Yu Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Yu Zhang has a total of 169 co-authors. Here are the top co-authors by number of shared publications:
- Jianmin Ji
University of Science and Technology of China, Hefei, China
18 shared publications - Yanyong Zhang
University of Science and Technology of China, Hefei, Anhui, China
18 shared publications - Hanqi Zhu
University of Science and Technology of China, Hefei, China
5 shared publications Haowei Deng
School of Computer Science and Technology, University of Science and Technology of China, Hefei, China
5 shared publications- Houqiang Li
University of Science and Technology of China, Hefei, China
5 shared publications Ming Fu
University of Science and Technology of China
5 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5245-5258
Optimizing Unitary Coupled Cluster With Single and Double Kernels for Modern NISQ Architectures
The Unitary Coupled Cluster with Singles and Doubles (UCCSD) ansatz is a widely used kernel in quantum chemistry and variational quantum eigensolver (VQE) algorithms, but its execution on near-term superconducting quantum processors is hindered by high gate costs and limited qubit connectivity. Existing compilation approaches either adopt simultaneous synthesis, which efficiently diagonalizes excitations but ignores connectivity, or individual synthesis, which enforces connectivity-awareness but sacrifices optimization opportunities. In this work, we propose a connectivity-aware compiler framework for UCCSD kernels that combines the efficiency of simultaneous synthesis with the practicality of connectivity constraints. First, we introduce blockwise qubit mapping, which groups program qubits into excitation-specific blocks and maps them to internally connected regions of the hardware graph. Second, we design connectivity-aware synthesis rules that diagonalize excitations directly on the mapped hardware graph, reducing the number of SWAP gates. We implement our framework and evaluate it on molecular and random UCCSD kernels using Google Sycamore’s and IBMQ Manhattan’s connectivity. Compared with state-of-the-art compilers, our approach reduces CNOT count by up to 34% and circuit depth by 49%, while improving VQE accuracy under noise.
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 45 (11) , 5245-5258
Optimizing Unitary Coupled Cluster With Single and Double Kernels for Modern NISQ Architectures
The Unitary Coupled Cluster with Singles and Doubles (UCCSD) ansatz is a widely used kernel in quantum chemistry and variational quantum eigensolver (VQE) algorithms, but its execution on near-term superconducting quantum processors is hindered by high gate costs and limited qubit connectivity. Existing compilation approaches either adopt simultaneous synthesis, which efficiently diagonalizes excitations but ignores connectivity, or individual synthesis, which enforces connectivity-awareness but sacrifices optimization opportunities. In this work, we propose a connectivity-aware compiler framework for UCCSD kernels that combines the efficiency of simultaneous synthesis with the practicality of connectivity constraints. First, we introduce blockwise qubit mapping, which groups program qubits into excitation-specific blocks and maps them to internally connected regions of the hardware graph. Second, we design connectivity-aware synthesis rules that diagonalize excitations directly on the mapped hardware graph, reducing the number of SWAP gates. We implement our framework and evaluate it on molecular and random UCCSD kernels using Google Sycamore’s and IBMQ Manhattan’s connectivity. Compared with state-of-the-art compilers, our approach reduces CNOT count by up to 34% and circuit depth by 49%, while improving VQE accuracy under noise.
