arXiv subjects

Yue Yu

Publications and source records attributed to Yue Yu.

At least 127 records · Page 7Linked to original sources

arXiv2025

Cambrian-S: Towards Spatial Supersensing in Video

We argue that progress in true multimodal intelligence calls for a shift from reactive, task-driven systems and brute-force long context towards a broader paradigm of supersensing. We frame spatial supersensing as four stages beyond linguistic-only understanding: semantic perception (naming what is seen), streaming event cognition (maintaining memory across continuous experiences), implicit 3D spatial cognition (inferring the world behind pixels), and predictive world modeling (creating internal models that filter and organize information). Current benchmarks largely test only the early stages, offering narrow coverage of spatial cognition and rarely challenging models in ways that require true world modeling. To drive progress in spatial supersensing, we present VSI-SUPER, a two-part benchmark: VSR (long-horizon visual spatial recall) and VSC (continual visual spatial counting). These tasks require arbitrarily long video inputs yet are resistant to brute-force context expansion. We then test data scaling limits by curating VSI-590K and training Cambrian-S, achieving +30% absolute improvement on VSI-Bench without sacrificing general capabilities. Yet performance on VSI-SUPER remains limited, indicating that scale alone is insufficient for spatial supersensing. We propose predictive sensing as a path forward, presenting a proof-of-concept in which a self-supervised next-latent-frame predictor leverages surprise (prediction error) to drive memory and event segmentation. On VSI-SUPER, this approach substantially outperforms leading proprietary baselines, showing that spatial supersensing requires models that not only see but also anticipate, select, and organize experience.

cs.CV↗

arXiv2025

Origin of sublattice particle-hole asymmetry in monolayer FeSe superconductors

In iron-based superconductors, the two Fe atoms in the unit cell are typically related by crystal symmetries; therefore, we expect no intra-unit cell variations in the superconducting gap. However, recent experiments have challenged this expectation, reporting intra-unit cell variations in the gap with an unusual particle-hole asymmetry. Here, we examine the origin of this asymmetry between the two Fe sublattices in monolayer FeSe grown on SrTiO$_3$. We reveal that, in addition to the substrate-induced broken inversion symmetry, substrate nematic symmetry breaking is key to observing this asymmetry. We further identify two possible mechanisms through which this can occur. The first is through an odd-parity gap function that coexists with an extended $s$-wave function. The second is via a nodeless $d$-wave gap function that develops in the presence of a symmetry-breaking substrate. We argue that the latter mechanism is more physical. To test our theory, we performed scanning tunneling spectroscopy measurements across the nematic domain walls, which exhibit a clear enhancement of the asymmetry between the two Fe sublattices. In addition, we reveal that the observed sublattice particle-hole asymmetry is associated with odd-frequency pairing correlations, providing an experimental realization of this unusual pairing correlation.

cond-mat.supr-con↗

arXiv2025

HoloCine: Holistic Generation of Cinematic Multi-Shot Long Video Narratives

State-of-the-art text-to-video models excel at generating isolated clips but fall short of creating the coherent, multi-shot narratives, which are the essence of storytelling. We bridge this "narrative gap" with HoloCine, a model that generates entire scenes holistically to ensure global consistency from the first shot to the last. Our architecture achieves precise directorial control through a Window Cross-Attention mechanism that localizes text prompts to specific shots, while a Sparse Inter-Shot Self-Attention pattern (dense within shots but sparse between them) ensures the efficiency required for minute-scale generation. Beyond setting a new state-of-the-art in narrative coherence, HoloCine develops remarkable emergent abilities: a persistent memory for characters and scenes, and an intuitive grasp of cinematic techniques. Our work marks a pivotal shift from clip synthesis towards automated filmmaking, making end-to-end cinematic creation a tangible future. Our code is available at: https://holo-cine.github.io/.

cs.CV↗

arXiv2025

Universal Reconstruction of Complex Magnetic Profiles with Minimum Prior Assumptions

