arXiv subjects

Yutong He

Publications and source records attributed to Yutong He.

At least 19 recordsLinked to original sources

arXiv2026

Self-organized positron reorienting and pinching mechanism for the experimental detection of the linear Breit-Wheeler process

The linear Breit-Wheeler (LBW) process ($\gamma+\gamma\rightarrow e^{-}+e^{+}$) is a fundamental prediction of quantum electrodynamics, but yet to be observed under laboratory conditions using real photons. In recent years, a few experimental schemes utilizing high-intense ($\sim10^{22}$W/cm$^2$) laser-plasma interactions to observe the LBW process have been proposed. However, a high level of signal-to-noise-ratio are expected in these schemes, hindering the first-ever experimental detection of the LBW process by real photons. In this paper, we present a simple experimental setup which could enhance the expected positron signals by 2-3 orders of magnitude compared to previously proposed schemes, reaching the level of $10^{6}$MeV$^{-1}$str$^{-1}$. Moreover, such high positron signal is achieve in the direction opposite to the laser propagation, where a significantly quieter background is expected compared to the previously focused direction of laser propagation. The key to achieve this result is a newly discovered self-organized positron reorienting and pinching mechanism, enabled by the in-situ strong plasma fields from the laser-plasma interaction.

physics.plasm-ph↗

arXiv2026

ReasFlow: Assisting Reasoning-Centric Scientific Discovery in Applied Mathematics via a Knowledge-Based Multi-Agent System

Recent advances in Large Language Models have fueled autonomous AI agents capable of tackling complex scientific tasks, yet existing automated research systems remain predominantly focused on empirically driven domains with quantitative benchmarks, leaving theory-driven discovery, particularly in mathematically grounded disciplines requiring rigorous proofs and synthesis of domain knowledge, largely underexplored. Key challenges include the difficulty of verifying theoretical reasoning at scale, insufficient reasoning ability for autonomous frontier exploration, and a scarcity of procedural heuristics in the literature. We introduce ReasFlow, an end-to-end autonomous agent system for reasoning-centric scientific discovery that operationalizes a collaborative paradigm where the human expert acts as Principal Investigator while the agent executes rigorous derivations as a capable graduate student. ReasFlow incorporates (i) a robust internal verification loop that audits logical coherence and corrects fundamental errors prior to human inspection, and (ii) an automated knowledge retrieval and self-improvement mechanism that proactively surfaces both declarative facts and overlooked procedural heuristics, substantially reducing expert intervention. The system unifies literature synthesis, algorithm design, theorem proving, experimentation, and manuscript preparation in a single system. Deployed to autonomously generate five complete research papers with rigorous theoretical and empirical content from minimal prompts, ReasFlow consistently achieves the highest evaluation scores among state-of-the-art open-access baselines under a curated LLM-based review rubric. ReasFlow is publicly accessible via the ReasLab platform, providing a collaborative workspace for AI-assisted theoretical research. Github repo: https://github.com/reaslab/ReasFlow.git.

cs.AI↗

arXiv2026

Few-Step Boltzmann Generators via Scalable Likelihood Flow Maps

Recent progress in flow-based generative modeling has led to models that output high-quality samples while using only a small number of function evaluations. However, at present, there is a lack of similar advances in estimating the model likelihood. In particular, most existing methods either rely on restrictive architectures that enable exact calculations, or use stochastic approximations such as Hutchinson's trace estimator that introduce substantial variance. In this work, we introduce SCAlable LikeLihood distillation of flOw maPs (SCALLOP). SCALLOP builds on the recently proposed F2D2, a likelihood flow map model that can generate samples and their densities in a small number of function evaluations. While F2D2 uses Hutchinson's estimator during training, we introduce an alternative and more scalable likelihood distillation objective that is Hutchinson-free and admits a vectorized formulation. Empirically, we demonstrate the effectiveness of SCALLOP as a Boltzmann generator in molecular science, and further validate its benefit on image datasets. SCALLOP significantly reduces both training variance and training time while consistently improving performance compared to F2D2, and is competitive with the state-of-the-art while achieving up to 10x inference speedup over the fastest baseline.

