Sound
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Saturday 7 November
Multimodal LLMs have made remarkable progress in perceiving the physical world, while their perceptual scope remains largely human-centric, focusing on what humans can directly see or hear. This progress raises the prospect of perception beyond human limits, such as extending hearing to sounds that are inaudible to humans. More broadly, the physical world is full of waves, vibrations, and oscillations, only a small fraction of which are directly audible to humans. This invites the extension of multimodal LLMs toward a broader form of machine hearing, which remains largely underexplored. To investigate this direction, we introduce the Machine Hearing Benchmark (MHB), comprising 6,128 instances across 13 tasks and seven domains. MHB spans highly heterogeneous input regimes, with sampling rates from approximately $10^{-5}$ to $10^6$ Hz, sequence lengths from $10^1$ to $10^6$ samples, and both single- and multi-channel inputs. Evaluations of current open-source and proprietary Audio LLMs (ALLMs) reveal limited zero-shot transfer to machine hearing. Supervised adaptation on machine-hearing data alone does not close this gap, suggesting that the challenge extends beyond data exposure to a systematic mismatch between existing audio front ends and machine-hearing inputs. To address this mismatch, we introduce MachEar, a lightweight encoder extension that complements the pretrained Audio LLM for machine-hearing inputs. The results show that MachEar equips a pretrained Audio LLM with stronger machine-hearing capabilities while largely preserving its existing audio understanding performance.
Analysis of PDEs
Zero extension for the Dirichlet problem on graphs
Solutions of the Dirichlet problem on a finite part of a weighted graph are extended by zero, together with their source, to a larger connected part. A necessary and sufficient condition is proved for the extended function to remain a solution of the corresponding Dirichlet problem on the enlarged domain: at every newly included vertex adjacent to the original part, the weighted sum of the solution over its neighbors in that part must vanish. The admissible sources are characterized as the orthogonal complement of the harmonic extensions from the enlarged domain to the original part, their space has dimension equal to the number of vertices of the original part minus the rank of the interface coupling, and an exact energy identity for the extension error follows from a Schur complement of the graph Laplacian. The same reduction covers other boundary conditions, including the Neumann problem, and extends to higher powers of the Laplacian, where the boundary data must be compatible with the lower-order interface traces. The discrete criterion is compared with its continuous counterpart, and no result in this form has been reported on Riemannian manifolds.
Read the full abstract
Machine Learning
A Geometric Theory for Nonequilibrium Transformer Dynamics
We develop a variational and geometric theory based on statistical physics for the joint evolution of Transformer attention and token representations at fixed model parameters. To this end, we introduce a free energy for the attention and model the evolution of the attention weights by its Fisher–Rao gradient flow. At fixed token representations, we prove that the softmax matrix is the unique free-energy minimiser and that every positive attention trajectory converges to it. For prescribed time-dependent representations, we show that the attention dynamics becomes linear after a change to logit variables. This result yields tracking bounds relative to the instantaneous softmax and sufficient conditions for attention to converge to a fixed matrix. Hamiltonian dynamics for the representations and their momenta are then coupled to the attention relaxation driven by the same energy. This coupling results in an energy-dissipation law. We also develop two alternative Hamiltonians incorporating the attention value map. We prove global well-posedness of the attention flow and all three coupled systems.
Read the full abstract
Computer Vision
JoVox: Joint Hierarchical Flow Matching for Lightweight Voxel Generation
Generating high-fidelity voxels while keeping their lightweight nature remains an open problem in computer vision. Current generators are either high-fidelity but computationally heavy, or lightweight but of lower fidelity. Hierarchical methods keep the lightweight nature of voxels. However, their coarse levels, which decide where all finer cells can appear, are generated with the least information about the shape and are never revisited, so they are often plausible but wrong. Our key insight is to let finer cells take part in deciding coarse cells. To this end, we propose JoVox, a discrete flow matching framework that learns the joint distribution of all levels of the voxel hierarchy, instead of a separate distribution for each level, so that coarse cells can be predicted with finer cells already generated nearby. Within this framework, we further introduce a two-level prediction, which outputs the cells two levels finer at once instead of one level finer, halving the number of levels that the network processes and yielding voxels of higher fidelity. We also introduce an unmasking strategy that revisits early decisions once finer cells are generated and corrects inconsistencies in masked diffusion models. Our framework outperforms the baselines on both unconditional and image-conditioned voxel generation. Compared with the current state-of-the-art TRELLIS.2, we achieve 8x higher training throughput, 1/15 of the FLOPs in a forward pass, train on 4 vs. 32 GPUs, and obtain higher quality with an 11% lower Chamfer distance.
