Machine Learning (Statistics)
Bias-Aware Adjoint Calibration and Optimal Control of Stochastic Multiscale Bioprocess Digital Twins
Keilung Choy and Wei Xie
We develop a bias-aware digital-twin calibration and control framework for multiscale bioprocess models within a biological systems-of-systems (Bio-SoS) paradigm. The digital twin is represented by a
stochastic differential equation (SDE) model and calibrated from sparse observations using quasi-likelihood estimation and adjoint sensitivity analysis. SDE generator-based moment expansions characterize truncation-induced parameter bias, while forward-backward adjoints quantify how calibration uncertainty propagates to value functions and policy performance. The resulting parameter-error distribution supports both policy-directed data acquisition and uncertainty-aware policy optimization through a second-order Gaussian-averaged objective. We characterize the asymptotic behavior of the resulting exploration criterion and derive a physical-system performance under the optimized policy. To implement these ideas, we develop an Actor-Simulator
algorithm that jointly updates model parameters, selects informative experiments, and optimizes control policies. Numerical studies demonstrate improved calibration accuracy, sample efficiency, and control performance relative to state-of-the-art baselines.
Read the full abstract ▾
High Energy Physics (Theory)
Emergent $\mathcal N=3$ and Frozen $\mathcal N=2$ SCFTs from $\mathcal N=1$ Gauge Theory
Hongliang Jiang
Four-dimensional $\mathcal{N}=3$ superconformal field theories (SCFTs) are an exotic class of
quantum field theories that admit no conventional Lagrangian description. In this Letter, we construct an $\mathcal{N}=1$ supersymmetric gauge theory and argue that its infrared fixed point exhibits supersymmetry enhancement to an $\mathcal{N}=3$ SCFT. The ultraviolet theory has gauge group $SU(2)^3$, a set of
chiral multiplets, and a suitable superpotential. By performing $a$-maximization, we obtain infrared central charges $a=c=5/4$, matching those of the rank-one $\mathcal{N}=3$ $\mathbb{Z}_3$ S-fold SCFT. The ultraviolet gauge-theory description allows us to compute the full superconformal index with four independent fugacities. We then examine various limits of the full index and further propose a closed-form expression for the Macdonald index, reproducing all known results for the proposed $\mathcal{N}=3$ theory. More remarkably, we argue that further gauging the $\mathcal{N}=3$ theory yields the frozen rank-one $\mathcal{N}=2$ SCFT conjectured from Seiberg–Witten geometry, providing strong evidence for its existence and unlocking its previously inaccessible full superconformal index. Our construction thus opens a systematic route to exploring the refined protected structure of these exotic, intrinsically strongly coupled theories using conventional $\mathcal{N}=1$ Lagrangian techniques.
Read the full abstract ▾
General Relativity and Quantum Cosmology
Ringdown analysis of GW250114 with RingCWB
Achal Kumar, Poulami Dutta Roy and Sergey Klimenko
The binary black hole merger GW250114, the loudest gravitational-wave signal detected to date, offers an opportunity to perform a precise test of Einstein's general relativity (GR). Its exceptionally high signal-to-noise ratio enables accurate measurements of the quasi-normal modes (QNMs) and properties of the remnant black hole. In this work, we present a ringdown analysis of GW250114 using the RingCWB
algorithm. We measure the dominant $(2,2,0)$ QNM along with its first overtone and constrain the fractional deviations in frequency and damping time from their GR-predicted values, finding $\delta f_{220} = 0.004_{-0.023}^{+0.022}$ and $\delta\tau_{220} = 0.010_{-0.087}^{+0.096}$ , both consistent with zero within the quoted uncertainties. These results represent stringent constraints on deviations from GR. The analysis also places tight constraints on the remnant
spin, consistent with a binary black hole formed through
stellar evolution.
Read the full abstract ▾
Computer Vision
RT-DETR-World: Transferring Rich LLM Semantics to Real-Time Open-Vocabulary Detection
Yupeng Zhang, Ziyi Zhao, Juntao Cheng et al. (+4 more)
Open-vocabulary detection (OVD) recognizes categories unseen during training through textual category queries, yet achieving strong generalization with real-time efficiency remains challenging. Beyond vocabulary scaling, zero-shot generalization may benefit from reusable visual–semantic cues learned from seen data, including attributes, actions, states, and contextual relations. Existing real-time OVD methods primarily emphasize vocabulary coverage and efficient region/query–text matching; under strict efficiency constraints, compact detectors may struggle to absorb rich instance semantics and scene context. We propose RT-DETR-World, a compact DETR-style detector that transfers the rich semantics conveyed by descriptions during training while retaining lightweight query–text matching at inference. We construct GroundingCapv2 with three levels of supervision: category names for standard OVD, object descriptions conveying instance-level semantics, and image descriptions conveying object relations and scene context. These descriptions serve only as training-time semantic supervision. To help the compact detector absorb these semantics, we propose Dual-Path Description
Alignment (DDA), combining a deployment-consistent MiniLM pathway with a training-only LLM teacher. MiniLM provides query–category supervision and object-description
alignment, while offline teacher features supervise matched queries and global visual representations at the object and image levels, respectively. All teacher features are precomputed, and the teacher-side modules are removed after training. We further propose Relation-Aware Negative Relaxation (RNR), which uses teacher-derived semantic similarities to relax related negatives while preserving exact positives. Experiments demonstrate competitive zero-shot accuracy and a favorable accuracy–efficiency trade-off. The code will be released.
