Robotics
Training Generalist Navigation Policies via Action Self-Distillation
Bowen Feng, Keqin Wang, Yulong Yang et al. (+4 more)
Language provides a natural interface for humans to specify and reason about goals, but training
language-conditioned robotic policies and
reasoning typically requires expensive manual annotations. We propose Action Self-
Distillation (ASD), a method for learning
language-conditioned
reasoning for robotic navigation without human-annotated
language or
reasoning labels. ASD uses an open-source VLM, which contains broad world knowledge and
reasoning capabilities, as both a teacher network and a student policy. During training, the self-teacher receives privileged trajectory feedback and infers the underlying navigation goal together with goal-oriented
reasoning. The student is then trained to condition on the inferred navigation goal, predict future ego-frame trajectory, and provide a description of the scene semantics, plan, and
reasoning, from onboard observations alone. Across 13 challenging indoor and outdoor real-world navigation tasks, our method improves success rate over the strongest baseline by $30\%$ in indoor settings, $30\%$ on multi-stage tasks, and achieves $95\%$ success rate on outdoor tasks, validating that privileged trajectory feedback can enable self-
distillation for scalable generalist navigation policies without human annotations.
Read the full abstract ▾
All authors Bowen Feng, Keqin Wang, Yulong Yang, Amogh Joshi, Christine Allen-Blanchette, Dhruv Shah, Felix Heide
Quantum Physics
Universal entanglement detection with geometrically optimal witnesses
Xuanran Zhu, Keyi Huang, Chao Zhang et al. (+3 more)
Constructing optimal
entanglement witnesses for arbitrary
quantum systems remains a formidable challenge, hindered by the structural complexity of high-dimensional
quantum state spaces. In this work, we overcome this barrier by introducing a unified framework for constructing geometrically optimal witnesses capable of identifying arbitrary
entanglement structures. By leveraging the
geometric properties of a target
quantum state relative to a specific
entanglement structure in
Hilbert-Schmidt space, these witnesses are constructed in an efficient and effective manner. Among all nontrivial witnesses, they provide maximal distinguishability between the target state and the set of states without the desired
entanglement structure. As a result, our approach enables the universal detection of bipartite, multipartite, and high-dimensional
entanglement. To demonstrate its experimental feasibility, we apply the framework to a four-
qubit nuclear
spin quantum system. Compared to existing methods, our approach offers enhanced noise robustness, broader detection universality, and improved construction efficiency, providing a powerful and versatile tool for advancing the characterization and understanding of
entanglement in
quantum systems.
Read the full abstract ▾
All authors Xuanran Zhu, Keyi Huang, Chao Zhang, Xinfang Nie, Qi Zhao, Dawei Lu
Computer Vision
Physical-Field-Bridged Material Parameter Inference from Multi-View Deformation Videos
Linrui Shou, Jie Li, Yu Liu et al. (+1 more)
Estimating simulation-ready material parameters from visual observations is crucial for deformable-object manipulation but challenging because material properties are only indirectly revealed through physical responses. We introduce physical-field-bridged material inference, a single-pass framework that uses dense physical responses to connect multi-view RGB deformation
videos with constitutive-type classification and continuous parameter regression. Unlike inverse-physics methods requiring per-scene reconstruction, online tuning, or simulation-in-the-loop optimization, our model predicts force/contact masks, projected motion-flow fields, and projected stress-response maps, then conditions material inference on these fields to reduce reliance on appearance or category shortcuts. We construct DeformFieldBench, a 3DGS MPM learning environment comprising 5,000 simulations across 100 object shapes and five mechanically informative loading modes, with dense pixel-aligned physical supervision. Our method achieves the best median errors on all four continuous parameters, improves stress-hotspot recall from 0.0762 to 0.5121, and attains 0.8676 SSIM/36.82 dB PSNR in same-simulator replay on unseen objects. Under a protocol-matched GSO evaluation, inference and future-state prediction take approximately 30 seconds, compared with 15-120 minutes for per-scene optimization. The method also yields measurable replay agreement on five real-world sequences. These results support dense physical fields as an effective bridge for fast material inference, while cross-simulator and real-world tests reveal remaining domain and identifiability gaps.
Read the full abstract ▾
All authors Linrui Shou, Jie Li, Yu Liu, Tao Huang
Machine Learning
Revisiting the 16x Pretraining Compute Gap between Diffusion and Autoregressive Language Models
Guanghan Wang, Bing He, Zhan Shi et al. (+3 more)
Diffusion language models (DLMs) have attracted wide
attention for their fast, parallel decoding, but they are believed to be far less efficient to train than autoregressive
language models (ARMs): a widely cited study reports that a DLM needs roughly 16x the training compute of an ARM to match its likelihood, a figure that has made pretraining DLMs from scratch look prohibitive at scale. We find that the methodology behind this figure has fallen out of step with current DLM practice: it scores the model with its evidence lower bound (ELBO) and trains a full-sequence, decoder-only model. We therefore revisit the scaling laws of DLMs, adopting the exact likelihood under the model's own decoding order, block
diffusion training, an encoder-decoder architecture, and a sweep over the block sizes used in practice. Under this new protocol, we study across ten model variants, about 1,700 models in total, and find that pretraining DLMs is far cheaper than previously thought. Concretely, scoring with the exact likelihood reduces the compute multiple to 12.8x; block
diffusion training and an encoder-decoder architecture further reduce it to 7.8x at block size 32, and smaller block sizes narrow it to 4.6x at block size 8 and 3.0x at block size 4. We complement these scaling analyses with an empirical rule for the optimal decoder depth of encoder-decoder block
diffusion, a held-out compute budget beyond the fitting range at which the gap between variants persists, and downstream
benchmark results where likelihood improvements carry over to accuracy. Overall, the 16x figure substantially overstates the pretraining cost gap between DLMs and ARMs, and pretraining DLMs from scratch is far more viable than previously assumed.
