Computation and Language
Lightweight but Linguistically Fair? Quantisation Robustness across Bavarian, German, and English
Jophin John, Michael Hoffmann, Piroska Lendvai et al. (+1 more)
Post-training quantisation reduces the
memory cost of large
language models, but it remains unclear whether compression preserves
language skills of regional varieties as reliably as their standard-
language counterparts. We study this question for Bavarian, Standard German, and English using BAVGROUND, a multiple-choice
benchmark, where 206 source questions are aligned across all three
languages. Across 14 open-weight 7B–10B instruction models, we compare BF16 with 8-bit GPTQ and two 4-bit settings, GPTQ and AWQ. Eight-bit GPTQ remains close to BF16 and largely preserves the relative behaviour of the three
languages, while 4-bit quantisation was found to be much less stable. The two 4-bit methods reduce average accuracy by about 1.4–1.5 percentage points and change 11.9–16.0% of BF16 decisions, with changes concentrated on examples where the BF16 model is already uncertain. This instability does not translate into a consistent Bavarian disadvantage. Although Bavarian, a low-resource dialect, has lower baseline accuracy than German and English, the
language most affected at 4 bits varies across models and quantisation methods. Calibration further shapes these cross-
language effects: the relative Bavarian–German effect changes with calibration
language, while changing the calibration source has little average effect on GPTQ-4 but can affect AWQ-4 much more strongly. Free-form evaluation on five models further shows that 4-bit compression can degrade generated responses even in settings where multiple-choice accuracy changes only modestly, while 8-bit GPTQ remains comparatively stable. Overall, our results show that quantisation robustness for regional
language varieties depends on the model, quantisation method, calibration data, and evaluation setting, and cannot be measured by aggregate multilingual accuracy alone.
Read the full abstract ▾
All authors Jophin John, Michael Hoffmann, Piroska Lendvai, Florent Dufour
Computer Vision
CityCode: Building Agent-Operable 3D Cities from Sparse Aerial Observations
Jiahui Sun, Ran Bi, Yifei Liu et al. (+8 more)
Urban spatial intelligence calls for 3D cities that
agents can measure, interpret, and modify across successive tasks. Maintaining such cities as persistent, code-based states requires lightweight acquisition and support for consistent state updates without repeated dense reconstruction. We therefore formulate a new setting for grounded urban world construction: generating large-scale, instance-level parameterized 3D cities from one low-altitude aerial RGB image per region and known camera parameters, while recovering metrically scaled buildings and explicit road geometry and connectivity. We introduce CityCode, an
agent-operable representation that unifies metric geometry, urban
topology, visual semantics, and 3D assets through persistent entity identities, with dependency-aware versioned execution for validated updates and historical restoration. We further develop Air2Code to construct this representation by combining visual structure extraction with adapted metric depth estimation. To address occlusion and incomplete coverage, recovered structure and observed appearance guide generative completion and subsequent structure-aware 3D generation for buildings and backgrounds. Experiments on CitySample-Aerial and UrbanScene3D demonstrate substantially more accurate metric structure recovery and superior visual fidelity to the target cities compared with the evaluated baselines. Across 50 task chains of 50 steps each, CityCode supports continued querying, editing, planning, and historical restoration over evolving city states. Together, CityCode and Air2Code turn sparse aerial observations into persistent,
agent-operable urban worlds for long-horizon urban spatial intelligence.
Read the full abstract ▾
All authors Jiahui Sun, Ran Bi, Yifei Liu, Xiaolong Zhou, Yishuo Xu, Yuanqi Li, Congcong Zhu, Lei Yang, Yanwen Guo, Zhihang Zhong, Hongjie Zhang
Analysis of PDEs
Frequency-dependent capacitance matrix formulation for Fabry–Pérot resonances. Part II: One-dimensional periodic media
Habib Ammari, Bowen Li, Ping Liu et al. (+1 more)
We study Bloch resonances of one-dimensional periodic systems of high-contrast resonators beyond the subwavelength regime. An exact transfer-
matrix formulation identifies the limiting wavenumbers with the union of the interior Neumann frequencies and the Dirichlet frequencies of the separating intervals. At each fixed reference wavenumber, we introduce a periodic frequency-dependent capacitance
matrix whose positive
eigenvalues determine all leading square-root splittings in the contrast parameter. A direct cyclic matching expansion proves the approximation, including multiple
eigenvalues and small complex contrasts. A spectral exclusion result outside the limiting set yields a direct bandgap statement. At a common resonance of all intervals, we also derive uniform eigenmode approximations and distinguish the zero-frequency mode from nonzero double Bloch
eigenvalues. For positive contrast, exact linear crossings are characterized by the monodromy
matrix being proportional to the identity. A complete local classification is obtained for two resonators per
cell and, through band folding, for fully symmetric arrays with any number of resonators per
cell. We further formulate the weighted Zak phase for isolated bands and its quantization under reflection
symmetry. Explicit two-resonator formulas and numerical illustrations complement the analysis.
