Peer Revenue Play credits. Real papers.

Papers

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

All of arXiv

Tuesday 29 September

High Energy Physics (Phenomenology)

Pion structure in magnetized nuclear medium at finite temperature

Fathimath Shifa N K, Navpreet Kaur, Abi Jebarson A et al. (+3 more)

We investigate the structure of the pion in a magnetized nuclear medium at finite temperature by analyzing the in-medium modifications of its parton distribution functions (PDFs) and electromagnetic form factors (EMFFs). We use a unified framework combining the chiral SU(3) quark mean field (CQMF) model and the light-cone quark model (LCQM) where the in-medium quark masses are evaluated using the CQMF model and subsequently employed as inputs in the light-cone formalism. The study systematically explores the effects of magnetic field, baryon density, temperature, and isospin asymmetry on the pion structure. In addition, Dirac sea contributions are incorporated to account for vacuum polarization effects in the medium. Our results of PDFs and EMFFs provide valuable insights into the modification of pion properties under extreme conditions, highlighting the significant role of the surrounding medium in altering its internal structure.
Read the full abstract

All authors Fathimath Shifa N K, Navpreet Kaur, Abi Jebarson A, Arvind Kumar, Suneel Dutt, Harleen Dahiya

Solar and Stellar Astrophysics

On the frequency-dependent time lags of the intermediate polar V709 Cas

Zackery A. Irving, Noel Castro Segura, Diego Altamirano et al. (+9 more)

Cataclysmic variables (CVs) are binary systems in which a white dwarf (WD) primary accretes material from a late-type secondary, typically via a disc. To date, only six CVs, all of which are non-magnetic, have been found to exhibit lags (delayed variability in one band with respect to another). In all six cases, these lags are “red” (i.e., red lags blue) with a time lag of order seconds. While there is no generally accepted mechanism for producing red lags in CVs, current theories suggest reprocessing in the disc, either on the thermal or recombination time-scales, or inside-out shocks propagating through the disc. Using $g$-, $r$-, and $i$-band data from the OPtical TIming CAMera (OPTICAM), we report the discovery of frequency-dependent red lags from V709 Cas: a magnetic CV of the intermediate polar (IP) subclass. These red lags reach a maximum of $\sim$4–6 s on time-scales of 7.4–13.0 min, similar to previously-reported lags in non-magnetic CVs. The detection of lags in an IP is significant as the WD's magnetic field truncates the accretion disc, limiting disc-based lag mechanisms to large radii; as a result, characteristic accretion disc time-scales cannot explain the observed lags. However, recombination can occur on second time-scales, even in systems with truncated discs. For example, we show that recombination in the disc's bright spot, where overflowing material from the secondary meets the outer-edge of the disc, plausibly explains V709 Cas's optical spectrum, the prominence of optical spin-orbit beat pulsations, and the red lags found in this work.
Read the full abstract

All authors Zackery A. Irving, Noel Castro Segura, Diego Altamirano, Martina Veresvarska, Mariano Mendez, Federico Garcia, Domitilla de Martino, Simone Scaringi, Federico Vincentelli, Angel Castro, Federico Soto, Raul Michel

Artificial Intelligence

Accelerated surrogate dynamics for dynamical, stochastic system evolution

Marco Jochum, Ioannis Kouroudis, Gohar Ali Siddiqui et al. (+3 more)

Dynamic simulations are an entrenched way of gaining insight into the evolution of system dynamics. Their computational cost however is often prohibitively high, especially in cases of stochastic frameworks. Machine learning algorithms are especially suited as simulation surrogates. Nevertheless, they face some very distinct limitations. Firstly, the sheer dimensionality of these systems, however, precludes the use of traditional time series models who struggle with high dimensional feature spaces. Additionally, traditional time series focus exclusively on either long or short range effects, causing local or global drift given enough time. In this paper, we propose a framework that addresses those limitations. Our framework combines a Variational Autoencoder, with a convolutional or graph basis that reduces the dimensionality of the system. This latent vector is propagated in time using a Temporal Fusion Transformer model, which includes both long range and short range effect encoding, as well as static covariate support. We test our framework on three distinct cases, to prove its robustness and in all three we have achieved practically identical to the simulation results at a fraction of the time. Further, our framework is flexible enough to be adapted to any new system and provides an inbuilt uncertainty quantification for targeted experiment design.
Read the full abstract

All authors Marco Jochum, Ioannis Kouroudis, Gohar Ali Siddiqui, Taher Amine Hamzaoui, Manuel Gößwein, Alessio Gagliardi

Information Retrieval

HELIX: Purified and Unified - Rethinking Feature Interaction and Sequence Modeling for Large-Scale Recommendation

Yuntao Zheng, Miao Zhang, Yadong Ding et al. (+18 more)