Read the full abstract ▾
All authors Guanghan Wang, Bing He, Zhan Shi, Benoit Dumoulin, Hanqing Lu, Volodymyr Kuleshov
Number Theory
Roth's theorem on weighted projective lines
Sajad Salami
Let ${\mathbb P}^1_{\mathfrak q}$ be the weighted projective line with coprime weights ${\mathfrak q}=(q_0,q_1)$ and $L=q_0q_1$. The weighted Veronese map $\Phi\colon {\mathbb P}^1_{\mathfrak q}\to {\mathbb P}^1$ satisfies $h\circ\Phi=L\,{\mathfrak s}$. Pulling back Roth's
theorem gives the optimal exponent $2L$ for the full rational orbit set. For points represented by coprime integral coordinates, counting suggests the smaller exponent $q_0+q_1$. We formulate this as a
conjecture and prove an $S$-adic radical form under the $abc$
conjecture. For ${\mathfrak q}=(2,3)$ the resulting inequality at the target $[1:1]$ is the coprime Hall-type inequality, and Danilov–Elkies examples show that the exponent $5$ is sharp. We also prove the expected exponent unconditionally over characteristic-zero function fields for constant targets and
polynomial approximants, and for almost every real target. A weighted $abc$ formulation is introduced; in signature $(3,2)$ finitely many coefficient pairs are equivalent to $abc$, and weighted Belyi maps transfer this input to arbitrary algebraic targets. Applications include a pullback form of Schmidt's subspace
theorem, weighted Thue inequalities, and a counterexample to a naive weighted rescaling of Roth's index construction.
Read the full abstract ▾
Machine Learning
Taming Softmax at the Limits of Muon: A Unified Approach through Hilbert Geometry
Aditya Somasundaram, Charles Guille-Escuret, Alexander Moreno et al. (+2 more)
Muon controls weight updates through the RMS movement of a linear layer's output. This implicitly treats the linear output as a proxy for the full module, including the nonlinearity that follows. This proxy can be misleading for softmax, where the RMS shift in activations does not faithfully measure the change in the output distribution. We propose
Hilbert distance as a natural metric for bounding changes in these output distributions.
Hilbert distance bounds changes in cross-
entropy loss and limits probability redistribution. For softmax, it takes the form of the variation seminorm of the preactivation perturbation, allowing probability constraints to be expressed as parameter constraints. Formulating the steepest descent problem with a
Hilbert distance constraint brings two recently proposed methods under a unified mathematical framework. For the LM head, solving a row-wise approximation to the resulting diameter-constrained problem leads us to RowNorm \citep{pethick2025trainingdeeplearningmodels}. For scaled dot-product
attention, the same principle yields the first-order query–key steepest descent problem underlying Compositional
Muon \citep{keigwin2026compositional}. These connections motivate using
Hilbert distance as a guiding principle for optimizer design in softmax modules.
Read the full abstract ▾
All authors Aditya Somasundaram, Charles Guille-Escuret, Alexander Moreno, Zhengzhong Liu, Eric Xing
Optics
Inverse design of poles and zeros: tailoring phase control in metasurfaces
Junhe Zhang, Seoyoung Kim, Hyun Jung Kim et al. (+1 more)
We develop a physically interpretable inverse-design framework for generalized Huygens metasurfaces, using prescribed scattering poles and zeros to engineer resonance interference and transmission-phase control. Compared with direct optimization of transmission or reflection coefficients, this approach offers a physically transparent route to rationally controlling resonance properties, including linewidth and group delay, while reducing computational work. The resulting designs demonstrate how near-unity transmission and $2\pi$ phase excursion can arise from resonance interference. These signatures of the Kerker effect extend beyond the conventional
alignment of electric and
magnetic dipole resonances to other mechanisms, including guided-mode resonances and quasi-bound states in the continuum. We further optimize the Kerker condition across a continuum of refractive-index states in a reconfigurable phase change material metasurface, enabling high-transmission $2\pi$ phase tuning. Our results establish a versatile and practical approach to metasurface phase control through the targeted design of scattering poles and zeros.
Read the full abstract ▾
All authors Junhe Zhang, Seoyoung Kim, Hyun Jung Kim, Juejun Hu
Databases
High-Throughput LLM-Based Compression and What It Takes: A Systems Study of Inference Engines and Entropy Coding Optimizations
David Adams, Luisa Neubauer, Ping-Lin Kuo et al. (+5 more)
Large
language models are attractive text compressors because next-
token probabilities can be converted into code lengths. In practice, however, the main bottleneck is often not modeling quality but systems cost: a compressor must obtain logits quickly enough to compete with strong classical baselines. This paper examines systems optimizations for LLM-based compression, including multistream pipelined arithmetic coding and target-interval capture, and evaluates their trade-offs across inference engines, configurations and
datasets. We evaluate LLM-based text compression on three different inference engines:
Transformers, vLLM, and TensorRT-LLM. Notably, we obtain a speedup of 18.84x compared to prior work's
Transformers implementation and a maximum recorded inference speed of 181 kilobytes per second at a compression ratio of 7.17, when running on an NVIDIA A100.
