Category Theory
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Saturday 12 September
We construct a categorification of the functor of grouplike elements of a Hopf algebra in the setting of braided monoidal categories. Given a braided bistrong monoidal functor admitting a right adjoint, we define a functor assigning to each Hopf monoid a group object obtained as a subobject of its underlying object by means of finite limits. We prove that this construction is functorial and provides a right adjoint to the induced free Hopf monoid functor, thereby extending the classical correspondence between groups and grouplike elements of Hopf algebras to a categorical setting.
Applied Physics
Magnetic Particle Spectroscopy for Detecting Cell-Associated Zinc Ferrite Nanoparticles and Probing Their Relaxation Dynamics
Magnetic particle spectroscopy (MPS) enables sensitive detection of magnetic nanoparticles (MNPs) and characterization of their dynamic magnetization through higher-order harmonics. Here, we synthesized citrate-functionalized, 30 nm cubic Zn0.4Fe2.6O4 (ZFO) MNPs and investigated their association with SKOV3 ovarian cancer cells using MPS. The ZFO MNPs exhibited a crystalline spinel structure, a mean hydrodynamic diameter of 39.4 nm, strong room-temperature magnetization, low coercivity, and citrate-associated surface functional groups. Live/Dead imaging indicated good cytocompatibility after 24 h exposure at concentrations up to 0.5 mg/mL, while bright-field microscopy showed concentration-dependent cell-associated nanoparticle accumulation. Two MPS drive fields were compared using ZFO MNPs dispersed in DI water and agar as relatively unrestricted and strongly confined reference states. The 7.75 kHz, 20 mT condition retained more higher-order harmonics than 11.37 kHz and 10 mT and produced a larger and order-dependent spectral separation between the two states; it was therefore selected for the cellular measurements. After ZFO exposure and removal of unbound nanoparticles, MPS detected cell-associated MNPs in samples containing 0.1, 1, and 2x10^6 SKOV3 cells. The 3rd-, 5th-, and 7th-harmonic amplitudes increased proportionally with cell number, with power-law exponents of 1.03-1.09. We also report that most normalized harmonic ratios from the 1- and 2x10^6-cell samples fell between the water and agar references, indicating partial restriction of Brownian rotation in the cellular environment. These findings demonstrate that MPS can detect cell-associated MNPs while providing complementary spectral information about their ensemble-averaged physical confinement, supporting its application in magnetic cell labeling and cell-tracking studies.
Read the full abstract
Number Theory
Quantitative linear independence for square roots
We consider the problem of finding lower bounds for integer linear combinations of $\sqrt{a_1},\ldots,\sqrt{a_K}$, where $a_1,\ldots,a_K$ are positive integers such that their square roots are linearly independent over the rationals. We use a probabilistic approach and prove that for $K\geq 8$ and nonzero integers $m_1,\ldots,m_K$, $$ \bigg|\sum_{n\leq K} m_n\sqrt{a_n}\bigg| > e^{\frac{2^{K-1}-1}{K} - \frac{1}{2}} \bigg(\max_{n\leq K}|m_n|\sqrt{a_n}\cdot \sqrt{K}\bigg)^{-(2^{K-1}-1)}. $$ This inequality improves the dependence on $K$ in the classical product bound.
Read the full abstract
General Relativity and Quantum Cosmology
Quantum black hole ringdown
The ringdown of a black hole following a merger is a potential candidate for revealing the signatures of quantum gravity in the emerging gravitational waves. In quantum theory, black holes are expected to have a discrete area and energy spectrum, which conflicts with the classical notion of an horizon that absorbs all infalling perturbations. We propose that the quantum black hole dissipates the energy it cannot absorb by emitting “soft” gravitons that carry away the energy difference between the energy of the infalling perturbation and the quantum transition energy. We find that the ringdown spectrum is consequently enriched with low-frequency components, and that there exists a weak low-frequency flux that persists for a timescale much longer than the ringdown itself.
