arXiv subjects

Tao Chen

Publications and source records attributed to Tao Chen.

At least 163 records · Page 9Linked to original sources

arXiv2025

CoTune: Co-evolutionary Configuration Tuning

To automatically tune configurations for the best possible system performance (e.g., runtime or throughput), much work has been focused on designing intelligent heuristics in a tuner. However, existing tuner designs have mostly ignored the presence of complex performance requirements (e.g., the latency shall ideally be 2 seconds), but simply assume that better performance is always more preferred. This would not only waste valuable information in a requirement but might also consume extensive resources to tune for a goal with little gain. Yet, prior studies have shown that simply incorporating the requirement as a tuning objective is problematic since the requirement might be too strict, harming convergence; or its highly diverse satisfactions might lead to premature convergence. In this paper, we propose CoTune, a tool that takes the information of a given target performance requirement into account through co-evolution. CoTune is unique in the sense that it creates an auxiliary performance requirement to be co-evolved with the configurations, which assists the target performance requirement when it becomes ineffective or even misleading, hence allowing the tuning to be guided by the requirement while being robust to its harm. Experiment results on 162 cases (nine systems and 18 requirements) reveal that CoTune considerably outperforms existing tuners, ranking as the best for 90% cases (against the 0%--35% for other tuners) with up to 2.9x overall improvements, while doing so under a much better efficiency.

cs.SE↗

arXiv2025

CTTS: Collective Test-Time Scaling

Test-time scaling (TTS) has emerged as a promising, training-free approach for enhancing large language model (LLM) performance. However, the efficacy of existing methods, such as Best-of-N and Self-Consistency, is fundamentally constrained by the dominant single test-time scaling (STTS) paradigm, which relies on a single LLM agent interacting with a single reward model (SA-SR). Inspired by recent work showing that collective methods can surpass the performance ceiling of individual models, we introduce Collective Test-Time Scaling (CTTS). First, we systematically investigate three primary interaction paradigms of existing multiple models: single-agent-multi-reward (SA-MR), multi-agent-single-reward (MA-SR), and multi-agent-multi-reward (MA-MR). Extensive experiments reveal that the MA-MR paradigm is consistently superior. Based on this finding, we further propose CTTS-MM, a novel framework that operationalizes multi-agent and multi-reward collaboration. CTTS-MM integrates two key technical contributions: (1) for agent collaboration, an Agent Collaboration Search (ACS) that identifies the most effective combination of LLMs from a candidate pool; and (2) for reward model collaboration, a Mixture of Reward Models (MoR) strategy that leverages a Prior Reward model Ensemble Selection (PRES) algorithm to select the optimal ensemble. Evaluations across seven mainstream benchmarks demonstrate that CTTS-MM significantly outperforms leading STTS methods (+4.82% over Best-of-N) and surpasses even flagship proprietary LLMs (+7.06% over GPT-4.1) and open-source LLMs. These results highlight the substantial potential of collective scaling to push the frontier of LLM inference. Code will be released at https://github.com/magent4aci/CTTS-MM.

cs.CL↗

arXiv2025

Tensor Network Markov Chain Monte Carlo: Efficient Sampling of Three-Dimensional Spin Glasses and Beyond

Sampling the three-dimensional (3D) spin glass -- i.e., generating equilibrium configurations of a 3D lattice with quenched random couplings -- is widely regarded as one of the central and long-standing open problems in statistical physics. The rugged energy landscape, pronounced critical slowing down, and intrinsic ergodicity breaking render standard Monte Carlo methods severely inefficient, particularly for large systems at low temperatures. In this work, we introduce the Tensor Network Markov Chain Monte Carlo (TNMCMC) approach to address the issue. It generates large-scale collective updates in MCMC using tensor networks on the 2D slices of the 3D lattice, greatly improving the autocorrelation time and offering orders-of-magnitude speed-ups over conventional MCMC in generating unbiased samples of the Boltzmann distribution. We conduct numerical experiments on 3D spin glasses up to system size $64\times 64\times 64$ using a single CPU, and show that TNMCMC dramatically suppresses critical slowing down in large disordered systems, which usually require a supercomputer to perform MCMC simulations. Furthermore, we apply our approach to the 3-state Potts model up to system size $64\times 64\times 64$ using a single CPU, and show that the TNMCMC approach efficiently traverses the exponential barriers of the strong first-order transition, whereas conventional MCMC fails. Our results reveal that TNMCMC opens a promising path toward tackling long-standing, formidable three-dimensional problems in statistical physics.