Read the full abstract ▾
All authors David Adams, Luisa Neubauer, Ping-Lin Kuo, Mihail Stoian, Thomas Rückstieß, Josif Grabocka, Renata Borovica-Gajic, Andreas Kipf
Analysis of PDEs
Global existence of large strong solutions to the 3D compressible micropolar equations with density-dependent viscosities
Tiantian Liu and Xinying Xu
In this paper, we consider the Cauchy problem for the three-dimensional (3D) compressible micropolar equations with density-dependent viscosities. By classical $L^p$ estimates to elliptic system, defining suitable energy functions, after the elaborate index analysis, we establish the global existence of strong solutions to the system with all viscosities depending density as long as the initial density is large enough. By product, We also prove the existence of global large strong solutions with only large initial shear viscosity $\mu_1(\rho)$ and angular viscosity $\mu_2(\rho)$, which consists with the research work \cite{ZZ2014} of 3D compressible Navier-Stokes equations with constant viscosities and improve the recent result of \cite{HLZ} in some extent where they requiring both initial shear viscosity and bulk viscosity to be large.
Read the full abstract ▾
Artificial Intelligence
RAN-JEPA: An Action-Conditioned World Model for Energy-efficient Mobile Network Control
Binghan Wu, Kai Zeng, Kai Yang et al. (+3 more)
Operational mobile networks are rich in observations but poor in interventions: they continuously generate large volumes of passive telemetry, while controlled action–response trajectories remain comparatively scarce and costly to collect. This observation–intervention asymmetry is a central obstacle to building practical world models for RAN control. We address this problem with RAN-JEPA, an action-conditioned world model that separates representation learning from intervention learning. The model first learns a predictive latent representation from observation-only network data, adapts that representation to the target deployment domain, and then learns action-conditioned latent dynamics from controlled sleep/wake trajectories while keeping the encoder fixed. We evaluate RAN-JEPA using public operator performance data for pretraining and a real-equipment four-
cell 5G testbed for domain adaptation, post-training, and control. Across four future 15-minute intervals, RAN-JEPA achieves lower action-conditioned prediction error than a GPT-2 world-model baseline and a mechanistic rule, while remaining close to a strong persistence reference. What-if evaluation further shows that the model produces distinct and coherent future trajectories under matched SLEEP and KEEP plans. In closed-loop control, RAN-JEPA consistently supports higher sleep ratios than the two baselines at comparable congestion cost. These results show that observation-first, action-conditioned learning can turn abundant passive telemetry and limited intervention data into a practical predictive model for energy-aware RAN control.
Read the full abstract ▾
All authors Binghan Wu, Kai Zeng, Kai Yang, Ya-Qin Zhang, Joseph Sifakis, Ye Ouyang
Quantum Physics
Coherent optical control of a thousand qubits in a three-dimensional trapped-ion crystal
Yue Li, Yuxiang Huang, Yi Li et al. (+8 more)
Leading
quantum platforms are largely limited to one or two dimensions, whereas three-dimensional Coulomb
crystals of trapped ions self-assemble into dense
qubit arrays with no rearrangement overhead. Coherent optical control of such
crystals has nonetheless been out of reach, because the optical transition couples to nearly all $3N$ collective modes, i.e., a thermally occupied quasi-continuum bath. Here we cool about 3,600 modes of a
crystal of over 1,200 ions below single-
phonon occupation, using a continuous cooling band that spans more than five octaves in vibrational spectrum, from about 50 kHz to 2.1 MHz. The
crystal then serves as a register of optical
qubits, allowing global initialization, coherent manipulation and collective detection. We demonstrate ensemble optical coherence of 21.6 milliseconds and reduce the optical-transition frequency uncertainty from 389 Hz for a single ion to 12 Hz for over a thousand ions, approaching the expected projection-noise scaling. Our approach could be extended to even larger
crystals with
phonon-mediated many-body couplings, opening a route towards
entanglement-enhanced precision metrology and three-dimensional
quantum simulation, particularly with spatial-shaped illumination for region-resolved collective control.