Understanding intricate magnetic structures in materials is essential for advancing materials science, spintronics, and geology. Recent developments of quantum-enabled magnetometers, such as nitrogen-vacancy (NV) centers in diamond, have enabled direct imaging of magnetic field distributions across a wide range of magnetic profiles. However, reconstructing the magnetization from an experimentally measured magnetic field map is a complex inverse problem, further complicated by measurement noise, finite spatial resolution, and variations in sample-to-sensor distance. In this work, we present a novel and efficient GPU-accelerated method for reconstructing spatially varying magnetization density from measured magnetic fields with minimal prior assumptions. We validate our method by simulating diverse magnetic structures under realistic experimental conditions, including multi-domain ferromagnetism and magnetic spin textures such as skyrmion, anti-skyrmion, and meron. Experimentally, we reconstruct the magnetization of a micrometer-scale Apollo lunar mare basalt (sample 10003,184) and a nanometer-scale twisted double-trilayer CrI3. The basalt exhibits soft ferromagnetic domains consistent with previous paleomagnetic studies, whereas the CrI3 system reveals a well-defined hexagonal magnetic Moire superlattice. Our approach provides a versatile and universal tool for investigating complex magnetization profiles, paving the way for future quantum sensing experiments.

cond-mat.mes-hall↗

arXiv2025

Emergent continuous symmetry and ground-state factorization induced by long-range interactions

The spontaneous breaking of a $Z_2$ symmetry typically gives rise to emergent excitations possessing the same symmetry with a renormalized mass. Contrary to this conventional wisdom, we present a theory in which the low-lying excitation in the broken-symmetry phase acquires a continuous symmetry, even when the underlying symmetry of the system is discrete. In the presence of anisotropic long-range interactions, the order parameter renormalizes the relative strength of the particle-conserving and particle-nonconserving interactions. When one of the two renormalized interactions vanishes, a conservation law absent in the original Hamiltonian emerges, giving rise to a continuous symmetry. A striking consequence of the emergent continuous symmetry and conservation law is that it constrains quantum correlations in the ground-state to be zero, leading to the ground-state factorization in the presence of strong interactions. Our finding is a universal feature of quantum phase transitions in fully-connected systems and in their lattice generalizations; therefore, it can be observed in a wide range of physical systems.

quant-ph↗

arXiv2025

When Retrieval Succeeds and Fails: Rethinking Retrieval-Augmented Generation for LLMs

Large Language Models (LLMs) have enabled a wide range of applications through their powerful capabilities in language understanding and generation. However, as LLMs are trained on static corpora, they face difficulties in addressing rapidly evolving information or domain-specific queries. Retrieval-Augmented Generation (RAG) was developed to overcome this limitation by integrating LLMs with external retrieval mechanisms, allowing them to access up-to-date and contextually relevant knowledge. However, as LLMs themselves continue to advance in scale and capability, the relative advantages of traditional RAG frameworks have become less pronounced and necessary. Here, we present a comprehensive review of RAG, beginning with its overarching objectives and core components. We then analyze the key challenges within RAG, highlighting critical weakness that may limit its effectiveness. Finally, we showcase applications where LLMs alone perform inadequately, but where RAG, when combined with LLMs, can substantially enhance their effectiveness. We hope this work will encourage researchers to reconsider the role of RAG and inspire the development of next-generation RAG systems.

cs.CL↗

arXiv2025

MedAgentGym: A Scalable Agentic Training Environment for Code-Centric Reasoning in Biomedical Data Science

We introduce MedAgentGym, a scalable and interactive training environment designed to enhance coding-based biomedical reasoning capabilities in large language model (LLM) agents. MedAgentGym comprises 72,413 task instances across 129 categories derived from 12 authentic real-world biomedical scenarios. Tasks are encapsulated within executable sandbox environments, each featuring detailed task specifications, interactive feedback mechanisms, verifiable ground truth annotations, and scalable training trajectory generation. Extensive benchmarking of 29 LLMs reveals substantial performance disparities in biomedical data science between commercial and open-source LLMs. Leveraging efficient multi-threaded and multi-turn trajectory sampling in MedAgentGym, Med-Copilot achieves performance gains of +43.02% and +45.28% from offline and online reinforcement learning, respectively, demonstrating MedAgentGym as an effective training ground while establishing itself as a cost-effective, privacy-preserving alternative competitive with proprietary LLMs (gpt-4o). By offering a unified execution environment with a comprehensive benchmark and accessible, extensible training resources, MedAgentGym delivers an integrated platform to develop LLM-based coding assistants for advanced biomedical data science.

