Peer Revenue Now accepting positions

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Friday 6 November

Computer Vision

VLM as an Editing Pilot: VLM-Steered Velocity Control for Flow-Based Image Editing

Text-guided image editing requires realizing the requested semantic change while preserving content unrelated to the instruction. FlowEdit provides an efficient inversion-free solution, but its source-anchored coupling can become overly restrictive, causing zero-edit, where the output remains close to the source and the requested change fails to emerge. We find that source coupling directly governs the preservation–editing trade-off, while a fixed coupling schedule cannot adapt to timestep-specific editing needs. Motivated by this observation, we propose V2C-Edit, a training-free VLM-steered velocity control framework. V2C-Edit constructs complementary coupling-controlled velocities and dynamically selects among them using Instruction-Level Similarity (ILS), which evaluates whether the predicted transformation follows the editing instruction. Adaptive spatial masking further preserves regions unrelated to the edit. Experiments on MagicBrush and PIE-Bench with FLUX and Stable Diffusion 3 show that V2C-Edit consistently improves instruction alignment over FlowEdit while maintaining competitive structural preservation, and that ILS better reflects editing effectiveness than target-caption CLIP similarity. Our code is available at https://github.com/Kuanshen77/V2C-Edit.
Read the full abstract

All authors Kuan-Shen Chiu, Jia-Wei Liao, Yu-Chieh Chao, Li-Xuan Peng, Wei-Yun Ma, Cheng-Fu Chou

High Energy Physics (Phenomenology)

Proof of multiloop high-multiplicity integrand reduction coefficients in the 't Hooft-Veltman scheme

In the 't Hooft-Veltman (tHV) scheme, external momenta lie in a fixed integer-dimensional space, and the propagator denominators of high-multiplicity Feynman integrands obey affine-linear relations. Previous work used these relations to construct a partial-fraction reduction with a compact determinantal formula for its coefficients. We present two independent proofs of this formula. The first proof uses recursive partial-fraction decomposition and induction on the number of affine relations. The second adopts the multiple cut method. After clearing the common denominator, setting the complementary denominators to zero isolates one product, and Cramer's rule gives the complete coefficient. We also discuss the behavior of the formula at special kinematics and give an example in which terms with singular coefficients combine to preserve a finite polynomial identity as the denominator-product basis degenerates.
Read the full abstract

Commutative Algebra

Finiteness of local cohomology modules over arbitrary regular local rings

In this paper, we consider arbitrary regular local ring $S$ and an ideal $J\subset S$ whose minimal primes are cohomologically complete intersection. In Section 2, using the Mayer-Vietoris spectral sequence, we construct examples of local cohomology modules $H^i_J (S)$, whose set of associated primes is finite, where minimal primes satisfy the `Dropping Condition'. As a special case, we show that when minimal primes of $J$ are part of a specific regular system of parameters satisfy the `Dropping Condition', the set of associated primes of $H^i_J (S)$ is finite. In Section 3, we prove the main result without the `Dropping Condition' and as a special case, we show that when minimal primes of $J$ are part of a specific regular system of parameters, the set of associated primes of $H^i_J (S)$ is finite. These examples of local cohomology modules include the cases whose associated primes can contain prime number $p$, but still the set is finite.
Read the full abstract

Software Engineering

Done Is Not Correct: Measuring Silent Failures and Self-Verification Calibration When LLM Agents Take CAD Actions

When a large language model (LLM) agent writes parametric CAD code from a natural-language specification, syntactic success decouples from semantic correctness. We release a benchmark and harness that measure this gap directly: 13 specification tasks whose every requirement maps to machine-checkable properties of the produced B-rep, measured by the kernel against published tolerance bands, with reference solutions and the full graded attempt corpus. The exact, self-audited oracle isolates two quantities no prior CAD benchmark reports: the silent-failure rate (a valid, renderable solid, an explicit completion claim, and at least one failed semantic check) and completion-claim calibration. Across four models from four vendors (156 attempts per architecture), single-shot agents claimed completion on 98% of attempts and were silently wrong on 14.1%; a tool-use loop that enforces measure-then-claim cut silent failures to 3.2% and raised claim precision from 64% to 96%, and a four-model ablation attributes the change to the claim contract rather than the feedback. A planner-coder-checker pipeline matched that precision at lower coverage with no success gain. Stated confidence was miscalibrated single-shot (ECE 0.32) and became informative under verification for three of four models. Hardening our own oracle flipped 18 of 519 regraded verdicts from pass to fail and none in reverse; a blind expert audit overturned 3 of the 18 silent failures outside the one ambiguously specified task; and fourfold scaling of every tolerance band leaves the contrast intact.
Read the full abstract

Computer Vision

REFORM-3D: A Representation-Centric Evaluation Framework for 3D Medical Vision Foundation Models