Read the full abstract
Artificial Intelligence
RetinaGuard: Risk-Controlled Agentic Triage for Multi-Disease Retinal Screening
Retinal screening programmes need to know which automated decisions can be accepted without review. This paper presents RetinaGuard, an agentic triage framework in which quality, diagnostic, uncertainty, verification and explanation agents gather evidence for each fundus photograph, a learned failure predictor estimates the probability that the decision is unsafe, auxiliary tools are invoked only when their expected value of information exceeds their cost, and the acceptance threshold is certified by conformal risk control. On the RFMiD test split, RetinaGuard was compared with confidence and calibrated-uncertainty referral at matched referral budgets and equal certified safety, using two RETFound primary models of different strength. With the weaker model (macro-AUROC 0.896), verifier disagreement added significant information, RetinaGuard captured 119 of 280 unsafe decisions at a 20% budget against 88 for lowest-confidence referral, and at a nominal 5% unsafe level it accepted 37.0% of cases against 10.3%. With the stronger model (0.925) the verifier added nothing beyond uncertainty and the advantage shrank to a few percentage points. In both settings its coverage at the nominal 5% level stayed near 37.7%, gating cut computation by about 30%, and a quality agent reduced unsafe acceptances under most image corruptions. The benefit of agentic evidence gathering thus depends on how much the primary model's own uncertainty already reveals about its errors.
Read the full abstract
Distributed Computing
Performance Benchmarking of Federated Learning Frameworks for Medical Image Classification
Selecting a Federated Learning (FL) framework requires understanding the resource demands of its training and evaluation workflow. This work compares Flower, NVIDIA FLARE, and FedBioMed using two-dimensional MedNIST classification with ResNet-10 in a containerized, single-host environment. Following a dated review of repository activity and documented medical applications, we evaluated nine configurations per framework, with three random seeds, for a total of 81 executions. The configurations varied federation size, communication rounds, and batch size. We monitored execution time, CPU and RAM consumption, host-level GPU utilization, server network traffic rates, and model accuracy at the end of each execution. The implementations shared nominal learning settings while retaining differences in data splitting and federated evaluation. Configuration-blocked non-parametric analysis identified differences in execution time, client and server memory, GPU utilization, and network traffic rates. NVIDIA FLARE exhibited the lowest server-memory consumption and the highest host-level GPU utilization, Flower exhibited the highest average server network traffic rate, and FedBioMed required more client memory and longer execution time. No statistically significant difference in final accuracy was detected within the evaluated short training schedules; this does not establish convergence or equivalent predictive performance. The results characterize the configured workflows under shared hardware resources. They provide a concrete case study of system behavior, with conclusions bounded by workflow asymmetries, local networking, and the lightweight MedNIST workload.
Read the full abstract
Quantum Physics
Hidden Coherence and Controlled Quasihole Motion in a Fermionic Laughlin State on a Superconducting Quantum Processor
Fractional quantum Hall (FQH) states support fractionalized excitations and geometric phases, yet experimentally resolving the many-body coherence encoded in their wave functions remains challenging. Here, we prepare a three-electron $\nu=1/3$ Laughlin state on the Haldane sphere using seven superconducting qubits and reconstruct its many-body density matrix. Through Read's composite-boson construction, we reveal vortex-dressed coherence that remains appreciable over a larger separation range than electron coherence. The extended profile is suppressed in a Fock-diagonal reference constructed from the measured populations, demonstrating its dependence on off-diagonal many-body coherence. We further prepare a one-flux-added quasihole and characterize its fractional charge response against finite-size benchmarks. Controlled quasihole motion yields an effective geometric-phase signature from the reconstructed dominant-eigenstate trajectory. These complementary measurements provide microscopic access to the coherent structure and quasihole response of a finite fermionic Laughlin state, demonstrating an experimental route to probing strongly correlated FQH wave functions on programmable quantum processors.