Read the full abstract
Computer Vision
3D Gait-Based Autism Classification Using Attention-Enhanced Deep Learning with Cross-Fold Statistical Stability Analysis
Autism Spectrum Disorder (ASD) is a neurodevelopmental condition whose early diagnosis remains challenging because conventional clinical assessments are often subjective, time-consuming, and require expert evaluation. Gait provides a promising non-invasive behavioral biomarker for auto- mated ASD screening; however, existing studies have primarily relied on single-dataset evaluations, convolutional architectures, and descriptive summaries of cross-validation performance without formally assessing fold-to-fold stability. This study addresses these gaps with an attention-enhanced Transformer framework for ASD classification, evaluated on two structurally different 3D gait feature representations: precomputed statistical gait descriptors and raw biomechanical ground-reaction- force measurements. Under five-fold cross-validation, the proposed framework achieved 99.00% accuracy, 99.02% precision, 99.00% recall, 99.00% F1-score, and 99.00% specificity on the public Kinect-based benchmark, exceeding the performance of the compared state-of-the-art methods. On the independent private force-plate dataset, it achieved mean values of 95.00% accuracy, 93.81% precision, 96.67% recall, 95.13% F1-score, and 93.33% specificity.
Read the full abstract
Analysis of PDEs
On Periodic Solutions of the Zakharov-Rubenchik System
We study the one-dimensional Zakharov-Rubenchik system with periodic initial conditions. This system models the interaction of low-frequency waves with high-frequency waves in different physical phenomena. Using semigroup theory, we prove the global well-posedness of the associated linear equation in $H^q(\mathbb{T})\times H^s(\mathbb{T})\times H^{s+1}(\mathbb{T})$, $q,s \in \mathbb{R}$. Subsequently, we consider the modified Zakharov-Rubenchik system and, by means of a change of variables and the Banach Fixed Point Theorem, we obtain local well-posedness in the space $H^s(\mathbb{T})\times H^s(\mathbb{T})\times H^{s+1}(\mathbb{T})$, $s>3/2$.
Read the full abstract
Computation and Language
When Tools Get in the Way: The Effect of Unnecessary Tool Availability on LLM Answering
Large language models (LLMs) are increasingly deployed with external tools that extend what they can do beyond their own knowledge. Tools help on tasks that need external information, but their availability may also change how a model handles questions that do not need them. Prior work has mostly asked whether models select and use tools appropriately; whether an unnecessary tool changes the correctness of answers has received less attention. We ask whether making a related but unnecessary tool available affects a model's ability to answer from its own knowledge, and whether a preceding tool interaction changes this behaviour. We construct 500 query pairs across 10 knowledge domains. Each pair consists of a tool query, which needs the domain's tool, and a closed-domain query, which does not. Six LLMs are evaluated with the tool unavailable, available, and available after a prior tool call. Across 3,000 baseline trials the pooled answer rate is 98.2%. When an unnecessary tool is available it falls to 63.5%, with large differences between models. The decrease occurs even when the tool is rarely called, so it cannot be explained by unnecessary tool invocation alone. A one-sentence scope-aware system instruction recovers most of the lost answers.
Read the full abstract
Robotics
Visible Touch: Rendering Contact for Visuomotor Policies
Integrating contact information into visuomotor policies remains an open problem. Touch is essential to robust manipulation, yet most modern policies, including pretrained vision-language-action (VLA) models, operate from vision and proprioception alone. Existing approaches to closing this gap require specialized tactile hardware, add separate tactile encoders, or commit to non-image policy backbones, all incompatible with the modern paradigm of image-conditioned policies built on pretrained 2D visual representations. Our key insight is that the bottleneck is not the contact information itself, but how it is delivered: when contact signals are exposed in the same spatial frame as the scene the policy already attends to, they become directly usable by any image-conditioned policy without architectural changes. We operationalize this insight in Visible Touch, paired with a custom low-cost magnetic contact sensor that is open-sourced and fabricated from off-the-shelf parts via a parametric CAD-to-mold pipeline. Across the LIBERO benchmark, Visible Touch improves BC-Transformer success by 15.7 percentage points on average in the 2-view setting, with similar gains in the 1-view setting; controlled comparisons show that the contact-integration strategy strongly affects how effectively tactile information is used. The pattern holds when fine-tuning pretrained VLAs: miniVLA on LIBERO gains 25 percentage points on average, and $\pi_{0.5}$ on four real-world contact-rich tasks gains 30 percentage points with our custom sensor.
