) : 167 β 206. doi : 10. 1006 / hmat. 2001. 2317. mr 1849797. robson, e. ( 2002 ). " words and pictures : new light on plimpton 322 ". american mathematical monthly. 109 ( 2 ). washington : 105 β 120. doi : 10. 1080 / 00029890. 2002. 11919845. jstor 2695324. s2cid 33907668. robson, e. ( 2008 ). mathematics in ancient iraq : a social history. princeton university press. toomer, g. j. ( 1981 ). hipparchus and babylonian astronomy.
|
https://en.wikipedia.org/wiki/Babylonian_mathematics
|
it is shown that the supergravity moduli spaces of d1 - d5 and d2 - d6 brane systems coincide with those of the coulomb branches of the associated non - abelian gauge theories. we further discuss situations in which worldvolume brane actions include a potential term generated by probing certain supergravity backgrounds. we find that in many cases, the appearance of the potential is due to the application of the scherk - schwarz mechanism. we give some examples and discuss the existence of novel supersymmetric brane configurations.
|
arxiv:hep-th/0007083
|
we compute the three - graviton tree amplitude in type iib superstring theory compactified to six dimensions using the manifestly ( 6d ) supersymmetric berkovits - vafa - witten worldsheet variables. we consider two cases of background geometry : the flat space example r6xk3, and the curved example ads3xs3xk3 with ramond flux, and compute the correlation functions in the bulk.
|
arxiv:hep-th/0201027
|
the solar electron and proton telescope ( sept ) aboard the solar terrestrial relations observatory ( stereo ) is designed to provide the three - dimensional distribution of energetic electrons and protons with good energy and time resolution. each sept instrument consists of two double - ended magnet / foil particle telescopes which cleanly separate and measure electrons in the energy range from 30 kev to 400 kev and protons from 60 kev to 7000 kev. anisotropy information on a non spinning spacecraft is provided by two separate but identical instruments : sept - e aligned along the parker spiral magnetic field in the ecliptic plane along looking both towards and away from the sun, and sept - ns aligned vertical to the ecliptic plane looking towards north and south. the dual set - up refers to two adjacent sensor apertures for each of the four viewing directions sun, antisun, north, and south : one for protons, one for electrons. in this contribution a simulation of sept utilizing the geant4 toolkit has been set up with an extended instrument model in order to calculate improved response functions of the four different telescopes. this will help to understand and correct instrumental effects in the measurements.
|
arxiv:1801.09513
|
in many applications, it is needed to change the topology of a tensor network directly and without approximation. this work will introduce a general scheme that satisfies these needs. we will describe the procedure by two examples and show its efficiency in terms of memory consumption and speed in various numerical experiments. in general, we are going to provide an algorithm to add an edge to a tensor network as well as an algorithm to remove an edge unless the resulting network is a connected graph.
|
arxiv:1203.1503
|
we explore a new method to calculate the valence light - front wave function of a system of two interacting particles, which is based on contour deformations combined with analytic continuation methods to project the bethe - salpeter wave function onto the light front. in this proof - of - concept study, we solve the bethe - salpeter equation for a scalar model and find excellent agreement between the light - front wave functions obtained with contour deformations and those obtained with the nakanishi method frequently employed in the literature. the contour - deformation method is also able to handle extensions of the scalar model that mimic certain features of qcd such as unequal masses and complex singularities. in principle the method is suitable for computing parton distributions on the light front such as pdfs, tmds and gpds in the future.
|
arxiv:2112.04858
|
we study automorphisms and representations of quasi polynomial algebras ( qpas ) and quasi laurent polynomial algebras ( qlpas ). for any qlpa defined by an arbitrary skew symmetric integral matrix, we explicitly describe its automorphism groups at generic $ q $ and at roots of unity. any qlpa is isomorphic to the tensor product of copies of the qlpa of degree $ 2 $ at different powers of $ q $ and the centre, thus the study of representations of qpas and qlpas largely reduces to that of $ { \ mathcal l } _ q ( 2 ) $ and $ { \ mathcal a } _ q ( 2 ) $, the qlpa and qpa of degree $ 2 $. we study a category of $ { \ mathcal a } _ q ( 2 ) $ - modules which have finite covers by submodules with natural local finiteness properties and satisfy some condition under localisation, determining its blocks, classifying the simple objects and providing two explicitly constructions for the simples. one construction produces the simple $ { \ mathcal a } _ q ( 2 ) $ - modules from $ { \ mathcal l } _ q ( 2 ) $ - modules via monomorphisms composed of the natural embedding of $ { \ mathcal a } _ q ( 2 ) $ in $ { \ mathcal l } _ q ( 2 ) $ and automorphisms of $ { \ mathcal l } _ q ( 2 ) $, and the other explores a class of holonomic $ { \ mathcal d } _ q $ - modules for the algebra $ { \ mathcal d } _ q $ of $ q $ - differential operators.
|
arxiv:2203.00208
|
using optical orientation to manipulate magnetic moments in matter with light is a key objective in opto - spintronics, however, realizations of such control on ultrafast timescales are limited. here, we report ultrafast optical control of magnetic moment orientation in magnetically doped metal halide perovskites. employing intense pulses of circularly polarized light, we inject populations of spin - polarized charge carriers in pristine and manganese - doped mapbbr3 thin films. using transient faraday rotation spectroscopy, we probe the ultrafast magnetic moment dynamics following photoexcitation and find that light - induced magnetization in doped samples is increased by a factor of 10. we attribute this to photoexcited carriers acting on the magnetic moments of manganese dopant - ions via the sp - d exchange interaction, which forces them to align on picosecond timescales. our findings open new avenues for device structures that use hybrid metal halide perovskites for ultrafast optical manipulation and read - out of magnetic order with the potential for high switching rates.
|
arxiv:2406.19127
|
several simulations of turbulence in the large plasma device ( lapd ) [ w. gekelman et al., rev. sci. inst. 62, 2875 ( 1991 ) ] are energetically analyzed and compared with each other and with the experiment. the simulations use the same model, but different axial boundary conditions. they employ either periodic, zero - value, zero - derivative, or sheath axial boundaries. the linear stability physics is different between the scenarios because the various boundary conditions allow the drift wave instability to access different axial structures, and the sheath boundary simulation contains a conducting wall mode instability which is just as unstable as the drift waves. nevertheless, the turbulence in all the simulations is relatively similar because it is primarily driven by a robust nonlinear instability that is the same for all cases. the nonlinear instability preferentially drives $ k _ \ parallel = 0 $ potential energy fluctuations, which then three - wave couple to $ k _ \ parallel \ ne 0 $ potential energy fluctuations in order to access the adiabatic response to transfer their energy to kinetic energy fluctuations. the turbulence self - organizes to drive this nonlinear instability, which destroys the linear eigenmode structures, making the linear instabilities ineffective.
|
arxiv:1301.0536
|
nonsmooth nonconvex - concave minimax problems have attracted significant attention due to their wide applications in many fields. in this paper, we consider a class of nonsmooth nonconvex - concave minimax problems on riemannian manifolds. owing to the nonsmoothness of the objective function, existing minimax manifold optimization methods cannot be directly applied to solve this problem. we propose a manifold proximal gradient descent ascent ( mpgda ) algorithm for solving the problem. at each iteration, the proposed algorithm alternately performs one or multiple manifold proximal gradient descent steps and a proximal ascent step. we prove that the mpgda algorithm can find an $ \ varepsilon $ - game - stationary point and an $ \ varepsilon $ - optimization - stationary point of the considered problem within $ \ mathcal { o } ( \ varepsilon ^ { - 3 } ) $ iterations. numerical experiments on fair sparse pca and sparse spectral clustering are conducted to demonstrate the advantages of the mpgda algorithm.
|
arxiv:2505.02140
|
understanding the dynamics of truck volumes and activities across the skeleton traffic network is pivotal for effective traffic planning, traffic management, sustainability analysis, and policy making. yet, relying solely on average annual daily traffic volume for trucks cannot capture the temporal changes over time. recently, the traffic monitoring analysis system dataset has emerged as a valuable resource to model the system by providing information on an hourly basis for thousands of detectors across the united states. combining the average annual daily traffic volume from the highway performance monitoring system and the traffic monitoring analysis system dataset, this study proposes an elegant method of imputing information across the traffic network to generate both truck volumes and vehicle class distributions. a series of experiments evaluated the model ' s performance on various spatial and temporal scales. the method can be helpful as inputs for emission modeling, network resilience analysis, etc.
|
arxiv:2411.00789
|
despite significant progress in the theory and practice of program analysis, analysing properties of heap data has not reached the same level of maturity as the analysis of static and stack data. the spatial and temporal structure of stack and static data is well understood while that of heap data seems arbitrary and is unbounded. we devise bounded representations which summarize properties of the heap data. this summarization is based on the structure of the program which manipulates the heap. the resulting summary representations are certain kinds of graphs called access graphs. the boundedness of these representations and the monotonicity of the operations to manipulate them make it possible to compute them through data flow analysis. an important application which benefits from heap reference analysis is garbage collection, where currently liveness is conservatively approximated by reachability from program variables. as a consequence, current garbage collectors leave a lot of garbage uncollected, a fact which has been confirmed by several empirical studies. we propose the first ever end - to - end static analysis to distinguish live objects from reachable objects. we use this information to make dead objects unreachable by modifying the program. this application is interesting because it requires discovering data flow information representing complex semantics. in particular, we discover four properties of heap data : liveness, aliasing, availability, and anticipability. together, they cover all combinations of directions of analysis ( i. e. forward and backward ) and confluence of information ( i. e. union and intersection ). our analysis can also be used for plugging memory leaks in c / c + + languages.
|
arxiv:cs/0608104
|
starting from a recently proposed linear formulation in terms of auxiliary fields, we study $ n $ - field generalizations of born and born - infeld theories. in this description the lagrangian is quadratic in the vector field strengths and the symmetry properties ( including the characteristic self - duality ) of the corresponding non - linear theory are manifest as on - shell duality symmetries and depend on the choice of the ( homogeneous ) manifold spanned by the auxiliary scalar fields and the symplectic frame. by suitably choosing these defining properties of the quadratic lagrangian, we are able to reproduce some known multi - field born - infeld theories and to derive new non - linear models, such as the $ n $ - field born theory. we also discuss non - abelian generalizations of these theories obtained by choosing the vector fields in the adjoint representation of an off - shell compact global symmetry group $ k $ and replacing them by non - abelian, $ k $ - covariant field strengths, thus promoting $ k $ to a gauge group.
|
arxiv:1609.07399
|
recently, the static spherically symmetric solution of the gravitational field equations have been found in theories describing massive graviton with spontaneous breaking of the lorentz invariance. these solutions, which show off two integration constants instead of one in general relativity, are discussed. they are candidates for modified black holes provided they are stable against small perturbations. these solutions may have both attractive or repulsive behavior at large distances. therefore, these modified black holes may mimics the presence of dark matter or be a source of anti - gravity.