Read the full abstract
Optics
Nonlinear Optical Properties of MXenes for Photonic Applications
MXenes - a rapidly expanding family of two-dimensional transition-metal carbides, nitrides, and carbonitrides - have emerged as powerful nonlinear photonic materials owing to their metallic conductivity, ultrafast carrier dynamics, broadband optical response, and highly tunable surface chemistry. This Review provides the first comprehensive overview of the nonlinear optical (NLO) properties of MXenes and their growing impact on advanced photonic technologies. We discuss state-of-the-art synthesis strategies, from fluoride-based etching to molten-salt, electrochemical, and CVD-enabled routes, highlighting how structural quality and surface terminations govern optical behaviour. A detailed analysis of third-order nonlinearities including Kerr effects, two- and multi-photon absorption, saturable and reverse-saturable absorption, and harmonic generation is presented, supported by quantitative comparisons across MXene families. We further examine modelling approaches linking band structure to nonlinear response and survey application frontiers such as gas sensing, ultrafast lasers, all-optical limiters, and emerging photonic neural networks. Finally, we outline future research directions spanning material engineering, device integration, and quantum/nonlinear metasurface platforms. Together, these insights position MXenes as a transformative materials platform for next-generation nonlinear photonics and ultrafast all-optical information processing.
Read the full abstract
Quantum Physics
From Machine-Discovered Graphs to Analytic NOON State Families
Machine-discovered graphs generate NOON states after heralding and output-number selection, but their interference mechanisms and optical implementations are not immediately apparent. We derive analytic weight conditions for arbitrary photon number $N$ by counting pair-generation processes that cancel mixed output occupations. These conditions yield quantum experiments by path identity using a Hong-Ou-Mandel interferometer and photon-pair sources. Takagi decomposition of the pair-amplitude matrix gives an exact Gaussian implementation using single-mode squeezers followed by a passive interferometer. Any pure, zero-displacement Gaussian scheme with photon-number-resolved heralding requires at least $N-2$ herald photons, and both constructions attain this bound. At finite gain, the ideal Takagi implementation maintains unit fidelity after output-number selection, whereas fidelity decreases in path identity with fixed source settings.
Read the full abstract
Artificial Intelligence
Audit Before You Merge: Provenance, Probing, and Continual LoRA Composition
Post-hoc merging of public LoRA adapters is an attractive gradient-free path to continual adaptation, but reliable measurement of catastrophic forgetting in this setting depends on infrastructure that prior work has not addressed. We treat each adapter merge as a sequential task in a continual-learning setting, with the instruction-tuned base model's capabilities as Task 0, and study forgetting on Llama-3.1-8B-Instruct. Public adapter hubs contain modules that fail basic provenance checks: silent base-version mismatches and near-zero delta norms whose inclusion inflates measured interference by ~10pp. We propose a two-check audit (base-model match, non-trivial delta norm) and a leakage-free probing protocol as preconditions for any forgetting study in this setting. On an audited 4-adapter set, naive merging causes -10.5pp on GSM8K and -18.3pp on HumanEval, while Task Arithmetic at lambda=0.5 sits within a standard error of the unmerged baseline. A coordinate search including negative coefficients identifies the math adapter as probe-optimal at lambda=-0.5, outside the reach of positive-only mergers, suggesting forgetting has structure that negation-aware merging may exploit. The same protocol replicates on Mistral-7B-Instruct-v0.3, where 2 of 4 candidate adapters fail the audit and the audited merge reproduces the same merge-method ranking on GSM8K, supporting that our findings are not Llama-specific.
Read the full abstract
Computation and Language
TransMem: Transition-Aware Retrieval for Evolving Personal Memory
Long-term personalized agents are increasingly expected to answer queries over user histories in which preferences, goals, and constraints evolve across interleaved conversations. Existing retrieval-based and structured-memory systems can surface relevant facts, but often return isolated memory units that do not explain which earlier preference was refined, contradicted, or superseded by later evidence. We introduce TransMem, a transition-aware memory retrieval framework that stores preference updates as explicit retrieval units alongside ordinary factual memories. During memory construction, TransMem routes user interaction history into topic-structured memory graphs, detects shift events, and grounds them as transition records. At inference time, TransMem first retrieves factual anchors and then adaptively expands over transition records, producing a compact context of factual anchors and accepted transition records. Experiments on PersonaMem and PersonaBench across two Qwen backbones show that TransMem outperforms representative retrieval-based and structured-memory baselines. Controlled analyses further show that transition records and adaptive expansion contribute beyond factual anchoring, chronological expansion, flat transition retrieval, and fixed-hop traversal. These results suggest that preference-update records are useful retrieval units for long-term personalization.