Read the full abstract ▾
All authors Yupeng Zhang, Ziyi Zhao, Juntao Cheng, Sheng Wang, Ningnan Guo, Ruize Han, Liang Wan
Mathematical Physics
Matrix Product State Representatives of Symmetry-Protected Phases
Thomas Jackson
We provide a simple ansatz for finding integer
spin $SO(3)\times\Z_2^T\times\Z_2^R$-invariant
matrix product states representing a phase for each triple of $\Z_2$ topological indices studied in the literature by Tasaki \cite{tasakiheistop}\cite{tasakitop} and Ogata (corresponding to time-reversal and reflection
symmetries) \cite{ogatatime}\cite{ogatareflection}. Every combination of trivial and non-trivial indices has a
spin-$1$ $SU(2)$, time-reversal, and site-reflection symmetric MPS representative with injectivity length $l\leq 8$. We provide some exact examples for
spin-$1$ and a Fortran program which will find representatives of any triple for any integer
spin, as well as some data from random sampling for small
spin. The existence of such states agrees with the classification of Chen, Gu, and Wen \cite{chen}, and with the recent results of Tasaki \cite{tasakinew} equating the hidden-order and one of Ogata's indices, wherein the Ogata index is that associated to the dihedral $D_2$
symmetry.
Read the full abstract ▾
Astrophysics of Galaxies
Complex Organic Molecules in Protostars with ALMA Spectral Surveys (COMPASS) V. Tracing cavity walls and shocked knots with nonthermally desorbed CH$_3$OH in BHR71-IRS1
H.-S. Yun, J.-E. Lee, J. K. Jørgensen et al. (+16 more)
Complex organic molecules are detected in both cold molecular clouds and the warm inner regions of protostars. Whether they are eventually inherited by protostars and
planets or whether their chemistry is reset by in situ reactions remains unclear. Understanding the desorption mechanisms of these molecules is essential for tracing their chemical
evolution. As part of the ALMA Large Program Complex Organic Molecules in Protostars with ALMA Spectral Surveys (COMPASS), we investigate the CH$_3$OH emission that extends beyond the hot corino of BHR71-IRS1. We identified extended CH$_3$OH emission around BHR71-IRS1. We assessed characteristic emission structures by applying a principal component analysis. The physical properties of their line-emitting gas were derived under the local thermodynamic
equilibrium (LTE) assumption through both rotational diagram and spectral modeling analyses. In total, 26 extended CH$_3$OH lines have been detected. These emission lines are distributed across (1) X-shaped cavity walls, (2) northern and southern shocked
knots, and (3) eastern and southeastern components. The cavity walls exhibit excitation temperatures of $\sim$ 25~K and column densities of $\sim$ 10$^{14}$~cm$^{-2}$, while the
knots show relatively higher excitation temperatures (70$-$120~K) and column densities ($10^{14}-10^{16}$~cm$^{-2}$). The physical properties of eastern and southeastern components are not well constrained under LTE conditions. Plausible desorption mechanisms are evaluated by comparing the physical parameters with model predictions. The low excitation temperatures and narrow line widths in the cavity walls favor a reactive desorption scenario driven by far-ultraviolet irradiation from IRS1. In contrast, the
knots with higher excitation temperatures and column densities support the
dust sputtering scenario induced by C-type shocks.
Read the full abstract ▾
All authors H.-S. Yun, J.-E. Lee, J. K. Jørgensen, A. Coutens, A. L. Plunkett, M. N. Drozdovskaya, A. Belloche, J.-H. Jeong, P. Nazari, Y.-L. Yang, M. L. R. van 't Hoff, S. Maret, F. Cruz-Sáenz de Miera, D. Harsono, C.-H. Kim, Á. Kóspál, M. Lützen, S. Spezzano, S. Zeng
Physics Education
Using Learning Analytics to Study Novice Learners' Work on Quantum Circuit Simulator Tasks
Juuso Jaakola, Vlad Stirbu and Teiko Heinosaari
Interactive
quantum circuit simulators are increasingly used to support introductory
quantum computing education for learners from varied disciplinary backgrounds. These environments can also generate detailed response logs of submitted
circuits, but such data are rarely transformed into evidence that instructors can use for course redesign and development. We present a learning analytics framework for
quantum circuit simulator tasks and report its task-level branch on anonymised logs from an introductory
quantum computing course delivered in the TIM learning environment. The task branch characterises exercises through participation, completion, first-attempt accuracy, valid-attempt counts, sessionised active time, and observed answer-space reporting of common correct and incorrect constructions. The
dataset covers the 2024-2025 cohort of 492 learners, 47,533 valid submissions, and 21 obligatory
quantum circuit exercises. The framework identifies bottleneck tasks, broad solution spaces, and recurring wrong-answer patterns that are not visible from final scores alone. Its most distinctive output is a plain-text representation of the observed answer space, designed to make open-ended
circuit submissions directly interpretable to instructors. We show how response logs can reveal actionable task-design issues and support concrete course redesign. A learner-level branch is retained only as an ongoing research extension and is not a central empirical contribution of the present paper.