cs.LG↗

arXiv2026

DFSAttn: Dynamic Fine-grained Sparse Attention for Efficient Video Generation

Diffusion transformers have achieved remarkable success in high-quality video generation, yet their reliance on spatiotemporal 3D full attention incurs prohibitive computational cost due to the quadratic complexity of attention. Block sparse attention is a common approach to mitigate this by focusing computation on important regions. However, attention maps in DiTs exhibit inherently dynamic and fine-grained sparsity, which causes existing block sparse attention methods to degrade significantly in quality, especially at high sparsity ratios. In this paper, we revisit block sparse attention and derive a theoretical lower bound on attention recall to characterize the key factors governing its effectiveness. Guided by these insights, we propose DFSAttn, a training-free sparse attention framework that enables dynamic, fine-grained sparsification efficiently. DFSAttn incorporates three core designs: Hilbert curve-based token reordering to achieve fine-grained sparsity while preserving efficient GPU execution, hierarchical block scoring for accurate block importance estimation, and sparse mask caching with adaptive ratios to balance accuracy and efficiency. Experimental results demonstrate that DFSAttn consistently outperforms prior methods under high sparsity, achieving up to 2.1$\times$ end-to-end speedup while maintaining high generation quality. Our code is open-sourced and available at https://github.com/jessica-hujie/DFSAttn.

cs.CV↗

arXiv2026

RoPeSLR: 3D RoPE-driven Sparse-LowRank Attention for Efficient Diffusion Transformers

Diffusion Transformers (DiTs) have revolutionized high-fidelity video generation, yet their $\mathcal{O}(L^2)$ attention complexity poses a formidable bottleneck for long-sequence synthesis. While recent sparse-linear attention hybrids aim to mitigate this, their performance severely degrades at extreme sparsity due to the "RoPE Dilemma": standard linear attention fails to preserve the orthogonal relative-position structure of 3D Rotary Position Embeddings (RoPE), neutralizing vital distance awareness. To address this, we propose \textbf{RoPeSLR}, a 3D RoPE-guided Sparse-LowRank attention framework. We establish that under empirically validated assumptions, the DiT attention manifold admits a decoupling into a high-frequency semantic spike set (bounded by $\mathcal{O}(L^{3/2})$ sparsity) and an extreme low-rank ($\mathcal{O}(d_h \log L)$) background continuum. Guided by this structural prior, RoPeSLR eschews standard linear attention for a head-wise low-rank parameterization equipped with a learnable 3D Absolute Positional Embedding (PE) injection, seamlessly synthesizing long-range relative distance decay. By guaranteeing sub-quadratic sparsity and sub-linear rank growth, RoPeSLR is exceptionally suited for scaling to ultra-long video inference. Extensive evaluations validate this scalable superiority: at 90\% sparsity, RoPeSLR achieves up to $10\times$ fewer FLOPs on Wan2.1-1.3B and delivers a $2.26\times$ end-to-end inference speedup on the ultra-long 100K+ token sequences of HunyuanVideo-13B, all while maintaining near-lossless generation fidelity (less than 1.3\% average VBench degradation).

cs.CV↗

arXiv2026

FedSLoP: Memory-Efficient Federated Learning with Low-Rank Gradient Projection

Federated learning enables a population of clients to collaboratively train machine learning models without exchanging their raw data, but standard algorithms such as FedAvg suffer from slow convergence and high communication and memory costs in heterogeneous, resource-constrained environments. We introduce FedSLoP, a federated optimization algorithm that combines stochastic low-rank subspace projections of gradients, thereby reducing the dimension of communicated and stored updates while preserving optimization progress. On the theoretical side, we develop a detailed nonconvex convergence analysis under standard smoothness and bounded-variance assumptions, showing that FedSLoP is guaranteed to converge to a first-order stationary point at a rate of $O(1/\sqrt{NT})$. On the empirical side, we conduct extensive experiments on federated MNIST classification with heterogeneous data partitions, showing that FedSLoP substantially reduces communication volume and client-side memory while achieving competitive or better accuracy compared with FedAvg and representative sparse or low-rank baselines. Together, our results demonstrate that random subspace momentum methods such as FedSLoP provide a principled and effective approach to communication- and memory-efficient federated learning. Codes are available at: https://github.com/pkumelon/FedSLoP.git.

cs.LG↗

arXiv2026

Diamond Maps: Efficient Reward Alignment via Stochastic Flow Maps

Flow and diffusion models produce high-quality samples, but adapting them to user preferences or constraints post-training remains costly and brittle, a challenge commonly called reward alignment. We argue that efficient reward alignment should be a property of the generative model itself, not an afterthought, and redesign the model for adaptability. We propose "Diamond Maps", stochastic flow map models that enable efficient and accurate alignment to arbitrary rewards at inference time. Diamond Maps amortize many simulation steps into a single-step sampler, like flow maps, while preserving the stochasticity required for optimal reward alignment. This design makes search, Sequential Monte Carlo, and guidance scalable by enabling efficient and consistent estimation of the value function. Our experiments show that Diamond Maps can be learned efficiently via distillation from GLASS Flows, achieve stronger reward alignment performance, and scale better than existing methods. Our results point toward a practical route to generative models that can be rapidly adapted to arbitrary preferences and constraints at inference time.

