arXiv subjects

Changhao Li

Publications and source records attributed to Changhao Li.

At least 19 recordsLinked to original sources

arXiv2026

Extended Onsager's variational principle for nonlinear chemical reaction kinetics in active matter

Onsager's variational principle (OVP) constructs dissipative dynamics from free energy, dissipation, and input power, but deterministic rate laws alone do not identify fluctuations or chemical fuel consumption. We examine these distinctions for nonlinear reaction kinetics in active matter. Comparing Gaussian diffusion and Poisson reaction events clarifies when statistical fluctuations match a physical Rayleighian: the reaction dissipation potential must be consistent with counting statistics, local detailed balance links rate bias to chemical affinity, and kinetic scales remain constitutive inputs. Under these conditions, the normalized Rayleighian excess gives the leading rate-fluctuation cost in the frozen-state, many-event regime. Three models illustrate the consequences. Spatially uniform reactions exhibit protocol-dependent nonreciprocal response compatible with fixed-state OVP; different Rayleighian representations can select the same rate without specifying the same fluctuations. In an active emulsion, a finite fuel pool accounts for free-energy transfer and dissolution of a chemically maintained domain after replenishment stops. In a membrane-cytosol model, distinct reaction-channel realizations yield identical protein-density trajectories but different chemical input power. Under a separate kinetic assumption, an exact total-density equation still requires an initial rate and does not determine chemical work or fluctuations. These results clarify the physical and statistical inputs required for constructing reaction-transport models and interpreting their fluctuations.

physics.bio-ph↗

arXiv2026

Perspective on AI and quantum open-source software

Quantum open-source software provides essential infrastructure for quantum science, and artificial intelligence (AI) is reshaping its capabilities and transforming its development. However, verifying quantum software at scale remains difficult, and AI-generated contributions can increase the workload of maintainers. In this Perspective, we examine verification as a central requirement for reliable AI-assisted quantum software development and research. We analyze public repositories, records covering 110 repositories from five annual quantum open-source hackathons that we organize, and responses from more than 1600 participants in a community survey. Our analysis connects limitations in automated verification with the review burden on quantum software maintainers. We propose verification practices and contribution standards that require contributors to support their changes with reproducible evidence for maintainer review. We also examine how quantum open-source software can provide infrastructure for AI-driven research whose results the community can independently verify and build upon. We maintain an open-source companion repository to support reproducibility and help projects adopt and refine these proposals. Together, these contributions help the quantum open-source software community translate advances in AI into reliable quantum research through reproducible verification.

quant-ph↗

arXiv2026

HI3D 3.0 (Twinkle3D): Object-specific 3D Asset Generation with High Resolution

Image-to-3D generation has become increasingly capable of producing objects that closely resemble the input image, and an outstanding challenge is to reproduce the depicted object itself, including the specific geometry that defines it. Inscriptions, brand marks, and repeated structures are frequently distorted or lost, despite being critical to object identity. We present Hi3D 3.0, an image-to-3D generation system targeting object-specific fidelity, with Twinkle3D as its geometry model for generating watertight triangle meshes at $2048^{3}$ resolution. Twinkle3D advances high-fidelity geometry generation along four dimensions. First, while O-Voxel/FaithC offers high representational precision, it often suffers from poor surface quality and non-watertight geometry. We address both issues while retaining its $2048^{3}$-level precision. Second, we scale diffusion generation to sequences of up to 300K geometric tokens through a redesigned DiT architecture and large-scale distributed training optimizations, reducing training time per step from approximately ten minutes to ten seconds. Third, subsequent refinement cannot fully compensate for errors introduced during initial generation; we therefore strengthen both global shape and local detail in the initial generation stage, and the resulting single-stage model surpasses prior two-stage pipelines with $512^{3}$ refinement. Finally, we introduce a fine-grained image-3D cross-modal interaction mechanism that strengthens correspondence between visual evidence and geometric tokens, improving the recovery of object-specific structures. We evaluate geometric fidelity using alignment metrics derived from silhouettes and normal fields. Hi3D 3.0 outperforms four commercial systems across all reported metrics, recovering 82.1% of inscribed characters at 98.2% precision, compared with 21.7% recall for the strongest competitor.

cs.CV↗

arXiv2026

Does On-Policy Distillation Need to Complete Every Rollout?