Read the full abstract
Robotics
Burden, Bounty, or Bliss? Muscle-Aware Retargeting and Null-Space Exploration for Whole-Body Musculoskeletal Control
Learning whole-body musculoskeletal control from human motion is important for developing human-like robots, assistive devices, and rehabilitation systems. This task presents two main challenges, namely reference trajectories that are kinematically plausible but difficult to reproduce through muscle-driven dynamics, and a highly redundant 700-dimensional action space in which unstructured exploration can easily disrupt the target motion. In this paper, we propose a motion-to-policy pipeline that exploits muscle redundancy in two complementary stages. First, muscle-aware retargeting (MAR) performs bounded adaptation of kinematic trajectories subject to hard semantic, anatomical, and contact constraints. We motivate the constraint bounds by analyzing how reference-trajectory artifacts affect muscle-driven tracking. Second, null-space exploration (NSE) estimates a one-control-frame mapping from muscle excitations to generalized forces and injects bounded, anatomically structured perturbations along its null space into the actions of a value-guided flow policy. On a 24-clip motion-capture benchmark, MAR and NSE together solve every reference, reach the full horizon with half the samples of the baseline pipeline, and lower joint-center error by 44% without late collapse. To the best of our knowledge, this is the first end-to-end pipeline that transforms monocular video into a trained whole-body imitation policy controlling 700 muscles.
Read the full abstract
Information Retrieval
Formalizing Patient-Set Semantics for OMOP CDM-Based Clinical Trial Feasibility Assessment:A Neuro-Symbolic Methodology for Traceable Site-Feasibility Patient Counts
This study examines the interpretation and execution of natural-language eligibility criteria for clinical trial site-feasibility patient-count estimation using the OMOP Common Data Model (CDM). It treats each count as the size of a patient set produced under explicitly declared choices about clinical concepts, temporal conditions, exclusions, proxies, and available data. The study presents NS-CoT, a neuro-symbolic pipeline that uses an LLM to interpret and structure criteria, followed by symbolic processing and deterministic SQL compilation. Intermediate representations and execution records document how criteria are translated into patient sets and counts. Evaluation uses synthetic OMOP CDM data, controlled benchmarks, and a multi-site simulation. Comparisons with a direct-SQL baseline examine execution success, count agreement, patient-set agreement, and preservation of reference-defined site-feasibility labels and rankings. Performance varies across evaluation settings. Layer-wise inspection also supports the identification of implementation and policy errors, followed by correction and re-evaluation. The work investigates the reviewability and fidelity of cohort computation relative to declared reference definitions. It does not establish that these definitions capture a unique clinical ground truth or demonstrate improved recruitment, site selection, or trial outcomes in practice.
Read the full abstract
Group Theory
Marginal Congruences and Elements for Two Commutator Words in Completely Regular Semigroups
Motivated by Problem 13 posed by Araújo, Araújo and Kinyon \cite{Araujo2025} concerning marginal elements, subsemigroups and marginal equivalences associated with commutator words in completely regular semigroups, in this paper we investigate the two commutator words $ \omega_1(x,y)=x^{-1}y^{-1}xy$ and $ \omega_2(x,y)=(yx)^{-1}xy$ on completely regular semigroups. We prove that their marginal equivalences $\sim_{\omega_r}$ ($r=1,2$) coincide and are congruences. More precisely, for a completely regular semigroup $S$, we show that $a\sim_{\omega_r}b$ ($r=1,2$) if and only if $a^0=b^0=p$ and $b^{-1}a\in Z(pSp)$, where $Z(pSp)$ is the center of the monoid $pSp$, and so both words have the same set of marginal elements: \[ M_{\omega}(S) =\bigcup_{p\in E(S)} \bigl(Z(pSp)\cap H_p\bigr), \] where $E(S)$ is the set of idempotents in $S$. We also characterize when this set is a subsemigroup and establish alternative descriptions of the common marginal congruence and marginal elements.
Read the full abstract
Artificial Intelligence
RankShift: Evaluating Ranking Instability Across Benchmark Dimensions in Document Parsing
Document parsing benchmarks are often treated as stable measures of model quality, yet rankings may depend substantially on benchmark design choices. In this work, we present RankShift, a systematic study comparing OCR and vision-language models (VLMs) under controlled variation in dataset composition, matching strategy, and scoring methodology. Using OmniDocBench as a case study, we evaluate the ranking stability of 16 models across 3 dataset configurations, 6 matching strategies, and 3 scoring metrics. We quantify rank stability using pairwise rank correlation, top-k churn, and pairwise reversal rate. Our results show that the interaction between matching strategy and ground truth format is a substantial but understudied driver of instability: scoring against OmniDocBench's JSON annotations versus its Markdown rendering yields a mean Kendall's tau of approximately 0.33 under region-level matching strategies, with individual pairs as low as tau = 0.23, compared with tau = 0.77 when each page is scored as a single concatenated string. Dataset composition produces comparable instability across page-type strata (mean tau = 0.48) and layout-type strata (mean tau = 0.49), while the cross-dataset shift from OmniDocBench to Real5's Scanning reconstruction of the same pages is more stable (tau = 0.80). Metric choice is the most stable dimension, with mean tau of approximately 0.78 across NED, BLEU, and METEOR. These findings indicate that leaderboard position is not solely a property of model capability, motivating leaderboards that report not only headline scores but also the robustness of rankings across reasonable design choices.