Read the full abstract
Quantum Gases
Phonon Scharnhorst effect: Acoustic analogue of the vacuum-fluctuation-boosted speed of light
The Scharnhorst effect in quantum electrodynamics predicts a superluminal velocity shift $\delta c>0$ at low photon frequencies due to vacuum fluctuations between two mirrors. Although theoretically robust (no causality violation), the effect is extraordinarily small ($\delta c/c\approx 10^{-32}$ for mirrors $1\,\mu{\rm m}$ apart) and outside current experimental reach. Here we propose a condensed matter analogue, replacing the photons in the electromagnetic vacuum by phonon excitations of a Bose-Einstein condensate. In an idealized geometry, the two-dimensional surface of a cylinder (circumference $L$) at zero temperature, the axial speed of sound is increased due to zero-point fluctuations of the condensate by $\delta c/c\simeq 5\times 10^{-3}\,\tilde{g}$ at $L\simeq 2\,\xi$, to first order in the dimensionless interaction constant $\tilde{g}\simeq 10^{-1}$ of the Bose gas of healing length $\xi\simeq 1\,\mu{\rm m}$. This is 28 orders of magnitude larger than the photonic effect, basically because phonons have a self-interaction while photons only interact via massive electrons. We discuss the prospects for observation of the acoustic Scharnhorst effect, the main restriction being the low-temperature requirement.
Read the full abstract
High Energy Astrophysical Phenomena
The Galactic Black Hole mass distribution
The population of extragalactic black holes has been significantly expanded and better understood since the first detection of gravitational waves. Over the past decade, approximately 290 events have been confirmed to involve at least one black hole. In contrast, the galactic black hole population has grown at a much slower pace, with only 42 systems possessing mass estimates, while the vast majority remain as candidates. In this work, we present the first update of a Bayesian analysis of the mass distribution of galactic black holes since 2011, previously based on a sample of 15 low-mass X-ray binaries and 5 high-mass X-ray binaries. Our update includes 28 low-mass X-ray binaries, 6 high-mass X-ray binaries, 7 detached binaries and 1 isolated black hole, also present in the living catalog CatNoir which we introduce. We apply a Bayesian framework using five parametric models – Gaussian, two-Gaussian, log-normal, power-law, and decaying exponential – and compare their performance in fitting the observed data. Furthermore, we inspect the effect of removing model-dependent systems from the sample. Our results suggest that the power law better describes the Combined sample, while the Gaussian is preferred for the LMXB-only sample, closely followed by the two-Gaussian and log-normal. We derive the $1\%$-quantile of the preferred model's posterior predictive distribution to access information about the lower mass gap. For the Combined and LMXB Samples, we find $M_{1\%} > 4.06\,M_\odot$ and $M_{1\%} > 2.56\,M_\odot$ (at $90\%$ C.I.), respectively. The LMXB result allows for the existence of a sparsely populated Gap, instead of a completely empty one.
Read the full abstract
Quantum Physics
Programmable nonlinear function synthesis on a photonic processor with quantum signal processing
Programmable photonic integrated circuits are emerging as increasingly large and versatile interferometric processors operating at room temperature. However, their native operations are linear, while many computational tasks require nonlinear input-output transformations that typically rely on nonlinear optical materials or resonant devices. Here, we establish a direct mapping between quantum signal processing (QSP) and the native two-mode operations of programmable interferometric photonic integrated circuits, showing that the $SU(2)$ structure required by QSP can be realized through programmable phase control and mode mixing. We experimentally demonstrate this correspondence using dual-rail single-photon encoding on a 24-mode programmable photonic integrated circuit, realizing QSP sequences up to depth $L=11$ and synthesizing STEP, ReLU and SELU functions through programmable phase control. The optical transformation remains linear, while repeated encoding of the input variable and coherent interference produce a nonlinear dependence of the output probabilities on the encoded variable. Across all accessible circuit depths, the measured responses follow the programmed QSP transformations, with hardware-induced mean squared errors between $10^{-3}$ and $10^{-2}$, all remaining below the intrinsic finite-depth approximation error. These results establish an algorithm-to-hardware mapping of QSP on programmable photonic processors. They also demonstrate an algorithmic route to programmable nonlinear functions of encoded variables on linear photonic hardware.