On-policy distillation (OPD) transfers teacher knowledge to a student through feedback on student-generated rollouts, but generating complete rollouts and computing per-token teacher feedback are computationally expensive. Existing truncation methods save computation by stopping rollouts early. However, they do not compensate for truncated token gradients, which biases the gradient estimates and changes Vanilla OPD's original training objective. Furthermore, their truncation decisions rely on assumptions about supervision quality or criteria that may not generalize across settings. We introduce Stochastic Truncated On-Policy Distillation (STOPD), which accelerates OPD by randomly truncating student rollouts and correcting the resulting sampling bias with blockwise inverse-probability weighting (IPW). At each training step, we sample a stopping position for each rollout before generation and let the student generate until that position or natural termination. We obtain teacher feedback on every generated token and use blockwise IPW to compensate for the lower probability of retaining later tokens while controlling the weights. We prove that this weighting preserves Vanilla OPD's gradient sum in expectation and derive one-step update-error bounds for SGD, Adam, and Muon. We evaluate STOPD across multiple teacher-student model pairs, task domains, and optimizers. STOPD reduces generated tokens by 25-48% while maintaining accuracy comparable to Vanilla OPD, achieving 1.3-2.2x training speedups.

cs.LG↗

arXiv2026

Closing a single-copy gap to the Holevo bound with two-copy measurements

The Holevo Cramér-Rao bound sets the asymptotically attainable precision of quantum multiparameter estimation, but its attainability with finite quantum state copies remains unresolved in general. While a gap between the single-copy optimum and the Holevo bound persists at any finite copy number for two-parameter estimation models, the higher-parameter case has remained open. Here we disprove universal finite-copy gap persistence beyond two parameters by constructing a three-parameter rank-two model whose Holevo bound is unattainable with one copy but attained exactly with two. We further prove finite-copy persistence for full-rank models with any number of parameters. We then experimentally implement the two-copy collective measurement for the constructed model using superconducting qubits. We observe a two-copy estimation error below the single-copy and separate-measurement bounds determined from state tomography. By introducing controlled depolarization, we show that this precision advantage persists over a finite noise range. Our results connect the attainability of ultimate quantum precision to the structure of the multiparameter estimation model and the resources available for collective measurements.

quant-ph↗

arXiv2026

Quantum counterdiabatic driving enhanced by two-stage local control

Counterdiabatic driving can suppress diabatic losses during adiabatic ground state preparation. As its implementation necessitates the generation of an adiabatic gauge potential, which requires knowledge of the spectral gap of instantaneous Hamiltonians and involves highly non-local controls in many-body systems, local counterdiabatic driving with approximate adiabatic gauge potential has been widely adopted. In this work, using transverse-field Ising model as an example, we present an in-depth study of the performance of local counterdiabatic protocols and provide a closed-form analysis of ground-state-fidelity optimization. We then propose a two-stage protocol based on local counterdiabatic and simple local single-body control to further improve the performance. The practical implementation of these protocols does not require diagonalization of instantaneous Hamiltonians or spectral gaps, and only additional local single-body driving is involved. To benchmark their performance, we experimentally implement digitized adiabatic quantum evolution in a trapped-ion system. We characterize the quality of the prepared states and explore the scaling behavior with system size up to 14 qubits. Our demonstration of quantum shortcut to adiabaticity opens a path towards preparing ground states of complex systems with accessible local controls.

