Data Structures and Algorithms
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Saturday 7 November
Quantified Boolean formula (QBF), the quantified version of Boolean satisfiability (SAT), is a fundamental problem in constraint solving. Due to the $\mathsf{PSPACE}$-complete nature of QBF, designing theoretically efficient QBF algorithms is much harder and receives less attention in academia. In this work, we present a new algorithm based on memoized search, which solves QBF with $n$ variables and $m$ clauses in $1.685^m\cdot\text{poly}(n)$ time and space. This is the first improvement over the $1.709^m\cdot\text{poly}(n)$ bound by Williams in over 20 years. Our technique is based on the minimum vertex cover of the conflict graphs, which was proposed to study the parameterized complexity of SAT. We extend this idea to QBF and further show that QBF is slicewise polynomial-time solvable using this parameter.
Computer Vision
Latent-WAM++: Shaping World and Action Representations for End-to-End Autonomous Driving
We introduce Latent-WAM++, a representation-centered World Action Model (WAM) for end-to-end autonomous driving. Existing WAMs are primarily trained to predict future states or final actions, leaving the intermediate world and action representations that connect prediction to planning only indirectly constrained. Latent-WAM++ matches the supervision of each representation to its role in planning. World Representation Construction combines scene-ego future prediction, SIGReg-based distribution stabilization, and frozen geometric priors to learn a temporally predictive, non-degenerate, and geometry-aware world representation without an EMA target encoder. Trajectory MeanFlow Planner directly predicts multimodal trajectory proposals while being optimized with an Improved MeanFlow velocity-space objective, enabling one-step sampling. Visual-Action Representation Alignment further aligns the resulting pre-output action features with a stop-gradient target derived from current visual semantics. Latent-WAM++ achieves state-of-the-art performance on NAVSIM v2 and under zero-shot evaluation on HUGSIM, reaching 90.1 EPDMS on NAVSIM v2 and 47.0 average RC and 30.1 average HD-Score on HUGSIM. Training the planner end to end costs 190 GPU-hours for Latent-WAM++ versus 1,600 for Latent-WAM, a reduction of 88.1%.
Read the full abstract
Social and Information Networks
Place and Political Behavior Online during the Los Angeles Blaze
Understanding how online communication during extreme weather crises enables or constrains climate mobilization is critical. During a crisis, the online world connects distant audiences and complicates the relationship between place and political behavior. We analyze all sociopolitical content from Bluesky about the January 2025 Los Angeles wildfire event, reporting how participation in collective action and allocation of attention vary by geographic distance. Participation varies nonlinearly with proximity, with local and international users participating more than regional and national users. Attention concentrates around regional and national users engaging primarily in viral, non-participatory political discussion. Analysis reveals a geographic mismatch between participation and attention, illuminating how online communication concentrates attention away from collective action amid an exacerbating climate crisis.
Read the full abstract
Sound
Steer Unified Audio Editing Toward Reliable Source Selection
Unified audio editors support diverse operations, yet often struggle to modify the intended source while preserving non-target content. We show that source selection in audio editing is shaped by early denoising states, while incorrect early selections can hardly be corrected during subsequent denoising. Motivated by these findings, we propose Source-selection Training via Early-trajectory Emphasis and Recovery (STEER), a two-stage post-training algorithm for reliable selective audio editing. STEER-Emphasis prioritizes training on task-critical early states to reduce source-selection errors, while STEER-Recovery learns corrective denoising from self-generated rollout states to recover from early deviations. To systematically assess source-selection reliability, we introduce the Selective Audio Editing Benchmark, enabling large-scale automatic evaluation of target manipulation and non-target preservation. Applying STEER to our pretrained multitask base editor gives SteerSound, a unified audio editor with leading selective editing performance among open-source audio editors. Experiments on our base editor, UNISON, and SAM Audio demonstrate that STEER consistently improves both selective editing capability and overall editing performance. Early-trajectory emphasis further brings substantial improvements in transcript-conditioned target speaker removal and separation, highlighting the broader applicability of this training principle.