Read the full abstract ▾
All authors Habib Ammari, Bowen Li, Ping Liu, Yingjie Shao
Combinatorics
Prescribed and simultaneous failures of log-concavity in trees with perfect matchings
Heshun Fan
We construct trees with perfect matchings and independence sequences that fail log-concavity near the independence number $\alpha$. For every integer $k\geq1$, one family fails at $\alpha-k$ and has a witness of order $O(k^2\log(k+2))$. For every integer $t\geq2$, a single tree has simultaneous failures at $\alpha-1,\alpha-3,\ldots,\alpha-(2t-1)$, with a witness of order $O(t^3)$. All these trees satisfy $\alpha=N/2$, where $N$ is the number of vertices, and contradict
Conjecture 4.2 of Ramos and Sun. A sufficient condition uses a fixed tree and the numbers of attached leaves. For fixed $k$ or $t$, explicit bounds give
geometric decay of $a_j^2/(a_{j-1}a_{j+1})$ at these indices as the trees grow; $a_j$ counts independent $j$-sets. The proofs count independent sets by empty matching pairs. The Lean proof assistant checks formal proofs of the parameterized results and their quantitative bounds; exact 90- and 180-vertex examples illustrate the constructions.
Read the full abstract ▾
Solar and Stellar Astrophysics
The EBLM project - XX. Absolute dynamical masses of five low-mass eclipsing binaries, and an evaluation of mass estimation methods
Adam T. Stevenson, Yasmin T. Davis, Daniel Sebastian et al. (+8 more)
Eclipsing binaries consisting of a solar-type primary and an M dwarf secondary are key laboratories to study objects at the bottom of the main sequence. Fundamental parameters of such
stars have a direct impact upon exoplanetary studies, as searches for temperate Earth-sized
planets focus on M dwarf hosts. Eclipsing binaries are essential to measure these parameters. However for stellar masses $<$ 0.35 M$_\odot$ , few double-lined eclipsing systems are known. Thus, stellar
evolution models and empirical relations have had to rely on single-lined eclipsing binaries, for instance those from the EBLM project (Eclipsing Binary - Low Mass). Using high resolution spectroscopy to extract the weak signal of the secondary (flux ratio $<$ 1 per cent in the optical), we transform five more systems from single to double-lined, deriving absolute masses for both
stars. We then compare methods typically employed to estimate stellar masses for field
stars or single-lined binaries with the absolute dynamical masses we obtain, to assess their accuracy. Our results for SB1 and SB2 analyses are mostly consistent within 1-$\sigma$, though we identify a small systematic overestimation of SB1 masses derived from a combination of radial velocities and eclipse photometry. We show that the bias originates from primary radius overestimation during isochronal analysis, wherein the secondary light reduces photometric magnitudes and impacts the recovered radii by modifying the spectral energy distribution. With proper calibration, we can ensure the accuracy of using the full EBLM SB1 sample to constrain mass-radius-luminosity relationships for the smallest, yet most abundant,
stars in the
Galaxy
Read the full abstract ▾
All authors Adam T. Stevenson, Yasmin T. Davis, Daniel Sebastian, Amaury H. M. J. Triaud, Pierre F. L. Maxted, Alix V. Freckelton, Thomas A. Baycroft, Neda Heidari, David V. Martin, Nicola J. Miller, Alexandre Santerne
Solar and Stellar Astrophysics
The EBLM Project - XIX. The absence of radius inflation in a homogeneous analysis of 66 Northern low-mass eclipsing binaries
Yasmin T. Davis, Amaury H.M.J. Triaud, David V. Martin et al. (+30 more)
Precise and accurate measurements of host
stars are crucial in the search for Earth-like
exoplanets, and M-dwarfs are favourable targets in this quest. Low-mass eclipsing binaries offer unique opportunities to empirically determine stellar masses and radii for low-mass
stars where stellar models may be systematically biased. We present results from the EBLM-North sample, the largest homogeneous study of low-mass eclipsing binaries with secondary characterisation yet, combining high-resolution and high-precision radial velocities from SOPHIE and TESS photometry. We measure a median precision for stellar companions of 4.2% in radius and 8.6% in mass within a sub-sample of 50 high-quality binaries, all with orbital periods $P_{\rm bin} > 5\,\rm days$. We compare our data to the Baraffe and MIST isochrones, and find our sample is statistically consistent with those models, producing reduced chi-squared values, $\chi^2_\nu\sim 1$. Our measured radius inflation is $0.3\pm 1.6$% indicating no statistically significant trend. We likewise detect no significant trend with
metallicity, despite having the precision to identify such a relationship. As such, our observations do not appear to follow the general
metallicity trend predicted in MIST. A $\chi_\nu^2 \sim 1$ indicates the scatter in our results is commensurate with our uncertainties alone. These results constrain the scope for unmodelled effects on the mass-radius relation that would produce radius deviations greater than 4.2%. However, our secondary effective temperatures do not reproduce this high level of agreement with the models, and we do not recover the predicted trends with
metallicity.