quant-ph↗

arXiv2026

InterMASH: A Unified Geometric Representation for Grasp Synthesis

Grasp synthesis aims to generate stable and physically plausible hand--object interactions, and has become a fundamental problem in both human hand modeling and robotic manipulation. However, a unified representation across human and robotic hands is still lacking, mainly due to differences in hand morphology and surface modeling. Prior methods typically rely on either contact maps or dense implicit descriptors to represent interaction, but these representations are often incomplete or computationally expensive and redundant. We propose InterMASH, a unified geometric representation that establishes cross-embodiment correspondence using sphere-fixed anchors. At each anchor, low-degree spherical harmonics compactly encode local hand geometry, object geometry, and contact, forming an explicit and interpretable token sequence. Building on this natively tokenized structure, we introduce a conditional Diffusion Transformer that operates directly in the proposed InterMASH representation space and jointly generates hand geometry and contact, improving consistency and physical plausibility. Our method achieves competitive performance with state-of-the-art methods on key physical feasibility metrics in a large-scale ShadowHand benchmark, supports joint training across multiple hands, and shows that cross-embodiment fine-tuning with human grasp data can improve robotic grasp success and diversity. Project page is available at https://inter-mash.github.io/.

cs.RO↗

arXiv2026

Evaluating Verified Autonomy in Quantum Engineering

Reliable quantum engineering is essential for turning quantum phenomena into practical technologies. As quantum platforms grow in scale and complexity, their characterization and operation require increasing human effort and coordination. Scientific artificial intelligence agents, which can plan experiments, operate instruments, and analyze observations, offer a promising route towards autonomous quantum engineering. Yet whether current agents can perform reliably in this setting has not been systematically established. To fill this gap, we developed Quantum-Harbor, a virtual laboratory that provides a controlled execution environment for agents to interact with quantum systems. This design enables direct verification of both the actions taken and the conclusions drawn. Building on this framework, we introduce QIQCBench, a benchmark of $49$ expert-authored tasks spanning multiple layers including calibration and control, error correction and compilation, sensing and networking. Across $17$ frontier agentic systems, QIQCBench reveals wide variation in verified performance. These results expose a substantial gap between demonstrating capability and achieving reliable operation, and establish Quantum-Harbor as a foundation for measuring progress towards verified autonomy in quantum engineering.

quant-ph↗

arXiv2026

Control incompatibility in multiparameter quantum metrology

In practical applications like quantum sensing and quantum imaging, there is often a necessity to estimate multiple parameters simultaneously. Although the ultimate precision limits for single-parameter estimation are well established, the precision limit of multi-parameter estimation is much less understood. This is primarily due to the inherent incompatibility of the optimal strategies for the estimation of different parameters, particularly those pertaining to optimal control.In this study, we tackle the critical issue of control incompatibility in multi-parameter estimation by presenting explicit cases that expose this challenge. Our research not only pioneers the exploration of control incompatibility but also highlights its pivotal role in the field. Furthermore, our work offers valuable insights into how to minimize trade-offs induced by control incompatibility and enhance precision. This paves the way for future investigations into control strategies that enable optimal estimation of multiple parameters that are incompatible.

quant-ph↗

arXiv2026

Efficient Real-World Online Reinforcement Learning for Robot Manipulation via Centralized Training and Critic Decomposition

Real-world online reinforcement learning (RL) provides a promising approach for training robotic manipulation policies directly in the physical world, avoiding the sim-to-real gap and enabling continuous policy refinement through human-in-the-loop interaction. Recent methods have demonstrated sample-efficient learning through human intervention but remain limited to small randomization ranges and encounter challenges with the non-stationarity induced by concurrently training multiple agents. To address these limitations, we introduce a unified framework that combines centralized training with decentralized execution (CTDE) and a Hybrid Reward Architecture (HRA). This enables multiple actors to share a centralized multi-head critic. The critic is decomposed into task and grasp heads, corresponding to the sparse task reward and a potential-based grasping reward, respectively. We accordingly reformulate the critic and actor objectives to exploit the decomposed Q-values while explicitly accounting for the categorical action distribution of the discrete gripper policy. Experimental results demonstrate that the proposed framework substantially improves both sample efficiency and policy performance. We validate our approach on two robotic arms and a simulated humanoid robot across tennis ball and banana pick-and-place, pot reset, and simulated block relocation tasks under dimension-wise domain randomization, approximately 5-25x larger than those considered in prior work. Compared with a state-of-the-art baseline, our method improves the success rate from 60% to 80% on tennis ball pick-and-place, from 60% to 90% on banana pick-and-place, and from 25% to 95% on simulated block relocation, while also successfully accomplishing a task where the baseline consistently fails. Videos and more details are available at our project website: https://hil-harc.github.io/.