Read the full abstract
Analysis of PDEs
A nonlinear problem related to optimal insulation
We study a generalization of the classical optimal insulation problem by replacing the standard Laplacian with the \ti{p-Laplacian}, leading to a \ti{p-Poisson equation} with Robin boundary conditions. We also investigate the corresponding eigenvalue problem. This work extends and complements some results from \cite{bu1,pietra1,pietra2} to the nonlinear setting.
Read the full abstract
Machine Learning
Signatures of Steerability in Activation Space of Language Models
Steering language models using a set of contrastive representations has been a canonical and computationally efficient method for controlling model behavior. Despite this success in controlling certain model behaviors, the effectiveness of activation steering varies markedly across concepts; the generalization properties of steering vectors are often considered a function of the dataset used to construct them. We make this dataset-dependence claim more rigorous and show that simple separation metrics strongly correlate with the downstream steerability of language models across diverse settings, even after controlling for layers and dataset effects. Beyond prediction, we provide evidence from a synthetic superposition experiment that separation metrics are strongly correlated with alignment between the empirical and true feature direction. Our results suggest that simple separability statistics can serve as practical diagnostics for when steering vectors are likely to work.
Read the full abstract
Computational Geometry
Computing the minimal perimeter polygon for digital objects in the triangular tiling
This work presents an algorithm, together with its correctness proof, to determine the minimum perimeter polygon (MPP) for digital objects given as regular complexes in the triangular plane tiling. Such objects are edge-adjacency-connected sets of triangle tiles that have no end tiles, and the point set union of all their tiles forms a simple polygon. Nevertheless, the boundary paths of the objects are not assumed to be simple. Then the MPP is a weakly simple polygon that coincides with the relative convex hull (i.e., geodesic hull) of a set $A$ with respect to a simple polygon $B$, where $A\subset B$, but $A$ is not necessarily a polygon, in fact it is generally not connected. Our MPP algorithm relies on constructing and iteratively constraining cones of visibility through forthcoming boundary tiles, it uses the structure of the canonical boundary path, the MPP frontier is the shortest polygonal curve following this path. We also propose a boundary tracing algorithm to obtain such paths from the objects.
Read the full abstract
Number Theory
Patterns in the Markov numbers and their generalizations
Positive integer solutions to $x^2+y^2+z^2- x y z =D$ correspond to the important Markov (Markoff) numbers when $D=0$. From a given solution triple, three more are found with Vieta involutions, making an infinite tree of solutions. Starting instead with three real numbers greater than $2$ gives lengths of closed geodesics in a punctured torus. In this paper we study these tree structures for any real $D$. A continuous function, originally related to a norm on homology, encodes all the numbers on each of these trees. The properties of this function are developed here in general, with a self-contained exposition, showing that the usual $D=0$ case is part of a bigger picture. Encoding function graphs are shown to be convex for $D<4$, straight lines for $D=4$ and concave for $D>4$. Among other consequences are generalizations to all $D$ of: estimates for counting numbers on these trees, descriptions of the geometry of the corresponding lattice curves, uniqueness conditions, and identities of McShane and Hines.
Read the full abstract
Functional Analysis
A positive version of Miljutin's Theorem
In this paper we provide a positive version of Miljutin's theorem. Namely, we show that for any uncountable compact metric spaces $K$ and $L$ there exists a positive isomorphism $T:C(K)\longrightarrow C(L)$. Moreover, the inverse of $T$ is a regular operator, and $T$ may be chosen so that $\|T\|_\mathrm{r} \cdot \|T^{-1}\|_\mathrm{r} \leq 9+6\sqrt3$. The method also gives positive versions of classical nonmetrizable results for Miljutin and Dugundji compacta. In particular, if $G$ and $H$ are infinite compact groups of the same weight, then $C(G)$ and $C(H)$ are positively and regularly isomorphic with the same uniform quantitative estimate.