Read the full abstract ▾
All authors Yasmin T. Davis, Amaury H.M.J. Triaud, David V. Martin, Alix V. Freckelton, Jenni R. French, Adam T. Stevenson, Annelies Mortier, Karen Collins, Pierre F. L. Maxted, Daniel Sebastian, Aleyna Adamson, Thomas A. Baycroft, Isabelle Boisse, Gavin A. L. Coleman, Georgina Dransfield, Yilen Gómez Maqueo Chew, Leslie Hebb, Neda Heidari, Vedad Kunovac, James McCormac, Nikki J. Miller, Richard Nelson, Martin Nielsen, Alexandre Santerne, Madison G. Scott, Owen Scutt, Alexis M. S. Smith, Matthew R. Standing, Amalie Stokholm, Matthew I. Swayne, Mathilde Timmermans, Peter J. Wheatley, Emma Willett
Data Structures and Algorithms
A Deterministic Near-Linear Time Approximation Algorithm for Restricted Shortest Paths in Directed Graphs
Vikrant Ashvinkumar
In the restricted shortest paths problem, we are given an $n$-vertex $m$-edge directed graph $G$ whose edges are assigned two non-negative weights (or resources) called lengths and costs, a source vertex $s$, and a cost threshold $C$. The goal is to find, for each target vertex $t$, the length of the shortest $(s,t)$-path whose total cost is at most $C$. Ashvinkumar, Bernstein, and Karczmarz (SODA 2025) gave the first randomized bicriteria $(1+\varepsilon,1+\varepsilon)$-approximate
algorithm on directed graphs running in sub-$mn$ time, answering a problem posed by Bernstein (SODA 2012). We substantially improve this on two fronts, supplying a deterministic bicriteria $(1+\varepsilon,1+\varepsilon)$-approximate
algorithm running in near-linear time. Our approach straightforwardly generalizes to more than two resources. For $k$ resources, a deterministic $(1+\varepsilon_1,\ldots,1+\varepsilon_k)$-approximate
algorithm running in near-linear time is possible, with $\varepsilon_1=\ldots=\varepsilon_k=\varepsilon$.
Read the full abstract ▾
Multiagent Systems
Strategic Multi-Agent Resource Allocation via Statistical Contracts
Ben Chugg, Michael I. Jordan and Aaditya Ramdas
We study a class of strategic hypothesis testing problems in which a principal aims to allocate resources among competing
agents who possess data or data-generating mechanisms that may be relevant to the principal's goals. Each
agent can pay a participation cost to opt into a statistical contract, in which case they generate data from a private distribution and receive an allocation based on a predetermined (and publicly known) mechanism. Each
agent is either "null" or "non-null" and the principal's goal is to design the contract to limit the allocation to the former while maximizing transactions with the latter. Using e-values, we define a "continuous-closure," a set of allocations which prevents null
agents from profiting in expectation and provides post-hoc risk control for general loss functions under arbitrary dependence. When the loss is the fraction of resources allocated to null
agents, the region admits
polynomial-time linear optimization. We also give explicit, interpretable closed-form approximations, and we analyze participation from a non-null
agent's perspective, giving conditions under which entry into the contract is profitable. Finally, we extend the framework to support online arrival of the
agents, in which case the principal has to make irrevocable on-the-fly decisions, and provide a method which gives time-uniform error-control in this setting.
Read the full abstract ▾
Statistical Methodology
Multilevel Main Effects Matrix Factor Model
Kaixin Liu, Clifford Lam and Zetai Cen
We propose a multilevel main effects
matrix factor model (MMEFM) to extract meaningful row and column effects from groups of
matrix data. As a generalization of a multilevel
matrix factor model, we provide rigorous definitions of MMEFM and identifications, together with iterative
algorithms for the
estimators of all global and local components in the main effects and common component, including global/local main effects and core ranks, with rates of convergence spelt out. An important feature of the model is that all main effects can be nonstationary and still be consistently estimated at each time point. At the same time, our estimated global common factors are
alignment independent, and hence are truly global, and can be presented and given interpretations more easily. We propose a Bootstrap procedure for detecting the existence and identifying the groups of data that are driven by the global common factors. The power of the test, together with the accuracy of various
estimators, are demonstrated in extensive simulation settings. Estimation performance is also compared to other state-of-the-art methods for
matrix time series by simulations and a set of U.S. equity data. All numerical works are done using the R package MMEFM, which is available at CRAN.
Read the full abstract ▾
High Energy Physics (Phenomenology)
Cosmological Consequences of a Highly Decoupled Right-handed Neutrino
Stephen F. King, Rishav Roshan and Sudipta Show
In the minimal type-I seesaw, one right-handed
neutrino (RHN) can decouple from
neutrino mass generation. Such a state is usually neglected, yet its feeble couplings make it long-lived, and it can dominate the energy density of the Universe before decaying. We study this in a minimal $U(1)_{B-L}$ model, where the RHNs are produced through $B-L$ gauge interactions, for the cases of a decoupled lightest RHN $N_1$ and a decoupled heaviest RHN $N_3$. The resulting early matter-dominated era dilutes the baryon asymmetry, and for decoupled $N_3$ it also delays the equilibration of the charged-lepton Yukawa interactions, allowing a lighter $N_1$ to generate the asymmetry in the unflavoured regime. Breaking $U(1)_{B-L}$ forms
cosmic strings whose gravitational-wave background acquires a spectral break from the early matter-dominated era, with a frequency set by the decoupled RHN mass and the lightest
neutrino mass, within reach of future detectors such as SKA, THEIA, LISA, $\mu$ARES, DECIGO and ET.