- 10 September 2026
- Published by Springer Nature
- p. 147-164
Perceptual Projection Pruning: Diversity-Aware Video Token Pruning for Multimodal Large Language Models
- 10 September 2026
- Published by Springer Nature
- p. 147-164
Perceptual Projection Pruning: Diversity-Aware Video Token Pruning for Multimodal Large Language Models
- 1 May 2026
- Published by Springer Nature
- p. 599-611
Toward System- and Theory-Oriented Talent Cultivation in Computing
- 1 May 2026
- Published by Springer Nature
- p. 599-611
Toward System- and Theory-Oriented Talent Cultivation in Computing
- Open Access
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 938-952
UrgenGo: Urgency-Aware Transparent GPU Kernel Launching for Autonomous Driving
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 938-952
Open AccessUrgenGo: Urgency-Aware Transparent GPU Kernel Launching for Autonomous Driving
- 13 September 2025
- Published by Elsevier in Journal of Systems and Software
- Vol. 231
An empirical study of CGO usage in Go projects — Distribution, purposes, patterns and critical issues
- 13 September 2025
- Published by Elsevier in Journal of Systems and Software
- Vol. 231
An empirical study of CGO usage in Go projects — Distribution, purposes, patterns and critical issues
- Open Access
- 1 September 2025
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 5, 4570-4578
Improving Efficiency of Answer Set Planning with Rough Solutions from Large Language Models for Robotic Task Planning
Answer Set Programming (ASP) planning can be used to refine the rough solutions generated by Large Language Models (LLMs) to handle specific restrictions of actions, i.e., reconstruct the rough solutions to be executable, for robotic task planning. However, it is still challenging to efficiently solve ASP programs that have multiple variables with large domains, which prevents the above application of ASP planning from real-world task planning problems. In this paper, we consider how to reduce the domains of variables without losing possible solutions for ASP planning, while given these rough solutions from LLMs. Based on the above reduction, we introduce CLMASP, an approach that couples LLMs with ASP for robotic task planning. We evaluate CLMASP on the VirtualHome platform for common indoor tasks, demonstrating a significant improvement in the executable rate from under 10% to nearly 90% and reducing average ASP planning time from over 2 hours to under 5 seconds. Code is available at https://github.com/CLMASP/CLMASP.
- 1 September 2025
- Published by International Joint Conferences on Artificial Intelligence
- Vol. 5, 4570-4578
Open AccessImproving Efficiency of Answer Set Planning with Rough Solutions from Large Language Models for Robotic Task Planning
Answer Set Programming (ASP) planning can be used to refine the rough solutions generated by Large Language Models (LLMs) to handle specific restrictions of actions, i.e., reconstruct the rough solutions to be executable, for robotic task planning. However, it is still challenging to efficiently solve ASP programs that have multiple variables with large domains, which prevents the above application of ASP planning from real-world task planning problems. In this paper, we consider how to reduce the domains of variables without losing possible solutions for ASP planning, while given these rough solutions from LLMs. Based on the above reduction, we introduce CLMASP, an approach that couples LLMs with ASP for robotic task planning. We evaluate CLMASP on the VirtualHome platform for common indoor tasks, demonstrating a significant improvement in the executable rate from under 10% to nearly 90% and reducing average ASP planning time from over 2 hours to under 5 seconds. Code is available at https://github.com/CLMASP/CLMASP.
- 25 August 2025
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 133 (11) , 8022-8040
OA-DET3D: Embedding Object Awareness As A General Plug-in for Multi-Camera 3D Object Detection
- 25 August 2025
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 133 (11) , 8022-8040
OA-DET3D: Embedding Object Awareness As A General Plug-in for Multi-Camera 3D Object Detection
- Open Access
An Empirical Study of CGO Usage in Go Projects -- Distribution, Purposes, Patterns and Critical Issues
Multilingual software development integrates multiple languages into a single application, with the Foreign Function Interface (FFI) enabling seamless interaction. While FFI boosts efficiency and extensibility, it also introduces risks. Existing studies focus on FFIs in languages like Python and Java, neglecting CGO, the emerging FFI in Go, which poses unique risks. To address these concerns, we conduct an empirical study of CGO usage across 920 open-source Go projects. Our study aims to reveal the distribution, patterns, purposes, and critical issues associated with CGO, offering insights for developers and the Go team. We develop CGOAnalyzer, a tool to efficiently identify and quantify CGO-related features. Our findings reveal that: (1) 11.3% of analyzed Go projects utilize CGO, with usage concentrated in a subset of projects; (2) CGO serves 4 primary purposes, including system-level interactions and performance optimizations, with 15 distinct usage patterns observed; (3) 19 types of CGO-related issues exist, including one critical issue involving unnecessary pointer checks that pose risks of runtime crashes due to limitations in the current Go compilation toolchain; (4) a temporary solution reduces unnecessary pointer checks, mitigating crash risks, and (5) we submitted a proposal to improve the Go toolchain for a permanent fix, which has been grouped within an accepted proposal for future resolution. Our findings provide valuable insights for developers and the Go team, enhancing development efficiency and reliability while improving the robustness of the Go toolchain.