cond-mat.stat-mech↗

arXiv2025

Biology-Instructions: A Dataset and Benchmark for Multi-Omics Sequence Understanding Capability of Large Language Models

Large language models (LLMs) have shown remarkable capabilities in general domains, but their application to multi-omics biology remains underexplored. To address this gap, we introduce Biology-Instructions, the first large-scale instruction-tuning dataset for multi-omics biological sequences, including DNA, RNA, proteins, and multi-molecules. This dataset bridges LLMs and complex biological sequence-related tasks, enhancing their versatility and reasoning while maintaining conversational fluency. We also highlight significant limitations of current state-of-the-art LLMs on multi-omics tasks without specialized training. To overcome this, we propose ChatMultiOmics, a strong baseline with a novel three-stage training pipeline, demonstrating superior biological understanding through Biology-Instructions. Both resources are publicly available, paving the way for better integration of LLMs in multi-omics analysis. The Biology-Instructions is publicly available at: https://github.com/hhnqqq/Biology-Instructions.

q-bio.BM↗

arXiv2025

BatchTNMC: Efficient sampling of two-dimensional spin glasses using tensor network Monte Carlo

Efficient sampling of two-dimensional statistical physics systems remains a central challenge in computational statistical physics. Traditional Markov chain Monte Carlo (MCMC) methods, including cluster algorithms, provide only partial solutions, as their efficiency collapses for large systems in the presence of frustration and quenched disorder. The recently proposed Tensor Network Monte Carlo (TNMC) method offers a promising alternative, yet its original implementation suffers from inefficiencies due to the lack of scalable parallel sampling. In this work, we introduce BatchTNMC, a GPU-optimized and parallelized implementation of TNMC tailored for large-scale simulations of two-dimensional spin glasses. By leveraging batch processing and parallel sampling across multiple disorder realizations, our implementation achieves speedups of up to five orders of magnitude compared with the original serial scheme. Benchmarking on two-dimensional spin glasses demonstrates dramatic gains in efficiency: for instance, on a single GPU, BatchTNMC concurrently produces 1000 uncorrelated and unbiased samples across 1000 disorder realizations on $1024\times 1024$ lattices in just 3.3 hours, with an acceptance probability of 37%. These results establish BatchTNMC as a scalable and powerful computational framework for the study of two-dimensional disordered spin glass systems.

cond-mat.stat-mech↗

arXiv2025

HiPhO: How Far Are (M)LLMs from Humans in the Latest High School Physics Olympiad Benchmark?

Recently, the physical capabilities of (M)LLMs have garnered increasing attention. However, existing benchmarks for physics suffer from two major gaps: they neither provide systematic and up-to-date coverage of real-world physics competitions such as physics Olympiads, nor enable direct performance comparison with humans. To bridge these gaps, we present HiPhO, the first benchmark dedicated to high school physics Olympiads with human-aligned evaluation. Specifically, HiPhO highlights three key innovations. (1) Comprehensive Data: It compiles 13 latest Olympiad exams from 2024-2025, spanning both international and regional competitions, and covering mixed modalities that encompass problems spanning text-only to diagram-based. (2) Professional Evaluation: We adopt official marking schemes to perform fine-grained grading at both the answer and step level, fully aligned with human examiners to ensure high-quality and domain-specific evaluation. (3) Comparison with Human Contestants: We assign gold, silver, and bronze medals to models based on official medal thresholds, thereby enabling direct comparison between (M)LLMs and human contestants. Our large-scale evaluation of 30 state-of-the-art (M)LLMs shows that: across 13 exams, open-source MLLMs mostly remain at or below the bronze level; open-source LLMs show promising progress with multiple golds; closed-source reasoning MLLMs can achieve 6 to 12 gold medals; and most models still have a significant gap from full marks. These results highlight the performance gap between open-source models and top students, the strong reasoning abilities of closed-source models, and the remaining room for improvement. HiPhO, a human-aligned Olympiad benchmark for multimodal physical reasoning, is open-source at https://github.com/SciYu/HiPhO with a public leaderboard at https://phyarena.github.io/.