Evaluation of 3D medical foundation models is typically dominated by downstream task scores, which entangle representation quality with adaptation choices. This is especially critical in volumetric imaging, where spacing and anisotropy can substantially alter image statistics while the anatomy itself remains unchanged. We therefore introduce REFORM-3D, a Representation-centric Evaluation FramewORk for benchmarking frozen 3D medical vision foundation Models across three questions: Acquisition Robustness, asking whether features remain stable when spacing varies while anatomy is fixed; Cross-Modal Anatomical Alignment, asking whether labeled anatomical structures remain recoverable across modalities under balanced bidirectional retrieval; and Anatomical Transfer, asking whether organ identity remains recoverable from limited labeled support under case-disjoint evaluation. These are also pretraining questions, because what a 3D encoder learns depends on how volumes are geometrically presented during self-supervised pretraining. We further introduce Med3DINO encoder family, pretrained on $62.6$K unlabeled MR, CT, PET, CTA, and CBCT volumes under three explicit geometric assumptions: isotropic normalization, voxel-relative sampling, and spacing-aware resampling. Across Med3DINO and heterogeneous external encoders, REFORM-3D reveals module-specific strengths rather than a single model ordering. Within the controlled Med3DINO comparison, voxel-relative sampling yields lower embedding drift, whereas spacing-aware resampling favors anatomical transfer; semantic-transfer advantages depend on the dataset and feature type. Project link: \href{https://endarzboy.github.io/reform3d-website/}{\highlightLR{REFORM-3D}}.
Read the full abstract

All authors Maregu Assefa, Muzammal Naseer, Divya Velayudhan, Gedamu Alemu Kumie, Naoufel Werghi

Robotics

UniLatent: Cross-Embodiment Transfer via Latent Observation Alignment

Robot datasets are costly to collect and are usually tied to a specific embodiment, making it difficult to reuse demonstrations across robots with different geometry, appearance, and kinematics. We introduce UniLatent, a cross-embodiment transfer framework that reduces the embodiment-induced observation gap in a shared latent space. UniLatent renders paired source-target robot observations in simulation and aligns a source visual encoder to a frozen target encoder. A policy trained on encoded source demonstrations can then run directly on target observations, with no pixel-level image translation. Across three simulation tasks and two target robot arms, UniLatent reaches up to 93% zero-shot success, exceeding the strongest pixel-translation baseline by over 30 points on average. In real-world xArm-to-Franka transfer, fine-tuning with only 10 target demonstrations per task yields 27/30 successful rollouts. Code, videos and data are available at https://selina2023.github.io/unilatent/.
Read the full abstract

All authors Yufei Ding, Yi Du, Weiduo Yuan, Hao Jiang, Dylan Goetting, Lawrence Yunliang Chen, Daniel Seita, Yue Wang

Soft Condensed Matter

A transversely isotropic constitutive model for semisoft elasticity in main-chain liquid crystal elastomers

Monodomain main-chain liquid crystal elastomers exhibit pronounced stiffness anisotropy and broad semisoft plateaus following finite onset strains. At fixed chain anisotropy, the neo-classical model predicts equal principal Young's moduli, while its transversely isotropic extension with a constant semisoft modulus constrains the relation between reorientation onset and subsequent stress growth. We develop a finite-strain, isothermal elastic model combining transversely isotropic network elasticity with a deformation-dependent quadratic semisoft penalty. Within a weak-fluctuation approximation, a reduced statistical closure based on assumed linear contraction of effective local anisotropy fluctuations at fixed mean anisotropy yields an exponential transition from an initial to a residual semisoft modulus. The network energy preserves ideal soft modes, and the formulation recovers the de Gennes strain-director coupling in the infinitesimal limit. Analytical relations for perpendicular tension show that the rotating-stage stress depends on both the current semisoft modulus and its derivative with respect to the first deformation invariant. This permits a finite onset followed by a substantially flatter rotating response than in the constant-modulus limit. Oblique tension couples director rotation, shear and lateral contraction. Comparisons with orientation-dependent measurements using a common parameter set capture the principal stiffness contrast and broad perpendicular plateau, with discrepancies remaining in oblique stresses and intermediate-angle initial moduli.
Read the full abstract