Read the full abstract ▾
Machine Learning (Statistics)
On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning
Caixing Wang
Random features provide a standard approach to scaling kernel methods to large
datasets by approximating the kernel with a finite-dimensional feature map. Existing works of random feature methods primarily focus on well-specified models and light-tailed noise conditions, leaving a gap in understanding their performance under more general conditions, such as misspecified models and heavy-tailed noise. In this paper, we provide a comprehensive and improved analysis of kernel ridge regression with random features (KRR-RF) under general conditions. In the light-tailed noise setting, we improve upon the worst-case uniform bounds of the feature map used in the previous works and lead to the minimax optimal learning rates with $2r+\gamma>\alpha_0$, where $(r,\gamma,\alpha_0)$ denote the regularity of the target function, the effective dimension of the kernel, and the
embedding index of the RKHS, respectively. In the heavy-tailed noise setting, we leverage a recently developed
Hilbert-valued Fuk-Nagaev inequality and extend the same optimal results under finite weak moment conditions. The additional cost to achieve robustness is a mildly stronger explicit sample-size requirement than in the light-tailed case. For both settings, our results recover the conditions random features budgets (up to log factors) required in the previous works. In the most benign case when $\alpha_0=\gamma$, KRR-RF are minimax optimal for all $r \in (0,1]$. To the best of our knowledge, this is the first work to establish optimal learning rates for KRR-RF under weak moment conditions in the misspecified setting. Numerical experiments and a real data analysis validate the derived learning rates and demonstrate their practical relevance in the presence of mis-specification and heavy-tailed noise.
Read the full abstract ▾
Mathematical Finance
Portfolio Decarbonization under Incomplete Carbon and Transition Information
Olivier Féron and Ruben Haalebos
We study dynamic portfolio decarbonization under incomplete information about firms' carbon intensities and the underlying transition state. The investor maximizes expected utility of terminal wealth net of a penalty on perceived cumulative financed emissions. Asset returns and carbon intensities depend on a
stochastic transition factor, while the investor may need to infer the current transition state, its long-run mean and firms' carbon intensities from
market observations and periodic noisy disclosures. The optimal strategy combines a Merton-type component with
climate-dependent expected-return adjustments, a direct carbon intensity penalty and intertemporal hedging demands associated with
stochastic carbon intensities and learning. In a numerical application, incomplete information has limited effects on welfare and target satisfaction when prior beliefs are well specified. By contrast, biased beliefs generate welfare losses in either direction but asymmetric environmental consequences: optimistic and pessimistic beliefs induce opposite portfolio tilts and therefore affect the probability of exceeding the cumulative emissions target differently.
Read the full abstract ▾
Computation and Language
Probing the Geometry of Color Representations in Multilingual Language Models
Pablo Loyola, Edison Marrese-Taylor and Andrés Hoyos-Idrobo
Do multilingual
language models learn a universal internal geometry for color, and is it perceptually grounded or primarily linguistic? We decompose color representations into two
geometric axes: (P1)
alignment to perceptual color space, and (P2) cross-lingual
symmetry. Analyzing layerwise
embeddings across models and multilingual
datasets, we find that perceptual grounding is weak:
alignment to CIE-LAB peaks in intermediate layers, then collapses toward output layers. Cross-lingual
symmetry degrades with depth, suggesting later layers encode
language-specific structure. Functionally, we show that
geometric properties predict task performance: models achieve 96.7% accuracy on color classification at layers where P1
alignment peaks, with P1
alignment (r=0.70) significantly correlating with downstream capabilities. Our results reveal a dissociation: text-only LMs learn universal color-
language geometry, not universal color-perception geometry, providing diagnostics for where perceptual structure is preserved in
language models.
Read the full abstract ▾
Artificial Intelligence
AssetForge: Asset-Augmented Task Synthesis for Business Automation Agents
Xiangrui Liu, Shuo Tang and Siheng Chen
Training business automation
agents requires executable tasks that connect a natural-
language request to an application state and a verifiable end state. Traditional Rubrics requires Authors to recreate application interfaces, initial records, policy evidence, and outcome checks for each task, entangling business logic with repeated infrastructure work. We introduce AssetForge, an asset-augmented task-construction pipeline that pairs a reusable construction-asset library with task-specific Rubrics. The assets provide application interfaces, world-state components, policy/evidence carriers, and native assertion definitions; the Rubric directs their instantiation for a particular business task. Executable checks and an additional model-based review validate each candidate, return targeted feedback, and support bounded repair before acceptance. On AutomationBench v1.0.6 public-600, fine-tuning Qwen3.6-35B-A3B on 6,000 AssetForge examples yields 41.17% on the official strict pass@1 metric, 4.4 times the 9.33% base score (+31.83 points). With the same backbone and GLM-5.3 supervision, AssetForge reaches 41.17% versus 30.33% for traditional Rubrics, a 10.84-point gain (35.7% relative) at the same accepted-example scale (6,000 versus 6,000 examples). Both DeepSeek- and GLM-supervised AssetForge models improve AppWorld completion over the base model on Test-Normal and Test-Challenge. These results establish the advantage of AssetForge's complete construction pipeline on AutomationBench public: the coupled asset-Rubric interface directs Authors toward business logic, while executable checks and model-based review validate their tasks.
Read the full abstract ▾
Dynamical Systems
Nonuniform dichotomies via time rescaling
Davor Dragicevic and Cesar M. Silva
We establish characterizations of general nonuniform dichotomies for linear nonautonomous dynamical systems in terms of nonuniform exponential dichotomies of appropriately time-rescaled systems. In discrete time, for a class of slowly growing rates, such a characterization requires an additional condition involving a suitable notion of nonuniform ordinary dichotomy, which can be omitted for invertible systems. In continuous time, we obtain a more general characterization for a large class of growth rates, without imposing this additional condition. Our approach does not rely on adapted Lyapunov norms and applies to
evolution families on arbitrary Banach spaces. As applications, we show how several results concerning general nonuniform dichotomies can be derived from their counterparts for exponential dichotomies.