|
arxiv:0905.0819
|
in this study, we have shown autonomous long - term prediction with a spintronic physical reservoir. due to the short - term memory property of the magnetization dynamics, non - linearity arises in the reservoir states which could be used for long - term prediction tasks using simple linear regression for online training. during the prediction stage, the output is directly fed to the input of the reservoir for autonomous prediction. we employ our proposed reservoir for the modeling of the chaotic time series such as mackey - glass and dynamic time - series data, such as household building energy loads. since only the last layer of a rc needs to be trained with linear regression, it is well suited for learning in real time on edge devices. here we show that a skyrmion based magnetic tunnel junction can potentially be used as a prototypical rc but any nanomagnetic magnetic tunnel junction with nonlinear magnetization behavior can implement such a rc. by comparing our spintronic physical rc approach with energy load forecasting algorithms, such as lstms and rnns, we conclude that the proposed framework presents good performance in achieving high predictions accuracy, while also requiring low memory and energy both of which are at a premium in hardware resource and power constrained edge applications. further, the proposed approach is shown to require very small training datasets and at the same time being at least 16x energy efficient compared to the sequence to sequence lstm for accurate household load predictions.
|
arxiv:2304.03343
|
the two - dimensional homogeneous euclidean algorithm is the central motivation for the definition of the classical multidimensional continued fraction algorithms, as jacobi - perron, poincar \ ' e, brun and selmer algorithms. the rauzy induction, a generalization of the euclidean algorithm, is a key tool in the study of interval exchange transformations. both maps are known to be dissipative and ergodic with respect to lebesgue measure. here we prove that they are exact.
|
arxiv:1104.3980
|
physicists and philosophers are increasingly prone to regarding our current physical theories as providing ' effective descriptions ' of real - world systems. in the context of quantum gravity research, this fuels a common view that the classical spacetime theory of general relativity provides effective descriptions where it is successfully applied. that common view of general relativity, in turn, encourages an ' effective ' understanding of spacetime emergence. but descriptions of spacetime in general relativity irreducibly include global physical content, which is not effective. recognizing this fact reigns in the interpretive scope of the common view of general relativity and specifically undermines our thinking about spacetime emergence effectively.
|
arxiv:2410.00932
|
we study the effect of magnetism and perpendicular external electric field strengths on the energy gap of length confined bilayer graphene nanoribbons ( or nanoflakes ) as a function of ribbon width and length using a \ textit { first principles } density functional electronic structure method and a semi - local exchange - correlation approximation. we assume ab ( bernal ) bilayer stacking and consider both armchair and zigzag edges, and for each edge type, we consider the two edge alignments, namely, $ \ alpha $ and $ \ beta $ edge alignment. for the armchair nanoflakes we identify three distinct classes of bilayer energy gaps, determined by the number of carbon chains in the width direction ( { \ it n } = 3 { \ it p }, 3 { \ it p } + 1 and 3 { \ it p } + 2, { \ it p } is an integer ), and the gaps decrease with increasing width except for class 3 { \ it p } + 2 armchair nanoribbons. metallic - like behavior seen in armchair bilayer nanoribbons are found to be absent in armchair nanoflakes. class 3 { \ it p } + 2 armchair nanoflakes show significant length dependence. we find that the gaps decrease with the applied electric fields due to large intrinsic gap of the nanoflake. the existence of a critical gap with respect to the applied field, therefore, is not predicted by our calculations. magnetism between the layers plays a major role in enhancing the gap values resulting from the geometrical confinement, hinting at an interplay of magnetism and geometrical confinement in finite size bilayer graphene.
|
arxiv:0910.2719
|
trypanosoma cruzi and giardia intestinalis are two human pathogens and protozoan parasites responsible for the diseases chagas disease and giardiasis, respectively. both diseases cause suffering and illness in several million individuals. the former disease occurs primarily in south america and central america, and the latter disease occurs worldwide. current therapeutics are toxic and lack efficacy, and potential vaccines are far from the market. increased knowledge about the biology of these parasites is essential for drug and vaccine development, and new diagnostic tests. in this thesis, high - throughput sequencing was applied together with extensive bioinformatic analyses to yield insights into the biology and evolution of trypanosoma cruzi and giardia intestinalis. bioinformatics analysis of dna and rna sequences was performed to identify features that may be of importance for parasite biology and functional characterization. this thesis is based on five papers ( i - v ). paper i and ii describe comparative genome studies of three distinct genotypes of giardia intestinalis ( a, b and e ). paper iii describes a genome comparison of the human infecting trypanosoma cruzi with the bat - restricted subspecies trypanosoma cruzi marinkellei. paper iv describes the repertoire of small non - coding rnas in trypanosoma cruzi epimastigotes. paper v describes transcriptome analysis using paired - end rna - seq of three distinct genotypes of giardia intestinalis ( a, b and e ).
|
arxiv:1211.4210
|
a d5 elliptic fibration is a fibration whose generic fiber is modeled by the complete intersection of two quadric surfaces in p3. they provide simple examples of elliptic fibrations admitting a rich spectrum of singular fibers ( not all on the list of kodaira ) without introducing singularities in the total space of the fibration and therefore avoiding a discussion of their resolutions. we study systematically the fiber geometry of such fibrations using segre symbols and compute several topological invariants. we present for the first time sen ' s ( orientifold ) limits for d5 elliptic fibrations. these orientifolds limit describe different weak coupling limits of f - theory to type iib string theory giving a system of three brane - image - brane pairs in presence of a z _ 2 orientifold. the orientifold theory is mathematically described by the double cover the base of the elliptic fibration. such orientifold theories are characterized by a transition from a semi - stable singular fiber to an unstable one. in this paper, we describe the first example of a weak coupling limit in f - theory characterized by a transition to a non - kodaira ( and non - ade ) fiber. inspired by string dualities, we obtain non - trivial topological relations connecting the elliptic fibration and the different loci that appear in its weak coupling limit. mathematically, these are surprising relations relating the total chern class of the d5 elliptic fibration and those of different loci that naturally appear in the weak coupling limit. we work in arbitrary dimension and our results don ' t assume the calabi - yau condition.
|
arxiv:1110.6177
|
in this letter we consider the fluctuation induced force exerted between two plates separated by a distance $ l $ in a fluid with a temperature gradient. we predict that, for a range of distances $ l $, this non - equilibrium force is anomalously large compared to other casimir forces. the physical reason is that correlations in a non - equilibrium fluid are generally of longer range than other correlations, even than those near an equilibrium critical point. this giant casimir force is related to a divergent burnett coefficient that characterizes an onsager cross effect between the pressure and the temperature gradient. the predicted casimir force should be detectable with currently available experimental techniques.
|
arxiv:1302.4704
|
currently, there are a number of light - shining - through - walls experiments searching for hidden photons - - light, sub - ev - scale, abelian gauge bosons beyond the standard model which mix kinetically with the standard photon. we show that in the case that one of these experiments finds evidence for hidden photons, laser communications through matter, using methods from free - space optics, can be realized in the very near future, with a channel capacity of more than 1 bit per second, for a distance up to the earth ' s diamater.
|
arxiv:0903.5300
|
the hot intracluster plasma in clusters of galaxies is weakly magnetized. mergers between clusters produce gas compression and motions which can increase the magnetic field strength. in this work, we perform high - resolution non - radiative magnetohydrodynamics simulations of binary galaxy cluster mergers with magnetic fields, to examine the effects of these motions on the magnetic field configuration and strength, as well as the effect of the field on the gas itself. our simulations sample a parameter space of initial mass ratios and impact parameters. during the first core passage of mergers, the magnetic energy increases via gas compression. after this, shear flows produce temporary, mpc - scale, strong - field " filament " structures. lastly, magnetic fields grow stronger by turbulence. field amplification is most effective for low mass ratio mergers, but mergers with a large impact parameter can increase the magnetic energy more via shearing motions. the amplification of the magnetic field is most effective in between the first two core passages of each cluster merger. after the second core passage, the magnetic energy in this region gradually decreases. in general, the transfer of energy from gas motions to the magnetic field is not significant enough to have a substantial effect on gas mixing and the subsequent increase in entropy which occurs in cluster cores as a result. in the absence of radiative cooling, this results in an overall decrease of the magnetic field strength in cluster cores. in these regions, the final magnetic field is isotropic, while it can be significantly tangential at larger radii.
|
arxiv:1904.10024
|
torus orbifolds are topological generalization of symplectic toric orbifolds. we give a construction of smooth orbifolds with torus actions whose boundary is a disjoint union of torus orbifolds using toric topological method. as a result, we show that any orientable locally standard torus orbifold is equivariantly cobordant to some copies of orbifold complex projective spaces. we also discuss some further equivariant cobordism results including the cases when torus orbifolds are actually torus manifolds.
|
arxiv:1905.07924
|
in this letter, i present the motivation and an example analysis method for discovering the elastic production of top - quark pairs at the lhc using forward proton tags, including an overview of the current theoretical tools and experimental acceptance. i show that it is possible to discover the semi - elastic process with only 300 pb $ { ^ { - 1 } } $ of data but that the fully - elastic case is currently out of reach. i also illustrate how the use of forward proton tags can result in limits on the branching ratio for flavor changing neutral current decays of the top quark of the form $ t \ rightarrow u \ gamma $ and $ t \ rightarrow c \ gamma $ of $ < 0. 39 \ cdot10 ^ { - 5 } $ and $ < 0. 97 \ cdot10 ^ { - 5 } $, respectively, both of which would surpass the existing world limits by at least an order of magnitude.
|
arxiv:2008.04249
|
embryonic development involves pattern formation characterized by the emergence of spatially localized domains characterized by distinct cell fates resulting from differential gene expression. the boundaries demarcating these domains are precise and consistent within a species despite stochastic fluctuations in the morphogen molecular concentration that provides positional information to the cells, as well as, the intrinsic noise in molecular processes that interpret this information to guide fate determination. we show that local interactions between physically adjacent cells mediated by receptor - ligand binding utilizes the asymmetry between the fate - determining genes to yield a switch - like response to the global signal provided by the morphogen. this results in robust developmental outcomes with a consistent identity of the gene that is dominantly expressed at each cellular location, thereby substantially reducing the uncertainty in the location of the boundary between distinct fates.
|
arxiv:2110.12927
|
growing evidence shows that proactive content moderation supported by ai can help improve online discourse. however, we know little about designing these systems, how design impacts efficacy and user experience, and how people perceive proactive moderation across public and private platforms. we developed a mobile keyboard with built - in proactive content moderation which we tested ( n = 575 ) within a semi - functional simulation of a public and private communication platform. where toxic content was detected, we used different interventions that embedded three design factors : timing, friction, and the presentation of the ai model output. we found moderation to be effective, regardless of the design. however, friction was a source of annoyance while prompts with no friction that occurred during typing were more effective. follow - up interviews highlight the differences in how these systems are perceived across public and private platforms, and how they can offer more than moderation by acting as educational and communication support tools.