Read the full abstract
Computer Vision
MGE: A Membership Guided Data Extraction Attack against Diffusion Models
Recently, diffusion models have achieved remarkable performance in image generation. However, their inherent memory properties introduce a potential safety risk of training data leakage, i.e., data extraction attack. To illustrate such risk, we propose a membership guided extraction (MGE) framework to effectively extract training data from a pre-trained diffusion model. Our core idea is to formulate the data extraction as a conditional generation, where the condition signal is from a membership inference classifier. To improve the success rate and efficiency of MGE, we propose adversarial gradient and parallel branch decoding. Our experiments on image generation benchmarks demonstrate that our MGE can extract hundreds of images, which shows that more attention should be paid to privacy-preserving of diffusion model.
Read the full abstract
Machine Learning
Safe Offline Reinforcement Learning using Behavior Regularisation and Latent Feasibility Guidance
Safe offline reinforcement learning (RL) seeks to learn policies that maximize return while satisfying safety constraints, using a static dataset. To characterize state-wise safety, reachability analysis identifies the feasible region, defined as the set of states from which constraints can be satisfied at every future step. Policies are commonly extracted from learned value functions with advantage-weighted regression (AWR). However, AWR relies on exponentiated weights that amplify value estimation errors, whereas behavior-regularized policy gradient methods directly utilize the action gradient of the value function and often make more effective use of learned critics. We propose Feasibility-Optimized Conditional Actor Learning (FOCAL), a method that incorporates behavior-regularized policy gradients into a conditional variational autoencoder (CVAE) actor. FOCAL optimizes this policy using a first-order objective partitioned by feasibility, updating the actor to increase return on transitions estimated to be feasible and to reduce constraint violations on transitions estimated to be infeasible. During inference, FOCAL dynamically adjusts the scale of the latent prior according to the estimated feasibility of the state to adjust the conservatism of action sampling. Theoretically, we show that for AWR, the policy update variance depends exponentially on the advantage error, whereas for FOCAL it scales linearly with the variance of the value estimation error. Across the DSRL benchmark, FOCAL satisfies the constraints on all evaluated tasks while achieving competitive or higher returns compared to baselines.
Read the full abstract
Number Theory
Restricted Quantum unique ergodicity under Generalized Lindelöf conjecture
Assume the generalized Lindelöf hypothesis. Let $u$ be an even Hecke–Maass cusp form on $\mathrm{SL}_2(\mathbb{Z})\backslash\mathbb{H}$, normalized in $L^2$ with respect to hyperbolic area, with Laplace eigenvalue $1/4+T^2$. For every fixed $\psi\in C_c((0,\infty);\mathbb{C})$ and every $\varepsilon>0$, we prove that \[ \int_0^\infty \psi(y)|u(i y)|^2\frac{d y}{y} =\frac{6}{\pi}\int_0^\infty \psi(y)\frac{d y}{y} +O_{\psi,\varepsilon}(T^{-1/4+\varepsilon}) \] as $T\to\infty$, uniformly in $u$. In particular, the restriction mass equidistributes on every fixed compact segment of the imaginary axis in $\mathbb{H}$.
Read the full abstract
Superconductivity
Sub-Thouless energy scale in interface-dominated unconventional superconductor junctions
We derive the conductance in unconventional superconductor - normal metal - normal metal junctions, taking into account the resistance between the normal metals. We show that applying Tanaka–Nazarov boundary conditions at both superconductor and normal metal interfaces reveals a new characteristic low-energy scale which is sub-Thouless in the experimentally relevant strong-interface regime. This scale governs, among other quantities, the curvature of the local density of states (LDOS) and the width of conductance dips near the zero-bias conductance peak in triplet superconductor junctions, thereby replacing the Thouless energy as the dominant low-energy parameter. We derive universal $s$-wave proximitized diffusive metal expressions for the zero-bias LDOS and differential conductance which solely depend on the interface and bulk resistances, independent of the superconducting gap and Thouless energy scale. This framework provides a possible origin of the anomalous low-bias conductance features observed in unconventional superconductors, offering a refined mechanism for interpreting interface and correlated electron physics in experimental hybrid junctions.