Read the full abstract ▾
Computer Vision
SpatialUQ: Post-Hoc Uncertainty Quantification from Spatial Consistency in Black-Box Vision Models
Md Kawsher Mahbub, Milon Biswas, Mirza Niaz Morshed et al. (+1 more)
Clinical
vision models are often deployed as frozen black boxes with no access to internals, retraining, or ground truth at inference time. We introduce SpatialUQ, a post-hoc uncertainty method using only output probabilities. It measures the Jensen-Shannon divergence between the global prediction and the mean of five fixed spatial crops in six deterministic forward passes. The premise is simple, trustworthy predictions are spatially consistent. On NIH ChestX-ray14 (DenseNet-121, $N{=}25{,}596$), our Multicrop Uncertainty Score (MUS) reaches $0.784$ failure-detection AUC versus $0.664$ for MC-Dropout ($p{<}10^{-6}$) at one-fifth the compute, with native calibration ($\text{SCE}{=}0.049$ vs.\ $0.127$ for $\ell_1$), the best-calibrated among methods above 0.78 AUC. A supervised fusion of MUS with
entropy, confidence, and $\ell_1$ reaches $0.832$, outperforming a five-member ensemble ($0.813$). MUS scales with model quality, reaching $0.899$ with BiomedCLIP ($\rho = 0.846$), while this relationship remains meaningful in-distribution ($\rho = 0.523$) but breaks down under severe distribution shift (VinBigData, $\rho = 0.027$). MUS is well-suited to diffuse findings but is less dependable for small focal lesions such as nodules. Code and experimental materials are publicly available at https://huggingface.co/
datasets/kawsher11/SpatialUQ.
Read the full abstract ▾
All authors Md Kawsher Mahbub, Milon Biswas, Mirza Niaz Morshed, Wei Yu
Artificial Intelligence
Ream: Unfolding Mutual Awareness in Human-Agent Workspaces
Peiling Jiang, Sangho Suh, Varsha Kishore et al. (+6 more)
As AI
agents work alongside humans in shared workspaces, a mutual awareness challenge arises:
agents act at speeds that outpace human monitoring, and users' evolving interests are not always expressed in chat. This challenge is especially pressing in literature review, where both parties retrieve, read, and synthesize a growing body of papers. We present Ream, a literature review workspace that supports mutual awareness through structured artifacts, bidirectional engagement tracking, and localized visualizations. Users can see each party's activity within these documents, and
agents can retrieve the same history to guide their work. In studies with eighteen researchers, participants used these traces to inspect evidence, steer
agents, communicate through annotations, and reflect on their research focus. Shared histories also helped
agents build on earlier work. These findings inform how engagement traces within shared documents can support transparency, personalized assistance, and coordination in human-
agent knowledge work.
Read the full abstract ▾
All authors Peiling Jiang, Sangho Suh, Varsha Kishore, Jonathan Bragg, Haijun Xia, Pao Siangliulue, Daniel S. Weld, Amy X. Zhang, Joseph Chee Chang
Robotics
Sparse Feature Policy Unlearning Mitigates State Hallucination in Vision-Language-Action Models
Jiho Lee, Jeongeun Park, Heayoun Choi et al. (+2 more)
Vision-
Language-Action (VLA) models have shown strong generalization in robotic manipulation by leveraging rich representations from pretrained
vision-
language models. However, their deployment in real-world environments remains limited by recurring unreliable behaviors. In this work, we study state
hallucination, a recurring failure pattern in which a VLA continues acting as if an unrealized
robot-object state had been achieved. Our analyses find that state
hallucination coincides with weakened
attention to task-relevant visual regions, and a mechanistic interpretation via sparse autoencoders reveals that
hallucination-associated sparse features are activated when these failures occur. Based on this analysis, we propose SOUL (Sparse feature pOlicy UnLearning), which selectively unlearns policy knowledge associated with state
hallucination behaviors, where sparse features identified from
hallucination failures and successful behaviors serve as explicit forgetting and retention targets, respectively. Experiments across VLA architectures in simulated and real-world environments show that our method substantially reduces hallucinated failures and improves task success without substantially compromising the existing manipulation capabilities. These results suggest that interpretable feature analysis provides a practical basis for selectively modifying undesirable knowledge in
robot policies.
Read the full abstract ▾
All authors Jiho Lee, Jeongeun Park, Heayoun Choi, Taekyung Kim, Eunwoo Kim
Artificial Intelligence
Cognitive Schemas, Laws and Tasks
Antal Jakovác and András Telcs
This paper asks how explicit representations can support reusable cognitive schemas in knowledge-based problem solving. We develop a structural framework in which schemas are organized by the information and relations required for their use, rather than introduced as unrelated primitives. The framework also distinguishes context-dependent relations from more stable structures that can be reused across different representations. Tasks are described through the information available, the unknowns to be determined, and the constraints that admissible solutions must satisfy. This makes it possible to separate limitations of the representation from limitations of the solving procedure. In particular, we distinguish inconsistency, underdetermination, and contextual insufficiency, where the current representation lacks distinctions or relations required by the external task meaning. We also show formally when a reduction of representation preserves the task-relevant solution structure. The resulting task–schema interface offers a structured way to describe representational conditions relevant to problem solving. It supports the reuse and stabilization of derived knowledge while remaining independent of the particular mechanism used to generate candidate solutions. This may provide a useful component for future solver architectures that combine structured knowledge, verification, and learned proposal mechanisms.