cs.CL↗

arXiv2025

AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforced Self-Play

Search-augmented LLMs often struggle with complex reasoning tasks due to ineffective multi-hop retrieval and limited reasoning ability. We propose AceSearcher, a cooperative self-play framework that trains a single large language model (LLM) to alternate between two roles: a decomposer that breaks down complex queries and a solver that integrates retrieved contexts for answer generation. AceSearcher couples supervised fine-tuning on a diverse mixture of search, reasoning, and decomposition tasks with reinforcement fine-tuning optimized for final answer accuracy, eliminating the need for intermediate annotations. Extensive experiments on three reasoning-intensive tasks across 10 datasets show that AceSearcher outperforms state-of-the-art baselines, achieving an average exact match improvement of 7.6%. Remarkably, on document-level finance reasoning tasks, AceSearcher-32B matches the performance of the DeepSeek-V3 model using less than 5% of its parameters. Even at smaller scales (1.5B and 8B), AceSearcher often surpasses existing search-augmented LLMs with up to 9x more parameters, highlighting its exceptional efficiency and effectiveness in tackling complex reasoning tasks. Our code will be published at https://github.com/ritaranx/AceSearcher and https://huggingface.co/AceSearcher.

cs.CL↗

arXiv2025

VMDiff: Visual Mixing Diffusion for Limitless Cross-Object Synthesis

Creating novel images by fusing visual cues from multiple sources is a fundamental yet underexplored problem in image-to-image generation, with broad applications in artistic creation, virtual reality and visual media. Existing methods often face two key challenges: coexistent generation, where multiple objects are simply juxtaposed without true integration, and bias generation, where one object dominates the output due to semantic imbalance. To address these issues, we propose Visual Mixing Diffusion (VMDiff), a simple yet effective diffusion-based framework that synthesizes a single, coherent object by integrating two input images at both noise and latent levels. Our approach comprises: (1) a hybrid sampling process that combines guided denoising, inversion, and spherical interpolation with adjustable parameters to achieve structure-aware fusion, mitigating coexistent generation; and (2) an efficient adaptive adjustment module, which introduces a novel similarity-based score to automatically and adaptively search for optimal parameters, countering semantic bias. Experiments on a curated benchmark of 780 concept pairs demonstrate that our method outperforms strong baselines in visual quality, semantic consistency, and human-rated creativity.

cs.CV↗

arXiv2025

Schrodingerization based quantum algorithms for the time-fractional heat equation

We develop a quantum algorithm for solving high-dimensional time-fractional heat equations. By applying the dimension extension technique from [FKW23], the $d+1$-dimensional time-fractional equation is reformulated as a local partial differential equation in $d+2$ dimensions. Through discretization along both the extended and spatial domains, a stable system of ordinary differential equations is obtained by a simple change of variables. We propose a quantum algorithm for the resulting semi-discrete problem using the Schrodingerization approach from [JLY24a,JLY23,JL24a]. The Schrodingerization technique transforms general linear partial and ordinary differential equations into Schrodinger-type systems--with unitary evolution, making them suitable for quantum simulation. This is accomplished via the warped phase transformation, which maps the equation into a higher-dimensional space. We provide detailed implementations of this method and conduct a comprehensive complexity analysis, demonstrating up to exponential advantage--with respect to the inverse of the mesh size in high dimensions~--~compared to its classical counterparts. Specifically, to compute the solution to time $T$, while the classical method requires at least $\mathcal{O}(N_t d h^{-(d+0.5)})$ matrix-vector multiplications, where $N_t $ is the number of time steps (which is, for example, $\mathcal{O}(Tdh^{-2})$ for the forward Euler method), our quantum algorithms requires $\widetilde{\mathcal{O}}(T^2d^4 h^{-8})$ queries to the block-encoding input models, with the quantum complexity being independent of the dimension $d$ in terms of the inverse mesh size $h^{-1}$. Numerical experiments are performed to validate our formulation.