Read the full abstract ▾
Data Structures and Algorithms
A Practical Combinatorial Interior Point Method for Minimum-Cost Flow
Emil Schätzle, Simon Meierhans, Rasmus Kyng et al. (+1 more)
In a recent breakthrough, Chen, Kyng, Liu, Peng, Probst Gutenberg, and Sachdeva (arXiv:2203.00671) designed the first almost-linear-time
algorithm for minimum-cost flow. Their approach uses an Interior Point Method (IPM) to achieve complexity $m^{1+o(1)}\log U \log C$. However, the $m^{1+o(1)}$ factor hides intricate recursive data structures, large constants and multiple log factors. In this paper, we give an implementation following the core ideas of their IPM with practical data structures and empirically obtained decision rules. We compare our solver with the cost scaling and network simplex solvers of LEMON on the instance families of the DIMACS Flow Challenge 2.0 core
benchmark. Our implementation scales competitively on five of twelve families and better on two. On grid_wide we observe close to linear scaling: $m^{1.04}$ against $m^{1.48}$ for LEMON. We read these results as first evidence that the almost-linear-time framework can be relevant in practice, discuss our insights gained during the implementation process, and motivate where future work should focus to bridge the theory-practice gap.
Read the full abstract ▾
All authors Emil Schätzle, Simon Meierhans, Rasmus Kyng, Maximilian Probst Gutenberg
Analysis of PDEs
Attainment of the sharp stability constant of Beckner conformal log-Sobolev inequality
Lu Chen and Bohan Wang
The existence of minimizer for stability of the fractional Sobolev inequality and the stability of Beckner conformal Log-Sobolev inequality (the endpoint case of fractional Sobolev inequality) have been established by König and Chen-Lu-Tang. The existence of minimizer for stability of the Log-Sobolev inequality is unsolved. In this paper, we solve this problem. For every $n\geq2$, we prove that the sharp $L^2$-stability constant of Beckner conformal log-Sobolev inequality on $\mathbb S^n$ is attained and is strictly less than $\frac{2\log2}{n}$. Moreover, the normalized stability constant of the fractional Sobolev inequality converges to the stability constant of the log-Sobolev inequality as $s\to0^+$. After Hersch normalization, every sequence of $L^{p_{s_k}}$-normalized minimizers for stability of the fractional Sobolev inequality, with $s_k\to0^+$, has a subsequence converging strongly in $L^2(\mathbb S^n)$ to an $L^2$-normalized minimizer of stability of the log-Sobolev inequality.
Read the full abstract ▾
Cryptography and Security
Jev-Guard: Fast Routing and Auditing Agent Actions over Editable Policy Trees
Wenbin Hu, Huihao Jing, Haochen Shi et al. (+9 more)
Agent harnesses require a low-latency guardrail model together with customizable audit policies. However, deploying an auto-regressive guardrail model remains challenging: such models incur high inference latency and are often poorly aligned with customized policies. To address these problems, we present Jev-Guard, a unified model for safety classification and policy routing. We further construct a policy tree over which Jev-Guard performs routing, spanning diverse domains including content moderation, cybersecurity,
privacy, and legal compliance. Specifically, Jev-Guard first navigates the policy tree to select the applicable audit policy given the caller's audit context, and then performs safety classification conditioned on the retrieved policies. On 7,033 evaluation cases, Jev-Guard achieves 93.72% safety accuracy, a mean latency of 58ms, and 93.10% hierarchical routing accuracy. The full routing-and-classification pipeline averages 240ms. Furthermore, we integrate Jev-Guard into the Hermes harness framework. Across
agent trajectories, its separate permission gate cuts mean review time per task from 4.80s for an SFT-LLM judge to 1.33s, with no policy violations are observed for either guard.
Read the full abstract ▾
All authors Wenbin Hu, Huihao Jing, Haochen Shi, Yuxuan Liu, Changxuan Fan, Sirui Zhang, Kawai Chung, Hanyu Yang, Zhenhuan Wang, Ziyi Chen, Haoran Li, Yangqiu Song
Computer Vision
ShoeFit: A New Dataset and Dual-image-stream DiT Framework for Virtual Footwear Try-On
Yuhan Li, Zhiyu Jin, Yifan Tong et al. (+5 more)
Virtual footwear try-on (VFTON), a critical yet underexplored area in virtual try-on (VTON), aims to synthesize faithful try-on results given diverse footwear and model images while maintaining 3D consistency and texture authenticity. Unlike conventional garment-focused VTON methods, VFTON presents unique challenges due to (1) Data Scarcity, which arises from the difficulty of perfectly matching product shoes with models wearing the identical ones, (2) Viewpoint Misalignment, where the target foot pose and source shoe views are always misaligned, leading to incomplete texture information and detail distortion, and (3) Background-induced Color Distortion, where complex material of footwear interacts with environmental lighting, causing unintended color contamination. To address these challenges, we introduce MVShoes, a multi-view shoe try-on
dataset consisting of 7305 well-annotated image triplets, covering diverse footwear categories and challenging try-on scenarios. Furthermore, we propose a dual-stream DiT architecture, ShoeFit, designed to mitigate viewpoint misalignment through Multi-View Conditioning with 3D Rotary Position
Embedding, and alleviate background-induced distortion using the LayeredRefAttention which leverages background features to modulate footwear latents. The proposed framework effectively decouples shoe appearance from environmental interferences while preserving high-quality texture detail through decoupled denoising and conditioning branches. Extensive quantitative and qualitative experiments demonstrate that our method substantially improves rendering fidelity and robustness under challenging real-world product shoes, establishing a new
benchmark in high-fidelity footwear try-on synthesis.
