Publications

Post-Selection-Free Quantum Automated Learning

Author: Wang, Junkai; Liu, Jin-Peng.
[First author.]

First-author preprint on arXiv (arXiv:2610.08219).

Summary: A post-selection-free quantum automated learning algorithm that implements a complete training path coherently and uses fixed-point amplitude amplification to obtain the learned model with high probability. The analysis provides output-error and learning-loss guarantees and identifies conditions under which coherent resources reduce the cost of restarting failed trajectories.

Recommended citation: Wang, Junkai; Liu, Jin-Peng. (2026). "Post-Selection-Free Quantum Automated Learning." arXiv. arXiv:2610.08219.
Article Link

Accelerating Quantum Dense-Output Simulation through Locality

Author: Yang, Songqinghao; Liu, Chang; Wang, Junkai; Zhang, Jingxuan; Liu, Jin-Peng.

Preprint on arXiv (arXiv:2609.40094).

Summary: A locality-based truncation framework for estimating time-integrated observables in lattice quantum simulations. Lieb-Robinson bounds control spatial truncation error and yield local-simulation query bounds for fermionic, bosonic, and Lindbladian systems, supported by applications and numerical experiments.

Recommended citation: Yang, Songqinghao; Liu, Chang; Wang, Junkai; Zhang, Jingxuan; Liu, Jin-Peng. (2026). "Accelerating Quantum Dense-Output Simulation through Locality." arXiv. arXiv:2609.40094.
Article Link

ScienceIDE: Turning World’s Scientific Codebase into Agent Learnable Environments

Author: Geng, Hejia; Huang, Zesen; Li, Haoyang; Li, Wenbin; Wu, Koutian; Zhou, Zihan; Pang, Yuanbo; Liu, Weihao; Xu, Zigong; Li, Zhiping; Zhang, Zongzheng; Dong, Chuanfei; Sun, Jiankai; Zheng, Tianzhe; Xie, Fengyu; Ma, Yue; Shi, Yueheng; Xie, Tong; Di, Zonglin; Liu, Xianrong; Gao, Qucheng; Liu, Yimin; Pan, Jiaming; Huang, Sheng; Ma, Xiao-Han; Yuan, Lanqing; Zhu, Zhenlin; Liu, Ziang; Xu, Ziyang; Wang, Junkai; Liang, Kangkai; Xian, Jiayi; Zhao, Zehong; Xu, Liuwei; Xie, Jingxu; Zhang, Peijin; Gao, Qiang; Xing, Chengyi; Zhao, Zhe; Wang, Xi; Xing, Yaopeng; Meng, Xing; Yin, Zhenfei; Wu, Yingcheng; Yang, Ling.

Preprint on arXiv (arXiv:2609.19134).

Summary: Infrastructure that converts scientific code repositories into executable learning environments with expert-defined cases and scientific verification. Verified interaction trajectories support supervised fine-tuning, reinforcement learning, and evaluation, with reported improvements in held-out scientific-code repair and selected general-purpose tasks.

Recommended citation: Geng, Hejia et al. (2026). "ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments." arXiv. arXiv:2609.19134.
Article Link

Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation

Author: Wu, Koutian; Zhou, Junjie; Shang, Ergan; Wang, Jiayu; Han, Pengqian; Wang, Junkai; Xu, Wanghan; Lu, Songyuanyi; Shi, Lin.

Preprint on arXiv (arXiv:2609.11115).

Summary: A continuously updated database and search engine for AI benchmarks, combining daily discovery with a searchable catalog, model-card references, and score histories. Source identities and citations support inspection of evaluation evidence, benchmark adoption, saturation, and the limits of score comparisons.

Recommended citation: Wu, Koutian et al. (2026). "Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation." arXiv. arXiv:2609.11115.
Article Link

LeanCat: A Benchmark Suite for Formal Category Theory in Lean (Part I: 1-Categories)

Author: Xu, Rongge; Dai, Hui; Fu, Yiming; Jiang, Jiedong; Nie, Tianjiao; Wang, Hongwei; Wang, Junkai; Yang, Holiverse; Yang, Jiatong; Zhang, Zhi-Hao.
[Equal contributions; authors listed by surname.]