General Mathematics

Prime supports and vanishing in restricted coprime representations

We study the prime supports that eliminate every representation of an even integer $N$ by two positive integers coprime to $6P$. Let $J(k)$ be the largest vanishing instance obtainable with at most $k$ factors of $P$ not dividing $N/2$. We prove $k(\log k)^2\ll J(k)\ll_\varepsilon k^{4+\varepsilon}$ for every $\varepsilon>0$. The upper bound follows by comparing a classical family of shifted-prime representations with a two-dimensional upper sieve for the loss at each additional factor. Retaining the same local density factor in both estimates yields a weighted survival criterion and positivity whenever the number of noncolliding factors above $N^\theta$ is sufficiently small compared with $N^\theta$, for fixed $0<\theta<1/4$. An exact reduction separates a minimum prime cover into forced prime-pair contributions and a residual problem, and gives the lower bound for $J$ along powers of two. Residue occupancies characterize the extremal concentration at a first zero. Explicit discrepancy estimates show that supports with at most eleven factors cannot eliminate every representation when $N>2780460$, nor supports with at most seven noncolliding factors when $N\ge2266292$. Positivity for the complete support of primes up to $\sqrt N$ remains open.
Read the full abstract

Software Engineering

From Delegation to Action Approval: An Exploratory Study of Developers' Practices in Collaborating with AI Coding Agents

Developers now collaborate with AI coding agents that write code and take actions on their behalf. Beyond reviewing this code, developers decide what to delegate and which agent actions to approve. Some of these actions, such as running shell commands or reading files, can have effects that never appear in the resulting code. How developers make these decisions and how much effort this requires are not well understood. To better understand this collaboration, we surveyed 203 developers who use AI coding agents. We found that careful code review does not necessarily extend to action approval: 61% of developers who carefully reviewed most or all generated code still often approved at least one type of agent action without close review. Developers aware of security or privacy incidents or concerning practices did so more often than those who were not. We also found that autonomy varied more between developers than between development activities. Although developers reported a lighter workload, supervising agents remained mentally demanding. When asked what would make them grant more autonomy, developers wanted more than reliability: more than half of their responses called for agents that are bounded, observable, or reversible. Our findings suggest that as agents take on more of the coding, the challenge shifts toward supporting the developers who supervise them.
Read the full abstract

All authors Yinan Wu, Dominik Wermke, Ze Shi Li, Kathryn Thomasset Stolee, Bowen Xu

Data Analysis and Statistics

Domain Adaptive Graph Neural Networks for Neutrino Experiments

Weakly interacting particles such as neutrinos require highly efficient and precise detection methods to enable searches for new physics. Deep learning, and particularly Graph Neural Networks (GNNs), has shown promise for reconstruction tasks in liquid argon time projection chamber (LArTPC) neutrino experiments. However, differences between simulations and experimental data introduce domain shifts that can degrade model performance and potentially mimic or obscure new-physics signals. Here, we demonstrate that robust learning across datasets can still be achieved by using domain adaptation during training. We introduce, to our knowledge for the first time, domain-adaptation-enabled GNNs for robust LArTPC reconstruction tasks. We use MicroBooNE open dataset as our labeled source domain, and introduce four perturbed datasets as our unlabeled target domains– including a fixed scale variation, moderate and severe gaussian noise, and a signal variation as a function of the position in the detector. The inclusion of adversarial-based domain adaptation during training improves performance for all four target domains, without a substantial drop in performance in the source domain. Precision in the target domain was improved up to 0.19 and recall up to 0.37, depending on the variation and the network task. These results demonstrate a potential of domain adaptation to bridge the gap between simulations and experiments, and build more trustworthy models for current and future LArTPC neutrino experiments.
Read the full abstract

All authors Hemalata Nayak, Giuseppe Cerati, Aleksandra Ćiprijanović, Tammy Walton

Cryptography and Security

Safety Direction Adapter: A Training-Free Post-Hoc Safety Patch for Fine-Tuned Open-Weight LLMs

Open-weight LLMs are routinely fine-tuned from public repositories, yet task-specific adaptation can silently erode safety alignment inherited from instruct-tuned bases. This is a supply-chain risk when organizations ship third-party checkpoints without full re-alignment. Prior work rarely audits diverse public fine-tunes under a shared multi-axis safety-utility protocol or offers training-free mitigation at deployment time. We propose Safety Direction Adapter (SDA), a training-free post-hoc safety patch that extracts a safety direction once from an aligned base and transfers it one-to-many within that base family as a small PEFT-compatible weight update, evaluated under the same multi-axis protocol. Across six direction extractors, we evaluate 14 checkpoints from two 1-2B lineages and probe 8B and 27B fleets (n=5 each) on 15 benchmarks (12538 prompts). On Llama, SDA improves the four-axis mean by about 2.3 points against a 4.2-point instruct-relative gap (about 55 percent recovery; partial Avg), with Safety and Cyber recovering toward or beyond instruct; Safety gains about 8.5 points against about 4.9 points mean degradation. On Qwen, degradation is larger and excess-driven; SDA still lifts Avg (about 36 percent of the gap) while trading Safety and Cyber against Excess and General. We report dual-use risks of publishing direction adapters and do not claim robustness to adaptive or white-box attacks.
Read the full abstract