Read the full abstract ▾
Analysis of PDEs
Logarithmic Hessian estimates for non-uniformly elliptic equations in non-divergence form
Jongmyeong Kim and Se-Chan Lee
We establish uniform logarithmic Hessian estimates for solutions of non-uniformly
elliptic equations in non-divergence form. The proof is based on a measure estimate from convex analysis without localization or covering arguments. Moreover, we construct a counterexample that rules out uniform power-type Hessian estimates and shows the sharpness of the logarithmic exponent in the plane. Finally, we extend our estimates to strong solutions and apply them to obtain regularity results for Grushin-type equations.
Read the full abstract ▾
Artificial Intelligence
The Attribution Blind Spot: Layerwise Trajectory Diagnostics for Source Reliance in Retrieval-Augmented Language Models
Zhe Yu, Wenpeng Xing, Yunzhao Wei et al. (+4 more)
A
retrieval-augmented model can match a document without relying on it. Controlled knowledge conflicts make source choice observable and let us ask a second question that prediction alone cannot answer: which internal-state properties define useful intervention directions? We study paired hidden-state changes with Latent Trajectory Shift (LTS), a signed projection onto a training-fitted first principal component (PC1), and keep verified training exposure separate from behavioral source choice. Across the evaluated conflicts, state-change magnitude is often the stronger predictor, whereas signed PC1 is the stronger selective controller: equal-norm interventions change source preference while better preserving non-target behavior, and the frozen direction transfers across the tested
datasets and aligned model pairs. A same-system OLMo study further combines positive choice and control results with inconclusive exposure detection at the achieved power. The central result is a separation: representations that diagnose what a model will choose need not be the representations that best control that choice.
Read the full abstract ▾
All authors Zhe Yu, Wenpeng Xing, Yunzhao Wei, Bo Yang, Chen Ye, Gaolei Li, Meng Han
Computer Vision
Gaussian Material Fields for Volumetric Multi-Energy CT Decomposition
Jian Lin, Jiancheng Fang, Hongming Shan et al. (+3 more)
Volumetric material decomposition in multi-energy computed
tomography requires a representation that organizes multiple three-dimensional material fields in a common spatial domain while retaining differences in composition and local structure. We observe that spatial primitives can be shared across materials without tying their coefficients, but their local capacity must respond to material-specific reconstruction needs. We introduce Gaussian material fields, which represent multiple material distributions with shared anisotropic 3D Gaussian primitives and independent nonnegative material coefficients. The shared geometry defines a continuous spatial basis, while the coefficients determine each primitive's contribution to the individual material fields. To reconstruct this representation from multi-energy projections, a differentiable spectral forward model combines Gaussian material path integrals with a calibrated basis
matrix, enabling joint optimization of spatial geometry and material composition. Material-aware adaptive density control retains material-specific refinement evidence before aggregation and adjusts local representation capacity to accommodate both spatially extensive components and sparse details. Experiments use synthesized multi-energy projections generated from pseudo-reference material maps constructed by conventional methods from publicly available CT data. Across 15 cases, our approach improves average PSNR by 4.03 dB and SSIM by 4.96% over the strongest baseline, while reducing NRMSE by 33.45%. Material-wise comparisons and component ablations support improved recovery of localized structures, while runtime and
memory measurements show favorable computational scaling. These results establish Gaussian material fields as an explicit, adaptive representation for volumetric multi-material reconstruction.
Read the full abstract ▾
All authors Jian Lin, Jiancheng Fang, Hongming Shan, Shaoyu Wang, Yang Chen, Qiegen Liu
Information Theory
Function-Correcting Lee-distance Codes for Symbol-Pair Read Channels
Hareesh K. and B. Sundar Rajan
Function-correcting codes (FCCs) protect specified function evaluations of messages against errors while reducing the redundancy required for reliable communication. We study FCCs for symbol-pair read channels, motivated by high-density storage systems that read overlapping symbol pairs. Phase-shift keying (PSK) modulation is well-suited to such systems due to its bandwidth efficiency and noise robustness. While FCCs for symbol-pair read channels have been studied under the Hamming metric, the Lee metric is a more appropriate error model for $q$-ary PSK and, hence for $q$-ary symbol-pair read channels. We generalize the symbol-pair Lee distance, previously defined only over $\mathbb{Z}_4$, to $\mathbb{Z}_q$, $q\ge2$, and introduce function-correcting symbol-pair Lee-distance codes (FCSPLCs) over $\mathbb{Z}_q$, specializing to $q=2^m$, $m\ge1$, for $2^m$-ary PSK constellations. We investigate their redundancy requirements by introducing irregular-pair Lee-distance codes and relating the optimal redundancy of FCSPLCs to the shortest length of such codes. We derive Plotkin-type and Gilbert–Varshamov-type bounds on the optimal redundancy through lower and upper bounds on the shortest length of irregular-pair Lee-distance codes. For bijective functions, we obtain corresponding Plotkin-type and Gilbert–Varshamov-type bounds for classical Lee metric codes for symbol-pair read channels over $\mathbb{Z}_q$, which, to the best of our knowledge, are the first such bounds for the Lee metric symbol-pair setting. We then specialize the FCSPLC framework to pair-locally bounded functions, the Pair-Lee weight function, and the Pair-Lee weight distribution function, giving explicit constructions and corresponding bounds on the optimal redundancy. Finally, for linear functions, we derive a Plotkin-type lower bound on the optimal redundancy.