Read the full abstract ▾
All authors Yue Li, Yuxiang Huang, Yi Li, Xu Cheng, Zihan Xie, Qingyuan Mei, Chang-Ling Zou, Ya Wang, Xinhua Peng, Yiheng Lin, Jiangfeng Du
Machine Learning
Chasing Label Shifters: A Change-Aware Framework for Dynamic Graph Node Classification
Junghoon Kim, Seungyoon Choi, Hyunsung Kim et al. (+1 more)
Entities in real-world systems often evolve over time: users shift between information consumption patterns, researchers migrate between fields, and firms transition between financial risk states. When such systems are modeled as dynamic graphs, these transitions correspond to nodes whose class labels change between consecutive snapshots, which we call shifters. Correctly classifying shifters is often more consequential than classifying stable nodes with persistent labels, as detecting such transitions enables timely intervention in high-stakes settings. Yet, we identify a systematic failure mode across all major dynamic graph neural network architectures: shifter performance consistently lags behind stable nodes, with errors concentrated on the old label. Our structural analysis traces this failure to
embedding inertia: neighborhood aggregation keeps shifter representations anchored to the old-class centroid, and this effect is strongest precisely where it is hardest to correct, namely for shifters whose neighborhoods remain aligned with the old class. Guided by this finding, we propose CHASE, a model-agnostic CHange-Aware framework for Shifting nodEs that wraps any dynamic GNN with targeted components for detecting label shifts and overriding stale neighborhood signals. CHASE consistently improves shifter performance across all tested models and
datasets (up to 114.9% and 202.5% improvement on accuracy and F1 score, respectively) while preserving stable-node performance. We additionally contribute three dynamic graph
benchmarks with naturally shifting labels, filling a gap in existing resources. The source code is available at https://github.com/jhkim611/CHASE
Read the full abstract ▾
All authors Junghoon Kim, Seungyoon Choi, Hyunsung Kim, Chanyoung Park
Data Structures and Algorithms
Query Complexity of Densest Sub(hyper)graph Problems
Sinan G. Aksoy, Eli Friedman, Nate Veldt et al. (+1 more)
Rubinstein et al. introduce the study of connectivity and minimum cut problems in the cut query model, where an
algorithm can only access the edges of a graph through CUT queries. In this work, we examine the query complexity of densest subgraph problems. Our first contribution is to generalize the uniform sparsification approach for Densest Subgraph and other heavy subgraph problems to weighted graphs, achieving the following results. >A $(1+\varepsilon)$-approximation for weighted Densest Subgraph in $O(n\log^2 n/\varepsilon^2)$ queries, roughly matching the $\Omega(n/\log n)$ lower bound we prove for randomized
algorithms. >A $(1+\varepsilon)$-approximation for weighted Max Cut in $O(n\log^2 n/\varepsilon^2)$ queries. This improves upon the bound of $O(n\log^3 n(\log n+1/\varepsilon^2))$ queries achieved by Plevrakis et al., making significant progress towards their lower bound of $\Omega(n/\log n)$ queries. In contrast, we rule out approximation schemes with truly subquadratic query complexity for Densest $k$-Subgraph and settle its exact query complexity in both the unweighted and deterministic settings. For hypergraphs, we show CUT queries alone cannot achieve $o(\sqrt{n})$-approximation for Densest Subhypergraph or $o(n)$-approximation for Densest $k$-Subhypergraph, regardless of the number of queries. Augmenting CUT queries with the EDGE oracle of Chen et al., we overcome this barrier for Densest Subhypergraph and obtain a $(1+\varepsilon)$-approximation on rank-$r$ hypergraphs using $O(nr\log n/\varepsilon^2)$ queries. We also extend our Densest $k$-Subgraph hardness results to hypergraphs and prove exponential query lower bounds for exact Densest $k$-Subhypergraph. Lastly, confirming a
conjecture of Chakrabarty and Liao, we show that simulating an ADD query using CUT and IS-EDGE$^?$ queries requires exponentially-many queries.
Read the full abstract ▾
All authors Sinan G. Aksoy, Eli Friedman, Nate Veldt, Stephen J. Young
Statistics Theory
Nonhomogeneous Subordinators and Lévy Copulas for Reliability Analysis of Time-Dependent Degradation Processes
Yu Shi and Qianmei Feng
Many systems in engineering applications may experience complex multi-dimensional degradation processes involving random jumps. Analyzing the reliability of such degradation processes presents significant challenges. Previously developed models have utilized multi-dimensional Lévy subordinators and Lévy copulas to describe these processes, with reliability functions expressed through characteristic functions and Laplace transforms (Shu et al., 2015, 2016, Shi et al., 2020, Shi and Feng, 2024). However, these models are restricted to stationary linear paths and homogeneous Lévy subordinators. In real-world scenarios, increments in a degradation process are often time-dependent and nonhomogeneous. To address this limitation of previous studies, this work extends existing degradation models by introducing nonhomogeneous Lévy subordinators, which are designed to capture time-varying complexities inherent in degradation processes. Unlike conventional degradation models that often assume a constant rate of jumps, nonhomogeneous Lévy subordinators aim to provide a more flexible representation of degradation behaviors that evolve under varying external conditions and internal system dynamics. Additionally, the joint behaviors of multiple degradation processes are characterized using Lévy copulas, enabling a more comprehensive analysis of system reliability. Numerical examples are provided to illustrate the applicability of these models in lifetime analysis.
Read the full abstract ▾
Machine Learning (Statistics)
Risk-Averse Inference-Time Diffusion Model Alignment with Finite-Sample Guarantees
Yuchen Liang and Xian Yu
Inference-time
alignment steers
diffusion models toward downstream objectives without costly fine-tuning, but existing methods typically optimize expected cost (or reward) without explicitly controlling tail risk. In this work, we propose a risk-averse framework for inference-time
alignment of
diffusion models that balances expected cost and Conditional Value-at-Risk (CVaR) under trajectory-level Kullback-Leibler regularization. We characterize the optimal path distribution and reverse transitions as functions of the CVaR threshold, reducing the
alignment problem to a one-dimensional optimization problem over the threshold. We show that this one-dimensional optimization problem can be nonconvex in general, establish sufficient conditions for convexity, and introduce quadratic regularization to enable efficient derivative-free ternary search. Under suitable assumptions, we provide finite-sample guarantees for threshold optimization and value estimation, and combine them into a terminal total-variation bound relative to the regularized target under exact resampling. Experiments on synthetic data and image generation using Stable
Diffusion demonstrate improved tail performance over risk-neutral baselines while maintaining competitive average cost.