|
arxiv:2401.10627
|
a method for bounding the rate of bit - stuffing encoders for 2 - d constraints is presented. instead of considering the original encoder, we consider a related one which is quasi - stationary. we use the quasi - stationary property in order to formulate linear requirements that must hold on the probabilities of the constrained arrays that are generated by the encoder. these requirements are used as part of a linear program. the minimum and maximum of the linear program bound the rate of the encoder from below and from above, respectively. a lower bound on the rate of an encoder is also a lower bound on the capacity of the corresponding constraint. for some constraints, our results lead to tighter lower bounds than what was previously known.
|
arxiv:0906.2372
|
let $ x $ be the constrained random walk on $ { \ mathbb z } _ + ^ 2 $ with increments $ ( 1, 0 ) $, $ ( - 1, 0 ) $, $ ( 0, 1 ) $ and $ ( 0, - 1 ) $ ; $ x $ represents, at arrivals and service completions, the lengths of two queues working in parallel whose service and interarrival times are exponentially distributed with arrival rates $ \ lambda _ i $ and service rates $ \ mu _ i $, $ i = 1, 2 $ ; we assume $ \ lambda _ i < \ mu _ i $, $ i = 1, 2 $, i. e., $ x $ is assumed stable. without loss of generality we assume $ \ rho _ 1 = \ lambda _ 1 / \ mu _ 1 \ ge \ rho _ 2 = \ lambda _ 2 / \ mu _ 2 $. let $ \ tau _ n $ be the first time $ x $ hits the line $ \ partial a _ n = \ { x \ in { \ mathbb z } ^ 2 : x ( 1 ) + x ( 2 ) = n \ } $. let $ y $ be the same random walk as $ x $ but only constrained on $ \ { y \ in { \ mathbb z } ^ 2 : y ( 2 ) = 0 \ } $ and its jump probabilities for the first component reversed. let $ \ partial b = \ { y \ in { \ mathbb z } ^ 2 : y ( 1 ) = y ( 2 ) \ } $ and let $ \ tau $ be the first time $ y $ hits $ \ partial b $. the probability $ p _ n = p _ x ( \ tau _ n < \ tau _ 0 ) $ is a key performance measure of the queueing system represented by $ x $ ( probability of overflow of a shared buffer during system ' s first busy cycle ). stability of $ x $ implies $ p _ n $ decays exponentially in $ n $ when the process starts off $ \ partial a _ n. $ we show that, for $ x _ n = \ lfloor nx \ rfloor $, $ x \ in { \ mathbb r } _ + ^ 2 $, $ x ( 1 ) + x ( 2 ) \ le 1 $, $ x ( 1 ) > 0 $, $ p _ { ( n - x _
|
arxiv:1806.00686
|
in multimedia broadcasting, no - reference image quality assessment ( nr - iqa ) is used to indicate the user - perceived quality of experience ( qoe ) and to support intelligent data transmission while optimizing user experience. this paper proposes an improved no - reference light field image quality assessment ( nr - lfiqa ) metric for future immersive media broadcasting services. first, we extend the concept of depthwise separable convolution ( dsc ) to the spatial domain of light field image ( lfi ) and introduce " light field depthwise separable convolution ( lf - dsc ) ", which can extract the lfi ' s spatial features efficiently. second, we further theoretically extend the lf - dsc to the angular space of lfi and introduce the novel concept of " light field anglewise separable convolution ( lf - asc ) ", which is capable of extracting both the spatial and angular features for comprehensive quality assessment with low complexity. third, we define the spatial and angular feature estimations as auxiliary tasks in aiding the primary nr - lfiqa task by providing spatial and angular quality features as hints. to the best of our knowledge, this work is the first exploration of deep auxiliary learning with spatial - angular hints on nr - lfiqa. experiments were conducted in mainstream lfi datasets such as win5 - lid and smart with comparisons to the mainstream full reference iqa metrics as well as the state - of - the - art nr - lfiqa methods. the experimental results show that the proposed metric yields overall 42. 86 % and 45. 95 % smaller prediction errors than the second - best benchmarking metric in win5 - lid and smart, respectively. in some challenging cases with particular distortion types, the proposed metric can reduce the errors significantly by more than 60 %.
|
arxiv:2412.07079
|
we establish necessary optimality conditions for variational problems with a lagrangian depending on a combined caputo derivative of variable fractional order. the endpoint of the integral is free, and thus transversality conditions are proved. several particular cases are considered illustrating the new results.
|
arxiv:1501.02082
|
let m be a six dimensional manifold, endowed with a cohomogeneity one action of g = su _ 2 x su _ 2, and m _ reg its subset of regular points. we show that m _ reg admits a smooth, 2 - parameter family of g - invariant, non - isometric strict nearly kaehler structures and that a 1 - parameter subfamily of such structures smoothly extend over a singular orbit of type s ^ 3. this determines a new class of examples of nearly kaehler structures on ts ^ 3
|
arxiv:1011.4681
|
quantization of electrodynamics in curved space - time in the lorenz gauge and with arbitrary gauge parameter makes it necessary to study green functions of non - minimal operators with variable coefficients. starting from the integral representation of photon green functions, we link them to the evaluation of integrals involving gamma functions. eventually, the full asymptotic expansion of the feynman photon green function at small values of the world function, as well as its explicit dependence on the gauge parameter, are obtained without adding by hand a mass term to the faddeev - - popov lagrangian. coincidence limits of second covariant derivatives of the associated hadamard function are also evaluated, as a first step towards the energy - momentum tensor in the non - minimal case.
|
arxiv:hep-th/0310049
|
a proof of the quantum $ h $ - theorem taking into account nonextensive effects on the quantum entropy $ s ^ q _ q $ is shown. the positiveness of the time variation of $ s ^ q _ q $ combined with a duality transformation implies that the nonextensive parameter $ q $ lies in the interval [ 0, 2 ]. it is also shown that the equilibrium states are described by quantum $ q $ - power law extensions of the fermi - dirac and bose - einstein distributions. such results reduce to the standard ones in the extensive limit, thereby showing that the nonextensive entropic framework can be harmonized with the quantum distributions contained in the quantum statistics theory.
|
arxiv:0908.0965
|
. 0 skin on top. the phone comes with qualcomm snapdragon 8 + gen 1 chipset. = = = = phone ( 2a ) = = = = nothing announced its budget smartphone, the " phone ( 2a ) ", on 5 march 2024. it came with the mediatek dimensity 7200 pro chipset running android 14 and with user interface nothingos 2. 5, a 6. 7 - inch 120 - hz oled display, and a 5000 mah battery. phone ( 2a ) surpassed 100, 000 units sold in its first day after release. the nothing phone ( 2a ) plus, released in 2024, is an upgraded version of the standard phone ( 2a ) with enhanced features aimed at improving performance, display quality, and photography capabilities. key upgrades include a mediatek dimensity 7350 pro processor for smooth multitasking and better gaming performance, achieving impressive benchmark scores compared to its predecessors. = = = = phone ( 3a ) and phone ( 3a ) pro = = = = nothing phone ( 3a ) and nothing phone ( 3a ) pro were announced on march 4th 2025. both phones have amoled 6. 77 inches 120hz displays, both exist in 128gb and 256gb storage variants, both have the same 5000 mah battery and the same snapdragon 7s gen 3 ( 4 nm ) chip. the main difference is in camera type ( the pro version has telephoto with 3x optical zoom vs only 2x zoom on 3a variant ). = = = audio products = = = = = = = ear 1 = = = = the nothing ear 1, stylized as the " ear ( 1 ) ", is nothing ' s first product. announced on 27 july 2021, the ear 1 is a set of wireless earbuds. the earbuds can be connected by bluetooth and have up to 34 hours of battery life when used with the charging case, and up to 5. 7 hours of battery life with anc off ; with 24 hours with the case used and up to four hours for the earbuds themselves with anc on. the earbuds went on sale on 17 august 2021, at $ 99 / Β£99 / β¬99. a black version was also announced on 6 december 2021, and went on sale on 13 december. nothing also announced on that day that the ear 1 earbuds are now carbon neutral. on 18 october 2022, nothing ' s ceo carl pei announced on x that the ear
|
https://en.wikipedia.org/wiki/Nothing_(company)
|
at the molecular scale there are strong attractive interactions between surfaces, yet few macroscopic surfaces are sticky. extensive simulations of contact by adhesive surfaces with roughness on nanometer to micrometer scales are used to determine how roughness reduces the area where atoms contact and thus weakens adhesion. the material properties, adhesive strength and roughness parameters are varied by orders of magnitude. in all cases the area of atomic contact rises linearly with load, and the prefactor rises linearly with adhesive strength for weak interactions. above a threshold adhesive strength, the prefactor changes sign, the surfaces become sticky and a finite force is required to separate them. a parameter - free analytic theory is presented that describes changes in these numerical results over up to five orders of magnitude in load. it relates the threshold strength to roughness and material properties, explaining why most macroscopic surfaces do not stick. the numerical results are qualitatively and quantitatively inconsistent with classical theories based on the greenwood - williamson approach that neglect the range of adhesion and do not include asperity interactions.
|
arxiv:1311.1178
|
deep neural network ( dnn ) has achieved great success on datasets of closed class set. however, new classes, like new categories of social media topics, are continuously added to the real world, making it necessary to incrementally learn. this is hard for dnn because it tends to focus on fitting to new classes while ignoring old classes, a phenomenon known as catastrophic forgetting. state - of - the - art methods rely on knowledge distillation and data replay techniques but still have limitations. in this work, we analyze the causes of catastrophic forgetting in class incremental learning, which owes to three factors : representation drift, representation confusion, and classifier distortion. based on this view, we propose a two - stage learning framework with a fixed encoder and an incrementally updated prototype classifier. the encoder is trained with self - supervised learning to generate a feature space with high intrinsic dimensionality, thus improving its transferability and generality. the classifier incrementally learns new prototypes while retaining the prototypes of previously learned data, which is crucial in preserving the decision boundary. our method does not rely on preserved samples of old classes, is thus a non - exemplar based cil method. experiments on public datasets show that our method can significantly outperform state - of - the - art exemplar - based methods when they reserved 5 examplers per class, under the incremental setting of 10 phases, by 18. 24 % on cifar - 100 and 9. 37 % on imagenet100.