Read the full abstract
Genomics
RAGCell: Retrieval-Augmented Generation as Supervision for Versatile Single-cell Analysis
Single-cell foundation models (scFMs) are transforming computational biology by enabling generalizable, task-agnostic representations for versatile single-cell analysis. Despite their progress in facilitating rapid deployment for downstream tasks, off-the-shelf scFMs still have some overlooked concerns: (I) (Pretraining Cost.) Pretrain-based scFMs necessitate pretraining on a vast volume of cells, rendering it draining resources in applications. (II) (Heterogeneous Gap.) Large Language Models (LLM)-based scFMs ignore the tremendous heterogeneous gap between LLM textual and raw cellular spaces, leading to insufficient capability when facing downstream tasks. To this end, we introduce RAGCell, a versatile single-cell analysis framework that achieves a double-win in both cost-effectiveness and high performance. The success of RAGCell lies in two key aspects: Leveraging LLMs to construct cell-level and feature-level knowledge databases, which serve as supervision signals for training the cell model and significantly reduce the training cost ($>$pretrain-based scFMs). Aligning cell representations with text embeddings from the bi-level knowledge databases, enabling knowledge transfer from textual spaces to cellular spaces and effectively mitigating the heterogeneous gap ($>$LLM-based scFMs). Through extensive experiments on six downstream single-cell analysis tasks, we demonstrate that RAGCell achieves outstanding performance compared to state-of-the-art scFMs while operating at less than $\sim$1/10 the cost of pretrain-based scFMs.
Read the full abstract
Robotics
When Faster VLA Deployment Changes Closed-Loop Behavior: Task Success-Latency Analysis of SmolVLA Across PyTorch and ONNX Variants
Vision-language-action (VLA) deployment can reduce inference latency while changing closed-loop task behavior. We evaluate HuggingFaceVLA/smolvla_libero on an RTX 2060 (6 GB) in LIBERO Spatial and Object (MuJoCo 3.3.2, LeRobot 0.6.1, seed 42), comparing PyTorch+AMP with ONNX Runtime CUDA Execution Provider (CUDA EP). The main evaluation uses 100 episodes/suite; a paired rollout uses 300 episodes/suite. PyTorch+AMP reaches 70.0%/88.0% Spatial/Object success at 1181 ms p99. Requested-FP16 and requested-INT8 ONNX reduce tether-inspect p99 to 601 ms and 532 ms, while Spatial success falls to 41.0% and 40.0% and Object remains at 89.0%. A graph audit shows those artifacts are byte-identical FP32 graphs, so the requested-INT8 row is not operator-level INT8 quantization. A static language-width ablation (16/24/32 tokens) yields Spatial success of 41.0%, 75.0%, and 71.0%; widths 24 and 32 recover much of the Spatial drop while Object success and uniform-bench latency stay approximately stable. Width-24 ONNX Spatial success is comparable to the PyTorch+AMP baseline at roughly half the latency (Wilson intervals overlap; two-proportion chi-squared p=0.53). Context width is an important contributor in this stack; it does not account for every PyTorch-vs-ONNX difference. Deployment evaluation should jointly report latency, artifact inspection, interface constraints, and closed-loop success. Code: https://github.com/rafiqul713/smolvla-libero-onnx.
Read the full abstract
Sound
A New Transformer-Based Approach for Audio-Based Kinship Verification and a New Uncontrolled Mandarin Kinship Speech Dataset
Kinship verification is a task involving determining whether two individuals share a first-order kin relation. To tackle this task, we propose CONVTRAP-TN, a new architecture for audio-based kinship verification, and conduct an ablation study on the proposed model. To the best of our knowledge, we are the first to apply the successful transformer architecture to the task of audio-based kinship verification. Furthermore, we also collect a custom speech dataset, ARKIN, which accurately reflects everyday recording conditions. We do this because only a few speech datasets with kinship labels currently exist, all of which either source extremely noisy in-the-wild data from the internet, or instruct speakers to record in specific environments. These settings fail to reflect real-world scenarios where users record on personal devices under unrestrained conditions. Additionally, we perform a series of preliminary baseline experiments on the collected dataset, including speaker verification and recognition, speech recognition, age estimation, and kinship verification, as well as cross-dataset kinship verification experiments to show that existing methods are not robust across datasets.