Read the full abstract ▾
Software Engineering
Cross-Project Bug Localization: What Works, What Doesn't, and Why
A Eashaan Rao and Sridhar Chimalakonda
Bug localization models require substantial within-project training data that newly active or small open-source projects rarely have. Cross-project learning (CPL) promises to close this gap by reusing labeled bugs from mature projects, but prior CPL studies evaluate a single model against ample-data within-project baselines that data-scarce projects cannot reach. We ask whether CPL's effectiveness is a property of the transfer setting or depends on model architecture. We evaluate three architecturally distinct models,
embedding-based BLAZE, convolutional TRANP-CNN, and GNN-based
adversarial COOBA, on 63 source to target pairs from 13 open-source Python projects under identical bug corpora, data splits,
retrieval candidates, and baselines. The primary baseline is within-project limited training (WP-small, 20% of target labels, the same target-side share as CP-transfer). CPL outperforms WP-small most clearly for TRANP-CNN (Cohen's d = 0.95, win rate 81.0%, p < 0.001, robust to clustering pairs by target project), and less so for BLAZE (d = 0.49, win rate 60.3%, p = 0.002; not significant once pairs are clustered by target). COOBA, under our re-implementation, shows a small gain that does not survive multiple-comparison correction (d = 0.27, p = 0.027). Negative transfer affects 9.5% of TRANP-CNN pairs and 30.2% of BLAZE's and COOBA's. Cross-model direction agreement is only 44-54%: the same pair can help one architecture and harm another, although the models' gains are positively correlated (rho = 0.24-0.42). CPL outcomes are therefore architecture-dependent: practitioners should validate CPL for their chosen model, and CPL studies should report results per architecture.
Read the full abstract ▾
Astrophysics of Galaxies
Non-equilibrium chemistry and radiation hydrodynamics in metal-poor and metal-rich interstellar gas
Shyam H. Menon, Piyush Sharda, Blakesley Burkhart et al. (+1 more)
We present a flexible implementation of non-
equilibrium chemical networks with the $\texttt{KROME}$ framework, coupled to a nonlocal variable Eddington tensor-closed radiation transport scheme ($\texttt{VETTAM}$). Our fiducial network is designed to explicitly capture the formation pathways of important molecular species such as CO, OH, and $\rm H_2O$, intermediaries like $\rm H_3^+$, and the destruction of these species through
photon and
cosmic-ray dissociation. We validate our scheme through
benchmark tests, and demonstrate its success in 1D and 3D problems covering a broad range of
metallicities ($0 \leq Z \lesssim \rm Z_{\odot}$) and densities ($n \lesssim 10^{17} \, \rm cm^{-3}$). We confirm earlier works that suggest that OH and $\rm H_2O$ are the key molecular species in dense gas ($n \gtrsim 10^{5} \, \rm cm^{-3}$) for $Z\lesssim 0.01 \rm Z_{\odot}$, dominate the cooling at these densities, and demonstrate that our scheme can reproduce their abundances with similar accuracy as dedicated static photo-dissociation region (PDR) codes. We also find that the dominant fine-structure coolant in dense ($n \gtrsim 10^3 \, \, \rm cm^{-3}$) PDRs transitions from CI to CII to OI at increasingly lower $Z$. Finally, we find that in turbulent media, rapid mixing at PDR interfaces renders the chemical species out of
equilibrium, and show that this could potentially alleviate the long-standing problem of (static) PDR models producing excess CI in the diffuse ISM over UV absorption-line measurements. We conclude by presenting and comparing with alternative implementations of reduced but faster chemical networks in our scheme, discussing the problem-dependent performance-accuracy tradeoffs and future prospects.
Read the full abstract ▾
All authors Shyam H. Menon, Piyush Sharda, Blakesley Burkhart, Tommaso Grassi
Computation and Language
Dissociating Answerability from Extraction in LLM Question Answering over Radiology Reports
Priyanshu Sinha and Kelsey McDonald
Answering questions from radiology reports involves two decisions: whether the report contains an answer and which span gives it. A false answer presents an unsupported claim as fact; an over-abstention hides what the report says. Yet RadQA results are reported as a single F1 that merges both. We propose an evaluation method that separates them, reporting false-answer and over-abstention rates beside F1 and splitting F1 changes into extraction and answerability terms. We apply it to 13 LLMs on the RadQA test set, zero-shot and with three cumulative demonstration sets, the last adding an unanswerable example. Models with similar F1 differed more than twofold in false-answer rate. The unanswerable example changed when models answered, not how well they extracted: 12 of 13 models abstained more often, and three-quarters of the answers they withdrew had F1 $\geq$ 0.5. Within models, false answers showed no clear association with extraction quality, but acting on the decision itself helped: keeping an answer only when enough other models also answered lowered the zero-shot false-answer rate by 4 points without a detectable F1 change. Because the two decisions can change independently, we suggest radiology QA evaluations report them separately, not through F1 alone.
Read the full abstract ▾
Astrophysics of Galaxies
Mergers may not be the main channel for galaxy stellar mass growth at redshift $z\gtrsim5$: Predictions from cosmological hierarchical models and simulations
Hao Fu, Mohammadreza Ayromlou, Francesco Shankar et al. (+13 more)
Mergers are a crucial channel of
galaxy growth, but their role relative to
star formation remains uncertain at high
redshift. Here we investigate merger rates and merger-driven stellar mass assembly for
galaxies with stellar mass $M_\star \gtrsim 10^8 \, {\rm M_{\odot}}$ at $z\simeq 5-11$, by combining complementary theoretical approaches within $\Lambda$CDM. These include the DECODE semi-empirical model, the TNG100 and EAGLE hydrodynamical simulations, and the L-
Galaxies semi-analytical model, together with the Millennium and TNG100-Dark simulations used to examine dark matter subhalo statistics and merger timescales. This compares models with different treatments of baryonic physics and
galaxy-
halo connections. We find that dark matter-only simulations predict merger timescales $\sim 30\%$ longer than those inferred from cosmological simulations including baryons.