Open AccessAn Empirical Study of CGO Usage in Go Projects -- Distribution, Purposes, Patterns and Critical Issues
Multilingual software development integrates multiple languages into a single application, with the Foreign Function Interface (FFI) enabling seamless interaction. While FFI boosts efficiency and extensibility, it also introduces risks. Existing studies focus on FFIs in languages like Python and Java, neglecting CGO, the emerging FFI in Go, which poses unique risks. To address these concerns, we conduct an empirical study of CGO usage across 920 open-source Go projects. Our study aims to reveal the distribution, patterns, purposes, and critical issues associated with CGO, offering insights for developers and the Go team. We develop CGOAnalyzer, a tool to efficiently identify and quantify CGO-related features. Our findings reveal that: (1) 11.3% of analyzed Go projects utilize CGO, with usage concentrated in a subset of projects; (2) CGO serves 4 primary purposes, including system-level interactions and performance optimizations, with 15 distinct usage patterns observed; (3) 19 types of CGO-related issues exist, including one critical issue involving unnecessary pointer checks that pose risks of runtime crashes due to limitations in the current Go compilation toolchain; (4) a temporary solution reduces unnecessary pointer checks, mitigating crash risks, and (5) we submitted a proposal to improve the Go toolchain for a permanent fix, which has been grouped within an accepted proposal for future resolution. Our findings provide valuable insights for developers and the Go team, enhancing development efficiency and reliability while improving the robustness of the Go toolchain.
- Open Access\(X\)-evolve: Solution space evolution powered by large language modelsWhile combining large language models (LLMs) with evolutionary algorithms (EAs) shows promise for solving complex optimization problems, current approaches typically evolve individual solutions, often incurring high LLM call costs. We introduce \(X\)-evolve, a paradigm-shifting method that instead evolves solution spaces \(X\) (sets of individual solutions) - subsets of the overall search space \(S\). In \(X\)-evolve, LLMs generate tunable programs wherein certain code snippets, designated as parameters, define a tunable solution space. A score-based search algorithm then efficiently explores this parametrically defined space, guided by feedback from objective function scores. This strategy enables broader and more efficient exploration, which can potentially accelerate convergence at a much lower search cost, requiring up to two orders of magnitude fewer LLM calls than prior leading methods. We demonstrate \(X\)-evolve's efficacy across three distinct hard optimization problems. For the cap set problem, we discover a larger partial admissible set, establishing a new tighter asymptotic lower bound for the cap set constant (\(C \ge 2.2203\)). In information theory, we uncover a larger independent set for the 15-vertex cycle graph (\(\mathcal{C}_{15}^{\boxtimes 5}\), size 19,946), thereby raising the known lower bound on its Shannon capacity. Furthermore, for the NP-hard online bin packing problem, we generate heuristics that consistently outperform standard strategies across established benchmarks. By evolving solution spaces, our method considerably improves search effectiveness, making it possible to tackle high-dimensional problems that were previously computationally prohibitive.Open Access\(X\)-evolve: Solution space evolution powered by large language modelsWhile combining large language models (LLMs) with evolutionary algorithms (EAs) shows promise for solving complex optimization problems, current approaches typically evolve individual solutions, often incurring high LLM call costs. We introduce \(X\)-evolve, a paradigm-shifting method that instead evolves solution spaces \(X\) (sets of individual solutions) - subsets of the overall search space \(S\). In \(X\)-evolve, LLMs generate tunable programs wherein certain code snippets, designated as parameters, define a tunable solution space. A score-based search algorithm then efficiently explores this parametrically defined space, guided by feedback from objective function scores. This strategy enables broader and more efficient exploration, which can potentially accelerate convergence at a much lower search cost, requiring up to two orders of magnitude fewer LLM calls than prior leading methods. We demonstrate \(X\)-evolve's efficacy across three distinct hard optimization problems. For the cap set problem, we discover a larger partial admissible set, establishing a new tighter asymptotic lower bound for the cap set constant (\(C \ge 2.2203\)). In information theory, we uncover a larger independent set for the 15-vertex cycle graph (\(\mathcal{C}_{15}^{\boxtimes 5}\), size 19,946), thereby raising the known lower bound on its Shannon capacity. Furthermore, for the NP-hard online bin packing problem, we generate heuristics that consistently outperform standard strategies across established benchmarks. By evolving solution spaces, our method considerably improves search effectiveness, making it possible to tackle high-dimensional problems that were previously computationally prohibitive.