Read the full abstract ▾
Analysis of PDEs
Interior $C^2$ Estimates for 2-Sum Hessian Equations
C.Y. Ren and X.L. Sun
This paper establishes an interior $C^2$ estimate for solutions of \[ \sigma_2(D^2u)+\alpha(x)\sigma_1(D^2u)=f(x,u,Du)>0, \] on the positive branch $\Delta u>0$, under the assumptions $\alpha\ge0$ and $D^2\alpha=\beta I$. Using one extra variable and a quartic correction, the equation is reduced to a quadratic Hessian equation in one higher dimension. The terms involving $D^2u$ cancel, so the interior estimate of Li–Wu \cite{LW} applies. The gradient-independent case is also treated. Finally, an example shows that an $L^\infty$ bound on $\alpha$ alone is not enough.
Read the full abstract ▾
Computation and Language
Goldsmith: Gold-Loss-Guided Definition Optimization with an Agentic Annotation Harness
Yihan Li, Hanyi Zhang, Xiaoxi Jiang et al. (+1 more)
Many annotation projects begin before experts have a stable guideline or enough labels to train a task-specific model. We present Goldsmith, an agentic pipeline that turns a small gold set—expert-annotated calibration examples representing the intended task boundaries—into a reusable structured annotation definition. Goldsmith treats this definition as a trainable textual object. Candidate definitions are run on the same gold examples and scored with an executable structured loss, while the output schema, formatting,
retrieval, repair, judging, and human review remain in an external harness. A large
language model (LLM) editor converts the highest-loss failures into textual-gradient revisions, which are accepted only when the measured loss decreases. In
prompt-optimization comparisons, Goldsmith improves over direct rewriting, OPRO, APE, and PromptBreeder under matched evaluation protocols. The resulting definition also improves downstream annotation when combined with
retrieval, score-based routing, and human review across typed span, pair-level relation, and fixed-trigger event-argument tasks. These results show that scarce expert supervision can support both task-definition learning and scalable annotation.
Read the full abstract ▾
All authors Yihan Li, Hanyi Zhang, Xiaoxi Jiang, Man Guo
Robotics
SiGNgapore - An Interactive Dataset for Sign-based Visual Navigation
Nicky Zimmerman, Joel Loo, Zishuo Wang et al. (+1 more)
Robots navigating open-world, human-centered environments must look beyond prebuilt high-fidelity maps and leverage navigational aids embedded in their surroundings, particularly, signs. To advance sign-guided visual
navigation, we introduce SiGNgapore, an interactive,
multimodal dataset designed to
benchmark sign-guided multi-step navigational decision-making. SiGNgapore covers diverse public spaces in Singapore, including shopping malls, public transportation hubs, and hospitals, featuring multi-floor, multi-building layouts and indoor-outdoor transitions. It consists of 56 long-horizon
navigation missions and 456 scenes, each containing a navigational sign and its surrounding environment. We provide synchronized RGB, depth, IMU, and odometry data for each scene. We additionally provide the layout map and the GPS-aligned scene graph for selected mission sites. Beyond navigational decision-making, SiGNgapore supports a variety of spatial-semantic
reasoning tasks, including navigational sign parsing and visual sign grounding, with the ultimate aim of driving progress in mapless
robot navigation.
Read the full abstract ▾
All authors Nicky Zimmerman, Joel Loo, Zishuo Wang, David Hsu
Databases
Correspondences as Decisions: JevNexus for Decision-Centric Schema Matching
Runze Li, Hanchen Wang, Ying Zhang et al. (+1 more)
Schema matching increasingly uses generative
language models to rerank retrieved column candidates, although the underlying task is a bounded correspondence decision. We present JevNexus, which combines typed pairwise decisions with schema/instance evidence and invokes listwise refinement only when the evidence disagrees and the fused margin is small. The evaluation covers 561 cases from six
benchmark families. JevNexus obtains
dataset-macro MRR and Hits@1 of 0.930 and 0.909, compared with 0.926 and 0.903 for Magneto, while reducing mean latency from 123.452 to 15.929 seconds (7.750). Paired analysis finds no statistically significant difference in either MRR or Hits@1. The gate invokes listwise refinement for only 5.665% of source columns and avoids the degradation caused by unconditional refinement. Code and experimental artifacts are available at https://github.com/RazeenLI/JevNexus.
Read the full abstract ▾
All authors Runze Li, Hanchen Wang, Ying Zhang, Wenjie Zhang
Audio and Speech Processing
Mitigating Accent-Language Confusion in Self-Supervised Speech Representations for Language Identification
Minu Kim, Jihwan Lee, David R. Mortensen et al. (+1 more)
Spoken
language identification (LID) aims to recognize the target
language regardless of accent. In practice, however, LID models fine-tuned from self-supervised
speech representations frequently confuse accents with
languages, misclassifying non-native (L2)
speech as the speaker's first
language (L1). We show that non-native
speech representations lie between native target-
language and native L1 poles, causing systematic misclassification. To address this, we introduce a
geometric projection that estimates an L1-bias direction solely from native
speech and removes it before the frozen LID head. Across five MMS-LID models and non-native corpora, this projection substantially improves target
language identification for L2-accented
speech while preserving predictions for native
speech. These results show that accent-induced L1 bias can be corrected directly within the representation space without L2 training data or model adaptation.