Read the full abstract ▾
All authors Yuhan Li, Zhiyu Jin, Yifan Tong, Wenxiang Shang, Benlei Cui, Xuanhong Chen, Hangcheng Zhu, Bingbing Ni
Algebraic Geometry
Lifting Prym Divisors on a Bielliptic Loop of Loops
Phew Aung, Eric Burkholder, Joan Dalton et al. (+5 more)
Let $C$ be a bielliptic curve over an algebraically closed nonarchimedean field of odd characteristic or characteristic zero, and let $f \colon \widetilde{C} \to C$ be an étale double cover. We prove that, if $f$ specializes to a certain double cover of a loop of loops, then every Prym divisor on the loop of loops lifts to a Prym divisor of the same rank on $\widetilde{C}$. As consequences, we compute the dimension of the Prym-Brill-Noether variety $V^r (C,f)$, show that it is reduced and irreducible when the dimension is positive, and compute its cohomology class.
Read the full abstract ▾
All authors Phew Aung, Eric Burkholder, Joan Dalton, David Jensen, Jacob Letamendi, Brianna Nguyen, Andrew Radicker, Doel Rivera Laboy
Computer Vision
Omni-RS: Unifying Multi-Task Understanding and Generation in Remote Sensing
Pan Yin. Shuai Yang, Yuanhao Liu, Datao Tang et al. (+6 more)
Practical remote sensing applications often require both understanding and generation capabilities for the same geographic scene. However, most existing methods address either understanding or generation in isolation, while existing unified remote sensing models' generative ability is limited to text-to-image. To bridge this gap, we propose Omni-RS, a unified framework supporting four remote sensing understanding tasks and five generation tasks. Specifically, we introduce the "Scene-Guided Two-Stage Training" strategy, which first adapts a
multimodal large
language model to remote sensing understanding and then uses its hidden states to guide a
diffusion transformer for image generation. To enable joint learning across diverse generation tasks, we construct UniGen, a unified multi-task generation
dataset, by organizing heterogeneous remote sensing data into a unified format. Furthermore, we introduce "Detailed Visual Conditioning", which uses visual conditions encoded by a VAE to guide the generation process, thereby supporting additional generation tasks that require reference images. Extensive experiments demonstrate that Omni-RS performs exceptionally well in understanding tasks and has achieved outstanding results in a variety of generative tasks.
Read the full abstract ▾
All authors Pan Yin. Shuai Yang, Yuanhao Liu, Datao Tang, Jialin Li, Hao Wang, Pei Liu, Kaiyu Li, Cong Zhao, Xiangyong Cao
High Energy Astrophysical Phenomena
Two sides of the same spin: tracking optical and radio pulsations of SAX J1808.4-3658 from outburst to quiescence
C. Ballocco, A. Miraval Zanon, A. Papitto et al. (+25 more)
High-time-resolution optical and UV observations have revealed bright coherent pulsations from accreting millisecond
pulsars, exceeding the luminosity expected from thermal emission from NS hotspots or X-ray reflection by the
accretion disk. Characterizing optical pulsations over a wide range of
accretion rates through coordinated multi-wavelength observations can shed light on their origin. We present a timing study of the AMSP SAX J1808.4-3658 during its 2022 outburst and quiescence, using optical TNG/SiFAP2 observations, simultaneous NICER X-ray data, LCO optical monitoring, and MeerKAT
radio observations in quiescence. During the outburst, we detect coherent optical pulsations with pulsed luminosities of 1.3x10^(30)erg/s. Combined with previous detections, our results suggest that the optical pulsed amplitude increases as the X-ray flux decreases. The X-ray and optical pulsed luminosities show opposite trends with X-ray luminosity, suggesting different responses to changes in the
accretion rate. We measure X-ray-to-optical delays of 0.88+-0.04ms and 0.51+-0.03 ms in two outburst observations, indicating that the relative timing between the signals is not constant. Together with the spectral energy distribution, these results support different emission mechanisms: non-thermal emission from relativistic particles for optical pulsations and channeled
accretion onto the NS surface for X-ray pulsations. During quiescence, we detect no coherent optical pulsations, obtaining a 95% confidence upper limit of <2.2x10^(29)erg/s on the optical pulsed luminosity, corresponding to an optical efficiency comparable to those of spider
radio millisecond
pulsars. Simultaneous MeerKAT observations reveal no
radio pulsations down to ~10{\mu}Jy, significantly improving previous constraints but not excluding an active rotation-powered
pulsar whose
radio emission may be absorbed by intra-binary material.