Galaxy major merger ($M_{\rm \star,sat} / M_{\rm \star,cen} > 0.25$) rates are high at
cosmic dawn and increase towards earlier epochs, reaching $\sim 1/{\rm Gyr}$ at $z\sim 5$ and peaking at $\sim 3/{\rm Gyr}$ at $z\sim 8$. Despite these high rates, mergers contribute only $\lesssim 10\%$ of the total
cosmic stellar mass density
evolution. The ex-situ stellar mass fraction ranges from a few percent to $\sim 20\%$ for all
galaxies at $z\gtrsim 5$, across different models and simulations. Their broad agreement implies that stellar mass assembly at
cosmic dawn is dominated by in-situ
star formation rather than mergers. These findings suggest that the high pair fractions identified by JWST may not directly trace efficient ex-situ stellar mass growth, highlighting the need for physically motivated merger timescales to interpret
galaxy assembly in the early Universe.
Read the full abstract ▾
All authors Hao Fu, Mohammadreza Ayromlou, Francesco Shankar, Lars Hernquist, Andrea Lapi, Daniel Roberts, Feng Yuan, Maulik Bhatt, Evgenii Chaikin, Simona Gallerani, Luca Graziani, Antonio La Marca, George Mountrichas, Yingjie Peng, Haocheng Wang, Kai Wang
Robotics
TacSpin: Tactile Tracking of In-Gripper Object Rotation on SO(3) via Axis-Magnitude Estimation
Yuhao Wu, Linhao Song and Huan Yin
Tracking object motion within a gripper provides essential feedback for robotic manipulation. However, limited contact coverage and uneven deformation make tactile estimation challenging. In this study, we present TacSpin, a method that tracks in-gripper object rotation on SO(3) using only two
vision-based tactile sensors. The key idea is to estimate the axis and magnitude of the rotation response separately, rather than fitting a rigid-body model by least squares. Each finger's response is further refined with contact-boundary cues. The two fingers are then fused according to their agreement. On standard objects, TacSpin achieves lower mean checkpoint errors. Qualitative demonstrations on real-world objects, using calibration transferred from a standard sphere, illustrate tracking under different contact geometries.
Read the full abstract ▾
Computer Vision
Benchmarking Self-Supervised 3D Medical Vision Foundation Models and Transferability Estimation
Jiaqi Tang, Yu Wang, Yang Liu et al. (+7 more)
Self-supervised pretraining has expanded the pool of 3D medical foundation models, but published evaluations use different
datasets, endpoints and fine-tuning protocols. Models also differ in architecture, pretraining objective and data, while downstream targets range from voxel-level masks to patient-level diagnoses and outcomes. As a result, published findings provide little directly comparable evidence for choosing a model for a new task. Direct comparison of adaptation performance requires fine-tuning candidates on labelled target data, at substantial computational cost. We
benchmark 23 candidates across the same 14 segmentation, classification and regression tasks spanning multiple anatomies and modalities. No candidate ranked among the top three across all segmentation tasks or across all classification and regression tasks; a convolutional encoder could outperform a
transformer trained with the same objective on one task but underperform it on another. We
benchmark transferability
estimators against the fine-tuned rankings and find that their agreement varies across dense segmentation, classification and regression. InsightRank uses task-appropriate frozen-feature statistics and architecture-aware pairwise comparisons to rank candidates before fine-tuning. Across the 23-model pool, mean Kendall tau-b was 0.431 for segmentation and 0.352 for classification and regression. The segmentation Top 3 contained a task winner on four of nine tasks; the image-level Top 3 contained none on the five tasks, despite better full-ranking agreement than the matched conventional
estimators. A three-model shortlist reduced recorded extraction-plus-training job time by 85.8% across seven segmentation tasks. Together, the
benchmark provides a common reference for model selection and InsightRank offers a lower-cost way to prioritize fine-tuning, with shortlist performance that differs by target type.
Read the full abstract ▾
All authors Jiaqi Tang, Yu Wang, Yang Liu, Jing Tao, Fandong Zhang, Zhou Zhen, Xiahai Zhuang, Qingchao Chen, Yining Yang, Yizhou Yu
Robotics
EgoCPT: Learning Dexterous Manipulation from a Single Ego-Centric Video with Contact-Aware Optimization
Zhixin Sun, Kairui Shi, Tianao Ren et al. (+3 more)
While expert demonstrations on real
robots are difficult to collect, human ego-centric
videos are easy to collect and abundant on the internet. However, learning dexterous manipulation policies from human
videos remains challenging—one of the most important challenges being the difficulty of reconstructing physically grounded
robot trajectories due to the large embodiment gap. In this paper, we propose EgoCPT, a novel paradigm for learning high-quality retargeted dexterous hand trajectories directly from an ego-centric
video, which can then be deployed directly on real
robots. The only inputs to our method are an ego-centric RGB
video and a reference hand pose trajectory, which can be easily obtained through off-the-shelf software or specialized cameras during recording. Our method is particularly robust to noisy hand poses attributed to our novel contact-aware optimization framework, which iteratively optimizes the contact points between the
robot hands and objects. Across five manipulation tasks, our method achieves overall success rates of 67.3% on noisy EgoDex hand-pose reconstructions and 92.0% on high-quality TACO reconstructions collected using motion-capture hardware, outperforming multiple baselines on both
benchmarks.