Read the full abstract ▾
All authors Minu Kim, Jihwan Lee, David R. Mortensen, Shrikanth Narayanan
Quantum Physics
The Geometry of Lee-Yang Tensors
Jake Hofgard
We provide a
geometric description of Lee-Yang tensors, building upon several recent results of Bravyi, Gosset, Liu, and Wong. By applying tools from complex Perron-Frobenius theory, we generalize one of their key lemmas to bound the spectral ratio of non-normal operators with Lee-Yang radius greater than one. As a direct consequence, we provide a local-to-global principle for bounding spectral gaps of Suzuki-Fisher
Hamiltonians, a family of $2$-local
Hamiltonians, using only the Lee-Yang radius of each local operator to bound the gap from below. On the other hand, we develop a simple perturbative technique to show that the Gibbs states of $k$-local
Hamiltonians, $k \geq 3$, cannot have Lee-Yang radius at least one for all inverse temperatures $\beta \geq 0$. We apply this same method to conclude that Suzuki-Fisher
Hamiltonians are precisely those
Hamiltonians whose Gibbs states are Lee-Yang at all inverse temperatures $\beta \geq 0$.
Read the full abstract ▾
Artificial Intelligence
MIMESIS: Learning User Simulators as Training Environments for Interactive Agents
Hoang Phan, Dat Huynh, Andrey Zhmoginov et al. (+7 more)
Training and evaluating interactive
language agents typically requires rich user interactions, yet collecting human feedback is expensive and difficult to scale. Simulated users offer a scalable alternative, but they must both resemble real user behavior and provide useful learning experiences for
agents. In contrast, most
agent-training frameworks rely on off-the-shelf assistant LLMs, whose helpfulness can make them overly cooperative, explicit, and behaviorally homogeneous compared with real users. We introduce MIMESIS, a purpose-built user simulator trained on human conversations with explicit
reasoning supervision and 13 realistic behavioral patterns derived from real user interactions. Empirically, our 9B model achieves a SOUL-Index of 65.7, surpassing the strongest frontier model. Compared with Claude-Opus-5, the strongest baseline on RealUserSim and SimulatorArena, MIMESIS improves behavioral fidelity by 13.4 points and reduces Turing distance by 3.6 points, respectively. We then freeze the simulator and train an
agent by interacting with the frozen simulator using multi-turn reinforcement learning. Across eight environments, training with MIMESIS yields better
agent performance than training with GPT-5.5 under all nine unseen user simulators, demonstrating stronger generalization to new user simulators. Moreover, we propose Coached On-Policy Self-
Distillation (CSD), which leverages simulator-generated private
reasoning traces and subsequent utterances as feedback on how well the
agent addresses user needs. A coach converts this information into concise coaching notes that describe how the
agent can better anticipate user needs and adapt its behavior over the course of an interaction. CSD turns this feedback into dense,
token-level supervision beyond sparse task rewards, yielding further gains across all nine evaluation user models.
Read the full abstract ▾
All authors Hoang Phan, Dat Huynh, Andrey Zhmoginov, Qi Zeng, Wancen Mu, Yue Cao, Shengjie Bi, Yun He, Changdae Oh, Deren Lei
Combinatorics
Graph Sensitivity of Cartesian Products with Matched Bridges
Zhen-Mu Hong, Zi-Yi Wu and Zheng-Jiang Xia
For a graph $G$, let $f_t(G)$ denote the minimum of the maximum degree of an induced subgraph with $\alpha(G)+t$ vertices, where $\alpha(G)$ is the independence number, and write $f(G)=f_1(G)$. Huang's
theorem gives $f(Q_k)\ge\lceil\sqrt{k}\rceil$ for the $k$-dimensional hypercube $Q_k$. We extend this lower bound to Cartesian products of $k$ bipartite graphs with perfect matchings and prove that equality holds when the factors are connected and each has a matched bridge. In particular, we prove that $f(T_1\Box\cdots\Box T_k)=\lceil\sqrt{k}\rceil$ whenever each $T_i$ is a tree with a perfect matching. We determine the sensitivity of every Cartesian product of paths, settling the even-path case left open by Zeng and Hou [J. Graph Theory 107 (2024), 169–180]. For these tree products, with $D=\lceil\sqrt{k}\rceil$, we also prove that $f_t(T_1\Box\cdots\Box T_k)=D$ whenever $1\le t\le 2^{D-\lceil\log_2D\rceil-1}$. When $t=2$, this equality holds for all $k\ge 2$, provided that at least one factor is not $K_2$. Matching cuts give an additional exact range for products of even-order paths. Finally, we prove that $f_2(Q_k)=\lceil\sqrt{k}\rceil$ for every $k\ge 2$ with $k\not\in \{4,9\}$, whereas $f_2(Q_4)=3$ and $3\le f_2(Q_9)\le 4$.
Read the full abstract ▾
Dynamical Systems
Typical expansive measures for infinite-dimensional maps
Gabriel Lacerda
A classical
theorem by Mañé states that compact metric spaces admitting expansive homeomorphisms must be finite-dimensional. It is a natural question, raised by Morales and Sirvent, whether this dimensional constraint extends to the measure-theoretic notion of expansiveness. In this paper, we answer this question in the negative. We prove that the shift map on the
Hilbert cube is expansive with respect to the product Lebesgue measure. Furthermore, we establish a general sufficient condition: any homeomorphism admitting a full-support expansive measure is densely measure-expansive. As a consequence, we show that the set of expansive and topologically stable measures for the shift map contains a dense $G_\delta$ subset, revealing a rich measure-theoretic structure despite the infinite dimensionality of the phase space.