Read the full abstract ▾
All authors C. Ballocco, A. Miraval Zanon, A. Papitto, M. Burgay, M. C. Baglio, C. Malacaria, F. Ambrosino, A. Possenti, G. Illiano, E. D. Barr, R. P. Breton, L. Burderi, S. Campana, C. J. Clark, F. Coti Zelati, D. de Martino, T. Di Salvo, M. Kramer, R. La Placa, F. Lewis, A. Marino, D. M. Russell, A. Sanna, L. Stella, B. Stappers, J. D. Turner, M. Cecconi, A. Ghedina
Robotics
CoSMO: Conformal Semantic Mapping for Open-World Safe Robot Navigation
Shuqi Wang, Yiwei Zheng, Pingrui Zhang et al. (+2 more)
Safe autonomous navigation requires not only
geometric obstacle avoidance but also semantic awareness. Different object categories demand different safety clearances, and the set of relevant categories may expand as the
robot explores novel environments.
Vision-
language models (VLMs) offer rich semantic representations for building maps of unknown scenes, yet their confidence scores are poorly calibrated, making raw predictions unreliable for safety-critical planning. Conformal prediction (CP) can provide coverage guarantees, but existing methods assume a fixed label set and stationary data, limiting their use in open-world settings. We propose Conformal Semantic Mapping for Open-World Safe
Robot Navigation (CoSMO), a semantic safety mapping framework that transforms VLM
embeddings into calibrated coverage sets and safety-aware obstacle dilations over a dynamically expanding hazard vocabulary, even under distribution shift. Experiments on the Matterport3D (MP3D)
dataset show that CoSMO reduces semantic collisions by 61.6% relative to a geometry-only baseline, while maintaining formal coverage guarantees even under delayed expert feedback. Real-world experiments on a quadruped
robot further demonstrate generalization by achieving long-run empirical coverage of 90.73% against the 90% target. Project website: https://yangming911.github.io/cosmo-page/.
Read the full abstract ▾
All authors Shuqi Wang, Yiwei Zheng, Pingrui Zhang, Yue Gao, Xiang Yin
Statistical Mechanics
Charge-selective topological locking and ultraslow coarsening in magnetoelastic media
Tirthankar Banerjee and Amaresh Jaiswal
We demonstrate that the
symmetry of magnetoelastic coupling governs the topological organization of composite defects and can qualitatively alter phase-ordering kinetics. In a two-dimensional theory of an in-plane
magnetic vector coupled to elastic strain, we find that integer
magnetic vortices bind pairs of elementary elastic half-defects. Isotropic magnetostriction produces a core-induced elastic winding independent of the
magnetic charge, giving defect binding without charge locking and leaving the
magnetic sector with essentially XY-like coarsening. In contrast, quadrupolar magnetoelastic coupling locks
magnetic and elastic defect charges: continuous rotational
symmetry favors $q_\phi=q_\theta$, whereas crystalline anisotropy can reverse the locking to $q_\phi=-q_\theta$. We reveal that both locking sectors display anomalously slow
vortex annihilation: defect density shows a power-law decay $t^{-2/5}$, while the average defect-velocity scales as $t^{-4/5}$. These results are consistent with a dynamic exponent $z\simeq 5$. We show that this coarsening slowdown is associated with large anisotropic elastic friction. An adiabatic elimination of the elastic field reveals that isotropic magnetostriction merely renormalizes local
magnetic interactions, whereas orientational coupling induces effective nonlocal interactions between
magnetic spins. Our experimentally testable results thus establish magnetoelastic coupling
symmetry as a means of controlling defect
topology and ordering kinetics.
Read the full abstract ▾
Number Theory
Untrodden pathways in the study of partitions with parts separated by parity
Nargish Punia and Rahul Kumar
In this paper, we study a partition function with parts separated by parity from a family of partition functions introduced by Andrews in 2019. We first show that its generating function admit representation in terms of mock modular forms of weight $3/2$. We then initiate a study of smallest-part partition function analogues of partitions with parts separated by parity, which, to the best of our knowledge, has not previously been investigated in the literature. We show that the generating functions of these smallest-part partition functions are likewise expressible in terms of mock modular forms, mock Maass theta functions and Zagier's $q$-series $H(q)$, which he describes as a “half-derivative of the Dedekind eta-function". Our study also leads to new combinatorial interpretations of the coefficients of the aforementioned mock modular form, function $\sigma^*(q)$, and mock theta functions $\psi(q)$ and $A(q)$ in terms of odd Ferrers graphs. Moreover, we provide a
quantum $q$-series identity involving Ramanujan's famous $\sigma(q)$ function. We also derive the leading asymptotic term of one of these partition functions, whose generating function we first express in terms of $\sigma^*(q)$.
Read the full abstract ▾
Machine Learning
Over-parameterization Is Neither Robust nor Fragile in Matrix Sensing: A Geometric Theory of Solution Robustness
Runze Shi and Ziye Ma
Over-parameterization is ubiquitous in modern machine learning, yet its effect on robustness remains poorly understood. We study this question in
matrix sensing, a canonical non-convex problem that can also be viewed, under an appropriate parameterization, as training a two-layer quadratic neural network. We develop a rank-aware robustness theory for the Burer-Monteiro factorization when the search rank k may exceed the target rank r_
star, thereby capturing over-parameterization. Our analysis shows that robustness depends jointly on the degree of over-parameterization and the local geometry of the attained second-order stationary point: increasing k generally worsens sensitivity, but stationary points with the same search rank can exhibit substantially different robustness. Motivated by this observation, we propose a trajectory-aware
algorithm that uses spectral balancing and randomized sketching to steer optimization toward less sensitive solutions. In
matrix-sensing experiments, the method reduces paired sensitivity by 18-29 percent for isotropic operators and 87-89 percent for weak-tail operators, without degrading reconstruction accuracy. These results demonstrate that parameter count alone does not determine robustness and identify solution geometry and the optimization trajectory as important levers for improving it.