Read the full abstract
Computation and Language
One Size Does Not Fit All: Setting Inference Depth from the Questions a Deployment Actually Asks
A transformer language model is trained to respond to any prompt, but each deployment asks only a narrow range of questions: a support assistant sees delivery complaints, a coding tool sees Python. Every deployment nonetheless pays the same computation per token. This paper measures how much of that cost is avoidable when the range of prompts is known in advance. The mechanism examined is early exit: a small, trained component - called a readout - is attached to an intermediate layer and proposes a token, and a confidence test decides whether to emit it or to run the remaining layers. The models are frozen, and the only supervision used is the model's own output on ordinary traffic. Three findings are reported. First, achievable savings depend strongly on the kind of traffic: at half depth on a 1.5-billion-parameter model, 96 percent of tokens could be emitted early for arithmetic word problems and 8 percent for Chinese-language explanations, at matched token-level fidelity to the full model (a measure whose limits the third finding exposes). Second, of three ways a deployment might use knowledge of its traffic, only customizing the threshold for exiting early is worthwhile: calibrating it per deployment raised exit rates by up to 59 percentage points across three models, and by more than 10 points on most corpora tested. Third, token-level fidelity - the standard evaluation measure in the early-exit literature - fails in domains where tokens can be checked against ground truth: on arithmetic word problems, three models each answered sixty questions correctly when run in full, and between 10 and 28 correctly under early exit, in the configuration that scored highest on fidelity. The intended setting is small models on personal devices, where generation is limited by memory bandwidth rather than computation.
Read the full abstract
Solar and Stellar Astrophysics
Reconstruction of pretelescopic and early telescopic solar activity cycles from auroral records
The historical record of low-latitude aurorae is essentially a poorly sampled record of the largest space weather events (SWEs). Its use for the identification of individual solar cycles is hindered by the low event rate and by the fact that the solar cycle profile of the occurrence of SWEs does not closely follow the variation of sunspot numbers. Based on recent studies of the solar cycle dependence of the occurrence rates of large SWEs, here we construct Monte-Carlo simulations of a large number of activity cycles to identify the optimal procedure to infer the characteristics of underlying solar cycles from the sparse record. We find that a reliable reconstruction of the cycle phase ($>90$% of reconstructed minima corresponding to actual minima within $\pm 2$ years) is possible whenever the long-term mean event rate (annual mean number of space weather events resulting in low-latitude auroral sightings) reaches or exceeds a value around 3. This condition is found to be satisfied during most of the the Early Modern Active Period (EMAP), a century-long period of normal solar activity between the Spörer and Maunder Minima. For the numbering of solar cycles in the EMAP we introduce the "telescopic era", where T$n$ denotes the $n$th cycle from the first telescopically observed cycle, T$0$, ongoing in 1610. Using our optimal procedure we reconstruct a series of 8 solar activity cycles from T$-5$ to T$2$ (1560-1640). Earlier cycles starting from 1540 can be reconstructed with a somewhat lower degree of reliability. Comparing our results with radionuclide-based reconstructions and sunspot observations we find a good overall correspondence, with the exception of the last cycle before the Maunder Minimum.
Read the full abstract
Machine Learning
T-SMART: Mechanism-Level Attribution for Tool-Augmented Time-Series Question Answering
Large language models (LLMs) can struggle with time-series question answering (TS-QA), especially when numerical signals are serialized as text and require explicit computation. Tool-augmented approaches improve performance, but existing systems often intertwine language reasoning, computation, and perception, making it difficult to determine which components drive the gains. We present T-SMART, a neurosymbolic framework that separates these roles: a frozen LLM interprets questions and selects operations, deterministic tools perform numerical computation, and structured perception is invoked only when needed. Controlled paired ablations show that deterministic computation provides the dominant benefit, improving accuracy by 31.7 percentage points over direct LLM reasoning on serialized time series, while language understanding and perception offer smaller complementary gains. These results indicate that tool-augmented TS-QA benefits primarily from reliable numerical execution rather than additional language-model reasoning and provide a controlled framework for analyzing component contributions in neurosymbolic time-series systems.