Computer Vision

Video-Coder: Reinforcing Agentic Coding for Video Generation

Recent years have witnessed remarkable progress in video generation in terms of visual quality and temporal consistency. However, generating videos with precise spatial structure, coherent motion dynamics, and controllable scene composition remains challenging. In this paper, we introduce Video-Coder, the first attempt to train a code-driven video generation agent, which autonomously writes and executes code to construct a dynamic 3D scene (Draft Video) that provides structure and motion signals to downstream video generators. To support this paradigm, we design a dedicated data pipeline and curate two high-quality datasets, Video-Coder-SFT-6k and Video-Coder-RL-2k, which contain diverse prompts and ground-truth reference videos. We further develop the Video- Coder Harness, a unified interactive environment integrating code execution, 3D scene rendering, and execution feedback for agent training and inference. We then train Video-Coder with SFT fol- lowed by harness-native agentic reinforcement learning with dual reward feedback, which combines execution-based and video-based rewards. Additionally, we design an annealing training strategy that gradually reduces the execution reward over time, allowing the model to rely more on the video reward in the later training stages. Experiments show that Video-Coder achieves significant gains, improving Minimax-H3 by approximately 7 points on PhyGenBench and 24 points on VBench-2.0. We hope this work can serve as an open foundation for coding agents in video generation, and we fully open-source our data, models, and code
Read the full abstract

All authors Chaoyang Wang, Sicheng Gao, Junan Hu, Kunyang Li, Dongyang Chen, Yiyang Zhou, Huaxiu Yao, Yong Jae Lee, Radu Timofte, Yuzhang Shang

High Energy Physics (Phenomenology)

Challenges for minimal warm inflation in the Standard Model and beyond

A minimal realisation of warm inflation, even within the Standard Model, assumes that the inflaton $\phi$ couples via the effective operator $\phi\, G\tilde G$. We show that the SM solutions are in tension with the plasma calculation (as each sphaleron transition releases an energy much larger than the temperature) and with the assumed effective operator description. Embedding $\phi$ as the Goldstone boson of a renormalizable Peccei-Quinn theory requires an anomaly coefficient $\gtrsim100$, implying a Landau pole for QCD. These tensions are relaxed for more strongly coupled dark gauge sectors.
Read the full abstract

Robotics

On Agentic Manipulation Policies

Recent work on robot manipulation has focused on specialist policies trained on demonstrations. We show that generalist AI models can control manipulators more reliably than specialist policies such as vision-language-action models (VLAs) and world-action models (WAMs). We systematically analyze agentic manipulation policies (AMPs), in which a generalist AI model commands the manipulator directly, adapting its actions as the task unfolds. We present AMP1, a reference harness that runs unchanged across single-arm, bimanual, and mobile manipulation. AMP1 achieves state-of-the-art results on the RoboLab-120 and CaP-Gym benchmarks, with no demonstrations or task-specific training. On RoboLab-120, AMP1 reaches an 85.6% success rate zero-shot, compared to 49.2% and 42.9% for leading VLA and WAM policies. The agent succeeds zero-shot in 104 of 108 real-world trials spanning twelve tasks and three robot instances, reducing failure rates by 20x and 15x relative to strong VLA and WAM baselines. The basic harness supplies no grasping, pushing, or placing skills; the agent develops its own strategies, such as searching for hidden objects, using objects as tools, and calibrating its cameras. With the addition of a persistent memory store, the agent can also learn from experience: an augmented harness with a memory representation that is carried across episodes lowers the failure rate on RoboDojo bimanual manipulation tasks by 4.6x relative to zero-shot deployment. Our analysis suggests that AMPs are substantially more general and robust than specialist manipulation policies, providing a promising foundation for robot manipulation. See https://apple-aiml-research.github.io/ml-amp/ for code and videos.
Read the full abstract

All authors Chen Chen, Omid Bagherieh, Naveen Nikamanth Appiah Balaji, Jesse T. Gonzalez, Negin Heravi, Fahad Islam, Xueting Li, Jean-Baptiste Regli, Kejia Ren, Petr Švec, David J. Yoon, Matthias Müller, Vladlen Koltun

Artificial Intelligence

CPU-ARC: Agentic Reasoning Caching for on-CPU Reasoning Reuse

Large language models reinvent the wheel all the time. They conflate task-solving with the execution of the solutions, which forces them to solve every problem from scratch. We show that once a task has been reasoned through, the application of its solution to a new problem instance can be done very efficiently and with little involvement from language models. We introduce agentic reasoning caching, a method for constructing robust CPU-executable agentic reasoning caches (ARCs) from LLM reasoning traces. ARCs show on-par performance with frontier LLMs across a range of difficult mathematical and scientific tasks, are fully explainable, continually trainable, robust to input perturbations, and 16-291x cheaper to invoke than the source model.
Read the full abstract