Industrial recommendation ranking models typically scale along two modeling axes: feature interaction over heterogeneous user, item, context, and cross features, and sequence modeling over long, informative, and multi-type user behavior histories. We find that scaling either capability in isolation is insufficient, as each exhibits a limited scaling ceiling and a suboptimal scaling-law slope. We conjecture that achieving a more favorable scaling-law slope requires jointly scaling both axes. To support this, we present HELIX, a purified and unified architecture for large-scale recommendation. HELIX interleaves sequence retrieval and feature interaction while enforcing one-way information flow from reusable sequence states to candidate-conditioned mix-tokens. This design preserves cross-depth communication between the two modeling axes while keeping user-side sequence computation amortizable, enabling flexible and asymmetric scaling of sequence modeling and feature interaction. Deployed in TikTok's e-commerce recommendation system, HELIX consistently improves offline CTR AUC, CVR AUC, and other ranking metrics. In online A/B tests, it achieves an approximately 6% increase in e-commerce video GMV per user.
Read the full abstract

All authors Yuntao Zheng, Miao Zhang, Yadong Ding, Yanchuan Tang, Lixiyu Chen, Hao Wang, Quan Li, Shiying Cai, Yue Lin, Jiayu Li, Yu Feng, Wentao Yang, Rongkun Xing, Jiekai Wang, Mingge Zhang, Feiling Gong, Xiang Gao, Jinyu Dong, Yajing Zhang, Pengfei Ren, Yinzhou Wang

Mesoscale and Nanoscale Physics

Engineering Vacuum Fluctuations in Hyperbolic Heterostructures

Jie-Cheng Feng, Xinle Cheng, Itai Keren et al. (+5 more)

Vacuum fluctuations can be engineered using optical cavities, resonators, and surface-polaritonic modes, but extending such control deep inside a material remains challenging. Here we show that the vacuum field inside a hyperbolic material can be reshaped by a much thinner overlayer with opposite hyperbolicity. This effect requires overlap between the hyperbolic bands of the two materials and modifies the coupling between the vacuum field and matter excitations within the bottom material, without changing its intrinsic dielectric properties. As an application, we use Eliashberg theory to study a two-dimensional superconducting layer inside the bottom material, showing that the overlayer can remotely change its critical temperature. Our results may shed light on recent experiments reporting changes in superconductivity arising from a frequency-matched hyperbolic overlayer, and suggest an on-chip route to long-range control of matter excitations and collective material properties.
Read the full abstract

All authors Jie-Cheng Feng, Xinle Cheng, Itai Keren, Dante M. Kennes, Abhay N. Pasupathy, Dmitri N. Basov, Emil Vinas Bostrom, Angel Rubio

Robotics

EgoHumanoid-V2: Human-to-Humanoid Transfer of Coordinated Whole-Body Skills for Loco-Manipulation

Jin Chen, Yiming Jiang, Chongyang Xu et al. (+8 more)

Human demonstrations capture diverse scenes and rich whole-body skills without requiring robot teleoperation. Prior work on egocentric transfer has emphasized scene generalization in loco-manipulation under decoupled control, leaving direct transfer of coordinated whole-body skills less explored. We present EgoHumanoid-V2, the first egocentric human-to-humanoid skill transfer framework for coordinated whole-body loco-manipulation. At its core, coarse-to-fine action alignment combines kinematic reference correction with dynamics-aware refinement. It improves end-effector pose accuracy while preserving whole-body coordination. We also use robot-arm rendering and training-time image augmentation to reduce the visual embodiment gap and improve viewpoint robustness. On four real-world tasks, vision-language-action (VLA) policies trained on aligned human data show zero-shot skill transfer without target-task robot demonstrations. Task scores are comparable to those of policies trained on teleoperation data at a lower collection cost. These results support human data as direct skill supervision.
Read the full abstract

All authors Jin Chen, Yiming Jiang, Chongyang Xu, Modi Shi, Shijia Peng, Li Chen, Tianyu Li, Mu Xu, Yilun Chen, Steven Hoi, Hongyang Li

Complex Variables

Polynomial Counterexamples to Shanks' Conjecture in Dirichlet-Type Spaces over the Bidisc

Oliver Navío and Daniel Seco

We study optimal polynomial approximants in spaces of Dirichlet-type over the bidisc. Given a nonzero function f, the linear optimal polynomial approximant (opa) to its reciprocal is the affine polynomial p for which pf is closest to the constant function 1 in the space norm. We prove that, for every positive value of the Dirichlet-type parameter, there exists a symmetric polynomial f with no zeros on the closed bidisc, while the corresponding linear optimal approximant has a zero inside the bidisc. This provides polynomial counterexamples to the version of the Weak Shanks Conjecture on these spaces. The construction also extends to higher-dimensional polydiscs and to anisotropic Dirichlet-type spaces.
Read the full abstract

General Relativity and Quantum Cosmology

Closing the Loop: from EPRL-FK spinfoams to Regge dynamics

Matteo Bruno, Pietro Donà and Gowrisankar Sreeram

We study the semiclassical limit of the EPRL-FK model on a fixed simplicial complex. We ask what additional condition is needed for its variational principle to reproduce the equations of motion of Regge calculus, the standard discretization of General Relativity. This limit differs and complements the definition of the semiclassical regime in which, for suitable boundary states and a sufficiently refined simplicial complex, spinfoam amplitudes approximate gravitational transition amplitudes to a prescribed accuracy. We recast the EPRL-FK model in wedge-holonomy variables and identify geometric closure as the missing condition. We prove that imposing geometric closure as an additional constraint alongside local flatness allows us to reconstruct curved geometries. On a regular, nondegenerate Lorentzian branch with spacelike tetrahedra, these constraints define a surface equivalent to the space of length-Regge geometries, modulo gauge. The wedge action restricted to this surface becomes the Regge action, and variations tangent to this surface yield the length-Regge equations. We propose a way to implement geometric closure directly in the spinfoam amplitude. With this additional constraint, the EPRL-FK model is equivalent to Regge calculus in its semiclassical regime on a fixed simplicial complex.
Read the full abstract