cs.RO↗

arXiv2026

Matryoshka Agent: Unfolding Sub-Agents for Long-Horizon Machine Learning Engineering

Machine learning engineering (MLE) tasks require long-horizon decision making over iterative solution debugging and refinement, under expensive and feedback-driven environment interactions. Developing and training a monolithic agent for such tasks is fundamentally challenging, as it must simultaneously manage extremely long and noisy contexts, explore vast solution spaces, and remain effective under limited model capacity and computational budgets. To address these challenges, we propose Matryoshka Agent, a unified hierarchical agent framework for complex long-horizon tasks. Matryoshka Agent decomposes agentic problem solving into a coordinated hierarchy of decision making and execution: a high-level Orchestrator maintains compact, long-horizon exploration states and issues strategic instructions, while lower-level Sub-Agents execute concrete solution attempts through direct environment interaction, mediated by standardized Tool interface. This design decouples strategic exploration from costly execution, substantially reducing the burden of long-context reasoning and enabling efficient iterative refinement. We further develop an efficient training paradigm for Matryoshka Agent. Experimental results on a broad range of MLE tasks with diverse model types and scales demonstrate that Matryoshka Agent is an effective and scalable paradigm for long-horizon MLE tasks and complex agentic problem solving. Notably, Matryoshka Agent enables Qwen3-4B-Instruct to reach Orchestrator performance comparable to o4-mini. Applying Matryoshka Agent to Qwen3-30B-Coder results in at most 36.7% relative performance gain.

cs.AI↗

arXiv2026

Sample Is Feature: Beyond Item-Level, Toward Sample-Level Tokens for Unified Large Recommender Models

Scaling industrial recommender models has followed two parallel paradigms: \textbf{sample information scaling} -- enriching the information content of each training sample through deeper and longer behavior sequences -- and \textbf{model capacity scaling} -- unifying sequence modeling and feature interaction within a single Transformer backbone. However, these two paradigms still face two structural limitations. Firstly, sample information scaling methods encode only a subset of each historical interaction into the sequence token, leaving the majority of the original sample context unexploited and precluding the modeling of sample-level, time-varying features. Secondly, model capacity scaling methods are inherently constrained by the structural heterogeneity between sequential and non-sequential features, preventing the model from fully realizing its representational capacity. To address these issues, we propose \textbf{SIF} (\emph{Sample Is Feature}), which encodes each historical Raw Sample directly into the sequence token -- maximally preserving sample information while simultaneously resolving the heterogeneity between sequential and non-sequential features. SIF consists of two key components. The \textbf{Sample Tokenizer} quantizes each historical Raw Sample into a Token Sample via hierarchical group-adaptive quantization (HGAQ), enabling full sample-level context to be incorporated into the sequence efficiently. The \textbf{SIF-Mixer} then performs deep feature interaction over the homogeneous sample representations via token-level and sample-level mixing, fully unleashing the model's representational capacity. Extensive experiments on a large-scale industrial dataset validate SIF's effectiveness, and we have successfully deployed SIF on an industrial food delivery platform.