cs.AI↗

arXiv2025

Sequential Token Merging: Revisiting Hidden States

Vision Mambas (ViMs) achieve remarkable success with sub-quadratic complexity, but their efficiency remains constrained by quadratic token scaling with image resolution. While existing methods address token redundancy, they overlook ViMs' intrinsic Limited Directional Sequential Dependence (LDSD) - a critical information flow mechanism revealed in our analysis. We further identify Mamba's selective scan enables gradual information aggregation in hidden states. Based on these insights, we propose Sequential Token Merging (STM), featuring: 1) Bidirectional nearest neighbor merging to preserve sequential dependencies through symmetric spatial aggregation, and 2) Hidden states protection to stabilize the hidden states around the class token. STM strategically leverages Mamba's layer-wise loss convergence to convert temporal forgetfulness into stability. Experiments demonstrate STM's superiority: 1.0% accuracy drop for ViM-Ti at 20% token reduction, and only 1.4% degradation for ViM-S at 40% reduction. Our method achieves state-of-the-art efficiency with minimal complexity, while providing new insights into state-space model dynamics. Codes will be released soon.

cs.CV↗

arXiv2025

Dynamically Optimized Super-Robust Nonadiabatic Holonomic Quantum Gates Based on Superconducting Circuits

Nonadiabatic holonomic quantum computation (NHQC) leverages non-Abelian geometric phases within a nonadiabatic framework to achieve fast and robust quantum gate operations. However, the practical implementation of NHQC is challenged by the imperfect control inherent in experimental environments. Ensuring deep suppression of control error is critical. In this work, we propose a dynamically optimized NHQC scheme to construct universal super-robust holonomic quantum gates. The proposed scheme is implemented by strategically designing a set of dynamically correcting pulses to achieve cyclic evolution, while ensuring that unwanted and disruptive dynamical phase elements, including previously neglected cross-coupling terms, are not accumulated. This constructed super-robust NHQC scheme efficiently safeguards the cyclic evolution process and makes the holonomic gate immune to control error by effectively correcting the error up to the fourth order. Furthermore, when integrated with decoherence-free subspace (DFS) encoding in superconducting quantum circuits, our scheme can achieve high-fidelity holonomic gates, and demonstrate robust resilience against both control errors and collective dephasing errors. Consequently, our work offers a promising pathway toward the realization of scalable and fault-tolerant holonomic quantum computation.

quant-ph↗

arXiv2025

QualityFM: a Multimodal Physiological Signal Foundation Model with Self-Distillation for Signal Quality Challenges in Critically Ill Patients

Photoplethysmogram (PPG) and electrocardiogram (ECG) are commonly recorded in intesive care unit (ICU) and operating room (OR). However, the high incidence of poor, incomplete, and inconsistent signal quality, can lead to false alarms or diagnostic inaccuracies. The methods explored so far suffer from limited generalizability, reliance on extensive labeled data, and poor cross-task transferability. To overcome these challenges, we introduce QualityFM, a novel multimodal foundation model for these physiological signals, designed to acquire a general-purpose understanding of signal quality. Our model is pre-trained on an large-scale dataset comprising over 21 million 30-second waveforms and 179,757 hours of data. Our approach involves a dual-track architecture that processes paired physiological signals of differing quality, leveraging a self-distillation strategy where an encoder for high-quality signals is used to guide the training of an encoder for low-quality signals. To efficiently handle long sequential signals and capture essential local quasi-periodic patterns, we integrate a windowed sparse attention mechanism within our Transformer-based model. Furthermore, a composite loss function, which combines direct distillation loss on encoder outputs with indirect reconstruction loss based on power and phase spectra, ensures the preservation of frequency-domain characteristics of the signals. We pre-train three models with varying parameter counts (9.6 M to 319 M) and demonstrate their efficacy and practical value through transfer learning on three distinct clinical tasks: false alarm of ventricular tachycardia detection, the identification of atrial fibrillation and the estimation of arterial blood pressure (ABP) from PPG and ECG signals.