Earth and Planetary Astrophysics

Chondrule size averaging in two and three dimensions

Emmanuel Jacquet

Chondrule sizes are an important discriminant of chondrite groups and chondrule formation theories. Yet most studies report 2D sizes without attempting the somewhat tedious corrections to 3D proposed in the past literature. I show here that 3D size averages can be simply calculated directly from 2D averages of appropriate powers of sizes (and vice-versa), without having to reconstruct 3D distributions, although explicit formulas to that end are also provided. This illuminates the comparison between 3D and 2D arithmetic size averages, which are confirmed to be little different for the published datasets. Yet I suggest another type of average, the "sphere-weighted average" (i.e. weighted by the chondrule equal-volume sphere area), a measure of the surface/volume ratio of the chondrule population, which is more robust against omission of small chondrules. While larger than the arithmetic averages, it is best in line with previous literature estimates based on less exhaustive surveys of chondrules.
Read the full abstract

Computer Vision

Sparse cubical complexes for efficient topology-preservation in image data

Alexander H. Berger, Marco Fontana, Daniel Rueckert et al. (+3 more)

Persistent homology (PH) is a frequently used tool for extracting and preserving topological information from image data, particularly in image segmentation, where preservation of topological structures is important. However, despite its general applicability across dimensionality, domains, and target structures, the runtime cost of PH-based methods often makes their practical use infeasible. In this work, we argue that this runtime cost is largely driven by processing information that is unimportant for downstream application (e.g. as optimization objective). We propose sparse cubical filtrations as an alternative foundation for PH computation, reducing subsequent computational costs by factors of up to 100 on real datasets. We show close agreement with the optimization signal of the dense counterpart and empirically evaluate our solution's effectiveness as an optimization objective in realistic training regimes where other PH-based objectives can practically not operate (i.e., 3D data with large patch sizes). We show how our solution improves topological accuracy by up to 80% across six diverse datasets while maintaining pixel- and region-based accuracy.
Read the full abstract

All authors Alexander H. Berger, Marco Fontana, Daniel Rueckert, Johannes C. Paetzold, Laurin Lux, Ulrich Bauer

Artificial Intelligence

Absorbed in Inertia: Activation Analysis for Computer-Use Agents

Giulio Segalini, Zhi Wen Soi, Jérémie Decouchant et al. (+1 more)

Computer-use agents have become increasingly capable of executing tasks on live desktops through natural-language instructions, based on trajectories of screenshots, actions, and reasoning. We discover that they can stealthily exhibit inertia, in which they repeat fruitless actions despite recognizing that these actions are ineffective. We hypothesize that inertia is reflected in the agent's internal state, i.e., the activation values of the agent's underlying model, and propose a protocol to measure the relationship between the two. Extensive analysis of high-dimensional activation states shows that inertia corresponds to an absorbing region of activation space, where activation values become stale across actions and even after attempts to steer them. We conjecture that drastically changing the agents' activations by re-initializing them is necessary to escape inertia. Specifically, we propose R$^3$ (Reset, Reroute, Restore), which temporarily resets the agent's context trajectory to escape the absorbing region and then restores the historical context to effectively complete the task. Our approach yields 17-55% lower measured inertia across models relative to unmodified agents. These results suggest that changing the context can interrupt recurrence more effectively than directly steering the resulting activations. Our code is available at https://anonymous.4open.science/r/vlm-agent-defense-D076
Read the full abstract

All authors Giulio Segalini, Zhi Wen Soi, Jérémie Decouchant, Lydia Chen

Computation and Language

VLM Fine-Tuning for End-to-End Combinatorial Optimization

Qingsong Yan, Xia Jiang, Yaoxin Wu et al. (+3 more)

Large language models (LLMs) have provided a unified interface for end-to-end combinatorial optimization (CO), but textual serialization alone may obscure spatial and relational structures that are important for generating effective CO solutions. This paper presents a general-purpose vision-language solver that augments textual instance descriptions with input-derived visual representations. A single vision-language model (VLM) is applied across different CO tasks and trained using supervised fine-tuning followed by verifier-guided reinforcement learning. While the visual inputs contain no gold solutions or solution-derived information, our experiments show that the VLM generally improves solution quality over its text-only counterpart, with particularly clear gains on more complex CO problems such as CVRP and JSSP. The advantage of visual information is more pronounced at large problem scales.
Read the full abstract

All authors Qingsong Yan, Xia Jiang, Yaoxin Wu, Wen Song, Lu Zhang, Yingjie Zhou

Statistical Mechanics

Accurate ground-state entropies from population Monte Carlo: The antiferromagnetic Ising model on the Shastry-Sutherland lattice