Read the full abstract
Optimization and Control
Riemannian ascent–descent for nonconvex nonconcave minimax landscapes: convergence to basin saddle points and applications to distributionally robust optimization
We study a class of distributionally robust optimization (DRO) problems for the statistical risk problem, formulated as minimax problems over the product of a Euclidean space and a Riemannian manifold. Because the resulting minimax landscape is nonconvex nonconcave in general, no globally convergent first order method is known to be available. We instead introduce the notion of a basin saddle point, a Nash equilibrium defined locally on the Cartesian product of a $\delta$ basin around a connected component of the local minima critical set and a geodesic ball on the measure manifold. We develop an abstract convergence framework for a Riemannian gradient ascent multistep descent iteration to a basin saddle point under a local Łojasiewicz type growth condition, with exponent $\beta \in (1,2]$, in the $\delta$ basin around connected components of the local minima critical sets. Under Lipschitz regularity of critical sets we establish linear convergence for $\beta = 2$ and polynomial convergence for $\beta \in (1,2)$ to a basin saddle point, with explicit dependence on the sectional curvature of the manifold. We then instantiate this framework for the statistical risk DRO problem over Gaussian measures, where the ambiguity set is naturally modeled as the product of Euclidean space and the Bures Wasserstein manifold of covariance matrices, which we relax to a penalized DRO formulation. We derive nonasymptotic Hessian estimates for the resulting Lagrangian, establish existence and local uniqueness of its maximizer, and prove that an alternating Riemannian gradient scheme converges to a basin saddle point of the penalized DRO problem, recovering the linear and polynomial rates of the abstract theory with all constants explicit in terms of data dimension, loss moments, and the reference covariance.
Read the full abstract
Dynamical Systems
Limit Cycles in a Discontinuous Generalized Liénard Systems with Bivariate Perturbation
We study the number of limit cycles of the piecewise smooth differential system \[ \dot{x}=y, \quad \dot{y}=-x-\varepsilon\bigl(f(x)y+\operatorname{sgn}(x)g(x,y)\bigr),\] where $f(x)$ is a polynomial of degree $n\geq 1$, $g(x,y)$ is a bivariate polynomial of degree $m\geq 2$, and $\varepsilon$ is a sufficiently small parameter. Using the first-order averaging theorem for discontinuous systems, we obtain that the number $H_{m,n}=\lfloor n/2\rfloor+\lfloor m/2\rfloor$ is a lower bound for the number of limit cycles bifurcating from the linear center $\dot{x}=y, \ \dot{y}=-x$. The bound is sharp, and we provide an explicit example that attains $H_{m,n}$ limit cycles.
Read the full abstract
Sound
Robust Cross-Domain Speech-Based Alzheimer's Disease Detection via Iterative Adversarial Self-Training
As Alzheimer's disease (AD) has increasingly become a major global public health issue, speech-based AD detection has attracted widespread attention. However, most existing methods are trained and evaluated on a single dataset, often leading to severe cross-domain performance degradation due to reliance on dataset-specific artifacts rather than disease-related speech cues. In real-world applications, reliable Alzheimer's disease detection requires models that are robust to variations in recording environments, speakers and data collection conditions. To address this challenge, this paper adopts unsupervised domain adaptation to learn robust, domain-invariant feature representations in the absence of target-domain diagnosis labels. On this basis, a novel unsupervised domain adaptation method, Iterative Adversarial Self-Training (IAST), is proposed. Results demonstrate that IAST significantly improves the generalization ability and robustness under various cross-domain settings.
Read the full abstract
Machine Learning
LIMBO: Lifelong Inference-Time Memory and Budget Optimization for LLM Agents
As LLM agents become integrated into increasingly complex workflows, they must continually acquire new capabilities while retaining competence on previously learned tasks. Lifelong agents address this through experience replay, injecting past interactions into the prompt to leverage prior experience during inference. However, replay is not free: every replayed trajectory competes with retrieval, reasoning, tool use, and verification for the same limited prompt and compute budget, making effective resource allocation essential. Existing approaches allocate these resources using fixed replay policies, regardless of whether replay is beneficial for the current task. We identify this as inference-time memory allocation, a distinct problem class for lifelong agents, and introduce LIMBO: the first online framework to our knowledge that treats memory as a controllable inference-time resource and jointly optimizes memory strategy and inference budget for each incoming task. Unlike prior approaches that fix the replay policy or require model weights, teacher supervision, or offline retraining, LIMBO learns this allocation online in a single pass, explicitly balancing task performance and inference cost without modifying the underlying agent. Across three LLM backbones on LifelongAgentBench, LIMBO achieves better cost-accuracy tradeoffs than state-of-the-art memory-augmented baselines and nearly matches all strongest such baselines at up to ~83% lower inference cost (~53% on average). LIMBO adapts its policy across models and environments without retraining, demonstrating that effective allocation can be learned online rather than manually specified.
