Selected work

Representative Papers

Quantum minimum description of density matrices

Oct 05, 2026

This study addresses the minimum memory cost of compressing multiple copies of a density matrix when its spectrum is known but its eigenbasis is not. The proposed approach integrates irreducible representations of GL(d,ℂ) with quantum information theory, constructing an achievable scheme by generalizing the Werner cloning map and establishing a matching lower bound via Koashi–Imoto incompressibility, with formal verification completed using the Lean proof assistant. This work precisely determines the minimal description length constant for fixed dimensions, revealing its deep connections to universal lossless coding overhead and free entropy. Furthermore, it derives finite trace-distance bounds and provides complete machine-checkable proof certificates.

1 citationsRead paper

Expert Coupling in MoE Pretraining: Reducing All-to-All Overhead with Correlated Placement and Token Shuffling

Oct 06, 2026

This study addresses the substantial All-to-All communication overhead in Mixture-of-Experts (MoE) expert parallelism, which accounts for 45%–60% of training step time. By observing significant intra-layer and inter-layer correlations in expert selection during early pre-training stages, this work proposes a routing-correlation-based communication optimization method. Specifically, correlation-aware expert placement strategies and token shuffling mechanisms are designed within the Megatron-LM framework to effectively reduce cross-GPU data transfers. Experimental results demonstrate that the proposed approach reduces All-to-All communication latency by 1.16× to 2.63× and achieves up to a 1.41× end-to-end training step speedup, substantially improving the distributed training efficiency of MoE models.

0 citationsRead paper

Continuity of Regularized Channel Rényi Divergences

Sep 23, 2026

This paper investigates the asymptotic properties of finite-dimensional quantum channel discrimination. It first establishes that the regularized sandwiched Rényi divergence converges to the relative entropy as the order approaches one. Subsequently, by integrating the hockey-stick divergence, Stinespring approximation, and Gour’s method, it derives exponential decay bounds under higher-order thresholds. The core contribution lies in elevating asymptotic bounds to an exponential strong converse theorem, thereby establishing a zero-one testing law and the asymptotic equipartition property (AEP) for subchannels. Furthermore, this work provides a unified characterization of exponential strong converse results and AEP across both parallel and adaptive channel discrimination scenarios.

0 citationsRead paper

When Recursive Models Finish Computing

Sep 22, 2026

研究解决了递归模型在名义推理预算内未完成计算的问题，通过延长计算步骤并分析潜态动态，提高了数独难题的解决率。

0 citationsRead paper

ZUNA1.1: A more flexible EEG foundation model for Denoising and Super-resolution

Jul 29, 2026

This work addresses the limited flexibility of existing EEG denoising and super-resolution methods with respect to sequence length, number of channels, electrode placement, and temporal segments. To overcome these constraints, the authors propose ZUNA1.1, a 380-million-parameter diffusion autoencoder that, for the first time, enables unified reconstruction of EEG signals under arbitrary missing patterns across temporal, channel, and spatial dimensions. The model supports variable-length inputs (up to 30 seconds), arbitrary numbers of channels, flexible electrode configurations, and can recover data from any temporal segment within a channel. Experimental results demonstrate that ZUNA1.1 matches or exceeds the performance of ZUNA1 across diverse reconstruction tasks and significantly outperforms conventional approaches such as spherical spline interpolation, exhibiting exceptional capabilities in both denoising and super-resolution. The code is publicly available under the Apache 2.0 license.

0 citationsRead paper

Recent publications

Latest Papers

Expert Coupling in MoE Pretraining: Reducing All-to-All Overhead with Correlated Placement and Token Shuffling

Oct 06, 2026

This study addresses the substantial All-to-All communication overhead in Mixture-of-Experts (MoE) expert parallelism, which accounts for 45%–60% of training step time. By observing significant intra-layer and inter-layer correlations in expert selection during early pre-training stages, this work proposes a routing-correlation-based communication optimization method. Specifically, correlation-aware expert placement strategies and token shuffling mechanisms are designed within the Megatron-LM framework to effectively reduce cross-GPU data transfers. Experimental results demonstrate that the proposed approach reduces All-to-All communication latency by 1.16× to 2.63× and achieves up to a 1.41× end-to-end training step speedup, substantially improving the distributed training efficiency of MoE models.

0 citationsRead paper

Quantum minimum description of density matrices

Oct 05, 2026

This study addresses the minimum memory cost of compressing multiple copies of a density matrix when its spectrum is known but its eigenbasis is not. The proposed approach integrates irreducible representations of GL(d,ℂ) with quantum information theory, constructing an achievable scheme by generalizing the Werner cloning map and establishing a matching lower bound via Koashi–Imoto incompressibility, with formal verification completed using the Lean proof assistant. This work precisely determines the minimal description length constant for fixed dimensions, revealing its deep connections to universal lossless coding overhead and free entropy. Furthermore, it derives finite trace-distance bounds and provides complete machine-checkable proof certificates.

1 citationsRead paper

Continuity of Regularized Channel Rényi Divergences

Sep 23, 2026

This paper investigates the asymptotic properties of finite-dimensional quantum channel discrimination. It first establishes that the regularized sandwiched Rényi divergence converges to the relative entropy as the order approaches one. Subsequently, by integrating the hockey-stick divergence, Stinespring approximation, and Gour’s method, it derives exponential decay bounds under higher-order thresholds. The core contribution lies in elevating asymptotic bounds to an exponential strong converse theorem, thereby establishing a zero-one testing law and the asymptotic equipartition property (AEP) for subchannels. Furthermore, this work provides a unified characterization of exponential strong converse results and AEP across both parallel and adaptive channel discrimination scenarios.

0 citationsRead paper

When Recursive Models Finish Computing

Sep 22, 2026

研究解决了递归模型在名义推理预算内未完成计算的问题，通过延长计算步骤并分析潜态动态，提高了数独难题的解决率。

0 citationsRead paper

ZUNA1.1: A more flexible EEG foundation model for Denoising and Super-resolution

Jul 29, 2026

This work addresses the limited flexibility of existing EEG denoising and super-resolution methods with respect to sequence length, number of channels, electrode placement, and temporal segments. To overcome these constraints, the authors propose ZUNA1.1, a 380-million-parameter diffusion autoencoder that, for the first time, enables unified reconstruction of EEG signals under arbitrary missing patterns across temporal, channel, and spatial dimensions. The model supports variable-length inputs (up to 30 seconds), arbitrary numbers of channels, flexible electrode configurations, and can recover data from any temporal segment within a channel. Experimental results demonstrate that ZUNA1.1 matches or exceeds the performance of ZUNA1 across diverse reconstruction tasks and significantly outperforms conventional approaches such as spherical spline interpolation, exhibiting exceptional capabilities in both denoising and super-resolution. The code is publicly available under the Apache 2.0 license.

0 citationsRead paper