|
arxiv:2308.02346
|
event cameras are novel neuromorphic vision sensors with ultrahigh temporal resolution and low latency, both in the order of microseconds. instead of image frames, event cameras generate an asynchronous event stream of per - pixel intensity changes with precise timestamps. the resulting sparse data structure impedes applying many conventional computer vision techniques to event streams, and specific algorithms should be designed to leverage the information provided by event cameras. we propose a corner detection algorithm, esusan, inspired by the conventional susan ( smallest univalue segment assimilating nucleus ) algorithm for corner detection. the proposed esusan extracts the univalue segment assimilating nucleus from the circle kernel based on the similarity across timestamps and distinguishes corner events by the number of pixels in the nucleus area. moreover, esusan is fast enough to be applied to celex - v, the event camera with the highest resolution available. based on esusan, we also propose the se - harris corner detector, which uses adaptive normalization based on exponential decay to quickly construct a local surface of active events and the event - based harris detector to refine the corners identified by esusan. we evaluated the proposed algorithms on a public dataset and celex - v data. both esusan and se - harris exhibit higher real - time performance than existing algorithms while maintaining high accuracy and tracking performance.
|
arxiv:2105.00480
|
for every mapping of a perturbed spacetime onto a background and with any vector field $ \ xi $ we construct a conserved covariant vector density $ i ( \ xi ) $, which is the divergence of a covariant antisymmetric tensor density, a " superpotential ". $ i ( \ xi ) $ is linear in the energy - momentum tensor perturbations of matter, which may be large ; $ i ( \ xi ) $ does not contain the second order derivatives of the perturbed metric. the superpotential is identically zero when perturbations are absent. by integrating conserved vectors over a part $ \ si $ of a hypersurface $ s $ of the background, which spans a two - surface $ \ di \ si $, we obtain integral relations between, on the one hand, initial data of the perturbed metric components and the energy - momentum perturbations on $ \ si $ and, on the other hand, the boundary values on $ \ di \ si $. we show that there are as many such integral relations as there are different mappings, $ \ xi $ ' s, $ \ si $ ' s and $ \ di \ si $ ' s. for given boundary values on $ \ di \ si $, the integral relations may be interpreted as integral constraints ( e. g., those of traschen ) on local initial data including the energy - momentum perturbations. conservation laws expressed in terms of killing fields $ \ bar \ xi $ of the background become " physical " conservation laws. in cosmology, to each mapping of the time axis of a robertson - walker space on a de sitter space with the same spatial topology there correspond ten conservation laws. the conformal mapping leads to a straightforward generalization of conservation laws in flat spacetimes. other mappings are also considered....
|
arxiv:gr-qc/0504041
|
suppose that $ k $ is a field of characteristic 0, $ p $ is an odd prime, $ r $ a positive integer, $ q = p ^ r $ a prime power. suppose that $ f ( x ) $ is a polynomial of degree $ n > 4 $ with coefficients in $ k $ and without multiple roots. let us consider the superelliptic curve $ c : y ^ q = f ( x ) $ and its jacobian $ j ( c ) $. assuming that $ k $ is a subfield of the field of complex numbers, we study the ( connected reductive algebraic ) hodge group $ hdg $ of the corresponding complex abelian variety $ j ( c ) $. in our previous paper ( arxiv : 0907. 1563 [ math. ag ] ) we studied the center of $ hdg. in this paper we study the semisimple part ( commutator subgroup ) of $ hdg $. assuming that $ p $ does not divide $ n $ and $ n - 1 $ is not divisible by $ q $, the galois group of $ f ( x ) $ over $ k $ is either the full symmetric group $ s _ n $ or the alternating group $ a _ n $, we prove that the semisimple part of $ hdg $ is " as large as possible ".
|
arxiv:0910.2676
|
in this paper we explore the digital security experiences of marginalised populations in lebanon such as lgbtqi + identifying people, refugees and women. we situate our work in the post - conflict lebanese context, which is shaped by sectarian divides, failing governance and economic collapse. we do so through an ethnographically informed study conducted in beirut, lebanon, in july 2022 and through interviews with 13 people with lebanese digital and human rights expertise. our research highlights how lgbtqi + identifying people and refugees are scapegoated for the failings of the lebanese government, while women who speak out against such failings are silenced. we show how government - supported incitements of violence aimed at transferring blame from the political leadership to these groups lead to amplified digital security risks for already at - risk populations. positioning our work in broader sociological understandings of security, we discuss how the lebanese context impacts identity and ontological security. we conclude by proposing to design for and with positive security in post - conflict settings.
|
arxiv:2306.10149
|
the affine grassmannian of $ sl _ n $ admits an embedding into the sato grassmannian, which further admits a pl \ " ucker embedding into the projectivization of fermion fock space. kreiman, lakshmibai, magyar, and weyman describe the linear part of the ideal defining this embedding in terms of certain elements of the dual of fock space called " shuffles ", and they conjecture that these elements together with the pl \ " ucker relations suffice to cut out the affine grassmannian. we give a proof of this conjecture in two steps : first we reinterpret the shuffles equations in terms of frobenius twists of symmetric functions. using this, we reduce to a finite dimensional - problem, which we solve. for the second step we introduce a finite - dimensional analogue of the affine grassmannians of $ sl _ n $, which we conjecture to be precisely the reduced subscheme of a finite - dimensional grassmannian consisting of subspaces invariant under a nilpotent operator.
|
arxiv:1708.06076
|
we show that monolayer graphene intrinsically hosts higher - order topological corner states, in which electrons are localized topologically at atomic sizes. the emergence of the topological corner states in graphene is due to a nontrivial product of the zak phases for two independent directions, which can be handily calculated graphically by using the bulk wavefunctions. we give an explicit expression that indicates the existence of topological corner states for various geometric edges and corner angles. we also demonstrate the nontrivial localization nature of the topological corner states in graphene by putting an imaginary onsite potential mask.
|
arxiv:2009.08195
|
the network of contacts in space - filling disk packings, such as the apollonian packing, are examined. these networks provide an interesting example of spatial scale - free networks, where the topology reflects the broad distribution of disk areas. a wide variety of topological and spatial properties of these systems are characterized. their potential as models for networks of connected minima on energy landscapes is discussed.
|
arxiv:cond-mat/0407779
|
we study ordinary differential equations in the complex domain given by meromorphic vector fields on k \ " ahler compact complex surfaces. we prove that if such an equation has a maximal single valued solution with zariski - dense image ( in particular, if it has an entire one ) then, up to a bimeromorphic transformation, either the vector field is holomorphic or it preserves a fibration.
|
arxiv:1603.02288
|
a distinguishability operator is defined for the continuous - time quantum walk ( ctqw ) of a bipartite quantum walker on two simply connected graphs, $ w _ { g _ i, g _ j } = u _ { g _ i } \ left ( t \ right ) \ otimes u _ { g _ j } \ left ( t ' \ right ) - u _ { g _ j } \ left ( t ' \ right ) \ otimes u _ { g _ i } \ left ( t \ right ) $, where $ u _ { g _ i } \ left ( t \ right ) $ is the unitary ctqw operator for a labeled graph $ g _ i $ over a time interval $ t $. the null space of $ w _ { g _ i, g _ j } $ defines the vector space of initial bipartite states whose time development is either constant or only dependent on $ t + t ' $ and is invariant to which quantum walker subsystem goes with each graph. the set of null spaces corresponding with a set of $ w _ { g _ i, g _ j } $ have interesting relations as subspaces, intersections between subspaces, and subspaces of intersections. these relations are depicted as euler diagrams for labeled graphs of three and four vertices.
|
arxiv:1610.08421
|
the rapid growth in user generated content on social media has resulted in a significant rise in demand for automated content moderation. various methods and frameworks have been proposed for the tasks of hate speech detection and toxic comment classification. in this work, we combine common datasets to extend these tasks to brand safety. brand safety aims to protect commercial branding by identifying contexts where advertisements should not appear and covers not only toxicity, but also other potentially harmful content. as these datasets contain different label sets, we approach the overall problem as a binary classification task. we demonstrate the need for building brand safety specific datasets via the application of common toxicity detection datasets to a subset of brand safety and empirically analyze the effects of weighted sampling strategies in text classification.
|
arxiv:2303.15110
|
we revisit the sgoldstino interpretation of the diphoton excess in the context of gauge mediation. while the bound on the gluino mass might seem to make the sgoldstino contribution to the diphoton excess unobservable, we show that the interpretation is viable in a thin, near critical region of the parameter space. this regime gives rise to drastic departures from the standard gauge mediation picture. while the fermion messengers lie in the ( 10 - 100 ) tev range, some scalar messengers are significantly lighter and are responsible for the sgoldstino production and decay. their effective coupling to the sgoldstino is correspondingly enhanced, and a non - perturbative regime is triggered when light and heavy messenger masses differ by a factor $ \ sim4 \ pi $. we also comment on the possible role of an r - axion and on the possibility to decouple the sfermions in this context.
|
arxiv:1603.05682
|
in this note we consider high energy eigenfunctions of the harmonic oscillator in $ \ mathbb { r } ^ d $ and prove that any invariant measure on the energy surface can be written as a weak limit of eigenfunctions.
|
arxiv:1905.07763
|
we study proton - lead collisions with a new model for the fock states of the incoming proton. the number of collisions which the proton experiences selects the appropriate fock state of the proton which generates a multiple of pp - like rapidity distributions. we take as input the pp maximum entropy distributions shifting the respective cm - rapidities and reducing the available energies. a comparison with existing data at 5 tev is made and results for 8 tev are presented. we also explore the high multiplicity data in this model.
|
arxiv:1705.04533
|
we show how to apply ideas from the theory of rough paths to the analysis of low - regularity solutions to non - linear dispersive equations. our basic example will be the one dimensional korteweg - - de vries ( kdv ) equation on a periodic domain and with initial condition in $ \ ff l ^ { \ alpha, p } $ spaces. we discuss convergence of galerkin approximations, a modified euler scheme and the presence of a random force of white - noise type in time.
|
arxiv:math/0610006
|
both uncorrelated ( " sequential " ) and correlated ( " nonsequential " ) processes contribute to the double ionization of the helium atom in strong laser pulses. the double ionization probability has a characteristic " knee " shape as a function of the intensity of the pulse. we investigate the phase - space dynamics of this system, specifically by finding the dynamical structures that regulate the ionization processes. the emerging picture complements the recollision scenario by clarifying the distinct roles played by the recolliding and core electrons. our analysis leads to verifiable predictions of the intensities where qualitiative changes in ionization occur, leading to the hallmark " knee " shape.
|
arxiv:0905.0181
|
in this work, we consider a linear age - structured problem with diffusion and non - homogeneous boundary conditions both for the age and the space variables. we handle this linear problem by re - writing it as a non - densely defined abstract cauchy problem. to that aim we develop a new result on the closedness of a commutative sum of two non - densely defined operators by using the theory of integrated semigroups. as an application of this abstract result, we are able to associate a suitable integrated semigroup to some age - structured problem with spatial diffusion and equipped with non - homogeneous boundary conditions. this integrated semigroup is characterized by the description of its infinitesimal generator. further applications of our abstract result are also given to the commutative sum of two almost sectorial operators, for which we derive a closedness results.