Read the full abstract ▾
Optics
Vector beams for real-time monitoring of crystalline symmetry on {\mu}m length scale
Yongbo Lv, Zidu Yu, Haotian Zhang et al. (+2 more)
Symmetry constrains and determines the physical properties of crystalline solids. Determining crystalline
symmetry often requires large-scale diffraction facility and advanced analysis software and therefore cannot be realized with real-time feedback. Alternatively, optical anisotropy has also been exploited for elucidating the rotation, mirror and inversion
symmetries of solids on smaller tabletop setups. Second harmonic generation, in particular, has demonstrated a remarkable sensitivity to subtle changes in these
symmetries during phase transitions. To access the full nonlinear susceptibility of a
crystal in a cryogenic or
vacuum environment, angle-by-angle rotation of the scattering plane of an obliquely incident and outgoing beam in co-ordination with their polarization states is required. This typically results in a bulky and wobbly setup, significant feedback latency, and a focal spot size > 20 {\mu}m unsuitable for microscopic samples such as few-layer 2D materials. In this work, we make use of vector beams to realize a non-rotating oblique-incidence setup, demonstrating a real-time (> 10 fps) visualization of crystalline
symmetry on a sample area of ~ 2 {\mu}m in diameter. Both the feedback latency and focal spot size represent more than an order of magnitude improvement over previous designs based on a Gaussian beam and mechanically-rotating stages. Here, the use of a vector beam greatly simplifies and reduces the cost of nonlinear optical anisotropy measurements, heralding widespread applications of this approach in both research and industrial settings, for example, for strain monitoring and
crystal alignment. Our work may also promote broader integration of structured optical field into solid state and condensed matter research in the future.
Read the full abstract ▾
All authors Yongbo Lv, Zidu Yu, Haotian Zhang, Liejia Qian, Hao Chu
High Energy Astrophysical Phenomena
Axion effects on hybrid stars: earlier hadron-quark phase transition and larger quark-matter cores
Yu-Heng Liu, He Liu, Hong-Ming Liu et al. (+3 more)
We investigate the effects of
axions on the properties of hybrid
stars and the possible
hadron-
quark phase transition in their cores within the Maxwell construction. The hadronic phase is described by the improved isospin- and momentum-dependent interaction (ImMDI) model, while the
quark phase is modeled by a three-
flavor Nambu–Jona-Lasinio (NJL) model with an
axion-modified Kobayashi-Maskawa-'t Hooft (KMT) interaction. Based on the obtained equation of state, we analyze the properties of hybrid-
star matter, including the speed of sound and the polytropic index, as well as the structural characteristics of hybrid
stars, including the mass-radius relation and the properties of the
quark core. Our study shows that the
axion field $a/f_a$ slightly stiffens the equation of state of
quark matter but significantly shifts the
hadron-
quark phase transition to lower
baryon densities. These two effects compete in influencing the equation of state of hybrid-
star matter, and may even soften it, which is different from the effect of the vector coupling interaction. Consequently, the
axion field has only a weak influence on the maximum mass of hybrid
stars, but significantly increases the mass and radius of the
quark core, especially for weak vector coupling. These results indicate that, within the present Maxwell construction, the primary role of the
axion is to promote an earlier
hadron-
quark transition at low densities and to enlarge the
quark core.
Read the full abstract ▾
All authors Yu-Heng Liu, He Liu, Hong-Ming Liu, Min Ju, Xu-Hao Wu, Peng-Cheng Chu
Robotics
Precise SE(3) End-Effector Tracking in Whole-Body Humanoid Control
Joohwan Seo, Xiaofeng Guo, Jinkun Cao et al. (+4 more)
Precise end-effector tracking during
humanoid whole-body motion is challenging due to floating-base oscillations,
gravity, dynamic coupling, and locomotion-induced disturbances. We propose ResGAC, a whole-body
humanoid controller for precise end-effector pose tracking that combines
geometric admittance control (GAC) with residual reinforcement learning. GAC provides structured $\SE$ task-space feedback and generates nominal arm joint-position targets, while residual RL compensates for unmodeled dynamics and coordinates locomotion and balance in the shared joint-position action space. The left-invariant
geometric formulation allows the same GAC law to be used across manipulation reference frames. This enables the use of a ground-attached heading frame that preserves planar locomotion while removing pelvis roll, pitch, and heave from the manipulation reference, thereby reducing reference-induced end-effector motion during locomotion. ResGAC is validated on a real Unitree G1
humanoid. Across four standing end-effector tracking
benchmarks, ResGAC consistently outperforms representative baselines, including SONIC, achieving lower translational and rotational errors. Real-world experiments further demonstrate reduced propagation of pelvis motion to the desired end-effector pose using the proposed ground-attached heading frame. ResGAC achieves $90\%$ success in a standing peg-in-hole task compared with $50\%$ for SONIC, and accurate world-frame $\SE$ end-effector pose tracking during lower-body motion. Experimental
videos are included in the supplementary material and are also available on the project website: https://resgac.github.io/ResGAC-website/.