Read the full abstract
Artificial Intelligence
DFT-Agent Bench: Scaffold Effects and Execution Provenance in Autonomous DFT
Large language models are increasingly deployed as autonomous agents for first-principles materials simulation. We report empirical findings from DFT-Agent Bench, a benchmark of 200+ Quantum ESPRESSO tasks spanning relaxation, electronic structure, equation-of-state, and elastic-tensor pipelines over multiple material tiers. Each task carries a QE-derived ground truth with public-reference cross-checks where available, and a provenance audit checks whether the required QE execution contract was met. Under one execution harness, we compare a near-bare ReAct baseline with a bundled v5.1 scaffold on two models. The complete scaffold improves bands+DOS by +0.08/+0.11 and d-electron vc-relax by +0.17 on both models, while high-scoring relaxation and equation-of-state cells leave little room for improvement. The elastic result is model-dependent (0.18 to 0.40 for deepseek-v4-pro and 0.62 to 0.59 for claude-opus-4-8). A fixed 100-task analysis with retrieval disabled supports the aggregate pattern: the fully frozen DeepSeek comparison gains +0.17, while the Opus comparison gains +0.10 against a historical pool baseline (p < 0.001). The provenance gate catches incomplete or unmarked executions that can receive high rubric scores; it is an execution-contract check, not a test for intentional fraud. These results characterize the evaluated DFT workflows. The anonymous review artifact contains the frozen diagnostic slice, prompts, scorer, and solver revisions: https://anonymous.4open.science/r/231e31re91287h/README.md.
Read the full abstract
Artificial Intelligence
Towards Identity-Preserving LLM Agents under Model Transition
When an LLM agent's backend model is upgraded or replaced, operators typically reload the same persona, memory, and rules onto the new model and assume the agent will behave as before. We show this assumption requires empirical validation-and that whether identity is preserved depends on the specific model pairing. Across 5,760 compliance-testing runs in 12 regulated domains with three open-weight models, we find that some transitions produce significant behavioral drift while others do not, even when the agent's external state remains identical. Crucially, this drift is invisible to verbal verification: every model claims full policy compliance, yet actual decisions under pressure differ across models. The same external state can reorder which model is most compliant-a large positive effect on one substrate, negligible on the other two. Practitioners should treat model transitions as behavioral changes requiring action-level empirical testing, not as transparent infrastructure operations.
Read the full abstract
Machine Learning
CONSTRAINER: Promptable Graph-Structured Optimization via Constraint Conditioning
Neural methods for graph-structured optimization have achieved strong performance within individual problem classes, but are typically trained for fixed formulations, limiting reuse across related problem variants. Recent progress in foundation models has motivated shared neural backbones trained across multiple graph-structured optimization tasks; however, existing approaches primarily share instance-level representations while treating feasibility constraints implicitly. A key source of shared structure across constrained optimization problems is the constraint set itself, which often exhibits substantial overlap across tasks. We propose CONSTRAINER, a constraint-conditioned framework for graph-structured optimization that treats constraints as structured inputs. Constraint specifications are parsed into operator graphs and encoded using a graph neural network to produce invariant constraint embeddings, which condition a shared backbone model for solution prediction. We evaluate our approach on power grid optimization and mixed-integer linear programming problems, where it shows improved solution optimality and enables efficient transfer learning and generalization compared to baseline methods.
Read the full abstract
Computer Vision
EndoSCOP-V: A Multi-Turn Video Understanding Evaluation Framework for Multimodal Models in Endoscopy Reporting and Clinical Reasoning
Endoscopic diagnosis unfolds over native video, where a single positive finding during traversal of the gastrointestinal tract sets off a multi-minute chain of reasoning over location, morphology, severity and intervention. Existing endoscopy multimodal benchmarks test only static images or single-disease videos, which miss the temporal and clinical reasoning that real-world reporting demands. We introduce EndoSCOP-V, a video-grounded multi-turn evaluation framework for structured endoscopy reporting and clinical reasoning. It contains 1,307 real-world endoscopy and colonoscopy videos (63.2 h; 97% 1080p) averaging about 174 s. Its 20,274 question-answer turns are instantiated deterministically from clinician-filled structured endoscopy-reporting EHRs, so no per-case annotation is synthetic or LLM-rephrased. They come from 1,033 clinician-reviewed templates, about 18 per disease, that span a clinically aligned 9-task x 30-sub-task reporting taxonomy over 54 World Endoscopy Organization (WEO) MST 3.0 diseases. 22% of procedures carry co-existing pathologies, up to six per case. A Markovian context-pruned oracle-forcing protocol evaluates the dialogue turn by turn, injecting the gold answer before each subsequent turn so that per-turn reasoning is isolated from cascading dialogue errors. A panel of 15 board-certified endoscopists sets a 78% strict-accuracy baseline on 1,500 questions. Across 14 open-weight LMMs, the best 27B-class model reaches 59.8% strict accuracy, falling to 53.4% once format-trivial disease-rejection items are excluded. On a between-subjects spot-check, the best of 3 frontier proprietary models trails the clinician panel by 23 pp on strict NDR accuracy. Sub-task analysis reveals consistent gaps in quantitative measurement, multi-variable grading and intervention reasoning on long-horizon video. Data: Hugging Face (EndoSCOP-V). Code: GitHub (abhi25t/EndoSCOP-V).