cs.IR↗

arXiv2026

Not Only NTP: Extending Training Signal Coverage for Generative Recommendation

Next-Token Prediction (NTP) carries two structural training signal limitations. First, NTP optimizes for single-step prediction only, placing no supervised pressure on learning longer-range behavioral structure -- we term this \textbf{temporal locality}. Second, in multi-domain sequences, each target item embedding receives gradient updates exclusively from the immediately preceding hidden state, with no explicit gradient pathway from cross-domain context -- we term this \textbf{spatial locality}. We propose \textbf{NONTP}, extending NTP's signal coverage along both dimensions through two auxiliary objectives. \textbf{TCL (Temporal Contrastive Learning)} uses a BYOL-style EMA teacher with InfoNCE to align hidden states against a $K$-step future trajectory in representation space. \textbf{TDL (Trans-Domain Learning)} mean-pools cross-domain hidden states and predicts through the shared prediction head, opening a second gradient pathway with no additional parameters. Both are discarded at inference: zero overhead. On a four-domain Meituan industrial dataset (full ranking), NONTP achieves HR@10 +34.3\% over NTP and +18.3\% over MBGR. On the public Amazon Movie-Book-CDs benchmark, HR@10 +2.8\% and NDCG@10 +3.7\%. Online A/B tests confirm CTR +1.8\% and GMV +2.1\% (both $p < 0.01$). Ablation studies confirm each component contributes independently, with gradient conflict analyzed as a direction for future work.

cs.IR↗

arXiv2026

Neural optimization of the most probable paths of 3D active Brownian particles

We develop a variational neural-network framework to determine the most probable path (MPP) of a 3D active Brownian particle (ABP) by directly minimizing the Onsager-Machlup integral (OMI). To obtain the OMI, we use the Onsager-Machlup variational principle for active systems and construct the Rayleighian of the ABP by including its active power. This approach reveals geometric transitions of the MPP from in-plane I- and U-shaped paths to 3D helical paths as the final time and net displacement are varied. We also demonstrate that the initial and final boundary conditions have a significant impact on the MPPs. Our results show that neural optimization combined with the Onsager-Machlup variational principle provides an efficient and versatile framework for exploring optimal transition pathways in active and nonequilibrium systems.

cond-mat.soft↗

arXiv2026

Skeletal-Anchored Dual Harmonics for Structured 3D Modeling

We present Skeletal-Anchored Dual Harmonics (SADH), a novel 3D shape representation that tightly couples local surface geometry with internal meso-skeletal organization. SADH represents a shape as a collection of compact surface patches rooted on internal anchors optimized directly inside the object volume. Each patch is parameterized using a dual-channel spherical harmonic (SH) formulation, where one channel models local radial geometry while the other defines adaptive patch support through a generalized viewing cone. Unlike isotropic primitives such as medial spheres or Gaussian kernels, SH patches directly encode anisotropic local surface geometry together with adaptive spatial support, enabling compact representation of detailed and directionally varying surface regions. Starting from unorganized point clouds, SADH jointly optimizes surface geometry, anchor locations, patch orientations, and structural connectivity through a staged optimization process that progressively forms a coherent meso-skeletal structure. A geodesic anchor graph further preserves structural relationships between neighboring patches. Experiments on complex 3D shapes demonstrate that SADH achieves accurate surface reconstruction together with compact and coherent skeletal organization across a wide range of geometries.