Preprint on arXiv (arXiv:2512.24796); preparing submission to ICML 2026.

Summary: First Lean benchmark dedicated to formal category theory: Statement-level Lean tasks across 8 clusters (e.g., adjunctions, limits-colimits, abelian categories, monads) to stress-test abstraction-heavy, library-mediated reasoning; tough baselines (8.25% pass@1 / 12.00% pass@4) expose the natural-to-formal bottleneck, and LeanBridge + Lean-Explore (retrieve-generate-verify with compiler feedback) improves performance on a subset.

Recommended citation: Xu, Rongge; Dai, Hui; Fu, Yiming; Jiang, Jiedong; Nie, Tianjiao; Wang, Hongwei; Wang, Junkai; Yang, Holiverse; Yang, Jiatong; Zhang, Zhi-Hao. (2025). "LeanCat: A Benchmark Suite for Formal Category Theory in Lean (Part I: 1-Categories)." arXiv. arXiv:2512.24796. Preparing submission to ICML 2026.
Article Link

Quantum Gravity Meets DESI: Dynamical Dark Energy in Light of the Trans-Planckian Censorship Conjecture

Author: Li, Chunyu; Wang, Junkai; Zhang, Dongdong; Saridakis, Emmanuel N.; Cai, Yi-Fu.
[Co-first author responsible for the original idea and the theory; another responsible for the numerics.]

Published in Journal of Cosmology and Astroparticle Physics, 2025

Summary: Using DESI DR2 hints of evolving dark energy, the work shows alignment with the Swampland Transplanckian Censorship Conjecture and uses swampland TCC criteria to constrain EoS parameterizations (CPL/BA/JBP/EXP/LOG) and test modified gravity, implying viable scenarios must asymptotically transition to deceleration.

Recommended citation: Li, Chunyu; Wang, Junkai; Zhang, Dongdong; Saridakis, Emmanuel N.; Cai, Yi-Fu. (2025). "Quantum Gravity Meets DESI: Dynamical Dark Energy in Light of the Trans-Planckian Censorship Conjecture." Journal of Cosmology and Astroparticle Physics.
Article Link

Development of Gravity Theories in the View of TRAPPIST-1e

Author: Wang, Nan; Lu, Lu-Yao; Liu, Hui-Gen; Chen, An-Dong; Lu, Tiger; Cui, Ao-Ran; Wang, Jun-Kai.

Published in Research in Astronomy and Astrophysics, 2025

ISSN: 1674-4527; 2397-6209.

Summary: From a TRAPPIST-1e observer’s viewpoint, N-body integrations generate ephemerides and observable sky phenomena to assess what “discovering gravity” looks like in a compact exoplanet system; Keplerian regularities are readily testable, while relativistic effects (e.g., perihelion precession) are too small to stand out, suggesting GR would be hard to infer locally.

Recommended citation: Wang, Nan; Lu, Lu-Yao; Liu, Hui-Gen; Chen, An-Dong; Lu, Tiger; Cui, Ao-Ran; Wang, Jun-Kai. (2025). "Development of Gravity Theories in the View of TRAPPIST-1e." Research in Astronomy and Astrophysics. 25(1):015003.
Article Link

Form & Function: A Linguistic-Semiotic Investigation on Modern Physics

Author: Wang, Junkai. [Sole author.]

Published in 5th National Symposium on Philosophy of Physics, 2021

Conference paper presented at the 5th National Symposium on Philosophy of Physics, China.

Summary: As the youngest speaker to deliver a 30-min plenary talk, in this article we develop a compact framework for physics discourse using linguistic semiotics—Peirce’s sign–interpretant–object triad, Saussurean oppositions, and Hallidayan Systemic Functional Linguistics with multimodality—to model modern physics language as a structured, multimodal sign system. It operationalizes the framework via (i) a semiotic comparison of interpretations of quantum mechanics and (ii) an SFL/discourse analysis of Einstein’s 1905 On the Electrodynamics of Moving Bodies, yielding a reusable template for analyzing and standardizing scientific writing across languages.

Recommended citation: Wang, Junkai. (2021). "Form & Function: A Linguistic-Semiotic Investigation on Modern Physics." Conference paper, 5th National Symposium on Philosophy of Physics (China).