Read the full abstract ▾
Materials Science
Accurately Generalizing Deep-Learning Hamiltonian Methods to Plane-Wave Density Functional Theory
Chuntian Wu, Jiaxiang Li, Qiang Xu et al. (+4 more)
Deep neural networks that predict density functional theory (DFT)
Hamiltonians from material structures offer a route to efficient electronic-structure calculations. However, extending localized atomic-orbital (AO)
Hamiltonian learning to plane-wave (PW) calculations with hybrid functionals requires a compact AO representation that preserves the effects of nonlocal operators. Here we propose a projectability-guided construction of nonorthogonal AO
Hamiltonians from a small, selected set of PW eigenstates and
eigenvalues. High-projectability states are retained, and selected higher-energy states complete the AO subspace, avoiding an empirical shift and its arbitrary contribution to the learning targets. Because the construction uses PW eigenstates and
eigenvalues, it incorporates nonlocal contributions present in the reference calculation without explicitly reconstructing their AO
matrix elements.
Benchmarks for MoS2 and sodium phases show accurate reproduction of PW bands with smaller AO bases than real-space reconstruction. The resulting PBE and HSE
Hamiltonians can be learned by equivariant neural networks, connecting PW reference calculations with electronic-structure prediction. This approach broadens the range of PW DFT calculations available for AO
Hamiltonian learning and enables compact, accurate training targets for developing more widely applicable electronic-structure models.
Read the full abstract ▾
All authors Chuntian Wu, Jiaxiang Li, Qiang Xu, Xuecheng Shao, Congwei Xie, Yu Xie, Yanming Ma
Classical Analysis and ODEs
Endpoint power weight inequalities for spherical maximal functions
Marco Fraccaroli, Juyoung Lee, Nikou Lei et al. (+2 more)
We prove sharp $L^p(|x|^\alpha)$ estimate for spherical maximal functions. We resolve the remaining open problem for the full maximal operator for $\alpha=1-d$ and $p<2$ and prove a more general result involving a local version of Assouad dimension of the underlying dilation set. We also make a case study with endpoint bounds for a model case involving convex sequences in $[1,2]$.
Read the full abstract ▾
All authors Marco Fraccaroli, Juyoung Lee, Nikou Lei, Joris Roos, Andreas Seeger
High Energy Astrophysical Phenomena
Cell-based theoretical criteria on the onset and stability limit of thermonuclear detonation
Kazuya Iwata, Keiichi Maeda and Ken J. Shen
One of the key unresolved problems in thermonuclear explosions of white dwarfs (WDs), including type Ia
supernovae, is the condition for ignition and propagation of thermonuclear flame. Given the limitation of hydrodynamic simulations in resolving the flame front involved in the explosion process, theoretical considerations for these criteria merit accurate prediction of the onset of deflagration and detonation (deflagration-to-detonation transition or direct initiation) and subsequent stable propagation. Application of
cell-based theories, which are originally designed for terrestrial detonation incorporating the size of the cellular structure, to pure-helium detonation in a WD envelope was proven to be powerful. In the present work, we extend this for detonation in helium-carbon and carbon-oxygen mixtures, both of which are associated with the double-detonation model for type Ia
supernovae. Comparing with previous hydrodynamic studies in terms of the minimum initiation energy and critical stellar mass for stable propagation, the agreement is satisfactory. For the helium-carbon detonation, it is theoretically substantiated that even an extremely low-mass WD envelope $\sim 10^{-3}$ M$_{\odot}$ can accommodate stable detonation. For the carbon-oxygen detonation, the oxygen-scale
cell width is shown to be a good diagnosing indicator. In addition, it is indicated that direct initiation of the carbon-oxygen detonation is only available for a WD heavier than 0.6 M$_\odot$. The findings of this study enable first-order calibration of the explosion model by a simple calculation through the analogy between astrophysical and terrestrial burning, the latter of which can be tested and observed in a laboratory-scale experiment.
Read the full abstract ▾
Artificial Intelligence
Complex materials engineering via Provenance Action Graphs: potential, structure and demonstration
Yevgeny Rakita, Paul Todd, James R. Neilson et al. (+1 more)
This work discusses and tests the potential of reaction graphs for material synthesis, in which events consist of reaction steps (actions) and their outputs (observations). The importance comes from the requirement to synthesize technologically useful materials, especially those that are kinetically stabilized. Often, it is a science that follows years of trial and error, culminating in (published) successful recipes, but is truly driven by far more (unpublished) failures. The entry into the era of AI provides an opportunity to use learning
algorithms to encode rules from multi-dimensional synthesis pathways, regardless of their technological usefulness, and to decode both forward prediction and inverse learning modes. The main incentive is to accelerate material discovery by generating a global knowledge base that can be shared among data
agents and contributed to by them. We will focus on several scenarios in the common synthesis process during solid-state synthesis, primarily because these reactions are kinetic in nature and require closer monitoring and control of the reaction history.
Read the full abstract ▾
All authors Yevgeny Rakita, Paul Todd, James R. Neilson, Simon J. L. Billinge