Denis Gessert, Wolfhard Janke and Martin Weigel

We demonstrate how the population annealing Monte Carlo simulation method can be used to compute the ground-state entropy to high precision. This approach is tested for the classical antiferromagnetic Ising model on the Shastry-Sutherland lattice with diagonal couplings chosen twice as strong as the nearest-neighbor interactions. In the absence of an external magnetic field, the model is known to be disordered at all temperatures, with a macroscopically degenerate ground state. We find the ground-state entropy to be $0.458\,777\,78(10)$ per site, which is in good agreement with the estimate of a recent study using the corner transfer matrix renormalization group method. The sampled ground-state configurations are analyzed by considering the bond configurations, the local-energy configurations, and the spin-spin correlations. While all observations are consistent with the disordered nature of the ground state, the numerical data suggest a power-law decay of the size distribution of clusters of sites with local energies different from the ground-state level.
Read the full abstract

Systems and Control

Decentralized Continuous-Time Power Dispatch for Integrated Heat and Power Systems via Bernstein-Galerkin Equivalent Projection

Jie Deng, Zhigang Li, J. H. Zheng et al. (+1 more)

Decentralized power dispatch for integrated heat and power systems (IHPSs) involves coordinating electric power systems (EPSs) and district heating networks (DHNs) while preserving privacy. The existing methods rely mainly on discretetime DHN models and constant-flow operations, limiting the ability to exploit thermal flexibility. Enabling variable-flow and variable-temperature (VF-VT) operations requires thermal-hydraulic coupling to be addressed together with continuous-time thermal dynamics. However, the simultaneous variations in mass flows and temperature lead to a nonconvex IHPS model, for which conventional decomposition methods are difficult to apply. This paper proposes a decentralized continuous-time dispatch framework using Bernstein-Galerkin equivalent projection. For a prescribed mass flow trajectory, DHN thermal dynamics are formulated in the Bernstein space and projected onto boundary variables, yielding an equivalent feasible-region model without disclosing the internal DHN topology or states. The equivalent model and DHN subproblem are derived from a unified Bernstein-Galerkin formulation, ensuring a consistent thermal state representation throughout the EPS-DHN coordination procedure. A safeguarded Anderson prediction scheme further accelerates the alternating coordination process by using historical updates and fixed-point residuals to predict the next input without additional subproblem solving steps. Numerical results show the proposed method preserves thermal state consistency, exploits VF-VT flexibility, and improves economic and computational performance.
Read the full abstract

All authors Jie Deng, Zhigang Li, J. H. Zheng, Yue Chen

Artificial Intelligence

SimpleEvol: An Agent-Loop Framework for LLM-Driven Automated Heuristic Design with Minimal Human Priors

Jianghan Zhu, Cong Zhang, Rongjie Zhu et al. (+2 more)

Large language models (LLMs) have emerged as powerful tools for automated heuristic design (AHD), enabling iterative generation and refinement of heuristics. However, the dominant paradigm embeds LLMs as narrow, fixed components, such as crossover or mutation, within heavily hand-engineered evolutionary frameworks. We argue this misapprehends LLMs. It treats them as specialized tools rather than general reasoners, constrains them to low-level operations, and underutilizes their autonomy. Moreover, the extensive human priors in these frameworks violate the bitter lesson principle that general methods scaling with computation surpass hand-crafted solutions. This raises a key question: which AHD framework designs best convert stronger LLM capabilities into better heuristics? To address this, we propose metrics for LLM-driven AHD framework handcraftedness (AHI) and intelligence conversion efficiency (ICE). Evaluating ten LLMs across three challenging combinatorial optimization problems, we obtain a notable finding that frameworks with fewer human priors consistently yield higher ICE. Based on this finding, we propose SimpleEvol, an agent-loop framework for AHD which removes nearly all human priors and allows the LLM to operate autonomously. SimpleEvol consistently achieves the highest ICE, often by a large margin. Our results challenge the trend toward complex AHD pipelines and point to a lighter and more model-centric alternative, suggesting that reducing human priors is a more effective strategy to scale up with model intelligence. The source code is available at https://github.com/HenryZhu1029/SimpleEvol-Master.
Read the full abstract

All authors Jianghan Zhu, Cong Zhang, Rongjie Zhu, Chi Zhang, Zhiguang Cao

Computation and Language

Bridging Semantic Gaps in RAG through Generated Context Knowledge Fusion

Xinkai Du, Chao Lv, Yalin Sun et al. (+3 more)

Retrieval-Augmented Generation has established itself as a fundamental framework in natural language processing, seamlessly integrating information retrieval with the generative capabilities of large language models. However, this process is fundamentally constrained by a critical challenge: semantic space mismatch between queries and retrieved contexts. We propose Knowledge-Aware Semantic Bridging (KASB), a novel framework that improves passage selection quality through semantic space alignment between queries and retrieved documents through intelligent knowledge fusion. Our approach leverages the complementary strengths of generative and retrieval-based knowledge through a multistage process that enhances both relevance and accuracy. We evaluate KASB on three popular open-domain Question Answering datasets to demonstrate the effectiveness of our approach.
Read the full abstract