cs.GR↗

arXiv2026

Revisiting DAgger in the Era of LLM-Agents

Long-horizon LM agents learn from multi-turn interaction, where a single early mistake can alter the subsequent state distribution and derail the whole trajectory. Existing recipes fall short in complementary ways: supervised fine-tuning provides dense teacher supervision but suffers from covariate shift because it is trained on off-policy teacher trajectories; while reinforcement learning with verifiable rewards avoids this off-policy mismatch by learning from on-policy rollouts but with only sparse outcome feedback. We address this dilemma by revisiting Dataset Aggregation (DAgger) for multi-turn LM agents: the algorithm collects trajectories through a turn-level interpolation of student and teacher policies, and the student is then trained on these trajectories using supervised labels provided by the teacher. By directly interacting with environments, we expose the model to realistic states likely to be encountered during deployment, thereby effectively mitigating covariate shift. Besides, since the student is learned by mimicking the teacher's behavior, it receives rich feedback during learning. To demonstrate DAgger enjoys the benefits of both worlds, we tested the algorithm to train a software-engineering agent with 4B- and 8B-scale student models. On SWE-bench Verified, our DAgger-style training improves over the strongest post-training baseline by +3.9 points at 4B and +3.6 points at 8B. The resulting 4B agent reaches 27.3%, outperforming representative published 8B SWE-agent systems, while the 8B agent achieves 29.8%, surpassing SWE-Gym-32B and coming within 5 points of stronger 32B-scale agents. Together with consistent gains on the held-out SWE-Gym split, these results suggest the effectiveness of DAgger for modern long-horizon LM agents.

cs.LG↗

arXiv2026

Exploration-Driven Optimization for Test-Time Large Language Model Reasoning

Post-training techniques combined with inference-time scaling significantly enhance the reasoning and alignment capabilities of large language models (LLMs). However, a fundamental tension arises: inference-time methods benefit from diverse sampling from a relatively flattened probability distribution, whereas reinforcement learning (RL)-based post-training inherently sharpens these distributions. To address this, we propose Exploration-Driven Optimization (EDO), which extends reward-biasing style exploration objectives to iterative post-training and integrates them into standard RL objectives, encouraging greater diversity in sampled solutions while facilitating more effective inference-time computation. We incorporate EDO into iterative Direct Preference Optimization (iDPO) and Group Relative Policy Optimization (GRPO), resulting in two variants: ED-iDPO and ED-GRPO. Extensive experiments demonstrate that both ED-iDPO and ED-GRPO exhibit greater solution diversity and improved reasoning abilities, particularly when combined with test-time computation techniques like self-consistency. Across three in-distribution reasoning benchmarks, EDO achieves a 1.0-1.3\% improvement over the strongest baselines, and delivers an additional 1.5\% average gain on five out-of-distribution tasks. Beyond accuracy, EDO preserves model entropy and stabilizes RL training dynamics, highlighting its effectiveness in preventing over-optimization collapse. Taken together, these results establish EDO as a practical framework for balancing exploration and exploitation in LLM reasoning, especially in settings that rely on test-time scaling.

cs.LG↗

arXiv2026

Operational criteria for quantum advantage in latency-constrained nonlocal games

Remote entanglement enables coordinated decision making without communication and produces correlations beyond those achievable by any classical strategy, representing a practical quantum advantage in time-critical distributed decision-making problems. However, existing analyses of quantum-classical gaps in such latency-constrained tacit coordination (LCTC) have focused on idealized models that neglect the finite stationary window of the LCTC, finite operation times, and limited entanglement generation rates, leaving fundamental constraints unaccounted for. In this work, we develop a comprehensive framework to quantitatively analyze quantum advantage in LCTC that explicitly incorporates finite-duration and finite-rate operations, as well as generalized utility structures with a limited stationary window. These advances are made possible by adapting statistical certification methods for nonlocal games to the decision-making scenarios of LCTC, identifying operational criteria that must be satisfied by the hardware implementations to realize quantum advantage with sufficient statistical significance. To meet the stringent criteria, we propose time-multiplexed, event-ready operations of cavity-assisted trapped-atom quantum network nodes that provide a continuous stream of entangled qubit pairs, with decision latencies of a microsecond and decision rates of $8\times 10^3~\text{s}^{-1}$ per channel for a representative metropolitan-scale $50$-km fiber network to keep up with the fast-changing environment, such as financial markets and electric grid networks. These results bridge the gap between the theoretical notions of the quantum-classical gap in nonlocal games and concrete implementations that meet the stringent operational criteria for achieving robust quantum advantage in realistic coordination tasks.

quant-ph↗