cs.LG↗

arXiv2025

Joint Distillation for Fast Likelihood Evaluation and Sampling in Flow-based Models

Log-likelihood evaluation enables important capabilities in generative models, including model comparison, certain fine-tuning objectives, and many downstream applications. Yet paradoxically, some of today's best generative models -- diffusion and flow-based models -- still require hundreds to thousands of neural function evaluations (NFEs) to compute a single likelihood. While recent distillation methods have successfully accelerated sampling to just a few steps, they achieve this at the cost of likelihood tractability: existing approaches either abandon likelihood computation entirely or still require expensive integration over full trajectories. We present fast flow joint distillation (F2D2), a framework that simultaneously reduces the number of NFEs required for both sampling and likelihood evaluation by two orders of magnitude. Our key insight is that in continuous normalizing flows, the coupled ODEs for sampling and likelihood are computed from a shared underlying velocity field, allowing us to jointly distill both the sampling trajectory and cumulative divergence using a single flow map. F2D2 is modular, compatible with existing flow-based few-step sampling models, and requires only an additional divergence prediction head. Experiments demonstrate F2D2's capability of achieving accurate log-likelihood with few-step evaluations while maintaining high sample quality, solving a long-standing computational bottleneck in flow-based generative models. As an application of our approach, we propose a lightweight self-guidance method that enables a 2-step MeanFlow to outperform a 1024 step flow matching model with only a single additional backward NFE.

cs.LG↗

arXiv2025

Mixture-of-Channels: Exploiting Sparse FFNs for Efficient LLMs Pre-Training and Inference

Large language models (LLMs) have demonstrated remarkable success across diverse artificial intelligence tasks, driven by scaling laws that correlate model size and training data with performance improvements. However, this scaling paradigm incurs substantial memory overhead, creating significant challenges for both training and inference. While existing research has primarily addressed parameter and optimizer state memory reduction, activation memory-particularly from feed-forward networks (FFNs)-has become the critical bottleneck, especially when FlashAttention is implemented. In this work, we conduct a detailed memory profiling of LLMs and identify FFN activations as the predominant source to activation memory overhead. Motivated by this, we introduce Mixture-of-Channels (MoC), a novel FFN architecture that selectively activates only the Top-K most relevant channels per token determined by SwiGLU's native gating mechanism. MoC substantially reduces activation memory during pre-training and improves inference efficiency by reducing memory access through partial weight loading into GPU SRAM. Extensive experiments validate that MoC delivers significant memory savings and throughput gains while maintaining competitive model performance.

cs.LG↗

arXiv2025

An All-Reduce Compatible Top-K Compressor for Communication-Efficient Distributed Learning

Communication remains a central bottleneck in large-scale distributed machine learning, and gradient sparsification has emerged as a promising strategy to alleviate this challenge. However, existing gradient compressors face notable limitations: Rand-$K$ discards structural information and performs poorly in practice, while Top-$K$ preserves informative entries but loses the contraction property and requires costly All-Gather operations. In this paper, we propose ARC-Top-$K$, an {All-Reduce}-Compatible Top-$K$ compressor that aligns sparsity patterns across nodes using a lightweight sketch of the gradient, enabling index-free All-Reduce while preserving globally significant information. ARC-Top-$K$ is provably contractive and, when combined with momentum error feedback (EF21M), achieves linear speedup and sharper convergence rates than the original EF21M under standard assumptions. Empirically, ARC-Top-$K$ matches the accuracy of Top-$K$ while reducing wall-clock training time by up to 60.7\%, offering an efficient and scalable solution that combines the robustness of Rand-$K$ with the strong performance of Top-$K$.