Read the full abstract ▾
All authors Zhixin Sun, Kairui Shi, Tianao Ren, Ming Yang, Mark Cutkosky, James Jingxi Xu
Differential Geometry
On the existence problem for a double mean field equation and its application to prescribed curvature problems
Jiayu Li, Lei Liu, Yanjun Liu et al. (+1 more)
Let $(M,g)$ be a compact Riemann surface with smooth boundary $\partial M$. Let $K(x)>0$, $h(x)> 0$ or $h(x)\equiv 0$ be two smooth functions on $M$ and $\partial M$ respectively. In this paper, we consider the following double mean field type equation with a nonlinear Neumann boundary condition related to a prescribed
curvature problem \begin{equation}\label{equation:01} \begin{cases} \displaystyle -\Delta_g u+\frac{4\pi}{|M|} =2\rho\frac{Ke^u}{\int_MKe^u\,dv_g}, & \text{in }M,\\[1.2em] \displaystyle \partial_\nu u =2\rho'\frac{he^{u/2}}{\int_{\partial M}he^{u/2}\,ds_g}, & \text{on }\partial M, \end{cases} \end{equation} where $\rho> 0,\ \rho'\geq 0$ and $\rho+\rho'=2\pi$, $\nu$ is the outward unit normal vector to $\partial M$. When $h\equiv 0$, the second equation changes to $\partial_\nu u=0$ on $\partial M$. By introducing a boundary integration constant, we obtain a sufficient condition under which \eqref{equation:01} admits a solution. As a
geometric application, we solve the prescribed
curvature equation for both the Gauss
curvature and the geodesic
curvature in the case that $M$ is topologically a disk. When $(M,g)$ is the standard hemisphere $\mathbb S^2_+$, let $p\in \partial \mathbb S^2_+$ be a maximum point of $$ \Phi(q)=h(q)+\sqrt{h(q)^2+K(q)}, \qquad q\in\partial \mathbb S^2_+. $$ Suppose \begin{equation*} K_\nu(p)+2\Phi(p)(-\Delta_{\partial \mathbb S^2_+})^{1/2}h(p)<0. \end{equation*} Then the prescribed
curvature problem \begin{align*} -\Delta_g u+\frac{4\pi}{|M|}=2Ke^u, \text{in }\mathbb S^2_+,\ \ \partial_\nu u=2he^{u/2}, \text{on }\partial \mathbb S^2_+. \end{align*} admits a smooth solution.
Read the full abstract ▾
All authors Jiayu Li, Lei Liu, Yanjun Liu, Mingyuan Zhu
Information Theory
Puncturing-inclusion constructions of optimal Ferrers diagram rank-metric codes using semifield codes
Dengming Xu, Chengming Duan, Yong Jin et al. (+1 more)
Ferrers diagram rank-metric codes are essential components of the multilevel construction of constant dimension codes for random network coding. This paper aims to construct optimal Ferrers diagram rank-metric codes, focusing on the diagram \([3,3,3,5]\) with rank distance \(3\). We construct an optimal \([[3,3,3,5],6,3]_q\) Ferrers diagram rank-metric code for every
prime power \(q\), thereby answering an open problem posed by Etzion et al. in 2016~\cite{Etzion2016}. The construction first builds a \([3\times4,4,3]_q\) MRD code and then proves that the Delsarte dual of its puncturing on the last column contains a \([3\times3,3,3]_q\) MRD code, which comes from semifield codes. We further construct optimal Ferrers diagram rank-metric codes for two infinite families of Ferrers diagrams: the first relies on the \([[3,3,3,5],6,3]_q\) construction and holds for every
prime power \(q\), while the second is binary and obtained directly by the puncturing-inclusion construction.
Read the full abstract ▾
All authors Dengming Xu, Chengming Duan, Yong Jin, Yuhao Ning
Analysis of PDEs
An $\eps$-regularity criterion at one scale for 3-D Navier-Stokes equations and improved upper bound on the box dimension of the interior singular set
Yi Lu, Rongrong Chi and Wei Ren
In this paper, we obtain an $\varepsilon$-regularity criterion at one scale for 3D Navier-Stokes equations using a coarse-to-refined estimate argument based on mean-fluctuation decomposition. Making use of this $\varepsilon$-regularity criterion, we improve the upper bound on the box dimension of the interior singular set to $15/13$.
Read the full abstract ▾
Numerical Analysis
A Structure-Preserving Effective Linear Approximation of the Quantum Boltzmann Collision Operator
Shuang Hu, Bo Lin and Zhenning Cai
We construct a structure-preserving effective linear approximation of the
quantum Boltzmann collision operator for Bose and Fermi gases in the normal phase. For variable hard-sphere kernels, the resulting
Fourier spectral evaluation costs $O(N_v^4\log N_v)$, with $N_v$ velocity nodes per direction and no angular quadrature factor. The continuous operator conserves mass, momentum and energy and satisfies a linearized $H$-
theorem. We also use the operator at local
quantum equilibrium as an approximate collision Jacobian in a modified Newton solver that retains the full nonlinear residual. Numerical tests show smaller errors than the
quantum BGK model in homogeneous relaxation and fewer full collision evaluations for steady Couette and
Fourier flows than the same Newton solver using a
quantum BGK Jacobian.
