# Zhanghan Wang

Research homepage for Zhanghan Wang, a PhD student at New York University studying high-performance and reliable systems.

## Research

- [Research homepage](https://rabbitwhite1.github.io/): Publications, research projects, and experience.
- [Entangle: Formal Verification for Distributed Deep Learning](https://rabbitwhite1.github.io/projects/entangle/): Static verification of distributed model correctness, including tensor-parallel implementations. Entangle checks model refinement by inferring how sequential reference outputs can be reconstructed from distributed outputs, using iterative rewriting and equality saturation. Its outputs help localize correctness bugs.
- [Complete relation inference walkthrough](https://rabbitwhite1.github.io/projects/entangle/walkthrough.html): Static explanation of every demo step, from input relations through the final output relation, and Appendix A.1’s Bug 4 with missing off-diagonal products and a numerical counterexample.

## Papers

- [It Takes Two to Entangle](https://rabbitwhite1.github.io/projects/entangle/): Zhanghan Wang, Ding Ding, Hang Zhu, Haibin Lin, and Aurojit Panda. ASPLOS 2026. Published March 22, 2026.
- [DOI / ACM publication](https://doi.org/10.1145/3779212.3790178)
- [Paper PDF](https://rabbitwhite1.github.io/projects/entangle/paper.pdf)
- [Older version on arXiv](https://arxiv.org/abs/2508.09505)
- [BibTeX](https://rabbitwhite1.github.io/projects/entangle/entangle.bib)
- [Entangle code and artifact](https://github.com/nyu-systems/Entangle)