|
arxiv:2105.07479
|
using v band photometry of the wings survey, we derive galaxy luminosity functions ( lf ) in nearby clusters. this sample is complete down to mv = - 15. 15, and it is homogeneous, thus allowing the study of an unbiased sample of clusters with different characteristics. we constructed the photometric lf for 72 out of the original 76 wings clusters, excluding only those without a velocity dispersion estimate. for each cluster we obtained the lf for galaxies in a region of radius = 0. 5 x r200, and fitted them with single and double schechter ' s functions. we also derive the composite lf for the entire sample, and those pertaining to different morphological classes. finally we derive the spectroscopic cumulative lf for 2009 galaxies that are cluster members. the double schechter fit parameters are neither correlated with the cluster velocity dispersion, nor with the x - ray luminosity. our median values of the schechter ' s fit slope are, on average, in agreement with measurements of nearby clusters, but are less steep that those derived from large surveys, such as the sdss. early - - type galaxies outnumber late - types at all magnitudes, but both early and late types contribute equally to the faint end of the lf. finally, the spectroscopic lf is in excellent agreement with the ones derived for a2199, a85 and virgo, and with the photometric one at the bright magnitudes ( where both are available ). there is a large spread in the lf of different clusters. however, this spread is not caused by correlation of the lf shape with cluster characteristics such as x - - ray luminosity or velocity dispersions. the faint end is flatter than what previously derived ( alpha _ f = - 1. 7 ) at odds with what predicted from numerical simulations.
|
arxiv:1506.01201
|
community detection is of great importance for understand - ing graph structure in social networks. the communities in real - world networks are often overlapped, i. e. some nodes may be a member of multiple clusters. how to uncover the overlapping communities / clusters in a complex network is a general problem in data mining of network data sets. in this paper, a novel algorithm to identify overlapping communi - ties in complex networks by a combination of an evidential modularity function, a spectral mapping method and evidential c - means clustering is devised. experimental results indicate that this detection approach can take advantage of the theory of belief functions, and preforms good both at detecting community structure and determining the appropri - ate number of clusters. moreover, the credal partition obtained by the proposed method could give us a deeper insight into the graph structure.
|
arxiv:1501.01780
|
a long - standing conjecture asserts that there exists a constant $ c > 0 $ such that every graph of order $ n $ without isolated vertices contains an induced subgraph of order at least $ cn $ with all degrees odd. scott ( 1992 ) proved that every graph $ g $ has an induced subgraph of order at least $ | v ( g ) | / ( 2 \ chi ( g ) ) $ with all degrees odd, where $ \ chi ( g ) $ is the chromatic number of $ g $, this implies the conjecture for graphs with { bounded } chromatic number. but the factor $ 1 / ( 2 \ chi ( g ) ) $ seems to be not best possible, for example, radcliffe and scott ( 1995 ) proved $ c = \ frac 23 $ for trees, berman, wang and wargo ( 1997 ) showed that $ c = \ frac 25 $ for graphs with maximum degree $ 3 $, so it is interesting to determine the exact value of $ c $ for special family of graphs. in this paper, we further confirm the conjecture for graphs with treewidth at most 2 with $ c = \ frac { 2 } { 5 } $, and the bound is best possible.
|
arxiv:1707.04812
|
a real anti - symmetric tensor field was introduced to realize a holographic magnetic ordered phase in our previous works. however, a more careful analysis shows there is a vector ghost in the model. in this paper we present a modified lagrangian density for the anti - symmetric tensor, which is ghost free and causality is well - defined, and keeps all the significant results in the original model qualitatively. we show this modified lagrangian density could come from the dimensional compactification of $ p $ - form field in string / m - theory. for static curved space - time, we also prove that this modified model is ghost free and dose not violate causality. this new model offers a solid foundation for the application of antisymmetric tensor field in holographic duality, especially for the spontaneous magnetization.
|
arxiv:1504.00855
|
we re - analyse the bu \ v { z } ek - hillery universal quantum cloning machine protocol and show that it allows better values for fidelity and hilbert - schmidt norm than hitherto reported. this higher value for the fidelity is identical to the maximum fidelity of phase covariant quantum cloning of bru \ ss - cinchetti - d ' ariano - macchiavello. this value of fidelity has also been obtained by niu and griffiths in their work without machine states. this is the maximum possible fidelity obtainable in $ 1 \ rightarrow 2 $ qubits cloning. we then describe a different and new state dependent cloning protocol with four machine states where all non - exact copies of input states are taken into account in the output and we use the hessian method of determining extrema of multivariate functions. the fidelity for the best overall quantum cloning in this protocol is $ \ bar { f } = 0. 847 $ with an associated von - neumann entropy of $ \ bar { s } = 0. 825 $.
|
arxiv:1405.5529
|
a decidability proof for bisimulation equivalence of first - order grammars is given. it is an alternative proof for a result by s \ ' enizergues ( 1998, 2005 ) that subsumes his affirmative solution of the famous decidability question for deterministic pushdown automata. the presented proof is conceptually simpler, and a particular novelty is that it is not given as two semidecision procedures but it provides an explicit algorithm that might be amenable to a complexity analysis.
|
arxiv:1812.03518
|
we present chemical abundances for 17 elements in a sample of 11 red giant branch stars in ngc 6362 from uves spectra. ngc 6362 is one of the least massive globulars where multiple populations have been detected, yet its detailed chemical composition has not been investigated so far. ngc 6362 turns out to be a metal - intermediate ( [ fe / h ] = - 1. 07 \ pm0. 01 dex ) cluster, with its \ alpha - and fe - peak elements content compatible with that observed in clusters with similar metallicity. it also displays an enhancement in its s - process element abundances. among the light elements involved in the multiple populations phenomenon, only [ na / fe ] shows star - to - star variations, while [ al / fe ] and [ mg / fe ] do not show any evidence for abundance spreads. a differential comparison with m4, a globular cluster with similar mass and metallicity, reveals that the two clusters share the same chemical composition. this finding suggests that ngc 6362 is indeed a regular cluster, formed from gas that has experienced the same chemical enrichment of other clusters with similar metallicity.
|
arxiv:1703.00385
|
key - value ( kv ) caching has emerged as a crucial optimization technique for accelerating inference in large language models ( llms ). by allowing the attention operation to scale linearly rather than quadratically with the total sequence length, kv caching significantly enhances generation throughput. however, due to large context lengths in the modern llms, the memory footprint of the kv is a huge bottleneck for model deployment directly impacting the model ' s batch size, hindering its ability to deliver high - throughput. existing research addresses this challenge using several techniques, such as discarding low - attention tokens, quantization, and matrix approximation which typically lead to a negative impact on the model accuracy. in this paper, we propose kvcrush technology which can be combined with many kv compression technologies to improve the model accuracy at a much smaller memory. kvcrush provides an alternate representation scheme for key - value states, along with a low - overhead token pruning algorithm that accounts for the token distribution in the kv cache, which in turn allows for a a smaller footprint while maintaining the accuracy of the model. based on our results, kvcrush reduces longbench kv cache size by 4x with less than 1 % accuracy drop and achieves state - of - the - art average accuracy with minimal overhead, incurring less than 0. 5 % total inference latency. kvcrush not only outperforms the accuracy of state - of - the - art importance - based token retention schemes but is also compatible with typical practical llm deployments using kv cache paging schemes such as vllm and mixed precision quantization.
|
arxiv:2503.00022
|
we investigate experimentally and theoretically the dynamics of a crack front during the micro - instabilities taking place in heterogeneous materials between two successive equilibrium positions. we focus specifically on the spatio - temporal evolution of the front, as it relaxes to a straight configuration, after depinning from a single obstacle of controlled strength and size. we show that this depinning dynamics is not controlled by inertia, but instead, by the rate dependency of the dissipative mechanisms taking place within the fracture process zone. this implies that the crack speed fluctuations around its average value $ v _ m $ can be predicted from an overdamped equation of motion $ ( v - v _ m ) / v _ 0 = ( g - g _ c ( v _ m ) ) / g _ c ( v _ m ) $ involving the characteristic material speed $ v _ 0 = g _ c ( v _ m ) / g _ c ' ( v _ m ) $ that emerges from the variation of fracture energy with crack speed. our findings pave the way to a quantitative description of the critical depinning dynamics of cracks in disordered solids and open up new perspectives for the prediction of the effective failure properties of heterogeneous materials.
|
arxiv:1807.11302
|
we report on unusually very hard spectral states in three confirmed neutron - star low - mass x - ray binaries ( 1rxs j180408. 9 - 342058, exo 1745 - 248, and igr j18245 - 2452 ) at a luminosity between ~ 10 ^ { 36 - 37 } erg s ^ { - 1 }. when fitting the swift x - ray spectra ( 0. 5 - 10 kev ) in those states with an absorbed power - law model, we found photon indices of \ gamma ~ 1, significantly lower than the \ gamma = 1. 5 - 2. 0 typically seen when such systems are in their so called hard state. for individual sources very hard spectra were already previously identified but here we show for the first time that likely our sources were in a distinct spectral state ( i. e., different from the hard state ) when they exhibited such very hard spectra. it is unclear how such very hard spectra can be formed ; if the emission mechanism is similar to that operating in their hard states ( i. e., up - scattering of soft photons due to hot electrons ) then the electrons should have higher temperatures or a higher optical depth in the very hard state compared to those observed in the hard state. by using our obtained \ gamma as a tracer for the spectral evolution with luminosity, we have compared our results with those obtained by wijnands et al. ( 2015 ). we confirm their general results in that also our sample of sources follow the same track as the other neutron star systems, although we do not find that the accreting millisecond pulsars are systematically harder than the non - pulsating systems.
|
arxiv:1703.09497
|
the integration of artificial intelligence ( ai ) into the workplace is advancing rapidly, necessitating robust metrics to evaluate its tangible impact on the labour market. existing measures of ai occupational exposure largely focus on ai ' s theoretical potential to substitute or complement human labour on the basis of technical feasibility, providing limited insight into actual adoption and offering inadequate guidance for policymakers. to address this gap, we introduce the ai startup exposure ( aise ) index - a novel metric based on occupational descriptions from o * net and ai applications developed by startups funded by the y combinator accelerator. our findings indicate that while high - skilled professions are theoretically highly exposed according to conventional metrics, they are heterogeneously targeted by startups. roles involving routine organizational tasks - such as data analysis and office management - display significant exposure, while occupations involving tasks that are less amenable to ai automation due to ethical or high - stakes, more than feasibility, considerations - - such as judges or surgeons - - present lower aise scores. by focusing on venture - backed ai applications, our approach offers a nuanced perspective on how ai is reshaping the labour market. it challenges the conventional assumption that high - skilled jobs uniformly face high ai risks, highlighting instead the role of today ' s ai players ' societal desirability - driven and market - oriented choices as critical determinants of ai exposure. contrary to fears of widespread job displacement, our findings suggest that ai adoption will be gradual and shaped by social factors as much as by the technical feasibility of ai applications. this framework provides a dynamic, forward - looking tool for policymakers and stakeholders to monitor ai ' s evolving impact and navigate the changing labour landscape.