Read the full abstract
Statistics Applications
Regularity in Cryptocurrency Market Cycles: A Halving-to-Halving Analysis Using Data from Google Trends and Wikipedia
Bitcoin halvings provide a natural temporal structure for studying long-term dynamics of cryptocurrency markets. This paper presents a methodological framework for analyzing the regularity of Bitcoin and aggregate cryptocurrency market cycles by normalizing historical periods between consecutive halving events and projecting them onto a common relative-time scale expressed as the percentage of cycle completion. The proposed framework integrates market variables, namely Bitcoin price and aggregate cryptocurrency market capitalization, with measures of public attention derived from Google Trends search volume and Wikipedia pageviews. To obtain long-horizon daily attention indices, Google Trends data are reconstructed from overlapping 90-day windows linked by median scale ratios, following established approaches to combining high-frequency Google Trends samples. Rather than attempting to forecast future prices, the study aims to compare historical cycles, identify recurring temporal patterns, and evaluate the relationship between market dynamics and retail investor interest across different cycle-relative stages. Exploratory results point to a possible recurring structure. The maxima of the aggregate market capitalization in the two most recent completed cycles occurred at 39.0% and 37.9% of cycle completion, and the highest value of the ongoing cycle observed to date at 38.1% (534-547 days after the respective halvings), although the corresponding multiple of the halving-day value fell from 66 to 1.8. Public attention shows a similar compression of peak multiples, but not a similar timing. The paper also discusses current data limitations and outlines future research directions, including formal lead-lag analyses and expanded multi-asset investigations.
Read the full abstract
Systems and Control
Two-Timescale Trajectory Control, Task Reassignment and Resource Scheduling for Multi-UAV Cooperative Mobile Edge Computing
Uncrewed aerial vehicle (UAV)-assisted mobile edge computing (MEC) is a promising technology, leveraging the advantages of UAVs including flexible deployment, wide coverage, and reliable wireless communication. In this paper, we study the task reassignment and offloading problem in a multi-UAV cooperative operating environment, and consider the real occurrence that the computing tasks arrive more frequently than UAVs make flight decisions. In addition, we also take into account the position offset of the UAV during flight and the unavailability of the UAV due to external interference. We propose a two-timescale optimization framework, where the trajectory of the UAV is carried out on a large timescale, while the task reassignment and computing resource allocation are jointly optimized on a small timescale. The core objective is to maximize the long-term average system gain, which takes both energy consumption and task delay into account. To address the UAV trajectory optimization problem under large timescale constraints, we propose a constrained multi-agent deep reinforcement learning algorithm. For decision-making within small timescale, to jointly optimize task offloading decisions and computing resource allocation, we integrate convex optimization with a heuristic algorithm. Finally, we conduct extensive simulations to evaluate the proposed algorithms against the baseline, achieving a 24% reduction in latency and a 28% reduction in total system cost.
Read the full abstract
Computer Vision
Heterogeneous Asymmetric Contrastive Distillation
Knowledge distillation (KD) between architectural families, e.g., from a Transformer teacher to a CNN student, is difficult because the two networks learn different representations. Logits share a label space, but heterogeneous teachers differ in logit offset, scale, and sharpness, and their non-target classes are less semantically organized. Conventional logit distillation matches these values sample by sample, asking the student to imitate how another architecture expresses its predictions. We argue that a heterogeneous teacher should pass on the relational structure of its predictions, i.e., how it arranges samples relative to one another, and propose Heterogeneous Asymmetric Contrastive Distillation (HACD) to transfer this structure. HACD maps teacher and student logits to centered, temperature-softened probability directions on a shared unit hypersphere, which removes architecture-dependent offsets and lets the temperature control how sharply peaked outputs enter the representation while keeping informative non-target classes. Each student output must then identify its own teacher target within the batch. Because every teacher target also acts as a negative for the others, HACD computes the targets on a moderately augmented view and contrasts both a moderate and a strong student view against them. HACD adds no learnable parameters and needs one teacher pass per sample. It achieves the best or second-best accuracy on 11 of 12 heterogeneous pairs on CIFAR-100, without ground-truth labels, and on 14 of 15 pairs on ImageNet-1K, improving over the reported vanilla KD results by 8.82 and 1.33 percentage points on average. Checkpoints are available at https://github.com/BTAEYANG/HACD-Heterogeneous-Asymmetric-Contrastive-Distillation; the code will be released upon publication.