math.NA↗

arXiv2025

ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Data

Vision-Language Models (VLMs) have enabled computer use agents (CUAs) that operate GUIs autonomously, showing great potential, yet progress is limited by the lack of large-scale, open-source computer use data and foundation models. In this work, we introduce ScaleCUA, a step toward scaling open-source CUAs. It offers a large-scale dataset spanning 6 operating systems and 3 task domains, built via a closed-loop pipeline uniting automated agents with human experts. Trained on this scaled-up data, ScaleCUA can operate seamlessly across platforms. Specifically, it delivers strong gains over baselines (+26.6 on WebArena-Lite-v2, +10.7 on ScreenSpot-Pro) and sets new state-of-the-art results (94.4% on MMBench-GUI L1-Hard, 60.6% on OSWorld-G, 47.4% on WebArena-Lite-v2). These findings underscore the power of data-driven scaling for general-purpose computer use agents. We will release data, models, and code to advance future research: https://github.com/OpenGVLab/ScaleCUA.

cs.CV↗

arXiv2025

Actively Learning to Coordinate in Convex Games via Approximate Correlated Equilibrium

Correlated equilibrium generalizes Nash equilibrium by allowing a central coordinator to guide players' actions through shared recommendations, similar to how routing apps guide drivers. We investigate how a coordinator can learn a correlated equilibrium in convex games where each player minimizes a convex cost function that depends on other players' actions, subject to convex constraints without knowledge of the players' cost functions. We propose a learning framework that learns an approximate correlated equilibrium by actively querying players' regrets, \emph{i.e.}, the cost saved by deviating from the coordinator's recommendations. We first show that a correlated equilibrium in convex games corresponds to a joint action distribution over an infinite joint action space that minimizes all players' regrets. To make the learning problem tractable, we introduce a heuristic that selects finitely many representative joint actions by maximizing their pairwise differences. We then apply Bayesian optimization to learn a probability distribution over the selected joint actions by querying all players' regrets. The learned distribution approximates a correlated equilibrium by minimizing players' regrets. We demonstrate the proposed approach via numerical experiments on multi-user traffic assignment games in a shared transportation network.

cs.GT↗

arXiv2025

ReBaNO: Reduced Basis Neural Operator Mitigating Generalization Gaps and Achieving Discretization Invariance

We propose a novel data-lean operator learning algorithm, the Reduced Basis Neural Operator (ReBaNO), to solve a group of PDEs with multiple distinct inputs. Inspired by the Reduced Basis Method and the recently introduced Generative Pre-Trained Physics-Informed Neural Networks, ReBaNO relies on a mathematically rigorous greedy algorithm to build its network structure offline adaptively from the ground up. Knowledge distillation via task-specific activation function allows ReBaNO to have a compact architecture requiring minimal computational cost online while embedding physics. In comparison to state-of-the-art operator learning algorithms such as PCA-Net, DeepONet, FNO, and CNO, numerical results demonstrate that ReBaNO significantly outperforms them in terms of eliminating/shrinking the generalization gap for both in- and out-of-distribution tests and being the only operator learning algorithm achieving strict discretization invariance.

cs.LG↗

arXiv2025

Correlated helimagnetic configuration in a nonsymmorphic magnetic nodal semimetal

Nonsymmorphic magnetic Weyl semimetal materials such as ReAlX (Re=rare earth, X=Si/Ge) provide a unique opportunity to explore the correlated phenomena between Weyl fermions and nontrivial magnetic configurations. To be specific, we study a lattice model in which the magnetic configuration is determined by the competition among ferromagnetic (FM) interaction, the Dzyaloshinskii-Moriya interaction, and the Kondo coupling $K_0$ to the Weyl fermion. Both quantum and finite-temperature phase transitions between FM and correlated nesting helical configurations are found. Different from the uncorrelated helimagnet that decouples from the Weyl fermions, this correlated helimagnet induces a magnetic Brillouin zone with a $K_0$-dependent nesting in the band structure of the conduction electrons instead of the monopole-like Weyl cone. By measuring the current induced by the chiral magnetic effect on the conduction electron with nesting Weyl nodes, one can distinguish the correlated nesting helical order from the ferromagnetism because the chiral magnetic effect is considerably suppressed in the former case. These properties we find here may explain the experimental observations in ReAlX.