All authors Peter Belcak, Greg Heinrich, Saurav Muralidharan, Shizhe Diao, Pavlo Molchanov

Probability

Global Two-Sided Estimates for Time-Fractional Diffusion Kernels on Hyperbolic Space

We study time-fractional diffusion on real hyperbolic space $\mathbb H^n$, $n\ge2$, with Caputo order $0<\alpha<1$. We establish global two-sided pointwise estimates for the homogeneous fundamental kernel $Z_{\mathbb H}(t,r)$ and the inhomogeneous (Duhamel) kernel $Y_{\mathbb H}(t,r)$, uniformly over all $t>0$ and positive geodesic distances $r$. By combining inverse-stable subordination with global bounds for the hyperbolic heat kernel, we distinguish a short-time local regime, a large-time interior region, and an exterior region governed by a saddle-point problem with parameter $r/t$. Both kernels have the same exterior exponential rate, while their amplitudes are related by an explicit saddle-point factor. The estimates are uniformly comparable on the interfaces between regimes, recover the Euclidean time-fractional singularities at small scales, and identify the influence of hyperbolic geometry on large-scale decay. The kernel $Z_{\mathbb H}$ also describes the one-time marginal density of hyperbolic Brownian motion under an independent inverse stable clock. This representation yields a radial limit law, moment and quantile asymptotics, and a change in the mean square displacement from order $t^\alpha$ at small times to order $t^{2\alpha}$ at large times.
Read the full abstract

Cryptography and Security

SciCheck: A Framework for Verifying the Integrity of Deep Learning Training

Deep learning training is difficult to verify within a scientific workflow: its architecture, hyperparameters, and procedure are rarely fixed until training has been attempted, and two honest runs never produce the same weights, so recomputation cannot serve as a check. We present SciCheck, a framework in which the scientist explores training configurations freely and posts a blueprint to a provenance ledger for each candidate they want verified, and a verifying scientist judges the resulting run in two layers: a statistical judge model that compares per-step gradient statistics against the verifier's own reference run, and a spot-check that re-executes randomly chosen training segments whose checkpoints are committed to the ledger one epoch at a time. On an MLP, a CNN, and an autoencoder at 100k-500k parameters, under gradient sign flipping, label poisoning, and fabricated runs, SciCheck reaches F1 of 0.966, 0.906, and 0.976 while rejecting 0-4.4% of honest runs. For the attacks we test, the spot-check accounts for most of this accuracy, and we characterize what each layer can detect. The effective-signal statistic falls as $(1-2s)^2$ under a sign flip of severity $s$, so gradient statistics alone catch strong enough sign flips, while fabricated runs that replay genuine statistics look no more anomalous than honest runs to any statistical detector we test. The spot-check catches fully fabricated segments with probability at least $1-(1-s)^k$ for $k$ sampled segments when the draws stay hidden until the run is committed, which roughly matches our measurements and lets a verifier choose $k$ for a target soundness. Verification costs 1.5-2.3x one retraining, mostly for the verifier's reference run; each submission's spot-check takes under 1.5 s, and per-epoch commitments reduce checkpoint storage 5-17x compared with keeping every checkpoint.
Read the full abstract

All authors Alexander Nguyen, Dylan Gresham, Gaby G. Dagher, Min Long

Robotics

Scenario Orchestration for Closed-loop Policy Evaluation of Autonomous Driving

Scenario-based simulation is central to evaluating autonomous-driving policies, but a scenario scripted before execution (i.e., with predetermined, fixed actor routes and speeds) may not preserve its intended interaction once the tested ego policy responds in closed loop. Scripted scenarios cannot adapt to unanticipated ego behavior, while learned traffic actors may remain reactive without reliably realizing a particular scenario intent. We introduce a scenario orchestrator that continuously monitors the joint traffic state and adapts non-ego actors through re-timing, re-routing, and role reassignment for creating a specific scenario intent. At each step, the orchestrator solves a constrained optimization problem that preserves the intended interaction while limiting interventions and maintaining traffic feasibility. We further distinguish scenario realization from policy failure, allowing a test to first be judged based on whether the intended interaction occurred and then by how the ego policy responded. Experiments in the CARLA simulator across multiple scenario families and driving policies show that orchestration realizes targeted interactions more reliably than scripted and learned actor-level baselines, particularly under variations in initial conditions and ego behavior. These results support closed-loop scenario orchestration as a tool for controlled and reproducible policy evaluation.
Read the full abstract

All authors Frieda Rong, Zhaoyue Wang, Veer Kapadia, Vishaka Rajani, Aren A. Babikian, Xujie Si, Mahsa Golchoubian, Raquel Urtasun, Igor Gilitschenski