Read the full abstract
Lattice Field Theory
An Information-Theoretic Perspective on Spectral Reconstruction in Lattice QCD
Assessing the reliability of spectral reconstruction from finite, noisy lattice correlators requires understanding the constraints supplied by the data relative to prior assumptions. We develop an information-theoretic framework using Kullback–Leibler (KL) divergences to quantify the information supplied by the data through changes from prior to posterior probability distributions. Our analysis accounts for both changes in the probability distributions of spectral strength at individual energies and constraints encoded in correlations across the spectrum. We smear these information measures over neighboring energies using Gaussian weights to obtain a smoothed assessment at a specified energy scale. We demonstrate this framework using a hierarchical Bayesian reconstruction method with posterior sampling, applied to mock correlator data generated from a three-peak spectrum and a near-physical lattice proton correlator. These calculations show that the information measures help explain the properties of reconstructed spectra. They also provide a semiquantitative criterion for assessing how strongly reconstructed spectral structures are constrained by the data and how reliable they are in light of prior effects.
Read the full abstract
Quantum Physics
Demonstration of a partial knowledge quantum key distribution protocol
We demonstrate a novel QKD protocol, named \texttt{Bridge[$d,q$]}, which generalizes the SARG04 protocol to $d$ dimensions and $q$ bases. In this protocol, Alice and Bob gain only partial knowledge about each other's raw key. Nevertheless, this partial information is sufficient to distill a definite mutual encryption key. We present an experimental demonstration of the \texttt{Bridge[$2,3$]} protocol. The three unbiased polarization bases were switched using a single Pockels cell. The applied polarization basis switching method is relevant to any 3-basis QKD protocol and to other applications that require fast manipulation of photon polarization qubits.
Read the full abstract
Superconductivity
Identifying Contributions to the Superconducting Diode Effect through Nonreciprocal Superconducting Properties
Nonreciprocal superconducting responses provide a sensitive probe of superconducting states in systems where inversion and time-reversal symmetries are broken. However, in the superconducting diode effect (SDE), where the critical current differs for opposite current directions, nonreciprocity can also arise from vortex-induced contributions caused by minute magnetic field misalignment, complicating its interpretation. Here, we investigate a noncentrosymmetric $[Nb/V/Ta]_{40}$ superlattice and demonstrate nonreciprocity in all three fundamental superconducting properties: the critical current, critical field, and superconducting transition temperature. By measuring the angular dependence of the nonreciprocal critical field $\Delta B_{\mathrm{c} 2}$, we separate the contribution of misalignment-induced out-of-plane vortices from the in-plane response. We identify a sharply enhanced SDE emerging near $B_{\mathrm{c} 2}$, accompanied by nonreciprocal responses in $T_{\mathrm{c}}$. These observations exhibit phenomenology similar to that reported in layered superconductors, where an orbital-FFLO state with finite-momentum Cooper pairing has been reported. Our results establish a protocol for assessing out-of-plane vortex contributions to the superconducting diode effect and demonstrate that nonreciprocal responses provide complementary probes of unconventional superconducting states.
Read the full abstract
Machine Learning
Causal Heteroscedastic Structure Learning From Incomplete Temporal data
Discovering causal structures in multivariate time series (MTS) is critical in domains such as finance and neuroscience, yet remains challenging in practice due to three distinct obstacles. First, MTS are routinely corrupted by missing values, where naive impute-then-discover pipelines distort the underlying data distribution and yield spurious causal graphs. Second, identifying instantaneous effects is difficult even on complete data, and missingness further entangles imputation with graph recovery. Third, real-world data frequently exhibits heteroscedastic noise with variance driven by both instantaneous and lagged causes, violating the core assumptions of standard discovery methods. Existing approaches fail to address these challenges concurrently, typically presuming complete data, homoscedastic noise, or strictly time-lagged relationships. Crucially, we show through ablation studies that resolving these issues sequentially, rather than jointly, leads to a substantial degradation in performance. To overcome this, we introduce Cheesefill, an optimal transport framework for causal structure learning under missingness and heteroscedastic noise. Cheesefill parameterizes causal mechanisms using conditional normalizing flows and jointly learns a stochastic correction map that refines naive imputations into trajectories consistent with the inferred causal mechanisms. Theoretically, we prove that optimizing over stochastic correction maps is equivalent to a conditional kernel reformulation of the Kantorovich problem, from which our neural parameterization yields a tractable objective. Extensive experiments on synthetic and real-world benchmarks demonstrate that Cheesefill consistently outperforms existing baselines across diverse missingness regimes.