All authors Xinkai Du, Chao Lv, Yalin Sun, Quanjie Han, Lei Yao, Maosong Sun

Machine Learning

Interpolated Policy Distillation: A Controllable Continuum Between Off-Policy and On-Policy Distillation

Youxu Shi, Yifan Sun, Dacheng Yin et al. (+5 more)

Off-policy and on-policy distillation have traditionally been formulated as separate paradigms, each favoring a different property of distillation trajectories. Teacher-generated (off-policy) traces are typically high-quality but lie far from the student's distribution, whereas student-generated (on-policy) rollouts are more learnable but often contain erroneous reasoning. We view these paradigms as the endpoints of a policy continuum and posit that a more effective rollout policy may lie in between. We introduce Interpolated Policy Distillation (IPD), which defines the next-token distribution at every decoding step as an explicit linear interpolation between the student and teacher distributions. The interpolation operates at the distribution level, token by token, and its coefficient provides direct control over the balance between trajectory quality and student learnability. Naively sampling from this policy would require sequentially querying the teacher at every token and is thus expensive. To make IPD practical, we accelerate it with a new speculative-decoding rule while exactly preserving the interpolated next-token distribution.At the trajectory level, the resulting rollouts naturally interleave student- and teacher-generated segments. Unlike recent heuristic segment-interleaving methods, however, this interleaving is induced by an exactly realized token-level interpolated policy rather than by hand-designed switching rules. Across text-only and multimodal reasoning benchmarks, IPD consistently outperforms both endpoint policies (SFT and OPD), their conventional two-stage combination (SFT-then-OPD), and recent heuristic segment-interleaving methods, demonstrating that token-level policy interpolation better balances trajectory quality and student learnability.
Read the full abstract

All authors Youxu Shi, Yifan Sun, Dacheng Yin, Haomiao Tang, Guangting Wang, Fengyun Rao, Jing Lyu, Dong Liu

Machine Learning

Trajectory Soup: Pushing the Compute-Scaling Frontier of LLM Mid-training via Diverse Trajectories

Zhehao Huang, Changxin Tian, Qingyuan Yang et al. (+5 more)

Mid-training equips pretrained large language models with specialized and reasoning capabilities, but the returns of this stage are bounded since additional serial compute yields little further downstream improvement and can even degrade some capabilities, which places a practical ceiling on how much compute mid-training absorbs. We revisit how this compute should be allocated to a single run or multiple similar optimizations. We find that branches forked from a shared checkpoint under various controlled recipe reaches measurably different regions of parameter space, and establish a form of compatible diversity that extending one run cannot supply. Therefore, we introduce Trajectory Soup, which distributes a mid-training budget over several independent branches, and consolidates strongest checkpoints selected on validation through intra- and inter-trajectory averaging into a single model. A local bias and variance analysis separates the two averaging levels, showing that inter-trajectory averaging removes residual error beyond the reach of averaging within a trajectory, while checkpoint selection carries a bias that bounds how many checkpoints are worth merging. Across model scales, learning-rate schedules, token budgets, and trajectory counts, Trajectory Soup improves aggregate downstream performance over the strongest single-trajectory average under matched budgets and keeps improving as budgets expand, with the advantage preserved after an identical post-training pipeline. These results position trajectory allocation and merging as a practical way to extend the compute-scaling frontier of mid-training beyond serial saturation.
Read the full abstract

All authors Zhehao Huang, Changxin Tian, Qingyuan Yang, Kunlong Chen, Ziqi Liu, Zhiqiang Zhang, Xiaolin Huang, Jun Zhou

Econometrics

Are there asymmetries in euro area monetary policy transmission?

Michael Pfarrhofer and Anna Stelzer

We answer the question posed in the title with a nonlinear mixed-frequency vector autoregression, estimated with Bayesian additive regression trees. The model combines monthly macro-financial variables with quarterly bank lending survey data, and identifies the dynamic responses from high-frequency policy surprises. Sign asymmetry dominates; a tightening produces monetary policy transmission mostly in line with the theoretical predictions, whereas easing of any size produces mostly insignificant responses. Peak effects vary with initial conditions, while several common, predefined regime splits do not yield considerable differences in the propagation of the shocks.
Read the full abstract

Graphics

Length-varying Neural Motion Stitching via Cluster Transition Graph

Haemin Kim, Junghyun Nam, Seokhyeon Hong et al. (+2 more)

Motion stitching aims to create new character animations by seamlessly combining existing motion sequences. Existing approaches often require manual selection of transition range or assume fixed transition length, restricting the types of motions that can be connected. To broaden the diversity of motions that can be synthesized, it is essential to generate transitions of varying lengths, allowing the character sufficient time to adapt its pose when the input motions differ significantly. To this end, we propose a length-varying neural motion stitching method based on a cluster transition graph, which produces naturally connected motion sequences given two distinct input motions. Our framework consists of three stages: motion clustering, cluster pathfinding, and motion generation. First, motion clustering maps input motions to discrete clusters. Next, we identify the corresponding clusters in the cluster transition graph and search for a connecting path. In this graph, nodes represent motion clusters, and directed edges indicate valid transitions between them. The resulting path determines both the transition length and a guide sequence that informs motion generation. Finally, the path and input motions are provided to a Transformer encoder-based motion generator to produce the final transition poses. Experimental results demonstrate that our method adaptively adjusts the motion length and successfully generates plausible transitions between distinct motions, such as crawling, basketball shooting, and slow locomotion. We also show that using a graph structure effectively estimates transition durations and produces high-fidelity results compared to methods that assume a fixed transition length, or directly compute the time.
Read the full abstract