cs.LG↗

arXiv2025

Collective dissipation engineering of interacting Rydberg atoms

Engineered dissipation is emerging as an alternative tool for quantum state control, enabling high-fidelity preparation, transfer and stabilization, and access to novel phase transitions. We realize a tunable, state-resolved laser-induced loss channel for individual Rydberg atoms, in both non-interacting and strongly correlated settings. This capability allows us to reveal interaction-driven shifts of the exceptional point separating quantum Zeno and anti-Zeno regimes, and to demonstrate interaction-enhanced decay. By exploiting interaction-dependent energy level shifts, we observe a configuration-selective two-body Zeno effect that freezes target spin states. We theoretically show that when this mechanism is extended to many-body chains it allows for the dissipative distillation of unwanted spin configurations. These experimental studies establish a versatile approach for exploring strongly interacting, open quantum spin systems, and opens possible new routines for dissipative preparation of correlated quantum states in Rydberg atom arrays.

quant-ph↗

arXiv2025

PromiseTune: Unveiling Causally Promising and Explainable Configuration Tuning

The high configurability of modern software systems has made configuration tuning a crucial step for assuring system performance, e.g., latency or throughput. However, given the expensive measurements, large configuration space, and rugged configuration landscape, existing tuners suffer ineffectiveness due to the difficult balance of budget utilization between exploring uncertain regions (for escaping from local optima) and exploiting guidance of known good configurations (for fast convergence). The root cause is that we lack knowledge of where the promising regions lay, which also causes challenges in the explainability of the results. In this paper, we propose PromiseTune that tunes configuration guided by causally purified rules. PromiseTune is unique in the sense that we learn rules, which reflect certain regions in the configuration landscape, and purify them with causal inference. The remaining rules serve as approximated reflections of the promising regions, bounding the tuning to emphasize these places in the landscape. This, as we demonstrate, can effectively mitigate the impact of the exploration and exploitation trade-off. Those purified regions can then be paired with the measured configurations to provide spatial explainability at the landscape level. Comparing with 11 state-of-the-art tuners on 12 systems and varying budgets, we show that PromiseTune performs significantly better than the others with 42% superior rank to the overall second best while providing richer information to explain the hidden system characteristics.

cs.SE↗

arXiv2025

A general kinematic theory of fluid-element rotation and intrinsic vorticity decompositions

The present study proposes a general kinematic theory for fluid-element rotation and intrinsic vorticity decompositions within the context of vorticity and vortex dynamics. Both the angular velocities of material line and surface elements comprise a classical contribution driven by volume-element rotation (equal to half the local vorticity), and a strain-rate-induced specific angular velocity. Then, two direction-dependent vorticity decompositions (DVDs) are constructed, revealing the rigid rotation and spin modes of vorticity. We derive intrinsic coupling relations for orthogonal line-surface element pair, elucidating their complementary kinematic and geometric roles. Notably, we rigorously prove that the spin mode (in the surface-element-based DVD) is identical to the relative vorticity in the generalized Caswell formula, thereby faithfully accounting for surface shear stress in Newtonian fluids. Next, within a field-theoretic framework, vorticity decompositions are proposed based on streamline and streamsurface using differential geometry. The physical roles of the six rotational invariants in the characteristic algebraic description (including the the normal-nilpotent decomposition (NND) of the velocity gradient tensor (VGT) and the resulting invariant vorticity decomposition (IVD)) are clarified through unified analysis with the DVD, Caswell formula, and Helmholtz-Hodge decomposition. ... It is found that physical admissible DVD vorticity modes must be bounded by IVD modes in phase space, whereas enforcing a minimization principle naturally yields the Liutex formula. Finally, the effectiveness of theory is validated across diverse flows, from simple to complex. Results show that a coupled IVD-DVD analysis could enhance physical understanding of complex vortical flows under both algebraic and field-theoretic frameworks.