Read the full abstract
Number Theory
Trivial solutions in Generalized Fermat equations
The modular method is a powerful tool to study generalized Fermat equations. Trivial solutions often become an obstruction to the method. While using hypergeometric motives to construct Galois representations, some of these solutions give rise to newforms with complex multiplication. In the present article, assuming a standard conjecture in Darmon's program, we prove that asymptotically any newform with complex multiplication cannot be related to a non-trivial primitive solution. This allows to discard all such newforms, including those associated to trivial solutions.
Read the full abstract
Applied Physics
Parameterized Physics-Informed Neural Networks for Efficient Simulation of Embedded Microfluidic Chip Cooling
Embedded microfluidic cooling is a promising technology for dissipating the intense heat generated by high-performance electronic chips. However, the design and optimization of such coolers impose prohibitive computational costs when using conventional numerical solvers. Here, we present a parameterized physics-informed neural network (PINN) as a deep learning framework for rapidly and accurately predicting the thermohydraulic performance of embedded manifold microchannels. By incorporating four geometric dimensions along with three spatial coordinates as the network inputs, the proposed model enables multiphysics prediction across a continuous design space. Remarkably, our predicted velocity and temperature fields differ from finite-element simulations by relative errors of only 8.1% and 5.6%, respectively, while reducing the computational time and cost by over four orders of magnitude. Key to this framework is a composite strategy combining geometric ansatz embedding, the SOAP optimizer, and a thermal-conductivity-based curriculum learning scheme, which helps overcome the convergence challenges inherent in training PINNs for complex three-dimensional flows. Notably, by leveraging a highly optimized JAX backend, the network achieves convergence on a single consumer-grade GPU in less than four hours. Our approach offers a physics-consistent, real-time surrogate model that can substantially accelerate innovations in advanced microfluidic cooling of future electronics.
Read the full abstract
Computer Vision
Physically Guided TeX Generation for Low-Light Visual Tasks
Low-light conditions reduce visible contrast and obscure details, while infrared images lack color and texture, motivating the fusion of infrared and visible information for visual tasks. Existing fusion methods primarily optimize image appearance or task-specific features without explicitly disentangling the coupled factors underlying thermal observations, limiting their physical interpretation under thermal degeneracy. Hyperspectral thermal measurements enable physical decomposition, but they require specialized imaging hardware, while pixel-level physical annotations for supervision or evaluation incur additional costs. We propose TeX-Gen, a physically guided framework for TeX generation from paired visible and infrared images in low-light visual tasks. TeX Generation constructs complementary multi-view features and employs independent branch-specific gates to route cues into the Temperature, Emissivity, and Texture maps. TeXMetric integrates radiometric reconstruction, channel structure, and material consistency to assess the generated maps without requiring pixel-level physical labels. TeX Tune improves performance through task-guided adjustments, jointly optimizing the adjustment modules and task network while keeping the generator frozen. Experiments show that TeX-Gen achieves competitive task performance in low-light environments on the FLIR, LLVIP, M3FD, and MSRS datasets. Experiments on HADAR also examine the association between TeXMetric scores and downstream detection performance.
Read the full abstract
Robotics
Task-Critical Part Alignment Improves Imitation of Object Interactions
This paper presents enhancements to the DINOBot framework for imitation learning of object-object interactions. We introduce two key contributions aimed at improving the accuracy and efficiency of alignment in imitation-based robotic tasks. First, we integrate RANdom SAmple Consensus (RANSAC) for outlier rejection during the alignment phase, effectively mitigating the impact of noisy depth data and spurious feature matches. This improves robustness when transferring demonstrations to real-world applications. Second, we refine the semantic correspondence process by focusing on task-relevant interaction regions, ensuring that the learned representation prioritizes critical object features that influence successful manipulation. We evaluated our approach in a range of real-world manipulation tasks, including picking, insertion, shaking, and pouring, using a robotic system. The experimental results demonstrate an overall improvement of 13.8% in task success rates compared to the baseline, along with faster convergence in the alignment process. These results highlight the effectiveness of our enhancements in improving precision and efficiency for object-object interaction imitation in robotics.