Read the full abstract
Soft Condensed Matter
This Is Water: Relative Humidity as a Hidden Parameter in the Self-Assembly of CdSe Nanoplatelets into Threads
When nanocrystals order into large-scale assemblies, interparticle coupling can give rise to new properties. In the case of semi-conducting nanoplatelets (NPL), stacking them orderly into threads yields highly efficient, ultrafast Förster Resonance Energy Transfer (FRET) over distances of hundreds of nanometers. Producing such assemblies reliably and reproducibly therefore matters. Yet, the mechanism by which dispersed NPL form stacked threads under evaporative self-assembly remains unclear. Here, we show that the relative humidity (RH) under which the NPL dispersion evaporation occurs is a critical, and so far overlooked, experimental parameter. Using a humidity-controlled chamber, we find that micrometer-long NPL stacks form only when evaporation occurs at RH > 50% rather than under dry conditions and when free oleic acid is present in the dispersion. Because such values lie within typical laboratory conditions, humidity is an easily overlooked parameter. We discuss how adventitious water may modify colloidal forces between NPL, through changes in solvent polarity, depletion, capillary or solvation interactions. Our experiments demonstrate that relative humidity is a key parameter for nanoparticle self-assembly and, more broadly, point to colloidal interactions at the nanoscale.
Read the full abstract
Discrete Mathematics
A Simple Computer-Assisted Proof of the 5/6-Bound for Real-Period Pinwheel Scheduling
In pinwheel scheduling with real periods, an instance consists of $n$ tasks with real periods $1\le a_1\le a_2\le\cdots\le a_n$. The goal is to find a schedule that assigns one task to each day indefinitely such that, for every task $i\in\{1,\dots,n\}$ and every positive integer $\ell$, task $i$ occurs at least $\ell$ times in every block of $\lceil\ell a_i\rceil$ consecutive days. Given an instance $(a_1,a_2,\dots,a_n)$, its dimension is $n$, its density is $\sum_{i=1}^n1/a_i$, and its potential is $\sum_{i=1}^{n-1}1/a_i$. We prove that every real-valued instance of density at most 5/6 admits a periodic schedule of period length less than $4\cdot2^n$. For integer periods, the 5/6 density bound was a long-standing conjecture settled by Kawamura using a computer-assisted proof [STOC 2024, PNAS 2026]. Our result confirms the real-period conjecture of Fujiwara, Miyagi, and Ouchi [DMTCS 2026]. Our computer-assisted proof builds on Kawamura's ideas while using fewer explicit schedules. It combines dimension reduction by folding with nine satisfiability modulo theories (SMT) checks over linear real arithmetic. A set of 14287 explicit periodic schedules contains a schedule for every instance of dimension at most 8 and density at most 5/6, and for every instance of dimension 9 and potential at most 5/6. Folding is used to reduce every instance of larger dimension and potential at most 5/6 to an instance of dimension 9 and the same potential bound. We also prove that every real-valued instance with minimum period at least 4 and density at most 17/20=0.85 admits a periodic schedule of period length less than $19\cdot2^n$. The density bound 17/20 is tight and improves the 0.84-bound of Mishra, Rho, and Kleinberg [arXiv 2025] for integer periods. It also resolves the $a_1=4$ case of Kawamura's 2025 conjecture that integer-valued instances of density at most $1-(a_1-1)/(a_1(a_1+1))$ are schedulable.