physics.flu-dyn↗

arXiv2025

Dually Hierarchical Drift Adaptation for Online Configuration Performance Learning

Modern configurable software systems need to learn models that correlate configuration and performance. However, when the system operates in dynamic environments, the workload variations, hardware changes, and system updates will inevitably introduce concept drifts at different levels - global drifts, which reshape the performance landscape of the entire configuration space; and local drifts, which only affect certain sub-regions of that space. As such, existing offline and transfer learning approaches can struggle to adapt to these implicit and unpredictable changes in real-time, rendering configuration performance learning challenging. To address this, we propose DHDA, an online configuration performance learning framework designed to capture and adapt to these drifts at different levels. The key idea is that DHDA adapts to both the local and global drifts using dually hierarchical adaptation: at the upper level, we redivide the data into different divisions, within each of which the local model is retrained, to handle global drifts only when necessary. At the lower level, the local models of the divisions can detect local drifts and adapt themselves asynchronously. To balance responsiveness and efficiency, DHDA combines incremental updates with periodic full retraining to minimize redundant computation when no drifts are detected. Through evaluating eight software systems and against state-of-the-art approaches, we show that DHDA achieves considerably better accuracy and can effectively adapt to drifts with up to 2x improvements, while incurring reasonable overhead and is able to improve different local models in handling concept drift.

cs.SE↗

arXiv2025

Wisdom of the Crowd: Reinforcement Learning from Coevolutionary Collective Feedback

Reinforcement learning (RL) has significantly enhanced the reasoning capabilities of large language models (LLMs), but its reliance on expensive human-labeled data or complex reward models severely limits scalability. While existing self-feedback methods aim to address this problem, they are constrained by the capabilities of a single model, which can lead to overconfidence in incorrect answers, reward hacking, and even training collapse. To this end, we propose Reinforcement Learning from Coevolutionary Collective Feedback (RLCCF), a novel RL framework that enables multi-model collaborative evolution without external supervision. Specifically, RLCCF optimizes the ability of a model collective by maximizing its Collective Consistency (CC), which jointly trains a diverse ensemble of LLMs and provides reward signals by voting on collective outputs. Moreover, each model's vote is weighted by its Self-Consistency (SC) score, ensuring that more confident models contribute more to the collective decision. Benefiting from the diverse output distributions and complementary abilities of multiple LLMs, RLCCF enables the model collective to continuously enhance its reasoning ability through coevolution. Experiments on four mainstream open-source LLMs across four mathematical reasoning benchmarks demonstrate that our framework yields significant performance gains, achieving an average relative improvement of 16.72\% in accuracy. Notably, RLCCF not only improves the performance of individual models but also enhances the group's majority-voting accuracy by 4.51\%, demonstrating its ability to extend the collective capability boundary of the model collective.

cs.AI↗

arXiv2025

Robustness Enhancement of Universal Noncyclic Geometric Gates via Evolution Optimization

Noncyclic geometric gates aim to overcome the stringent constraints of conventional cyclic conditions and enhance the flexibility in evolution choice. Conceptually, they can also avoid the error problems arising from the violation of cyclicity, thus holding significance for improving the fault tolerance of quantum gates. However, current research on noncyclic geometric gates lacks a comprehensive exploration of their flexibility in evolution choice and validation of their effectiveness in resilience against multiple error sources present in practical quantum systems. In this paper, we systematically evaluate all noncyclic evolution conditions, elucidate their corresponding potential geometric trajectories, and identify optimal conditions for enhancing gate robustness by quantifying the resilience of constructed noncyclic geometric gates against universal systematic errors and residual crosstalk error. The optimized gates demonstrate significant robustness advantages over representative dynamical Rabi gates and conventional cyclic geometric gates. Furthermore, we validate the physical feasibility of high-fidelity noncyclic geometric gates in superconducting quantum circuits, with focused investigation into the impacts of intrinsic leakage errors and decoherence effects. Therefore, this work establishes a critical foundation for robust gate implementation toward practical fault-tolerant quantum processors.