Machine Learning

On the (non-)closedness of the set of ReLU networks

We prove that the set $\mathcal{N}_N$ of ReLU networks with one hidden layer and at most $N\in\mathbb N$ neurons is weakly closed in $W^{1,p}(\Omega)$, for a bounded connected open set $\Omega\subset\mathbb R^d$. Consequently, every coercive, weakly lower semicontinuous energy on this space attains its minimum over $\mathcal{N}_N$. This covers the Deep Ritz method with convex integrands, Sobolev training and Lipschitz learning. Moreover, as the weight decay parameter tends to zero, minimizers of the regularized problem converge, up to subsequences, to minimizers with minimal parameter norm. We also prove that the set of ReLU networks with two hidden layers of bounded widths is not closed in $W^{1,p}(\Omega)$ or in $C(\overline\Omega)$. This disproves a conjecture of Petersen, Raslan and Voigtlaender.
Read the full abstract

Algebraic Topology

Directed systems of A-spaces and weak cores

We prove that for any regular CW-complex $X$ there exists a quotient map $X\to Y$ that is a weak homotopy equivalence and which is minimal with that property, meaning that any quotient $Y\to Z$ that is a weak equivalence, must be a homeomorphism. The space $Y$ we obtain is an Alexandroff space: a space in which arbitrary intersections of open sets are open. The weak core $Y$ of $X$ is not unique in general. We study homotopy invariants preserved by filtered colimits of Alexandroff spaces. This can be seen as an analogous to Stong's theory of cores of finite topological spaces.
Read the full abstract

Analysis of PDEs

Unconditional Uniqueness of KdV and mKdV Green Solutions at Low Regularity

We prove unconditional uniqueness for the Korteweg–de Vries (KdV) equation in $C_tH^{-1}_x(\mathbb R)$ and the real defocusing modified KdV equation in $C_tL^2_x(\mathbb R)$ within the class of Green solutions, an intrinsically well defined solution notion that encodes the evolution at these low regularities through the diagonal Green functions of the associated Lax operators. A key contribution is automatic local smoothing: the Green identities imply $q\in L^2_{\mathrm{loc}}$ for KdV and $u_x,u^2\in L^2_{\mathrm{loc}}$ for mKdV in spacetime, with estimates uniform over spatial translates. The uniqueness results therefore identify every Green solution in these classes with the continuous extension of the corresponding Schwartz flow.
Read the full abstract

Earth and Planetary Astrophysics

VLT and Keck recovery and characterisation of 2025 SC$_{79}$: linking 'Ayló'chaxnim and Atira asteroids

Asteroids near the boundary between the Atira and 'Ayló'chaxnim classes can trace dynamical links between the two populations, but their small solar elongations make recovery and physical characterisation difficult. We recover 2025 SC$_{79}$, measure its optical colours, and investigate its likely source region and transitions between these populations. We combine Keck/LRIS astrometry and VLT/FORS2 four-filter photometry with Bus-DeMeo taxonomic comparisons, published source-model probabilities, and integrations of 10,000 clones drawn from the covariance of JPL solution #2 and followed for 5 Myr in each direction. The SDSS-equivalent colours are $g-r=0.8254_{-0.1339}^{+0.2136}$, $r-i=0.1764_{-0.0540}^{+0.0523}$ and $i-z=-0.4975_{-0.1162}^{+0.1055}$ mag. The V-type class mean gives the closest colour match. At the V-type median albedo $p_V=0.362$, $H=18.83\pm0.15$ mag implies a diameter of $0.379^{+0.027}_{-0.025}$ km. The source model assigns a combined probability of 95.4% to delivery channels associated with the inner Main Belt. The ensemble-mean 'Ayló'chaxnim/Atira residence fractions are 30.2%/69.8% over the preceding 100 kyr and 3.6%/96.0% over the following 100 kyr; the remaining 0.4% of future ensemble time follows collision removal. Clones spend an average of 231 kyr in the 'Ayló'chaxnim class over the preceding 500 kyr and 120 kyr over the following 500 kyr. In the 5-Myr forward integrations, 14.56% of clones collide with Venus, 3.77% with Mercury and 0.22% with Earth. The V-like colours and source-model probabilities favour an origin in the inner Main Belt for 2025 SC$_{79}$. Its clones move between the 'Ayló'chaxnim and Atira populations.
Read the full abstract

All authors Bryce T. Bolin, Leo Y. Alcorn, Davide Farnocchia, Marco Delbo, Martina Baratella, Henri M.J. Boffin, Max Brodheim, Alexa Gordon, Lordrick Kahinga, Michael Lundquist, Rita Morris, J. Xavier Prochaska

Algebraic Geometry

Analytic information encoded in the Bernstein-Sato polynomial of genus-one plane branches