Read the full abstract
Superconductivity
Ferromagnetic resonance signatures of edge states for distinguishing chiral and helical $p$-wave superconductors
We theoretically investigate ferromagnetic resonance (FMR) modulation at interfaces between ferromagnetic insulators and two-dimensional chiral and helical $p$-wave superconductors. We evaluate the surface dynamical spin susceptibility, including both edge and bulk quasiparticle excitations. For momentum-conserving interfacial exchange at zero temperature, the edge contribution to the Gilbert damping exhibits distinct frequency scalings for the excitation energies well above the spectral broadening, $\hbar\omega\gg2\delta$: $\delta/(\hbar\omega)^{2}$ for the chiral state and $1/\hbar\omega$ for the helical state. For a clean superconducting boundary, edge-induced Ising spin anisotropy persists even when interfacial exchange disorder randomizes momentum transfer. The low-energy edge response is suppressed when the magnetization aligns with the edge-state spin axis. These results establish FMR modulation as a probe complementary to existing methods for distinguishing the chiral and helical states and constraining their $d$-vector directions.
Read the full abstract
Cryptography and Security
CAROUSEL: GPU-Accelerated Private Vector Search via Homomorphic Sketching
Vector search on untrusted infrastructure exposes the query embedding, a faithful summary of the user's intent. Fully homomorphic encryption can hide the query, but existing approaches still require a full scan followed by ranking of encrypted scores. Scalable private systems avoid encrypted ranking by either sending scores from a single cluster to the client or moving the search to the client, which must first stream the entire vector index for preprocessing. We present Carousel, which instead scores the full corpus under encryption and compresses the results for client-side ranking. The server scores every vector under encryption, raises the scores to a power that preserves their order by magnitude while suppressing all but a handful, and compresses the resulting sparse vector into a logarithmic number of ciphertexts using homomorphic additions alone. Because every vector is scored, Carousel incurs no recall loss from discarding candidates. Since clients hold no database-dependent state, insertions and deletions take effect immediately at the cost of a single vector write. Carousel searches $1$ billion vectors of BIGANN SIFT1B across multiple GPUs and achieves recall@100 of $0.9715$. At $100$ million SIFT vectors, it achieves recall@100 of $0.9845$ with a query latency of $3.735$ seconds on a single GPU.
Read the full abstract
Statistical Mechanics
Dynamic Scaling and Aging in Hydrodynamic Coarsening of Vapor Bubbles in a Lennard-Jones Fluid
We investigate the coarsening and aging dynamics of vapor bubbles during liquid-vapor phase separation in a three-dimensional single-component Lennard-Jones fluid using large-scale molecular dynamics simulations. Following a temperature quench from a homogeneous high-temperature state into the two-phase coexistence region, vapor bubbles nucleate and grow in the liquid background. The characteristic bubble size follows a power-law growth, $\ell(t)\sim t^\alpha$, with a growth exponent $\alpha \simeq 2/3$, characteristic of inertial hydrodynamic coarsening. We also investigate the two-time autocorrelation function $C_{ag}(t,t_w)$ which follows waiting-time dependent relaxation, revealing aging. Furthermore, data at different waiting times collapse onto a scaling function when plotted as a function of the length ratio $\ell(t)/\ell(t_w)$, demonstrating dynamical scaling during aging. The corresponding scaling function exhibits an algebraic decay with an exponent $\lambda \simeq 3$. Interestingly, bubble and droplet growth exhibit distinctly different domain-growth and aging exponents despite both being disconnected morphologies.
Read the full abstract
Machine Learning
What Do Collusion Probes Read? Three Pitfalls in Evaluating Multi-Agent Collusion Monitors
Activation probes look like near-perfect monitors for secret collusion among large language model (LLM) agents. We show that they mostly detect that an agent was told to collude, not that it decided to. Our benchmark (2,500 committee and 1,476 code-review episodes) adds agents who receive the collusion instruction and refuse it, separating instruction from decision. Published detectors barely tell colluders from these refusers, and our best probe separates them mostly by reading the instruction: without it, agent AUROC falls from 1.00 to 0.61. Two further pitfalls inflate reported results: scored after the vote, a rule that only reads the votes reaches AUROC 0.990 but finds nothing before it; and in 57% of attempts no colluder argues for the target in public. The decision is still there: a probe trained on agents' own choices reads it (AUROC 0.73 without the instruction), and probes find colluders whose goal was fine-tuned into the weights (AUROC 0.995; pre-registered). We turn the controls into a protocol for testing any collusion monitor, and adapt scan statistics into an online monitor for groups of any size.