All authors Haemin Kim, Junghyun Nam, Seokhyeon Hong, Vanessa Tan, Junyong Noh

Chemical Physics

Adaptive Expansions of the Optimized Effective Potential in Physically Motivated Response Spaces

Gabriel Chirchir, Aditi Singh, Yan Lukashevich et al. (+4 more)

The optimized effective potential (OEP) method provides an exact framework for incorporating orbital-dependent exchange within Kohn-Sham (KS) density-functional theory (DFT). Its practical implementation requires representing the exchange-response potential in a suitable auxiliary space, yet conventional choices are not necessarily adapted to the physical structure of the OEP response. Here, we introduce a general response-space strategy in which physically motivated response functions from existing model exchange potentials are repurposed as adaptive auxiliary directions for the OEP equation. Specifically, response functions associated with the Becke-Johnson (BJ), R"as"anen-Pittalis-Proetto (RPP), Gritsenko-van Leeuwen-van Lenthe-Baerends (GLLB), Krieger-Li-Iafrate (KLI), and localized Hartree-Fock (LHF) constructions are incorporated into compact auxiliary spaces, while their coefficients are determined directly from the projected OEP equation rather than fixed by the underlying model potentials. This establishes a systematic connection between model exchange potentials and finite-basis OEP: model potentials provide physically informed response directions, while OEP determines their system-dependent amplitudes. The framework encompasses one- and multidimensional response spaces, including occupied-orbital and occupied-pair representations, without modifying the underlying OEP condition. We show that these physically adapted spaces capture the dominant spatial structures of the OEP exchange response with substantially fewer degrees of freedom than conventional auxiliary expansions. The proposed approach therefore provides a general route to compact, adaptive representations of exchange-only OEP potentials and a systematic framework for developing low-dimensional OEP approximations from physically motivated model response functions.
Read the full abstract

All authors Gabriel Chirchir, Aditi Singh, Yan Lukashevich, Igor Sawicki, Bogumiła Jezierska, Subrata Jana, Szymon Śmiga

Robotics

Disentangling Spurious Correlations in Vision-Language-Action Models via Predicting Domain-Invariant Latent Lookahead

Junghyun Kim, Ngseo Kim, ChungWoo Lee et al. (+7 more)

Vision-Language-Action (VLA) models remain brittle under visual distribution shifts, often relying on spurious correlations tied to domain-specific factors rather than task-relevant structure. We propose Domain-Invariant Latent Lookahead (DILL), a representation-learning framework that mitigates shortcut learning in VLA policies. Our key idea is to supervise policies with domain-invariant future latents learned from domain-transformed trajectory data. A Task-Domain Encoder is trained with contrastive objectives and Gaussian disentanglement regularization to separate task-relevant structure from domain-specific visual variation. The learned encoder then provides future latents for VLA policy learning through lookahead prediction and domain disentanglement, encouraging the policy to focus on task-relevant structure rather than incidental visual factors. Counterfactual task-view evaluations show that DILL reduces shortcut reliance, while LIBERO-Plus evaluations demonstrate improved visual robustness, with 69.1% average success, 11.4 percentage points above the strongest baseline. Real-world manipulation experiments further support DILL's applicability beyond controlled simulation. Complementary latent-space diagnostics show that these behavioral gains are accompanied by representations that better preserve task-consistent structure while suppressing domain-specific variation. Our project page is available at https://dill-vla.github.io/.
Read the full abstract

All authors Junghyun Kim, Ngseo Kim, ChungWoo Lee, Seoyeon Lee, Woo-Jeong Baek, Adam Zhou, Chip Huyen, Jun-Ki Lee, Gi-Cheon Kang, Byoung-Tak Zhang

Number Theory

Collective Contraction in the de Bruijn–Newman Heat Flow: Off-Zero Logarithmic Derivatives and Certified Barrier Refinements