We give an explicit formula recovering the Zariski invariant of a plane branch with one Puiseux pair from its Bernstein-Sato polynomial. Thus the Bernstein-Sato stratification refines the Zariski-invariant stratification. We determine the generic roots in each stratum from the value semigroup and the Zariski invariant, refining Yano's description for a fixed topological class. We also compute the maximal spectral-exponent discrepancy and relate it to the dimension of the generic Bernstein-Sato locus and the minimum Tjurina number.
Read the full abstract

Computer Vision

FlowMoP: Stochastic Multi-Person Motion Prediction

Predicting the future motion of multiple people is inherently stochastic: the same observed history admits many plausible continuations depending on intent, coordination, and social context. Yet the dominant paradigm in multi-person 3D motion prediction remains deterministic, with existing methods producing a fixed number of outputs per agent. In contrast, we present a conditional flow matching framework that jointly generates diverse future motions for multiple agents, modeling both global trajectories and root-relative poses. A B-spline reparameterization compresses the generative state space while enforcing temporal smoothness, and a dual-stream transformer conditions each agent's predictions on observed neighbor motion for interaction-aware forecasting. Our method achieves state-of-the-art performance on CMU-Mocap (UMPM) and 3DPW, outperforming deterministic baselines on joint position error, pose error, and final displacement error despite producing a full predictive distribution. We additionally establish the first motion prediction baseline on WorldPose, a large-scale professional soccer dataset, where social conditioning demonstrably concentrates predictive uncertainty along interaction-constrained directions.
Read the full abstract

Strongly Correlated Electrons

Unraveling square and kagome lattice frustrated magnetism by scalable tensor networks

We extend the scalability of variational infinite two-dimensional tensor network methods with application to investigate frustrated magnetism of the square-lattice $J_1$-$J_2$ model and kagome-lattice Heisenberg model. In the sign-problem-free limit of the square-lattice model, we achieve high ground-state energy precision comparable to quantum Monte Carlo methods. With no obstruction beyond this limit, our method obtains direct estimates of order parameters. For the square-lattice $J_1$-$J_2$ model, we determine isotropic plaquette order in the valence-bond solid region. For the model near $J_2/J_1=0.5$, and the kagome lattice Heisenberg model, our data indicate linear-dispersion gapless states. To probe the candidate spin-liquid nature of the highly frustrated cases, we go beyond inferring the absence of symmetry-breaking orders and analyze susceptibilities that characterize gapless spin liquids. By estimating scaling dimensions, our data indicate a stable gapless spin liquid on the kagome lattice, and competing scenarios for the intermediate regime of the square-lattice model. Our developments include parallelization and randomized low-rank methods and largely expand the frontier of high-precision variational numerics for quantum phases of matter.
Read the full abstract

Computer Vision

ZeoBench: A Benchmark for Self-Supervised Learning on 3D Zeolite Representations

Zeolites are an important class of porous crystalline materials with broad applications in gas storage, separations, and catalysis. Although hundreds of thousands of experimentally realized and hypothetical zeolite structures are known, identifying materials optimized for a target application remains challenging because property labels are expensive to obtain and available only for limited structure-molecule pairs. While prior work has demonstrated that 3D neural representations can predict zeolite properties effectively, it remains unclear which representation-learning strategies are most effective in the low-data regime relevant to materials discovery. We present ZeoBench, a benchmark for self-supervised learning on dense 3D representations of all-silica zeolites, and evaluate the resulting features on six existing downstream adsorption-property prediction tasks, across a range of training set sizes. Our study compares self-supervised 3D pretraining on volumetric zeolite data, adaptations of image-pretrained 2D ConvNets and vision transformers to 3D via multiview and channel-inflation strategies, and 3D ConvNets pretrained on out-of-domain volumetric datasets, alongside hand-crafted descriptors and supervised baselines. Using our pretraining and evaluation framework, we find self-supervised 3D representations outperform existing methods and achieve state-of-the-art performance across both in-domain label regimes and for the unseen domain of metal organic frameworks. Learned representations consistently outperform hand-crafted descriptors, and convolutional architectures are generally more effective than vision transformers in this setting. These results establish ZeoBench as a practical benchmark for 3D representation learning in zeolites and provide guidance for data-limited adsorption property prediction.
Read the full abstract