|
arxiv:2412.04924
|
plasmonic phenomena are exhibited in light - matter interaction involving materials whose real parts of permittivity functions attain negative values at operating wavelengths. however, such materials usually suffer from dissipative losses, thus limiting the performance of plasmon - based optical devices. here, we utilize an alternative methodology that mimics a variety of plasmonic phenomena by exploiting the well - known structural dispersion of electromagnetic modes in bounded guided - wave structures filled with only materials with positive permittivity. a key issue in design of such structures is prevention of mode coupling, which can be achieved by implementing thin metallic wires at proper interfaces. this method, which is more suitable for lower frequencies, allows designers to employ conventional dielectrics and highly conductive metals for which the loss is low at these frequencies, while achieving plasmonic features. we demonstrate, numerically and analytically, that this platform can provide surface plasmon polaritons, local plasmonic resonance, plasmonic cloaking and epsilon - near - zero ( enz ) - based tunneling using conventional positive - dielectric materials.
|
arxiv:1512.01066
|
we introduce a nonlocal control condition and the notion of approximate controllability for fractional order quasilinear control inclusions. approximate controllability of a fractional control nonlocal delay quasilinear functional differential inclusion in a hilbert space is studied. the results are obtained by using the fractional power of operators, multi - valued analysis, and sadovskii ' s fixed point theorem. main result gives an appropriate set of sufficient conditions for the considered system to be approximately controllable. as an example, a fractional partial nonlocal control functional differential inclusion is considered.
|
arxiv:1405.6591
|
neural machine translation ( nmt ) models are sensitive to small perturbations in the input. robustness to such perturbations is typically measured using translation quality metrics such as bleu on the noisy input. this paper proposes additional metrics which measure the relative degradation and changes in translation when small perturbations are added to the input. we focus on a class of models employing subword regularization to address robustness and perform extensive evaluations of these models using the robustness measures proposed. results show that our proposed metrics reveal a clear trend of improved robustness to perturbations when subword regularization methods are used.
|
arxiv:2005.00580
|
we continue our study of ends of non - compact manifolds, with a focus on the inward tameness condition. for manifolds with compact boundary, inward tameness, has significant implications. for example, such manifolds have stable homology at infinity in all dimensions. we show that these manifolds have ' almost perfectly semistable ' fundamental group at each end. that observation leads to further analysis of group theoretic conditions at infinity, and to the notion of a ' near pseudo - collar ' structure. we obtain a complete characterization of n - manifolds ( n > 5 ) admitting such a structure, thereby generalizing earlier work. we also construct examples illustrating the necessity and usefulness of new conditions introduced here. variations on the notion of a perfect group, with corresponding versions of the quillen plus construction, form an underlying theme.
|
arxiv:1509.00918
|
deep visual odometry, despite extensive research, still faces limitations in accuracy and generalizability that prevent its broader application. to address these challenges, we propose an oriented fast and rotated brief ( orb ) - guided visual odometry with selective online adaptation named orb - sfmlearner. we present a novel use of orb features for learning - based ego - motion estimation, leading to more robust and accurate results. we also introduce the cross - attention mechanism to enhance the explainability of posenet and have revealed that driving direction of the vehicle can be explained through the attention weights. to improve generalizability, our selective online adaptation allows the network to rapidly and selectively adjust to the optimal parameters across different domains. experimental results on kitti and vkitti datasets show that our method outperforms previous state - of - the - art deep visual odometry methods in terms of ego - motion accuracy and generalizability.
|
arxiv:2409.11692
|
in any periodic direction on the regular pentagon billiard table, there exists two combinatorially different billiard paths, with one longer than the other. for each periodic direction, mcmullen asked if one could determine whether the periodic trajectory through a given point is long, short, or a saddle connection. in this paper we present an algorithm resolving this question for trajectories emanating from the midpoints of the pentagon.
|
arxiv:2111.09856
|
a semiclassical approach for calculating shell effects, that has been used in atomic and plasma physics, is applied to describe the electronic supershells in metal clusters. using the spherical jellium model we give the analytical expression for the oscillating part of the binding energy of electrons as an explicit sum of contributions from supershells with quantum numbers 2n _ r + l, 3n _ r + l, 4n _ r + l,... this expression is written in terms of the classical characteristics of the motion of an electron with the fermi energy in a self - consistent potential. the conditions under which a new supershell appears and the relative contribution of this shell are studied as a function of the cluster size and form of the potential. specific calculations are performed for a square well.
|
arxiv:cond-mat/0001116
|
optimal locally repairable codes with information locality are considered. optimal codes are constructed, whose length is also order - optimal with respect to a new bound on the code length derived in this paper. the length of the constructed codes is super - linear in the alphabet size, which improves upon the well known pyramid codes, whose length is only linear in the alphabet size. the recoverable erasure patterns are also analyzed for the new codes. based on the recoverable erasure patterns, we construct generalized sector - disk ( gsd ) codes, which can recover from disk erasures mixed with sector erasures in a more general setting than known sector - disk ( sd ) codes. additionally, the number of sectors in the constructed gsd codes is super - linear in the alphabet size, compared with known sd codes, whose number of sectors is only linear in the alphabet size.
|
arxiv:2001.02909
|
this paper presents a new array response control scheme named complex - coefficient weight vector orthogonal decomposition ( $ \ textrm { c } ^ 2 \ textrm { - word } $ ) and its application to pattern synthesis. the proposed $ \ textrm { c } ^ 2 \ textrm { - word } $ algorithm is a modified version of the existing word approach. we extend word by allowing a complex - valued combining coefficient in $ \ textrm { c } ^ 2 \ textrm { - word } $, and find the optimal combining coefficient by maximizing white noise gain ( wng ). our algorithm offers a closed - from expression to precisely control the array response level of a given point starting from an arbitrarily - specified weight vector. in addition, it results less pattern variations on the uncontrolled angles. elaborate analysis shows that the proposed $ \ textrm { c } ^ 2 \ textrm { - word } $ scheme performs at least as good as the state - of - the - art $ \ textrm { a } ^ \ textrm { 2 } \ textrm { rc } $ or word approach. by applying $ \ textrm { c } ^ 2 \ textrm { - word } $ successively, we present a flexible and effective approach to pattern synthesis. numerical examples are provided to demonstrate the flexibility and effectiveness of $ \ textrm { c } ^ 2 \ textrm { - word } $ in array response control as well as pattern synthesis.
|
arxiv:1807.06716
|
information diffusion and virus propagation are fundamental processes taking place in networks. while it is often possible to directly observe when nodes become infected with a virus or adopt the information, observing individual transmissions ( i. e., who infects whom, or who influences whom ) is typically very difficult. furthermore, in many applications, the underlying network over which the diffusions and propagations spread is actually unobserved. we tackle these challenges by developing a method for tracing paths of diffusion and influence through networks and inferring the networks over which contagions propagate. given the times when nodes adopt pieces of information or become infected, we identify the optimal network that best explains the observed infection times. since the optimization problem is np - hard to solve exactly, we develop an efficient approximation algorithm that scales to large datasets and finds provably near - optimal networks. we demonstrate the effectiveness of our approach by tracing information diffusion in a set of 170 million blogs and news articles over a one year period to infer how information flows through the online media space. we find that the diffusion network of news for the top 1, 000 media sites and blogs tends to have a core - periphery structure with a small set of core media sites that diffuse information to the rest of the web. these sites tend to have stable circles of influence with more general news media sites acting as connectors between them.
|
arxiv:1006.0234
|
we consider the \ emph { focusing } nonlinear schr \ " odinger equation posed on the one dimensional line, with nonzero background condition at spatial infinity, given by a homogeneous plane wave. for this problem of physical interest, we study the initial value problem for perturbations of the background wave in sobolev spaces. it is well - known that the associated linear dynamics for this problem describes a phenomenon known in the literature as \ emph { modulational instability }, also recently related to the emergence of \ emph { rogue waves } in ocean dynamics. in qualitative terms, small perturbations of the background state increase its size exponentially in time. in this paper we show that, even if there is no time decay for the linear dynamics due to the modulationally unstable regime, the equation is still locally well - posed in $ h ^ s $, $ s > \ frac12 $. we apply this result to give a rigorous proof of the unstable character of two well - known nls solutions : the peregrine and kuznetsov - ma breathers.
|
arxiv:1608.08169
|
in this paper we show that evolution algebras over any given field $ \ bbbk $ are universally finite. in other words, given any finite group $ g $, there exist infinitely many regular evolution algebras $ x $ such that $ aut ( x ) \ cong g $. the proof is built upon the construction of a covariant faithful functor from the category of finite simple ( non oriented ) graphs to the category of ( finite dimensional ) regular evolution algebras. finally, we show that any constant finite algebraic affine group scheme $ \ mathbf { g } $ over $ \ bbbk $ is isomorphic to the algebraic affine group scheme of automorphisms of a regular evolution algebra.
|
arxiv:2002.03338
|
existence of a local trivialization. = = = maps between spheres = = = a large class of examples of submersions are submersions between spheres of higher dimension, such as f : s n + k β s k { \ displaystyle f : s ^ { n + k } \ to s ^ { k } } whose fibers have dimension n { \ displaystyle n }. this is because the fibers ( inverse images of elements p i n s k { \ displaystyle pins ^ { k } } ) are smooth manifolds of dimension n { \ displaystyle n }. then, if we take a path Ξ³ : i β s k { \ displaystyle \ gamma : i \ to s ^ { k } } and take the pullback m i β s n + k β β f i x β Ξ³ s k { \ displaystyle { \ begin { matrix } m _ { i } & \ to & s ^ { n + k } \ \ \ downarrow & & \ downarrow f \ \ i & x \ rightarrow { \ gamma } & s ^ { k } \ end { matrix } } } we get an example of a special kind of bordism, called a framed bordism. in fact, the framed cobordism groups Ο n f r { \ displaystyle \ omega _ { n } ^ { fr } } are intimately related to the stable homotopy groups. = = = families of algebraic varieties = = = another large class of submersions is given by families of algebraic varieties Ο : x β s { \ displaystyle \ pi : { \ mathfrak { x } } \ to s } whose fibers are smooth algebraic varieties. if we consider the underlying manifolds of these varieties, we get smooth manifolds. for example, the weierstrass family Ο : w t o a 1 { \ displaystyle \ pi : { \ mathcal { w } } to \ mathbb { a } ^ { 1 } } of elliptic curves is a widely studied submersion because it includes many technical complexities used to demonstrate more complex theory, such as intersection homology and perverse sheaves. this family is given by w = { ( t, x, y ) β a 1 Γ a 2 : y 2 = x ( x β 1 ) ( x β t ) } { \ displaystyle { \ mathcal { w } } = \ left \ { ( t, x, y ) \ in \ mathbb {
|
https://en.wikipedia.org/wiki/Submersion_(mathematics)
|
most published information systems research are of the behavioral science research ( bsr ) category rather than the design science research ( dsr ) category. this is due in part to the bsr orientation of many is doctoral programs, which often do not involve much technical courses. this includes is doctoral programs that train information and communication technologies for development ( ict4d ) researchers. without such technical knowledge many doctoral and postdoctoral researchers will not feel confident in engaging in dsr research. given the importance of designing artifacts that are appropriate for a given context, an important question is how can ict4d and other is researchers increase their is technical content knowledge and intimacy with the dsr process. in this paper we present, a process for reviewing dsr papers that has as its objectives : enhancing technical content knowledge, increasing knowledge and understanding of approaches to designing and evaluating is / it artifacts, and facilitating the identification of new dsr opportunities. this process has been applied for more than a decade at a usa research university.