cs.LG↗

arXiv2025

MISA: Memory-Efficient LLMs Optimization with Module-wise Importance Sampling

The substantial memory demands of pre-training and fine-tuning large language models (LLMs) require memory-efficient optimization algorithms. One promising approach is layer-wise optimization, which treats each transformer block as a single layer and optimizes it sequentially, while freezing the other layers to save optimizer states and activations. Although effective, these methods ignore the varying importance of the modules within each layer, leading to suboptimal performance. Moreover, layer-wise sampling provides only limited memory savings, as at least one full layer must remain active during optimization. To overcome these limitations, we propose Module-wise Importance SAmpling (MISA), a novel method that divides each layer into smaller modules and assigns importance scores to each module. MISA uses a weighted random sampling mechanism to activate modules, provably reducing gradient variance compared to layer-wise sampling. Additionally, we establish an $\mathcal{O}(1/\sqrt{K})$ convergence rate under non-convex and stochastic conditions, where $K$ is the total number of block updates, and provide a detailed memory analysis showcasing MISA's superiority over existing baseline methods. Experiments on diverse learning tasks validate the effectiveness of MISA. Source code is available at https://github.com/pkumelon/MISA.

cs.LG↗

arXiv2025

MeCeFO: Enhancing LLM Training Robustness via Fault-Tolerant Optimization

As distributed optimization scales to meet the demands of Large Language Model (LLM) training, hardware failures become increasingly non-negligible. Existing fault-tolerant training methods often introduce significant computational or memory overhead, demanding additional resources. To address this challenge, we propose Memory- and Computation-efficient Fault-tolerant Optimization (MeCeFO), a novel algorithm that ensures robust training with minimal overhead. When a computing node fails, MeCeFO seamlessly transfers its training task to a neighboring node while employing memory- and computation-efficient algorithmic optimizations to minimize the extra workload imposed on the neighboring node handling both tasks. MeCeFO leverages three key algorithmic designs: (i) Skip-connection, which drops the multi-head attention (MHA) module during backpropagation for memory- and computation-efficient approximation; (ii) Recomputation, which reduces activation memory in feedforward networks (FFNs); and (iii) Low-rank gradient approximation, enabling efficient estimation of FFN weight matrix gradients. Theoretically, MeCeFO matches the convergence rate of conventional distributed training, with a rate of $\mathcal{O}(1/\sqrt{nT})$, where n is the data parallelism size and T is the number of iterations. Empirically, MeCeFO maintains robust performance under high failure rates, incurring only a 4.18% drop in throughput, demonstrating 5.0$\times$ to 6.7$\times$ greater resilience than previous SOTA approaches. Codes are available at https://github.com/pkumelon/MeCeFO.

cs.DC↗

arXiv2025

Greedy Low-Rank Gradient Compression for Distributed Learning with Convergence Guarantees

Distributed optimization is pivotal for large-scale signal processing and machine learning, yet communication overhead remains a major bottleneck. Low-rank gradient compression, in which the transmitted gradients are approximated by low-rank matrices to reduce communication, offers a promising remedy. Existing methods typically adopt either randomized or greedy compression strategies: randomized approaches project gradients onto randomly chosen subspaces, introducing high variance and degrading empirical performance; greedy methods select the most informative subspaces, achieving strong empirical results but lacking convergence guarantees. To address this gap, we propose GreedyLore--the first Greedy Low-Rank gradient compression algorithm for distributed learning with rigorous convergence guarantees. GreedyLore incorporates error feedback to correct the bias introduced by greedy compression and introduces a semi-lazy subspace update that ensures the compression operator remains contractive throughout all iterations. With these techniques, we prove that GreedyLore achieves a convergence rate of $\mathcal{O}(\sigma/\sqrt{NT} + 1/T)$ under standard optimizers such as MSGD and Adam--marking the first linear speedup convergence rate for low-rank gradient compression. Extensive experiments are conducted to validate our theoretical findings.

cs.LG↗

arXiv2025

CentroidKV: Efficient Long-Context LLM Inference via KV Cache Clustering

Large language models (LLMs) with extended context windows have become increasingly prevalent for tackling complex tasks. However, the substantial Key-Value (KV) cache required for long-context LLMs poses significant deployment challenges. Existing approaches either discard potentially critical information needed for future generations or offer limited efficiency gains due to high computational overhead. In this paper, we introduce CentroidKV, a simple yet effective framework for online KV cache clustering. Our approach is based on the observation that key states exhibit high similarity along the sequence dimension. To enable efficient clustering, we divide the sequence into chunks and propose Chunked Soft Matching, which employs an alternating partition strategy within each chunk and identifies clusters based on similarity. CentroidKV then merges the KV cache within each cluster into a single centroid. Additionally, we provide a theoretical analysis of the computational complexity and the optimality of the intra-chunk partitioning strategy. Extensive experiments across various models and long-context benchmarks demonstrate that CentroidKV achieves up to 75% reduction in KV cache memory usage while maintaining comparable model performance. Moreover, with minimal computational overhead, CentroidKV accelerates the decoding stage of inference by up to $1.92\times$ and increases the serving throughput by up to $4\times$.