All authors Aaron Sun, Yachan Liu, Ping Yang, Peng Bai, Subhransu Maji

Cryptography and Security

PhotoFinish: Portable Timerless Probing of Microarchitectural Latencies

Timing measurements are a fundamental building block of microarchitectural attacks, enabling software to convert otherwise invisible latency differences into architectural observations. Thus, restricting high-resolution timers is a common defense, especially in constrained environments such as web browsers. Timerless probing primitives provide an alternative approach. Instead of reading or constructing a timer, they observe architectural events that correlate with the microarchitectural state. However, such primitives typically rely on platform-specific behavior, and none generalizes across the wider computing landscape. In this paper, we present PhotoFinish, a portable timerless probing primitive that converts latency differences directly into an architectural value. PhotoFinish races an operation under test against a fixed reference in two parallel threads that then write unique values to the same shared location, resulting in a latency-dependent architectural value. PhotoFinish only requires parallel execution and shared memory, without constructing or reading a clock. We evaluate PhotoFinish on 38 CPUs spanning four ISAs and demonstrate reliable cache state classification across widely different microarchitectures, with comparable fidelity when compiled to WebAssembly and executed inside a browser. We further use PhotoFinish in AES T-table and Evict+Reload attacks, show that it applies beyond caches to TLB and execution-engine latencies, and demonstrate sufficient precision to identify the LLC slice of an address. Our results demonstrate shared-memory races as a generic and precise replacement for explicit timing, showing that restricting timers alone does not prevent software from observing fine-grained microarchitectural latency differences.
Read the full abstract

All authors Tristan Hornetz, Eduard Ebert, Lorenz Hetterich, Lukas Gerlach, Michael Schwarz

Computer Vision

Graded convolutional networks: scale invariance without the scale and robustness to unknown blur

The Gaussian jet of an image is a graded vector space: a dilation by $\lambda$ multiplies derivatives of order $k$ by $\lambda^{-k}$. Local graded convolutions are exactly homogeneous differential operators, and a local feature is dilation-invariant if and only if it factors through the scale-normalized jet or, without the scale, through the retraction onto the weighted sphere of the grading. Under unknown Gaussian blur the retraction changes only by a change of scale, whereas the scale-normalized jet acquires a grading factor that no radial normalization removes. At the zero jet the retraction separates image families that radial normalizations and continuous classifiers cannot. Networks built on the retraction are universal for invariant features. On MNIST such a network keeps 88.4% accuracy under a two-pixel blur, against 66.2% for the same network on scale-normalized derivatives; in Lindeberg's Gaussian derivative network the retraction raises accuracy at that blur from 91.5% to 95.1%.
Read the full abstract

Robotics

Locating Contact on Tendon-Driven Continuum Robots From Tendon Measurements Alone via Constrained Moving Horizon Estimation

Tendon-driven continuum robots can reach through constrained spaces, but contact along their bodies is invisible to the sensors at their motors. We show that this is a structural limitation rather than a tuning problem: the map from contact forces to tendon measurements has a null space of dimension at least seven, so no estimator that leaves the contact support free can locate contact. When the obstacle geometry is known, however, non-penetration and complementarity make the contact force locally identifiable, and we prove that the resulting system with unknown contact inputs is locally exponentially strongly detectable. We present a constrained moving horizon estimator (MHE) that encodes these conditions and runs in real time as one quadratic program per sample, and we give a local, mode-wise robust stability result for it. In a finite- element benchmark of a two-section robot pressing and sliding on a wall, the estimator located contact to 0.9 mm and estimated contact force to 0.22 N, compared with 65 mm and 0.72 N for a tuned extended Kalman filter (EKF) with state projection. The estimator tolerated millimeter- level errors in the obstacle position and unmodeled wall friction, under which the projected EKF failed. Under randomized model mismatch over 40 trials, including unmodeled tendon friction, it remained better in force and location in every trial. Without contact, the MHE offered no advantage over the EKF, which confirms that its benefit comes from the contact constraints. This report extends a letter submitted to IEEE Control Systems Letters with complete proofs and additional experiments.
Read the full abstract

Quantum Physics

Novikov-Magnus Master Equation for Driven Quantum Systems under Classical Correlated Noise

We introduce a modular framework for deriving time-local quantum master equations for driven systems subject to temporally correlated classical noise. The construction combines the Donsker-Furutsu-Novikov relation with the Magnus expansion: Functional response closes the noise average, while nested commutators retain the noncommuting chronological action of noise-induced perturbations along the controlled quantum trajectory. For zero-mean Gaussian noise, we derive the resulting Novikov-Magnus quantum master equation through fourth order in the noise strength. The same response hierarchy suggests a natural extension beyond Gaussian noise, with connected higher-order noise correlations weighting higher functional derivatives of the stochastic map. For a driven qubit under Ornstein-Uhlenbeck and band-limited $1/f$ noise, the second- and fourth-order maps remain CPTP throughout the perturbative regime and accurately reproduce stochastic simulations. Their diamond-norm errors are comparable to those of the cumulant-based pseudo-Lindblad master equation, with a modest fourth-order improvement for Ornstein-Uhlenbeck noise.
Read the full abstract

All authors Nada Khogeer, Jerison Parra, Caroline Desgranges, Hugo Ribeiro