arXiv subjects

Jinyan Chen

Publications and source records attributed to Jinyan Chen.

7 recordsLinked to original sources

arXiv2026

X-Stage: An Overlooked Pipeline Stage for Communication-Computation Overlap in DiT Inference

Fine-grained, device-initiated communication lets persistent GPU kernels in distributed diffusion transformer (DiT) inference issue remote stores and overlap data movement with Tensor Core computation. Existing systems schedule when communication is issued and when received data becomes consumable, but omit post-issue progress before remote-visible completion, making sender backpressure hard to predict. We identify X-Stage, a software-visible post-issue pipeline stage. Measurements on an eight-GPU node with a recent NVIDIA architecture show that short remote-store bursts drain as the issuer resumes work, whereas sustained injection exhausts finite outstanding capacity and delays later issues. A lightweight Burst-Gap model parameterized by backpressure-free issue time, effective drain rate, and outstanding capacity predicts issue overhead, recovery between bursts, and the onset of backpressure. Guided by the model, we redesign two communication-computation fused kernels. For DeepGEMM MegaMoE, interleaving Linear-1 and Linear-2 work across expert waves places computation between concentrated remote-store bursts, yielding a 1.18x geometric-mean and 1.62x maximum kernel speedup over the Expert-Wave baseline across 84 configurations. For Ulysses sequence-parallel attention, tile-granular fusion of the post-attention All-to-All with FlashAttention lets an output-tile owner issue remote stores and resume computation without a dedicated communication warp or streaming multiprocessor. FlashAttention-3 and FlashAttention-4 reach maximum sender-visible speedups of 1.43x and 1.42x over serial execution, and at long sequences their steady-state times approach those of FlashAttention alone. These results establish post-issue progress as a measurable scheduling lever for shaping bursts, avoiding backpressure, and hiding sender-side overhead.

cs.DC↗

arXiv2026

Experimental Tabletop Petz recovery of a photonic qubit

The quantum information lost in open evolutions cannot be fully recovered, but partial recovery is possible. The Petz recovery map guarantees almost optimal recovery, notably if the chosen reference state is close to the real one. This map has been widely used in theoretical studies, but has been the object of only a handful of experimental realisations, typically under a single fixed noise model. In this work, we describe and implement the Petz recovery map for a versatile class of qubit channels with tunable decoherence and dissipation. The setup we realize is also the first experimental example of ``tabletop reversibility'': for a good range of choices of the reference state, the Petz recovery map can be implemented with the same devices as the forward dissipative evolution, whose effect it is partially undoing. Our results demonstrate that the Petz recovery map can be resource-efficiently realized without requiring complex ancillary resources, providing a feasible pathway for mitigating information loss in quantum systems.

quant-ph↗

arXiv2025

The Petz recovery map for optical losses

Optical systems are a main platform for quantum information processing. A main challenge is information loss due to scattering in unmonitored modes. These losses are modeled as state-independent beam-splitter interactions, with a thermal state (for all practical purposes, the vacuum) in the second input port. The perfect correction of these Gaussian lossy channels with Gaussian operations alone is known to be impossible. In this work, we investigate the Petz recovery map as an approximate recovery. For single mode losses and Gaussian reference states, the Petz map is found to use either a beam-splitter or a state-independent amplifier, depending on the parameters. Then we study the recovery performance on several examples, showing that it is near-optimal among the considered class of protocols. We also obtain more specific comparisons: Petz is always better than just re-preparing the reference state; but it is worse than doing nothing if the reference state is far from the true state. Finally, we extend our study to losses on two modes, and compare the global Petz map to the local implementation on each mode separately.

quant-ph↗

arXiv2025

GSM8K-V: Can Vision Language Models Solve Grade School Math Word Problems in Visual Contexts