cs.CL↗

arXiv2025

Accelerating Diffusion Planners in Offline RL via Reward-Aware Consistency Trajectory Distillation

Although diffusion models have achieved strong results in decision-making tasks, their slow inference speed remains a key limitation. While consistency models offer a potential solution, existing applications to decision-making either struggle with suboptimal demonstrations under behavior cloning or rely on complex concurrent training of multiple networks under the actor-critic framework. In this work, we propose a novel approach to consistency distillation for offline reinforcement learning that directly incorporates reward optimization into the distillation process. Our method achieves single-step sampling while generating higher-reward action trajectories through decoupled training and noise-free reward signals. Empirical evaluations on the Gym MuJoCo, FrankaKitchen, and long horizon planning benchmarks demonstrate that our approach can achieve a 9.7% improvement over previous state-of-the-art while offering up to 142x speedup over diffusion counterparts in inference time.

cs.LG↗

arXiv2025

TAH-QUANT: Effective Activation Quantization in Pipeline Parallelism over Slow Network

Decentralized training of large language models offers the opportunity to pool computational resources across geographically distributed participants, but is often bottlenecked by network communication, particularly under pipeline parallel settings. While pipeline parallelism partitions model layers across devices to handle large-scale models, it necessitates frequent communication of intermediate activations, creating challenges when network bandwidth is limited. To address these issues, we propose TAH-Quant (Tile-wise Adaptive Hadamard Quantization), a novel activation quantization framework for pipeline parallelism. TAH-Quant integrates fine-grained tile-wise quantization, entropy-guided tile-wise adaptive bit allocation for optimal bit usage, and a Hadamard-based transformation with pivot swapping to effectively suppress outliers. Compared with token-level allocation, the tile-wise allocator assigns precision at the granularity of small channel windows within each token, reducing quantization error under the same bit budget. We prove that pipeline parallel training equipped with TAH-Quant maintains a convergence rate of O(1/sqrt(T)), matching that of vanilla stochastic gradient descent. Extensive experiments demonstrate that TAH-Quant achieves an aggressive activation quantization ratio of 3-4 bits, providing up to 4.3x throughput speedup over uncompressed FP32 and up to 1.33x wall-clock speedup over AQ-SGD, while preserving training convergence, avoiding AQ-SGD's activation-cache overhead, and generalizing well across various training scenarios.

cs.LG↗

arXiv2025

Counterpart-Induced Millisecond-Scale Truncation Mechanism of Fast Radio Bursts

The observed millisecond-scale duration is an essential yet mysterious feature of fast radio bursts (FRBs). In this Letter, we link the observed soft gamma-ray counterpart of FRB 200428 to electron-positron pair cascades driven by Compton scattering and the Breit-Wheeler process. We demonstrate that such pair cascades can truncate FRBs to durations down to millisecond-scale, thereby establishing millisecond-scale upper bounds on their durations. The physical processes involved in the truncation mechanism occur during the propagation of FRBs after their production. Consequently, this mechanism is independent of the specific production mechanism or origin of the FRBs, suggesting that it could potentially operate in all FRBs. Our results lift the constraint on FRB production mechanisms that they must inherently generate bursts lasting only milliseconds.

astro-ph.HE↗

arXiv2025

Performance-Aware Control of Modular Batteries For Fast Frequency Response

Modular batteries can be aggregated to deliver frequency regulation services for power grids. Although utilizing the idle capacity of battery modules is financially attractive, it remains challenging to consider the heterogeneous module-level characteristics such as dynamic operational efficiencies and battery degradation. In addition, real-time decision making within seconds is required to enable fast frequency response. In order to address these issues, this paper proposes a performance-aware scheduling approach for battery modules to deliver fast frequency response (FFR) support. In particular, the conduction loss and switching loss of battery packs as well as converters are captured within a mix-integer quadratic constrained program (MIQCP). The cycle-based aging model identifies the aging cost of battery modules during frequent cycling by introducing the aging subgradient calculation and linearization. Case studies based on real-world battery data show that the proposed scheduling approach can effectively reduce power loss cost by nearly 28%-57% and battery aging cost by 4%-15% compared to conventional methods, which can also enhance the SoC balance.

eess.SY↗