quant-ph↗

arXiv2025

DreamFrame: Enhancing Video Understanding via Automatically Generated QA and Style-Consistent Keyframes

Recent large vision-language models (LVLMs) for video understanding are primarily fine-tuned with various videos scraped from online platforms. Existing datasets, such as ActivityNet, require considerable human labor for structuring and annotation before effectively utilized for tuning LVLMs. While current LVLMs are primarily trained on existing datasets in broad, general-purpose settings, adapting them to specific downstream scenarios remains challenging, as collecting and annotating task-specific videos is highly labor-intensive and time-consuming. To address this issue, we propose a three-stage framework named DreamFrame for automatically generating style-consistent keyframes and corresponding question-answer (QA) pairs to support LVLM instruction tuning. DreamFrame generates datasets in a movie-like manner. First, we utilize an LLM to generate structured movie plots including movie prior information (like overview and style), frame descriptions and plot-related QA pairs, with a story expansion strategy to mitigate context length limitations.Then, to ensure visual consistency across generated frames, we design a Style Immobilization Process which maintains consistent style through an embedding learning strategy. Finally, frame descriptions and style embeddings are integrated to produce coherent keyframes. Using DreamFrame, we construct a dataset comprising approximately 1k stylized keyframe-like videos and 100k diverse QA pairs. Extensive fine-tuned experiments on various LVLM architectures demonstrate the effectiveness of the proposed dataset. Furthermore, based on the proposed dataset, we fine-tune a new LVLM named DreamFrame-7B, which significantly surpasses the previous similar-sized LVLMs across different benchmarks.

cs.CV↗

arXiv2025

Mechanistic Insight into BEOL Thermal Transport via Optical Metrology and Multiphysics Simulation

As integrated circuits continue to scale down and adopt three-dimensional (3D) stacking, thermal management in the back-end-of-line (BEOL) has emerged as a critical design constraint. In this study, we present a combined experimental and simulation framework to quantitatively characterize and mechanistically understand thermal transport in BEOL multilayers. Using the Square-Pulsed Source (SPS) method, a time-resolved optical metrology technique, we measure cross-plane thermal resistance and areal heat capacity in semiconductor chips at nanometer resolution. Two fabricated chip samples, polished to the M4 and M6 interconnection layers, are analyzed to extract thermal properties of distinct multilayer stacks. Results show that thermal resistance follows a series model, while areal heat capacity scales linearly with metal content. To uncover the underlying physical mechanisms, we perform finite element simulations using COMSOL Multiphysics, examining the influence of via connectivity and dielectric thermal conductivity on effective cross-plane heat transport. The simulations reveal that dielectric materials, due to their large volume fraction, are the primary limiting factor in BEOL thermal conduction, while the via structure plays a secondary but significant role. This combined experimental-simulation approach provides mechanistic insight into heat transport in advanced IC architectures and offers practical guidance for optimizing thermal pathways in future high-performance 3D-stacked devices.

cond-mat.mes-hall↗

arXiv2025

SC-Captioner: Improving Image Captioning with Self-Correction by Reinforcement Learning

We propose SC-Captioner, a reinforcement learning framework that enables the self-correcting capability of image caption models. Our crucial technique lies in the design of the reward function to incentivize accurate caption corrections. Specifically, the predicted and reference captions are decomposed into object, attribute, and relation sets using scene-graph parsing algorithms. We calculate the set difference between sets of initial and self-corrected captions to identify added and removed elements. These elements are matched against the reference sets to calculate correctness bonuses for accurate refinements and mistake punishments for wrong additions and removals, thereby forming the final reward. For image caption quality assessment, we propose a set of metrics refined from CAPTURE that alleviate its incomplete precision evaluation and inefficient relation matching problems. Furthermore, we collect a fine-grained annotated image caption dataset, RefinedCaps, consisting of 6.5K diverse images from COCO dataset. Experiments show that applying SC-Captioner on large visual-language models can generate better image captions across various scenarios, significantly outperforming the direct preference optimization training strategy.

cs.CV↗