Michel Planat

We study the de Bruijn–Newman family $H_\tau(z)$, with heat-deformation parameter $\tau$ and $H_0(z)=\tfrac18,\xi(\tfrac12+\tfrac{iz}{2})$. The moving zeros considered here are therefore zeros of $H_\tau$, not directly of $\zeta(s)$. Retaining the collective interaction term in the zero dynamics, we prove that for a simple nonreal zero $z=x+iy$ of maximal imaginary height, [ \frac{d}{d\tau}y^2\le -2-4y^2\mathcal G_\tau(z), ] where $\mathcal G_\tau\ge0$ is a projected interaction sum. A second exact inequality bounds $\mathcal G_\tau$ below by an off-zero logarithmic derivative $-\operatorname{Im}(H_\tau'/H_\tau)(x+i\eta)$, providing a direct interface with the effective Polymath approximation $H_\tau=B_\tau F_\tau$. At the frontier $X=6000000185827$, $\tau_0=129/800$, and $y_0^2=87677/2500000$, a directed Cauchy certificate gives $\mathcal G_\tau>3/2$ for $\tau_0\le\tau\le0.178$, $x\ge X$, and $|y|\le y_0$. Combining this contraction with certified upper and lower zero-free barriers, a two-envelope argument reduces the maximal nonreal height to $0.08$ by time $0.1747532546428610755\ldots$; the classical de Bruijn contraction then completes the landing. Relative to the publicly replayable but not yet peer-reviewed $0.1787854$ base certificate, this yields the audit-relative bound [ \Lambda\le0.1779532546428610755\ldots . ] The collective contraction, logarithmic-derivative bridge, and new high-$x$ certificates are independent contributions.
Read the full abstract

Optics

Optically-Induced Modulation and Programming of Diamond Photonic Cavities

Yuchun Zhu, Amirali Arabmoheghi, Nicolas Le Thomas et al. (+3 more)

Single-crystal diamond combines exceptional optical, thermal, and mechanical properties while hosting optically addressable and spin-coherent colour centres, making it a promising platform for integrated quantum and nonlinear photonics. However, practical post-fabrication mechanisms for dynamically controlling and tuning monolithic diamond photonic circuits remain limited. Here, we demonstrate all-optical modulation and long-lived tuning of suspended diamond Fabry–Perot nanobeam cavities containing nitrogen-vacancy centres. Under 532~nm illumination, the infrared (1000 to 1100~nm) cavity response is governed by two competing contributions: a well-understood thermo-optic red-shift, and a novel charge-mediated blue-shift that we attribute to the free-carrier plasma dispersion effect. With 10~mW of 20~kHz-modulated green light, we achieve an infrared modulation depth of approximately 45% of the available reflection contrast. Green illumination also produces quasi-permanent photo-refractive resonance blue-shifts, with a maximum observed tuning of 3.15~nm (0.87~THz), without measurable degradation of the cavity linewidth or contrast. Achieving the same cavity shift in a lithium-niobate Pockels modulator would require 200 to 500~V across a \(5~\mu\mathrm{m}\) electrode gap. These results establish a monolithic diamond nanophotonic platform combining agile and reversible modulation with long-lived optical reconfiguration, two key ingredients in photonic technologies.
Read the full abstract

All authors Yuchun Zhu, Amirali Arabmoheghi, Nicolas Le Thomas, Niels Quack, Valentin Goblot, Christophe Galland

Computer Vision

End-to-End Self-Supervised RGB-T Tracking without Modality Misleading

Shenglan Li, Rui Yao, Kunyang Sun et al. (+4 more)

RGB-T object tracking leverages the complementary characteristics of visible and thermal infrared modalities to improve robustness under adverse conditions. Existing supervised methods typically rely on costly modality-aligned bounding box annotations, while most self-supervised approaches follow a two-stage pseudo-labeling paradigm, making tracker training sensitive to pseudo-label quality and preventing joint end-to-end optimization. In this paper, we propose ESMTrack, a fully end-to-end self-supervised RGB-T tracking framework without offline pseudo-label generation or dense frame-level bounding box annotations. Given only the standard initial-frame annotation used in visual tracking, ESMTrack learns discriminative and temporally consistent representations through two complementary objectives: a grounding triplet loss on annotated initial frames and a cross-frame temporal triplet loss on unlabeled search frames, with reliable samples selected by forward-backward consistency. To address modality dominance bias, ESMTrack employs a three-branch architecture consisting of a fusion branch and two unimodal branches for RGB and thermal inputs. We quantify modality contributions using the Average Peak-to-Correlation Energy by measuring response discrepancies between the fusion and unimodal branches. The resulting reliability estimates guide a training-time modality decoupling mechanism that suppresses dominant-modality shortcuts and adaptively weights cross-modal contrastive learning for task-level alignment. Extensive experiments on five RGB-T tracking benchmarks show that ESMTrack achieves competitive state-of-the-art performance, strong cross-dataset generalization, and real-time inference speed. The source code is available at https://github.com/LiShenglana/ESMTrack.
Read the full abstract

All authors Shenglan Li, Rui Yao, Kunyang Sun, Hong Jia, Yong Zhou, Javen Qinfeng Shi, Xinyu Zhang

Machine Learning

The Vote Hides the Failure: Aggregation Choice and Noise Robustness in Heart Murmur Detection

Nicholaus Dismas Ladislaus, Olatunji Damilare Emmanuel and Samuel Chol Buol

Noise robustness in automated phonocardiogram (PCG) murmur detection, and how it is measured, remains underexamined despite growing interest in low-resource screening. We evaluate two independently reimplemented pipelines, Hierarchical Multi-Scale Convolutional Network (HMS-Net)–CNN, and Bidirectional Long Short-Term Memory (BiLSTM)–LSTM, under controlled, multi-severity noise with noise-augmented fine-tuning and held-out generalization testing. Under matched aggregation, the complete BiLSTM pipeline outperforms the complete HMS-Net pipeline across all conditions in accuracy and Weighted Accuracy. A stable aggregate accuracy score can misrepresent what individual predictions show: HMS-Net's native aggregation degrades under salt-and-pepper noise far less than majority-vote (MV) aggregation at the same severity, a gap reflecting window-level disagreement its native rule absorbs, while BiLSTM's MV accuracy rises after noise-augmented training even though its individual predictions do not improve. HMS-Net's training effect is significant under one accuracy metric but not another. Noise-robustness conclusions can depend as much on evaluation choices as on the models themselves.
Read the full abstract