|
arxiv:2408.07230
|
view augustine ' s scriptural commentaries as endorsing any particular cosmological model, endorsing instead the view that augustine shared the common view of his contemporaries that the earth is spherical, in line with his endorsement of science in de genesi ad litteram. c. p. e. nothaft, responding to writers like leo ferrari who described augustine as endorsing a flat earth, says that "... other recent writers on the subject treat augustine ' s acceptance of the earth ' s spherical shape as a well - established fact ". while it always remained a minority view, from the mid - fourth to the seventh centuries ad, the flat - earth view experienced a revival, around the time when diodorus of tarsus founded the exegetical school known as the school of antioch, which sought to counter what he saw as the pagan cosmology of the greeks with a return to the traditional cosmology. the writings of diodorus did not survive, but are reconstructed from later criticism. this revival primarily took place in the east syriac world ( with little influence on the latin west ) where it gained proponents such as ephrem the syrian and in the popular hexaemeral homilies of jacob of serugh. chrysostom, one of the four great church fathers of the eastern church and archbishop of constantinople, explicitly espoused the idea, based on scripture, that the earth floats miraculously on the water beneath the firmament. christian topography ( 547 ) by the alexandrian monk cosmas indicopleustes, who had traveled as far as sri lanka and the source of the blue nile, is now widely considered the most valuable geographical document of the early medieval age, although it received relatively little attention from contemporaries. in it, the author repeatedly expounds the doctrine that the universe consists of only two places, the earth below the firmament and heaven above it. carefully drawing on arguments from scripture, he describes the earth as a rectangle, 400 days ' journey long by 200 wide, surrounded by four oceans and enclosed by four massive walls which support the firmament. the spherical earth theory is contemptuously dismissed as " pagan ". severian, bishop of gabala ( d. 408 ), wrote that the earth is flat and the sun does not pass under it in the night, but " travels through the northern parts as if hidden by a wall ". basil of caesarea ( 329 β 379 ) argued that the matter was theologically irrelevant. =
|
https://en.wikipedia.org/wiki/Flat_Earth
|
although the conditional variational autoencoder ( cvae ) model can generate more diversified responses than the traditional seq2seq model, the responses often have low relevance with the input words or are illogical with the question. a causal analysis is carried out to study the reasons behind, and a methodology of searching for the mediators and mitigating the confounding bias in dialogues is provided. specifically, we propose to predict the mediators to preserve relevant information and auto - regressively incorporate the mediators into generating process. besides, a dynamic topic graph guided conditional variational autoencoder ( tgg - cvae ) model is utilized to complement the semantic space and reduce the confounding bias in responses. extensive experiments demonstrate that the proposed model is able to generate both relevant and informative responses, and outperforms the state - of - the - art in terms of automatic metrics and human evaluations.
|
arxiv:2209.09482
|
we study the formation of a sunspot penumbra in the active region noaa11024. we simultaneously observed the stokes parameters of the photospheric iron lines at 1089. 6 nm with the tip and 617. 3 nm with the gfpi spectropolarimeters along with broad - band images using g - band and caiik filters at the german vtt. the formation of the penumbra is intimately related to the inclined magnetic field. within 4. 5 h observing time, the magnetic flux of the penumbra increases from 9. 7e + 20 to 18. 2e + 20 mx, while the magnetic flux of the umbra remains constant at about 3. 8e + 20 mx. magnetic flux in the immediate surroundings is incorporated into the spot, and new flux is supplied via small flux patches ( sfps ), which on average have a flux of 2 - 3e + 18 mx. the spot ' s flux increase rate of 4. 2e + 16 mx / s corresponds to the merging of one sfp per minute. we also find that during the formation of the spot penumbra : a ) the maximum magnetic field strength of the umbra does not change, b ) the magnetic neutral line keeps the same position relative to the umbra, c ) the new flux arrives on the emergence side of the spot while the penumbra forms on the opposite side, d ) the average lrf inclination of the light bridges decreases from 50 to 37 deg, and e ) as the penumbra develops, the mean magnetic field strength at the spot border decreases from 1. 0 to 0. 8 kg. the sfps associated with elongated granules are the building blocks of structure formation in active regions. during the sunspot formation, their contribution is comparable to the coalescence of pores. a quiet environment in the surroundings is important for penumbral formation. as remnants of trapped granulation between merging pores, the light bridges are found to play a crucial role in the formation process. they seem to channel the magnetic flux through the spot during its formation. light bridges are also the locations where the first penumbral filaments form.
|
arxiv:1111.3189
|
canonical correlation analysis ( cca ) describes the associations between two sets of variables by maximizing the correlation between linear combinations of the variables in each data set. however, in high - dimensional settings where the number of variables exceeds the sample size or when the variables are highly correlated, traditional cca is no longer appropriate. this paper proposes a method for sparse cca. sparse estimation produces linear combinations of only a subset of variables from each data set, thereby increasing the interpretability of the canonical variates. we consider the cca problem from a predictive point of view and recast it into a regression framework. by combining an alternating regression approach together with a lasso penalty, we induce sparsity in the canonical vectors. we compare the performance with other sparse cca techniques in different simulation settings and illustrate its usefulness on a genomic data set.
|
arxiv:1501.01231
|
we present a data - driven analysis of the resonant s - wave $ \ pi \ pi \ to \ pi \ pi $ and $ \ pi k \ to \ pi k $ reactions using the partial - wave dispersion relation. the contributions from the left - hand cuts are accounted for using the taylor expansion in a suitably constructed conformal variable. the fits are performed to experimental and lattice data as well as roy analyses. for the $ \ pi \ pi $ scattering we present both a single - and coupled - channel analysis by including additionally the $ k \ bar { k } $ channel. for the latter the central result is the omn \ ` es matrix, which is consistent with the most recent roy and roy - steiner results on $ \ pi \ pi \ to \ pi \ pi $ and $ \ pi \ pi \ to k \ bar { k } $, respectively. by the analytic continuation to the complex plane, we found poles associated with the lightest scalar resonances $ \ sigma / f _ 0 ( 500 ) $, $ f _ 0 ( 980 ) $, and $ \ kappa / k _ 0 ^ * ( 700 ) $ for the physical pion mass value and in the case of $ \ sigma / f _ 0 ( 500 ) $, $ \ kappa / k _ 0 ^ * ( 700 ) $ also for unphysical pion mass values.
|
arxiv:2012.11636
|
the goal of this paper is text - independent speaker verification where utterances come from ' in the wild ' videos and may contain irrelevant signal. while speaker verification is naturally a pair - wise problem, existing methods to produce the speaker embeddings are instance - wise. in this paper, we propose cross attentive pooling ( cap ) that utilizes the context information across the reference - query pair to generate utterance - level embeddings that contain the most discriminative information for the pair - wise matching problem. experiments are performed on the voxceleb dataset in which our method outperforms comparable pooling strategies.
|
arxiv:2008.05983
|
location data is inherently uncertain for many reasons including 1 ) imprecise location measurements, 2 ) obsolete observations that are often interpolated, and 3 ) deliberate obfuscation to preserve location privacy. what makes handling uncertainty data challenging is the exponentially large number of possible worlds, which lies in o ( 2 ^ n ), for a database having n uncertain objects as it has been shown that general query processing in uncertain spatial data is np - hard. many applications using spatial data require counting the number of spatial objects within a region. an example is the k - nearest neighbor ( knn ) query : asking if an object a is a knn of another object q is equivalent to asking whether no more than k - 1 objects are located inside the circle centered at q having a radius equal to the distance between q and a. for this problem of counting uncertain objects within a region, an efficient solution based on generating functions has been proposed and successfully used in many applications, including range - count queries, knn queries, distance ranking queries, and reverse knn queries. this spatial gem describes the generating function technique for probabilistic counting and provides examples and implementation details.
|
arxiv:2112.06344
|
based on the assumption that there exists a neural network that efficiently represents a set of boolean functions between all binary inputs and outputs, we propose a process for developing and deploying neural networks whose weight parameters, bias terms, input, and intermediate hidden layer output signals, are all binary - valued, and require only basic bit logic for the feedforward pass. the proposed bitwise neural network ( bnn ) is especially suitable for resource - constrained environments, since it replaces either floating or fixed - point arithmetic with significantly more efficient bitwise operations. hence, the bnn requires for less spatial complexity, less memory bandwidth, and less power consumption in hardware. in order to design such networks, we propose to add a few training schemes, such as weight compression and noisy backpropagation, which result in a bitwise network that performs almost as well as its corresponding real - valued network. we test the proposed network on the mnist dataset, represented using binary features, and show that bnns result in competitive performance while offering dramatic computational savings.
|
arxiv:1601.06071
|
bosons carrying a conserved charge can form stable bound states if their lagrangian contains attractive self - interactions. bound - state configurations with a large charge $ q $ can be described classically and are denoted as q - balls, their properties encoded in a non - linear differential equation. here, we study q - balls in arbitrary polynomial single - scalar - field potentials both numerically and via various analytical approximations. we highlight some surprising universal features of q - balls that barely depend on the details of the potential. the polynomial potentials studied here can be realized in renormalizable models involving additional heavy or light scalars, as we illustrate with several examples.
|
arxiv:2211.00021
|
neutron stars could contain a mixture of ordinary nuclear matter and dark matter, such that dark matter could influence observable properties of the star, such as its mass and radius. we study these dark matter admixed neutron stars for two choices of dark matter : a free fermi gas and mirror dark matter. in addition to solving the multi - fluid tolman - oppenheimer - volkoff equations for static solutions and presenting mass - radius diagrams, we focus on two computations that are lacking in the literature. the first is a rigorous determination of stability over the whole of parameter space, which we do using two different methods. the first method is based on harmonic time - dependent perturbations to the static solutions and on solving for the radial oscillation frequency. the second method, which is less well - known, conveniently makes use of unperturbed, static solutions only. the second computation is of the radial oscillation frequency, for fundamental modes, over large swaths of parameter space.