Read the full abstract
Game Theory
On the Convergence of First-Order Methods in Signaling Games
We study first-order learning dynamics in Bayesian signaling games, a minimal extensive-form model of strategic interaction under asymmetric information. In these games, a Sender observes a private type and chooses a signal, after which a Receiver forms a posterior belief and selects an action. This Bayesian belief update makes the induced learning field fundamentally different from the affine game-gradient fields of normal-form games. Empirically, we find a sharp dichotomy: standard online learning algorithms reliably converge to strict Perfect Bayesian Equilibria (PE), but not to non-strict ones. We show that this behavior cannot be explained by the global geometric conditions commonly used to prove last-iterate convergence, such as monotonicity or the Minty variational inequality. Indeed, these conditions can fail even in binary signaling games. Instead, we identify the relevant local geometry. Our main result characterizes variational stability in finite signaling games: a PE is variationally stable if and only if it is strict. This yields local last-iterate convergence guarantees for online mirror descent and regularized dual averaging near every strict PE. We then study global average-iterate convergence and identify a class of signaling games in which coarse-trigger regret-minimizing dynamics converge to the unique PE. Finally, we complement the theory with experiments comparing projected gradient ascent, mirror-based methods, optimistic variants, and PPO across canonical signaling-game families. The results position signaling games as a compact testbed for understanding how Bayesian belief formation reshapes the convergence geometry of multi-agent learning.
Read the full abstract
Probability
Remarks on spectral gaps on Gaussian path space
In this paper, we discuss the relationship between spectral gaps of Ornstein–Uhlenbeck/$L^2$ operators on Gaussian path space and the underlying Volterra kernel $F$. This framework includes fractional Brownian motion $B^H$ with Hurst index $H\in(0,1)$. For the fractional Wiener measure $\mu_H$, the logarithmic Sobolev inequality with respect to the Ornstein–Uhlenbeck Dirichlet form holds for $H\geq\frac12$, whereas the corresponding weak Poincaré inequality fails for $0<H<\frac12$. The damped/$L^2$ Dirichlet forms satisfy logarithmic Sobolev inequalities for every $H\in(0,1)$.
Read the full abstract
Neurons and Cognition
NeuroHorizon: Long-Horizon Forecasting of Neural Population Activity with Hierarchical Memory
Long-horizon forecasting of neural population activity tests whether a model has captured dynamics that remain predictive well beyond the observed history. Prediction quality typically declines as the horizon grows, yet neither short-horizon accuracy nor teacher-forced evaluation, which supplies the true activity preceding each predicted step, reveals how well models sustain predictions from an initial history alone. We compare direct multi-step prediction and autoregressive rollout on spiking recordings across horizons and introduce NeuroHorizon, which pairs event-based spike encoding with a causal decoder for population firing rates. Its hierarchical prediction memory keeps recent predictions at full temporal resolution and compresses earlier ones into segment summaries, giving the decoder both local detail and longer-range context. Under teacher-forced training this memory holds ground-truth activity, whereas during rollout it holds the model's own predictions. To reduce this mismatch, we train with scheduled sampling, progressively exposing the decoder to its own predictions. From 500 ms of macaque motor-cortex activity, NeuroHorizon achieves the highest forward-prediction score among the evaluated autonomous models at all four horizons from 100 to 1000 ms, and its per-bin predictions remain above a mean-firing-rate baseline across nearly the full 1 s window. Complementary experiments on macaque sensorimotor and mouse multi-region recordings show that performance generally increases with model and data size, alongside improved prediction after cross-subject pretraining and fine-tuning. Together, these results clarify how forecasting performance changes across horizons and offer an effective approach to long-horizon forecasting of neural population activity.