Earth and Planetary Astrophysics

Astrometric Evidence for Outer Giant Planets in Known Exoplanet Systems

Clarissa R. Do Ó, William Thompson, Dori Blakely et al. (+19 more)

Outer giant planets can influence the formation and dynamical evolution of their planetary systems. However, their long orbital periods make them difficult to detect using transit and radial velocity (RV) techniques. Measurements from Hipparcos, Gaia DR2 and DR3 (G23H) can detect the stellar reflex motion induced by distant companions before the release of Gaia DR4 epoch astrometry. Using the G23H catalogue, we identify 29 outer companion candidates ($a_{median}$ $>$ 8.4 AU) in known transiting and RV systems. Starting from 170 exoplanet hosts with $P(\mathrm{companion})>0.75$ and posterior support for $0.3$–$13\,M_{\rm J}$ companions, we retain 29 systems where each selected component has a median semi-major axis beyond the outermost known planet with every known planet outside its 95% highest posterior density region in mass and semi-major axis. We incorporate published RV and imaging data from the literature to further constrain the outer candidate's mass and semi-major axis. In eight systems present in the sample, an RV variation or trend has already been reported in the literature. The reported candidates could potentially explain these RV variations, but joint astrometric and RV fits are needed to establish whether the same companions produce both signals. These results provide a set of high priority targets for further RV monitoring and joint orbit analysis with Gaia DR4. If these companions are confirmed, their orbital constraints will provide a starting point for studies of their influence on the dynamical evolution of the known inner planets.
Read the full abstract

All authors Clarissa R. Do Ó, William Thompson, Dori Blakely, Aniket Sanghi, Jerry W. Xuan, Dimitri Mawet, Jessica Speedie, Sarah Blunt, Simon Blouin, Jingwen Zhang, Doug Johnstone, Jean-Baptiste Ruffio, Eric Nielsen, Brendan P. Bowler, Alexandre Bouchard-Côté, Kyle Franson, William Roberson, Ryan Cloutier, Andre Fogal, Kaitlyn Hessel, Christian Marois, Alexandra Rochon

Machine Learning

GLASS: Global Latent Aggregation with Slot-based Set Decoding for Scalable All-Atom Crystal Generation

Hendrik Kraß, Seyed Mohamad Moosavi and Mathias Niepert

Generative models for crystals enable the discovery of novel structures, but scaling all-atom generation to larger systems such as metal–organic frameworks remains challenging. We connect this difficulty to the correspondence problem of particle-space generation. Even on a single fixed target set, index-free permutation-equivariant particle flows require substantially more training for reliable generation as set size and density increase, under both independent and optimal-transport couplings. To resolve this challenge, we introduce GLASS—Global Latent Aggregation with Slot-based Set Decoding, which encodes structures in a permutation-invariant global latent space and learns their distribution via flow matching. A learned-slot decoder constructs all atoms in parallel, removing atom-wise correspondence from generative transport. On MP20, GLASS is competitive with particle-space models, and flow training can reach the validity of the training data at every structure size. On a QMOF subset, GLASS generates MOFs with up to 150 atoms per unit cell without conditioning on building blocks, topology, or composition, and approaches the structural validity of the training data. On both datasets, flow training exposes a validity–novelty tradeoff, and MOF novelty remains limited by autoencoder generalization on the available data. These results show that separating correspondence assignment from generative transport provides a simple route toward high-validity generation of larger atomistic systems.
Read the full abstract

Artificial Intelligence

From Learner Behavior to Reusable Skills for Effective and Efficient Learner Simulation

Zijian Chen, Zheng Zhang, Miao Jia et al. (+3 more)

Learner simulation aims to reproduce how a particular learner behaves on new tasks. Although Large Language Models (LLMs) can generate increasingly fine-grained learning behaviors, existing approaches often need to repeatedly process a growing interaction history to reconstruct the learner. This introduces additional context and inference costs and makes the acquired learner-specific simulation capability difficult to reuse across different LLMs. We therefore propose Learner2Skill, which externalizes the simulation capability acquired from historical interactions into a persistent and reusable Simulation Skill. The Skill captures the learner's current learning state and recurring response patterns, evolves as new real interactions arrive, and can be adapted to a new LLM through lightweight executor calibration without reconstructing the learner from scratch. Experiments show that Learner2Skill more faithfully reproduces fine-grained learner behavior while reducing overall token cost, and that the same constructed Skills can be effectively reused across different LLM executors.
Read the full abstract

All authors Zijian Chen, Zheng Zhang, Miao Jia, Xingchen Hu, Weibo Gao, Linan Yue