cond-mat.mes-hall↗

arXiv2025

4D Visual Pre-training for Robot Learning

General visual representations learned from web-scale datasets for robotics have achieved great success in recent years, enabling data-efficient robot learning on manipulation tasks; yet these pre-trained representations are mostly on 2D images, neglecting the inherent 3D nature of the world. However, due to the scarcity of large-scale 3D data, it is still hard to extract a universal 3D representation from web datasets. Instead, we are seeking a general visual pre-training framework that could improve all 3D representations as an alternative. Our framework, called FVP, is a novel 4D Visual Pre-training framework for real-world robot learning. FVP frames the visual pre-training objective as a next-point-cloud-prediction problem, models the prediction model as a diffusion model, and pre-trains the model on the larger public datasets directly. Across twelve real-world manipulation tasks, FVP boosts the average success rate of 3D Diffusion Policy (DP3) for these tasks by 28%. The FVP pre-trained DP3 achieves state-of-the-art performance across imitation learning methods. Moreover, the efficacy of FVP adapts across various point cloud encoders and datasets. Finally, we apply FVP to the RDT-1B, a larger Vision-Language-Action robotic model, enhancing its performance on various robot tasks. Our project page is available at: https://4d-visual-pretraining.github.io/

cs.CV↗

arXiv2025

Odd-parity Magnetism Driven by Antiferromagnetic Exchange

Realizing odd-parity, time-reversal-preserving, non-relativistic spin splitting is a central goal for spintronics applications. We propose a group-theory-based microscopic framework to induce odd-parity spin splitting from coplanar antiferromagnetic (AFM) states without spin-orbit coupling (SOC). We develop phenomenological models for 421 conventional period-doubling AFM systems in non-symmorphic space groups and construct minimal microscopic models for 119 of these. We find that these AFM states can attain three possible competing ground states. These ground states all break symmetries in addition to those broken by the usual AFM order. Specifically, they give rise to either odd-parity spin-splitting, nematic order, or scalar odd-parity order related to multiferroicity. Our microscopic theories reveal that the odd-parity spin-splitting energy scale is generically large and further reveal that the scalar odd-parity order gives a non-zero Berry curvature dipole without SOC. We identify 67 materials in the Magndata database for which our theory applies. We provide DFT calculations on FeSe that reveal an $h$-wave spin splitting consistent with our symmetry arguments and apply our microscopic model to determine the non-relativistic Edelstein response for CeNiAsO.

cond-mat.str-el↗

arXiv2025

A Machine Learning Approach to Volumetric Computations of Solid Pulmonary Nodules

Early detection of lung cancer is crucial for effective treatment and relies on accurate volumetric assessment of pulmonary nodules in CT scans. Traditional methods, such as consolidation-to-tumor ratio (CTR) and spherical approximation, are limited by inconsistent estimates due to variability in nodule shape and density. We propose an advanced framework that combines a multi-scale 3D convolutional neural network (CNN) with subtype-specific bias correction for precise volume estimation. The model was trained and evaluated on a dataset of 364 cases from Shanghai Chest Hospital. Our approach achieved a mean absolute deviation of 8.0 percent compared to manual nonlinear regression, with inference times under 20 seconds per scan. This method outperforms existing deep learning and semi-automated pipelines, which typically have errors of 25 to 30 percent and require over 60 seconds for processing. Our results show a reduction in error by over 17 percentage points and a threefold acceleration in processing speed. These advancements offer a highly accurate, efficient, and scalable tool for clinical lung nodule screening and monitoring, with promising potential for improving early lung cancer detection.

eess.IV↗

arXiv2025

The Root Finding Problem Revisited: Beyond the Robbins-Monro procedure

We introduce Sequential Probability Ratio Bisection (SPRB), a novel stochastic approximation algorithm that adapts to the local behavior of the (regression) function of interest around its root. We establish theoretical guarantees for SPRB's asymptotic performance, showing that it achieves the optimal convergence rate and minimal asymptotic variance even when the target function's derivative at the root is small (at most half the step size), a regime where the classical Robbins-Monro procedure typically suffers reduced convergence rates. Further, we show that if the regression function is discontinuous at the root, Robbins-Monro converges at a rate of $1/n$ whilst SPRB attains exponential convergence. If the regression function has vanishing first-order derivative, SPRB attains a faster rate of convergence compared to stochastic approximation. As part of our analysis, we derive a nonasymptotic bound on the expected sample size and establish a generalized Central Limit Theorem under random stopping times. Remarkably, SPRB automatically provides nonasymptotic time-uniform confidence sequences that do not explicitly require knowledge of the convergence rate. We demonstrate the practical effectiveness of SPRB through simulation results.

math.ST↗