Read the full abstract
General Relativity and Quantum Cosmology
Scalar-tensor reconstruction of coherent quantum black holes
We construct a scalar-tensor realisation of spherically symmetric black hole geometries that effectively arise from quantum coherent states characterised by a finite core and finite extension. We determine the non-minimal coupling to the Einstein-Hilbert Lagrangian, kinetic coupling, and scalar potential and show that the kinetic term is positive only in finite domains. For fixed black hole mass, the boundaries of these domains provide a natural mechanism that relates the core size and spatial extension to the dynamics of the scalar-tensor realisation. We further show that the same geometry can also be obtained from a Horndeski action with derivative coupling and determine its domain of validity. Neither reconstruction includes the event horizon nor can it simultaneously reproduce arbitrarily large portions of the interior and exterior of the black hole, whose quantum origin therefore appears to remain partially outside the regime of applicability of semiclassical theories of gravity.
Read the full abstract
Quantum Physics
Simulating the quantum switch with quantum circuits is impossible
The quantum switch is a higher-order transformation that combines input quantum channels in an indefinite causal order. Whether quantum circuits can simulate its action, and how many calls to the input channels such a simulation requires, are natural questions about the power of quantum control of causal order. In this work, we prove that no quantum circuit with classically controlled causal order can exactly and deterministically simulate the quantum switch for arbitrary input channels using any finite number of calls. This impossibility holds already for qubit channels and extends to the complete $r$-switch, which coherently controls all permutations of $r$ input channels. These results settle the question of exact deterministic circuit simulation of the quantum switch. They demonstrate that processes with indefinite causal order cannot, in general, be efficiently simulated in this setting even by quantum circuits with classically controlled causal order.
Read the full abstract
Functional Analysis
Analytic criteria for criticality of subordinated Schrödinger operators
For a subordinated Schrödinger operator, we introduce a quantity that plays a role analogous to the bottom of the spectrum of a Schrödinger operator, and establish analytic criteria for subcriticality, criticality and supercriticality of a subordinated Schrödinger operator.
Read the full abstract
Machine Learning
Clinical Evidence-Guided Residual Learning for Medication Recommendation with Dynamic DDI Regularization
Medication recommendation must combine patient-specific treatment evidence with the interaction structure of a prescription. The central challenge is to determine which evidence supports each medication and how strongly to regularize interactions while preserving predictive quality. We propose EviDrug, a clinical evidence-guided residual learning framework. Its clinical-memory backbone integrates a self-trained visit-aware encoder, training-case memory and one medication-set interaction block to establish a patient-specific prescription. A gated refinement head combines the backbone states with clinical associations and history-conditioned medication identities through bounded logit corrections. Drug-drug interaction (DDI) feedback adapts regularization strength using training-query predictions. Across MIMIC-IV, NWICU and SICdb, EviDrug leads twelve published-method baselines in AP, F1 and Jaccard, improving AP by 2.26, 2.06 and 1.71 percentage points over the strongest comparator. These gains come at lower parameter and runtime costs than MetaDrug. Matched controls demonstrate DDI reductions with stable predictive quality. Code is available at https://github.com/LIZESHENG13/EviDrug.
Read the full abstract
Cryptography and Security
An Agentic Multi-Agent Framework for System Log Anomaly Detection
As enterprise systems grow in scale and complexity, manual log analysis and conventional SIEM tools struggle to keep pace with the volume and diversity of incidents. To address this challenge, this paper proposed a multi-agent AI framework supporting in analyzing the system log to trace the attack forensic in the system more effectively and efficiently. The framework is designed using n8n platform and Gemini large language model. The framework consists of 7 AI agent layers. Each agent work on a specific task. When the task is done, the agent will forward the result to the output-associated agent of the next layer. The multi agent model improves the analysis velocity, so it is able to support numerous log analysis requirement at the same time. We deployed 16 test cases to test operation of the multi agent workflow, and evaluate the accuracy per test case and the operative performance. The result findings are very promising.