|
arxiv:2102.08257
|
we present a calculation of the quark distribution amplitudes ( das ), the gegenbauer moments, and decay constants for $ \ pi, \ rho, k $ and $ k ^ * $ mesons using the light - front quark model. while the quark da for $ \ pi $ is somewhat broader than the asymptotic one, that for $ \ rho $ meson is very close to the asymptotic one. the quark das for $ k $ and $ k ^ * $ show asymmetric form due to the flavor su ( 3 ) - symmetry breaking effect. the decay constants for the transversely polarized $ \ rho $ and $ k ^ * $ mesons ( $ f ^ t _ \ rho $ and $ f ^ t _ { k ^ * } $ ) as well as the longitudinally polarized ones ( $ f _ \ rho $ and $ f _ { k ^ * } $ ) are also obtained. our averaged values for $ f ^ t _ v / f _ v $, i. e. $ ( f ^ t _ \ rho / f _ \ rho ) _ { \ rm av } = 0. 78 $ and $ ( f ^ t _ { k ^ * } / f _ { k ^ * } ) _ { \ rm av } = 0. 84 $, are found to be consistent with other model predictions. especially, our results for the decay constants are in a good agreement with the su ( 6 ) symmetry relation, $ f ^ t _ { \ rho ( k ^ * ) } = ( f _ { \ pi ( k ) } + f _ { \ rho ( k ^ * ) } ) / 2 $.
|
arxiv:hep-ph/0701177
|
we consider the recursion relation for loop integrands in planar n = 4 sym generated by an all - line shift of momentum twistors. we examine the behaviour of the rational loop integrands when the shift parameter becomes large, and find that a valid recursion relation may be obtained in all cases. the recursion relation is then formulated both in region momentum space and in momentum twistor space, and solved in detail for some one and two - loop examples. finally, we show that the general iterative solution of the recursion relation generates the mhv vertex expansion for all loop integrands, providing a proof of the mhv diagram formalism for all loop amplitudes in planar n = 4 sym.
|
arxiv:1010.5921
|
we study two basic problems regarding edit error, i. e. document exchange and error correcting codes for edit errors ( insdel codes ). for message length $ n $ and edit error upper bound $ k $, it is known that in both problems the optimal sketch size or the optimal number of redundant bits is $ \ theta ( k \ log \ frac { n } { k } ) $. however, known constructions are far from achieving these bounds. we significantly improve previous results on both problems. for document exchange, we give an efficient deterministic protocol with sketch size $ o ( k \ log ^ 2 \ frac { n } { k } ) $. this significantly improves the previous best known deterministic protocol, which has sketch size $ o ( k ^ 2 + k \ log ^ 2 n ) $ ( belazzougui15 ). for binary insdel codes, we obtain the following results : 1. an explicit binary insdel code which encodes an $ n $ - bit message $ x $ against $ k $ errors with redundancy $ o ( k \ log ^ 2 \ frac { n } { k } ) $. in particular this implies an explicit family of binary insdel codes that can correct $ \ varepsilon $ fraction of insertions and deletions with rate $ 1 - o ( \ varepsilon \ log ^ 2 ( \ frac { 1 } { \ varepsilon } ) ) = 1 - \ widetilde { o } ( \ varepsilon ) $. 2. an explicit binary insdel code which encodes an $ n $ - bit message $ x $ against $ k $ errors with redundancy $ o ( k \ log n ) $. this is the first explicit construction of binary insdel codes that has optimal redundancy for a wide range of error parameters $ k $, and this brings our understanding of binary insdel codes much closer to that of standard binary error correcting codes. in obtaining our results we introduce the notion of \ emph { $ \ varepsilon $ - self matching hash functions } and \ emph { $ \ varepsilon $ - synchronization hash functions }. we believe our techniques can have further applications in the literature.
|
arxiv:1804.05776
|
this paper explores the moral judgment and moral reasoning abilities exhibited by large language models ( llms ) across languages through the defining issues test. it is a well known fact that moral judgment depends on the language in which the question is asked. we extend the work of beyond english, to 5 new languages ( chinese, hindi, russian, spanish and swahili ), and probe three llms - - chatgpt, gpt - 4 and llama2chat - 70b - - that shows substantial multilingual text processing and generation abilities. our study shows that the moral reasoning ability for all models, as indicated by the post - conventional score, is substantially inferior for hindi and swahili, compared to spanish, russian, chinese and english, while there is no clear trend for the performance of the latter four languages. the moral judgments too vary considerably by the language.
|
arxiv:2402.02135
|
this book deals with functions allowing to express the dissimilarity ( discrepancy ) between two data fields or ' ' divergence functions ' ' with the aim of applications to linear inverse problems. most of the divergences found in the litterature are used in the field of information theory to quantify the difference between two probability density functions, that is between positive data whose sum is equal to one. in such context, they take a simplified form that is not adapted to the problems considered here, in which the data fields are non - negative but with a sum not necessarily equal to one. in a systematic way, we reconsider the classical divergences and we give their forms adapted to inverse problems. to this end, we will recall the methods allowing to build such divergences, and propose some generalizations. the resolution of inverse problems implies systematically the minimisation of a divergence between physical measurements and a model depending of the unknown parameters. in the context image reconstruction, the model is generally linear and the constraints that must be taken into account are the non - negativity as well as ( if necessary ) the sum constraint of the unknown parameters. to take into account in a simple way the sum constraint, we introduce the class of scale invariant or affine invariant divergences. such divergences remains unchanged when the model parameters are multiplied by a constant positive factor. we show the general properties of the invariance factors, and we give some interesting characteristics of such divergences. an extension of such divergences allows to obtain the property of invariance with respect to both the arguments of the divergences ; this characteristic can be used to introduce the smoothness regularization of inverse problems, that is a regularisation in the sense of tikhonov. we then develop in a last step, minimisation methods of the divergences subject to non - negativity and sum constraints on the solution components. these methods are founded on the karush - kuhn - tucker conditions that must be fulfilled at the optimum. the tikhonov regularization is considered in these methods. chapter 11 associated with appendix 9 deal with the application to the nmf, while chapter 12 is dedicated to the blind deconvolution problem. in these two chapters, the interest of the scale invariant divergences is highlighted.
|
arxiv:2003.01411
|
the effectiveness of intrusion detection systems ( ids ) is critical in an era where cyber threats are becoming increasingly complex. machine learning ( ml ) and deep learning ( dl ) models provide an efficient and accurate solution for identifying attacks and anomalies in computer networks. however, using ml and dl models in ids has led to a trust deficit due to their non - transparent decision - making. this transparency gap in ids research is significant, affecting confidence and accountability. to address, this paper introduces a novel explainable ids approach, called x - cba, that leverages the structural advantages of graph neural networks ( gnns ) to effectively process network traffic data, while also adapting a new explainable ai ( xai ) methodology. unlike most gnn - based ids that depend on labeled network traffic and node features, thereby overlooking critical packet - level information, our approach leverages a broader range of traffic data through network flows, including edge attributes, to improve detection capabilities and adapt to novel threats. through empirical testing, we establish that our approach not only achieves high accuracy with 99. 47 % in threat detection but also advances the field by providing clear, actionable explanations of its analytical outcomes. this research also aims to bridge the current gap and facilitate the broader integration of ml / dl technologies in cybersecurity defenses by offering a local and global explainability solution that is both precise and interpretable.
|
arxiv:2402.00839
|
exploiting long - range contextual information is key for pixel - wise prediction tasks such as semantic segmentation. in contrast to previous work that uses multi - scale feature fusion or dilated convolutions, we propose a novel graph - convolutional network ( gcn ) to address this problem. our dual graph convolutional network ( dgcnet ) models the global context of the input feature by modelling two orthogonal graphs in a single framework. the first component models spatial relationships between pixels in the image, whilst the second models interdependencies along the channel dimensions of the network ' s feature map. this is done efficiently by projecting the feature into a new, lower - dimensional space where all pairwise interactions can be modelled, before reprojecting into the original space. our simple method provides substantial benefits over a strong baseline and achieves state - of - the - art results on both cityscapes ( 82. 0 % mean iou ) and pascal context ( 53. 7 % mean iou ) datasets. code and models are made available to foster any further research ( \ url { https : / / github. com / lxtgh / gald - dgcnet } ).
|
arxiv:1909.06121
|
we introduce a renormalized 1pi vertex part scalar field theory setting in momentum space to computing the critical exponents $ \ nu $ and $ \ eta $, at least at two - loop order, for a layered parallel plate geometry separated by a distance l, with periodic as well as antiperiodic boundary conditions on the plates. we utilize massive and massless fields in order to extract the exponents in independent ultraviolet and infrared scaling analysis, respectively, which are required in a complete description of the scaling regions for finite size systems. we prove that fixed points and other critical amounts either in the ultraviolet or in the infrared regime dependent on the plates boundary condition are a general feature of normalization conditions. we introduce a new description of typical crossover regimes occurring in finite size systems. avoiding these crossovers, the three regions of finite size scaling present for each of these boundary conditions are shown to be indistinguishable in the results of the exponents in periodic and antiperiodic conditions, which coincide with those from the ( bulk ) infinite system.
|
arxiv:1103.1661
|
quivers play an important role in the representation theory of algebras, with a key ingredient being the path algebra and the preprojective algebra. quiver grassmannians are varieties of submodules of a fixed module of the path or preprojective algebra. in the current paper, we study these objects in detail. we show that the quiver grassmannians corresponding to submodules of certain injective modules are homeomorphic to the lagrangian quiver varieties of nakajima which have been well studied in the context of geometric representation theory. we then refine this result by finding quiver grassmannians which are homeomorphic to the demazure quiver varieties introduced by the first author, and others which are homeomorphic to the graded / cyclic quiver varieties defined by nakajima. the demazure quiver grassmannians allow us to describe injective objects in the category of locally nilpotent modules of the preprojective algebra. we conclude by relating our construction to a similar one of lusztig using projectives in place of injectives.
|
arxiv:0909.3746
|
the box - ball system ( bbs ) is a simple model of soliton interaction introduced by takahashi and satsuma in the 1990s. recent work of the authors, together with tsuyoshi kato and satoshi tsujimoto, derived various families of invariant measures for the bbs based on two - sided stationary markov chains. in this article, we survey the invariant measures that were presented in the latter work, and also introduce a family of new ones for periodic configurations that are expressed in terms of gibbs measures. moreover, we show that the former examples can be obtained as infinite volume limits of the latter. another aspect of the previous work was to describe scaling limits for the box - ball system ; here, we review these results, and also present scaling limits other than those that were covered there. one, the zigzag process has previously been observed in the context of queuing ; another, a periodic version of the zigzag process, is apparently novel. furthermore, we demonstrate that certain palm measures associated with the stationary and periodic versions of the zigzag process yield natural invariant measures for the dynamics of corresponding versions of the ultra - discrete toda lattice.
|
arxiv:1905.00186
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.