Read the full abstract ▾
All authors Joohwan Seo, Xiaofeng Guo, Jinkun Cao, Roberto Horowitz, Rocky Duan, Guanya Shi, Koushil Sreenath
Computer Vision
InstanceBench: Diagnosing Referential Reasoning and Target Identity in Referring Expression Segmentation
Yuchen Li, Shaoyang Zhou, Yiran Wang et al. (+5 more)
Referring Expression
Segmentation (RES) links natural-
language descriptions to pixel-level object masks. Yet standard evaluation provides limited insight into instance-level referential
reasoning: it does not systematically distinguish referential logics, test target preservation across valid grounding paths, or separate target-selection from mask-generation errors. We introduce InstanceBench, an instance-centered diagnostic
benchmark comprising 6,194 images, 9,264 target instances, and 25,077 human-verified expressions. Each target-centric expression set (TCES) fixes the image and target mask while pairing a minimal expression with a same-target variant that uses another valid cue or grounding path. A compact referential-logic taxonomy spans direct target evidence, same-class selection, relational and compositional grounding, and exclusion, while logic-critical construction suppresses simpler shortcuts. Identity-aware metrics measure target retention and set-level success while separating selection from mask-generation errors. Across 22 native-mask RES checkpoints from 18 model families, the strongest checkpoint reaches 67.1% mIoU but only 59.6% All@0.7. Controlled interventions confirm
language sensitivity, while failure decomposition identifies target selection rather than mask decoding as the main bottleneck. On a controlled training subset, matched supervision improves identity-aware performance, showing that the diagnosed capability responds to targeted supervision. Collectively, InstanceBench supports a measure-diagnose-improve cycle: measuring target consistency across grounding paths, localizing failure sources, and evaluating targeted interventions.
Read the full abstract ▾
All authors Yuchen Li, Shaoyang Zhou, Yiran Wang, Ruiyi Deng, Haoyu Wang, Ziru Wei, Zhen Zhao, Luping Zhou
High Energy Astrophysical Phenomena
Polarised X-ray emission from the accreting millisecond pulsar SAX J1808.4-3658 during an X-ray outburst and a type-I X-ray burst
A. Papitto, A. Di Marco, M.C. Baglio et al. (+22 more)
Modelling the X-ray pulse profiles of accreting millisecond
pulsars (AMSPs) is a key technique to constrain the masses and radii of
neutron stars. To this end, measuring the properties of polarised X-rays provides crucial information on the system geometry. We present the significant detection by IXPE of linearly polarised 2-8 keV X-rays during the decay of the 2025 outburst of the 2.5 ms AMSP SAX J1808.4-3658, with an average polarisation degree of $2.9\%\pm0.5\%$ and a polarisation angle of $46{\deg}\pm5{\deg}$, broadly consistent with the direction observed in the optical band. This level of polarisation is similar to that found in the only other source of this class that has been targeted so far. By using a pulse-phase timing solution obtained with IXPE and NuSTAR, which confirms the long-term
evolution of the
pulsar spin and orbit, we did not find any significant pulse phase-resolved variability in the Stokes parameters. Interpreting the polarised X-rays as originating from hot spots on the NS surface indicates they are almost aligned with the
spin axis. We also report an indication for linearly polarised emission with a degree of $16\%\pm6\%$ (2-6 keV) during the peak of a type-I thermonuclear X-ray burst, with an angle almost orthogonal to that of the persistent emission. If confirmed, this level would exceed expectations for X-rays reflected by a thin
accretion disc.
Read the full abstract ▾
All authors A. Papitto, A. Di Marco, M.C. Baglio, D. Baldi Tomei, C. Ballocco, G. Illiano, R. La Placa, C. Malacaria, A. Bobrikova, J. Poutanen, T. Salmi, A. Sanna, A. Veledina, F. Ambrosino, L. Burderi, S. Campana, F. Coti Zelati, D. de Martino, T. Di Salvo, G.L. Israel, F. La Monaca, A. Marino, A. Miraval Zanon, M. Pilia, L. Stella
Machine Learning
CHASE: Channel-Aligned Structure Exploitation for Geometry-Aware Model Engineering
Wei Wang, Wei Jiang and Ziran Liu
Geometric and Spectral
Alignment (GSA) characterizes trained networks through spectral concentration, physical-channel
alignment, support structure, and changes in singular bases. In this paper, we propose CHASE (Channel-Aligned Structure Exploitation) to use these structures in practical model design. CHASE covers six applications across model modification, reconfiguration, and compression. CORA, COEC, and CORAM apply GSA to parameter-efficient finetuning, structured-
pruning compensation, and model merging. We further develop three new methods. CAGA uses GSA to identify multi-head
attention heads that can share a KV representation and constructs the shared key and value heads through
geometric alignment and low-rank subspace extraction. SAKV uses GSA to determine which adjacent layers can share a low-rank KV-cache representation and the retained rank for each layer group. CAPS uses GSA spectral structure to group output
neurons and selects retained input channels separately for each group. Results from CORA, COEC, and CORAM establish the effectiveness of GSA for adaptation,
pruning compensation, and model merging. Experiments on CAGA show that
geometric shared-head construction substantially improves MHA-to-GQA conversion, and SAKV and CAPS improve over representative baselines for KV-cache compression and structured
pruning. These results show that the structures identified by GSA can be used directly to design methods for a range of model operations.
Read the full abstract ▾