Mathematical reasoning is a key capability for vision-language models (VLMs), yet current benchmarks mainly evaluate text-based or explicitly symbolic visual inputs. It remains unclear whether VLMs can reason mathematically when information must be perceived and inferred from images rather than read from explicit symbols. We introduce GSM8K-V, a benchmark transforming GSM8K into multi-image sequences with semantic equivalence preserved. By mapping text-based problems into visual form via an automated pipeline and human verification, we curate 1,319 high-quality samples. In GSM8K-V, quantities must be extracted through visual perception, and reasoning chains must be reconstructed by integrating implicit cues across scenes. Evaluation of 34 VLMs reveals a striking modality gap: while most models exceed 90\% on text, the best model achieves only 59\% on GSM8K-V, far below the 91\% human accuracy. Notably, models enhanced for visual math reasoning show no improvement on GSM8K-V despite large gains on existing benchmarks, confirming that it evaluates a distinct capability. Error analysis shows that the primary bottleneck lies in Implicit Visual Inference Error (IVIE), where models fail to recover visual semantics that are implied rather than explicitly stated. Our code and data are released at https://github.com/ZJU-REAL/GSM8K-V.

cs.CV↗

arXiv2025

STAR-Pose: Efficient Low-Resolution Video Human Pose Estimation via Spatial-Temporal Adaptive Super-Resolution

Human pose estimation in low-resolution videos presents a fundamental challenge in computer vision. Conventional methods either assume high-quality inputs or employ computationally expensive cascaded processing, which limits their deployment in resource-constrained environments. We propose STAR-Pose, a spatial-temporal adaptive super-resolution framework specifically designed for video-based human pose estimation. Our method features a novel spatial-temporal Transformer with LeakyReLU-modified linear attention, which efficiently captures long-range temporal dependencies. Moreover, it is complemented by an adaptive fusion module that integrates parallel CNN branch for local texture enhancement. We also design a pose-aware compound loss to achieve task-oriented super-resolution. This loss guides the network to reconstruct structural features that are most beneficial for keypoint localization, rather than optimizing purely for visual quality. Extensive experiments on several mainstream video HPE datasets demonstrate that STAR-Pose outperforms existing approaches. It achieves up to 5.2% mAP improvement under extremely low-resolution (64x48) conditions while delivering 2.8x to 4.4x faster inference than cascaded approaches.

cs.CV↗

arXiv2024

Even-parity precession protocol for detecting nonclassicality and entanglement

We introduce an even-parity precession protocol that can detect the nonclassicality of some quantum states using only measurements of a uniformly-precessing variable at different points in time. Depending on the system under study, the protocol may detect the Wigner negativity of a single quantum harmonic oscillator or of a single spin $j\geq 2$; the non-Gaussian entanglement of two harmonic oscillators; or genuine multipartite entanglement of a spin ensemble, whose total spin is an integer. Unlike in other nonclassicality tests, simultaneous or sequential measurements are not required. Our protocol can also detect states that commute with the parity operator, which were missed by similar protocols built from Tsirelson's original precession protocol. This work also closes a long-standing gap by showing the possibility of detecting the Greenberger-Horne-Zeilinger entanglement of an even number of qubits using only collective spin measurements.

quant-ph↗

arXiv2021

Synthesizing five-body interaction in a superconducting quantum circuit

Synthesizing many-body interaction Hamiltonian is a central task in quantum simulation. However, it is challenging to synthesize interactions including more than two spins. Borrowing tools from quantum optics, we synthesize five-body spin-exchange interaction in a superconducting quantum circuit by simultaneously exciting four independent qubits with time-energy correlated photon quadruples generated from a qudit. During the dynamic evolution of the five-body interaction, a Greenberger-Horne-Zeilinger state is generated in a single step with fidelity estimated to be $0.685$. We compare the influence of noise on the three-, four- and five-body interaction as a step toward answering the question on the quantum origin of chiral molecules. We also demonstrate a many-body Mach-Zehnder interferometer which potentially has a Heisenberg-limit sensitivity. This study paves a way for quantum simulation involving many-body interactions and high excited states of quantum circuits.

quant-ph↗