Read the full abstract
Machine Learning
Detent: Batch-Invariant LLM Serving and RL Rollouts at Near-Stock Speed
Reproducible evaluation, verifiable inference and on-policy reinforcement learning (RL) need an LLM engine to give a request the same output bits whichever requests share its batch. Batch-invariant kernels exist, but engines still select each kernel and its parameters from the batch. Two kernels that compute the same sum can round it differently. The engines' deterministic modes remove the choice by keeping one conservative kernel per operator and pay for it in latency. Many fast kernels are batch-invariant by themselves, so it is enough to fix the few batch-dependent values that kernel selection reads. Detent expresses kernel selection as a program that the engine executes. It finds the fewest values in this program that must be fixed and tunes them for the workload before deployment. On multi-GPU deployments of Llama-3.1-70B and Qwen3-30B-A3B in vLLM and SGLang, Detent is batch-invariant. Tuned per load, it serves at 1.04-1.17x the latency per output token of Stock (each engine with its standard settings), against 1.67-10.3x for the deterministic modes. In SGLang it matches the GSM8K accuracy of SGLang's deterministic mode. Detent covers multiple deployment combinations, including multiple engines, GPU generations, tensor-parallel degrees, and dense and MoE models. Seeded RL rollouts of models up to Llama-3.1-70B replay bit for bit under other batches and cache splits. A trainer on Detent's kernels and values matches every rollout logprob bit for bit at 1.03x Stock's rollout time per token. BF16 GRPO then stays stable where Stock's run collapses.
Read the full abstract
Artificial Intelligence
Efficient LLM Adaptation with Forward-Only Passes
Large language model (LLM) inference serving is undergoing rapid growth and large-scale deployment, motivating us to rethink how the inference process itself can be leveraged to enable efficient task-specific LLM adaptation. In this paper, we propose a forward-only approach for efficient LLM adaptation with forward-only passes in LLM inference serving. We exploit the angle concentration of activations induced by each singular value decomposition (SVD) component to measure its contribution-dispersed or concentrated-to the angle concentration of the hidden states. Based on this, our forward-only search then efficiently identifies the weight matrix with the highest dispersion that merits rank reduction. We then selectively remove the higher-order components and retain the lower-order components in SVD. Empirical results across diverse datasets demonstrate the competitive accuracy of our forward-only approach, while theoretical analysis shows lower peak memory and greater speedup than the gradient-based approach and greater speedup than the exhaustive search approach. The experiments further present the robustness and generalization of our forward-only approach to various LLMs with up to 57B parameters. The code is included in the supplementary material.
Read the full abstract
Computer Vision
Reason First, Retrieve What Matters: Hypothesis-Guided Keyframe Selection for Long Video Understanding
Long video understanding requires models to identify and reason over sparse, temporally distributed evidence within limited frame budgets. Existing keyframe selection methods typically retrieve query-relevant frames from a densely sampled candidate set before inference. However, global image-text matching incurs unnecessary computation when a small set of uniformly sampled frames already provides sufficient evidence. Query relevance alone may also overlook evidence that distinguishes competing answers, while one-shot selection offers no opportunity to correct earlier selection errors. We propose HypoKFS, a training-free, plug-and-play framework that couples keyframe selection with iterative inference by multimodal large language models (MLLMs). HypoKFS first reasons over sparsely sampled frames to construct an initial hypothesis space. The resulting answer distribution determines whether additional frame selection is needed. When needed, the retained hypotheses guide the selection of discriminative frames, followed by renewed inference and hypothesis refinement. This iterative process directs additional visual computation toward unresolved hypotheses while avoiding unnecessary global frame matching. Experiments on Video-MME, LongVideoBench, and MLVU demonstrate consistent improvements across multiple MLLM backbones. HypoKFS improves accuracy by up to 6.7 percentage points and achieves 72.9% accuracy on Video-MME at only 1.3% of the reference frame-hypothesis matching cost.
Read the full abstract
Artificial Intelligence
An LLM-Orchestrated Multi-Modal Chatbot for Tomato Disease Consultant
This paper introduces a chatbot developed to assist in diagnosing common diseases in tomato plants through natural language interaction. Users can describe symptoms in text. Then, the chatbot responds with the most likely disease, and basic treatment suggestions and related image-based reasoning. The system leverages pre-trained Mistral and Llama 3.2 large language models. The system is also built using LangChain and Streamlit to provide a lightweight and offline-friendly interface. This work targets to support tomato growers a fast and accessible tomato plant health guidance when the technical expertise is not available nearby, especially in the remote areas. The experimental results are promising and encourage further improvement of the chatbot for future real-world deployment.