Read the full abstract ▾
Numerical Analysis
Targeted Entropy-Stable Nonlinear Artificial Dissipation for High-Order Spectral-Element Summation-By-Parts Discretizations
Zelalem Arega Worku, Alex Bercik, David C. Del Rey Fernández et al. (+1 more)
Artificial dissipation improves the robustness of high-order methods for under-resolved flows, but can compromise accuracy and suppress physically relevant structures if its strength and spatial distribution are not appropriately controlled. We present a sensor-activated nonlinear artificial-dissipation framework, inspired by the residual correction method, for continuous and discontinuous summation-by-parts discretizations. In contrast to many existing spectral-element dissipation operators that either reduce the formal order of accuracy or target only the highest
polynomial modes, the proposed framework permits design- and higher-order dissipation with explicit control over its magnitude, spatial distribution, activation, and convergence rate under mesh refinement. This modular construction also recovers several existing methods through appropriate choices of its components. The dissipation preserves discrete conservation and semidiscrete
entropy stability whenever the base method satisfies these properties. It is dimensionally consistent, element local, and discretization agnostic, as the added term is constructed independently of the base spatial residual. For smooth solutions under suitable mesh assumptions, we establish parameter choices that retain the formal accuracy of the base discretization. Numerical experiments with summation-by-parts discretizations of the Burgers and Euler equations demonstrate substantially improved robustness for under-resolved flows and flows with shock waves while retaining high-order accuracy for smooth solutions, using the same parameter choices across several test problems. Overall, these results show that targeted dissipation can improve the robustness of high-order methods without excessive damping of resolved flow structures.
Read the full abstract ▾
All authors Zelalem Arega Worku, Alex Bercik, David C. Del Rey Fernández, David W. Zingg
Signal Processing
Learning or Modeling? Virtual XL-RIS Observation Reconstruction for Near-Field Sensing
Tuo Wu, Jie Tang, Jiacheng Wang et al. (+5 more)
A passive extremely large-scale reconfigurable intelligent surface (XL-RIS) offers a position-sensitive aperture for near-field sensing, but its element-wise signals are not directly observed. Recovering this virtual RIS observation (VOR) from fewer base-station measurements requires prior information. This paper investigates when calibrated propagation geometry or paired training data provides the more effective reconstruction prior. For a common sensing operator and target signal, we develop off-grid near-field sparse parametric reconstruction (OG-NF-SPR) and compare it with compact and DenseNet-based deep reconstructors. OG-NF-SPR uses reflected-path projection for position initialization, followed by
Bayesian estimation of all path gains and continuous position refinement. A causal extension incorporates temporal correlation without trajectory labels. Controlled simulations use near-field
UAV tracking as an illustration and separately examine noise, position-distribution shift, calibration error, and propagation mismatch. Under matched propagation at 20 dB, OG-NF-SPR achieves approximately 67% lower reconstruction NMSE than the two nominal deep networks. At low SNR, statistical averaging is more effective than selecting a single
geometric explanation. Under sufficiently strong repeatable mismatch, a same-size network trained on the mismatch family outperforms the nominal physical
estimator, whereas nominal training does not provide this benefit. Temporal regularization also reduces position RMSE from 0.653 m to 0.225 m on a shared trajectory. The results relate reconstruction performance to measurement quality, calibration accuracy, and training coverage, providing guidance on when geometry can replace supervised labels and when additional data are useful.
Read the full abstract ▾
All authors Tuo Wu, Jie Tang, Jiacheng Wang, Kangda Zhi, Maged Elkashlan, Matthew C. Valenti, Mérouane Debbah, Kenneth Tong
Rings and Algebras
On the Real Waring Rank of a Biharmonic Form
Hua-Lin Huang, Yiming Liu and Yilun Tang
We develop an effective method for determining the real Waring rank and the real border rank of biharmonic binary forms. By means of real apolarity, we convert both of these problems into membership questions for regions bounded by a hypocycloid and an epicycloid. We further provide a computable algebraic criterion for the relevant memberships, which involves only the signs of certain real-valued resultants.
Read the full abstract ▾
Earth and Planetary Astrophysics
Elastic Properties of Titan Tholins Revealed by Brillouin Spectroscopy
Mark A. Johnson, Jose R. Montes-Bojorquez, Adis Husic et al. (+7 more)
Titan tholins, complex organic solids synthesized by simulating photochemical processes in Titan's atmosphere, are widely regarded as analogs to organic surface materials on the Saturnian moon. Their mechanical properties are critical for understanding the
evolution of Titan's geomorphology and crustal layering. Here, we report the first direct experimental determination of the complete set of aggregate elastic moduli of tholins using Brillouin spectroscopy. Measurements conducted on tholin powder samples yielded a bulk modulus of 11 (2) GPa and a shear modulus of 3.9 (5) GPa, from which we derive a Young's modulus of 11 (1) GPa and a Poisson's ratio of 0.34 (3). Thin-film samples synthesized under slightly different conditions yield similar results. Compared with other Titan-relevant organic compounds, tholins exhibit the highest elastic stiffness measured to date, exceeding even H2O ice in both bulk and shear moduli. These results support the interpretation that complex organics contribute significantly to the mechanical stability and transport behavior of Titan's equatorial dune sediments. Measurements following prolonged atmospheric exposure reveal persistent stiffening of tholins that remains after
vacuum treatment, likely associated with minor chemical modification and physical reorganization, highlighting the importance of controlled sample handling for future laboratory investigations. This study also demonstrates the effectiveness of Brillouin spectroscopy as a non-destructive technique for characterizing planetary organic materials and provides new constraints relevant to future seismic and geophysical investigations by the Dragonfly mission.
Read the full abstract ▾
All authors Mark A. Johnson, Jose R. Montes-Bojorquez, Adis Husic, Xinting Yu, Wenhao Su, Jia-Sen Hu, Ryan Blase, Edward Patrick, Alan Whittington, Jin S. Zhang