Read the full abstract
Cryptography and Security
PixCrypt: Fast Fine-Grained FHE with Range-Aware Caching
Many analytics tasks require secure computation over encrypted data. In particular, fine-grained data such as pixel-level images require higher precision, as every pixel can directly affect outcomes in tasks like tumor segmentation and anomaly detection. While Multi-Party Computation (MPC) is interactive, Differential Privacy (DP) protects only aggregate values, and Partially Homomorphic Encryption (PHE) lacks multiplicative support, none of them can efficiently handle fine-grained data analytics. Fully Homomorphic Encryption (FHE) uniquely enables arbitrary operations on encrypted pixels but remains computationally expensive, posing significant challenges for both software and hardware accelerators. We present PixCrypt, a caching-based acceleration mechanism for fine-grained fully homomorphic encryption. PixCrypt replaces expensive fresh ciphertext generation with cache retrieval and coefficient-level operations across CKKS, BFV, and BGV, while randomized reconstruction ensures that ciphertexts do not repeat. Its linear noise growth reduces the need for bootstrapping and lowers NTT load, improving hardware accelerator efficiency. This design yields up to 35x faster fine-grained encryption and maintains IND-CPA (Indistinguishability under Chosen Plaintext Attack) security. Experiments on five real-world pixel-level image processing tasks show that PixCrypt significantly improves the practicality of FHE for privacy-preserving analytics.
Read the full abstract
Data Analysis and Statistics
Finite Asimov Sample Construction in Unbinned Neural Simulation-Based Inference
Expected sensitivity calculations in frequentist analysis using neural simulation-based inference often require large simulated samples to approximate an unbinned Asimov dataset. We construct a finite weighted reference sample, for analyses based on learned density ratios, for which the generating parameters globally maximize the weighted likelihood. A toy example with five-dimensional observable space based loosely on high-energy physics models shows that an Asimov dataset consisting of as few as 256 weighted reference events closely reproduce the expected test statistic scans obtained with two million simulated reference events. The resulting test statistic distributions are also shown to agree with independent simulator-based pseudo-experiments. We note that this agreement with the simulator depends on the accuracy of the learned ratios and must be validated explicitly.
Read the full abstract
Algebraic Geometry
Theta-duality and Prym-Torelli for cyclic covers of hyperelliptic curves
Let $f:\tilde{C}\to C$ be an étale cyclic cover of odd prime degree $d$ of a hyperelliptic curve of genus $g\geq 2$. We revisit the theta-duality reconstruction argument for the generic injectivity of the associated Prym map by Naranjo-Ortega-Pirola-Spelta and identify an additional residual locus arising from the fixed divisor of the line bundles occurring in that argument. For $d\ge5$, this locus does not affect reconstruction, giving injectivity under the numerical assumption $(d-1)(g-1)\geq 7$. For $d=3$, the geometry is governed instead by trigonal pencils on the quotient of $\tilde{C}$ by a lift of the hyperelliptic involution; this yields generic degree $2$ in genus $5$ and injectivity for $g\ge6$.
Read the full abstract
Strongly Correlated Electrons
Exact solutions of hidden free fermion models
Hidden-free-fermion (HFF) Hamiltonians exhibit free-fermion spectra despite lacking an explicit quadratic representation, yet their algebraic provenance and periodic-boundary behavior remain open questions. Here we establish that a broad class of HFF models arises from distinct Lax operator realizations of a twisted multistate Perk–Schultz \(R\) matrix. Its reflected degeneration generates a closed finite hierarchy of mutually commuting transfer matrices, whose functional relations determine the complete many-body spectrum. As a decisive example, we solve the inhomogeneous free-fermions-in-disguise (FFD) chain, which exhibits HFF, nested HFF, or interacting spectra, depending on its coupling. The homogeneous FFD chain with both one-wrap and periodic boundary conditions exhibits an emergent cubic finite-size gap while retaining the exact \(z=3/2\) bulk behavior. The construction extends to higher-state and face-type realizations, providing a common algebraic framework for coupling-dependent spectral organization and multiscale behavior in interacting quantum systems.