Read the full abstract
Risk Management
Worst-Case Contagion in Financial Networks: MILP Clearing and Robust Loss Analysis
We study worst-case contagion losses in financial networks with interbank liabilities and common exposures to marketable assets. Existing linear-programming approaches provide exact stress-test values only within an insolvency-resilience margin, where all institutions remain solvent with respect to the external sector. We show that this margin is computed exactly by a finite family of linear programs, one per vertex of the uncertainty ball, whereas a single linear program based on the componentwise worst deterioration of external resources yields only a certified lower bound. We extend this framework beyond that margin in three directions. First, we derive an exact mixed-integer linear programming (MILP) formulation for the maximal clearing vector under arbitrary signed external net positions, providing a clearing oracle for robust stress testing when insolvency can occur. Second, we develop certified bounds on the worst-case loss beyond the margin, including insolvency-pattern enumeration, scenario lower bounds, fixed-pattern dual upper bounds, localized certificates, and a regime-generation procedure with monotone gap certificates. Third, we analyze the worst-case loss curve as a function of the perturbation radius. For $\ell_1$ and $\ell_\infty$ price-perturbation sets, the curve is convex, nondecreasing and piecewise linear up to the margin, with breakpoints generated by switches in worst-case shock directions, LP bases, and bank-level activity patterns. Beyond the margin, it remains nondecreasing and piecewise affine, but convexity may fail, the curve may jump upward, and the worst case need not be attained. The framework provides a certifiable MILP/LP methodology for parametric stress testing and interpretable diagnostics identifying the banks responsible for changes in systemic loss.
Read the full abstract
Number Theory
On the error term in Gauss's mean value formulas for class numbers
Combining the frameworks of Chamizo–Iwaniec and Chamizo–Ubis with the recent character-sum and exponential-sum estimates of Bellotti and Kerr, we improve the error term in Gauss's mean value formulas for class numbers of binary quadratic forms from $O(N^{21/32+\eps})$ to $O(N^{55/84+\eps})$. If no Dirichlet $L$-function vanishes in $\mathrm{Re}\,s>7/8$, the exponent becomes $2274811/3487744$; under the Generalized Riemann Hypothesis for quadratic Dirichlet $L$-functions, it becomes $118/181$.
Read the full abstract
Soft Condensed Matter
Hyperuniformity in active matter
Disordered hyperuniform systems are characterized by an anomalously strong suppression of long-wavelength density fluctuations. As a result, these unusual materials behave as though virtually incompressible, even though the particle arrangement itself can remain fully disordered at short range. This peculiar state of matter has been the topic of intense research in physics and mathematics in recent years. Another flourishing research arena is active matter, composed of self-propelled agents that convert energy into motion. While the two fields remained largely disjunct for a long time, the paper by Lei and coworkers (2019) created a link between hyperuniformity and active matter by demonstrating that interacting circle swimmers self-organize into disordered hyperuniform structures. Inspired by this work, many other examples of hyperuniformity in active matter have been identified. We shall describe the impact and outgrowth of this development, and give a brief perspective on further links between hyperuniformity and active matter.
Read the full abstract
General Relativity and Quantum Cosmology
Generalized Second Law of Thermodynamics in $f(Q)$ Symmetric Teleparallel Gravity
We investigate the validity of the generalized second law (GSL) of thermodynamics on the cosmological apparent horizon in $f(Q)$ symmetric teleparallel gravity, using the Wald-type entropy prefactor $Af_Q/4G$ together with a correction term derived directly from the modified Friedmann equations. We derive, and verify by explicit symbolic computation, a closed-form expression for the apparent horizon entropy in $f(Q)$ gravity and show that its time derivative combines with the matter entropy rate into a remarkably compact master formula, $T_A\dot S_{\rm tot} = 4\pi\dot H(\dot H f_Q + H\dot f_Q)/H^4$ (in units $8\pi G=1$), which correctly reduces to the known general-relativistic limit. We apply this formalism to three observationally viable $f(Q)$ models — power-law, exponential, and logarithmic. We find that the GSL is satisfied throughout cosmic history for all three models, and relate this robustness to a general, theory-independent structural condition: for ordinary pressureless matter obeying the null energy condition, GSL validity reduces to the sign of $\dot H$, i.e., to the absence of a phantom crossing in the total effective expansion history. The geometric sector exhibits model-dependent energy conditions, power-law and exponential models violate only the strong condition, while the logarithmic model violates the null, weak, and dominant conditions for \(z>-1\), and the strong condition for \(z<3.84\).