AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement

Jiangjie Qiu, Wentao Li, Honghao Chen, Yi Shen Tew, Leyi Zhao, Xiaonan Wang

Submission Type: I want my submission to be considered for both oral and poster presentation.

Keywords: adsorbate placement, flow matching, rigid-body generation, OC20-Dense, catalysis

TL;DR: AdsorbFlow is an energy-conditioned rigid-body flow-matching model for adsorbate placement. Under OC20-Dense, it reaches 45.5% DFT success at N=1 and 61.4% at N=10 with 5-step sampling.

Confirmation Of Submission Requirements: I submit an abstract. It uses the template provided on the submission page and is no longer than 2 pages.

Submission Number: 187