![[2602.19289] AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19289] AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement
Summary
The paper introduces AdsorbFlow, a deterministic generative model that enhances the efficiency of adsorbate placement on catalytic surfaces by significantly reducing the number of iterative steps required compared to existing methods.
Why It Matters
This research addresses a critical challenge in computational heterogeneous catalysis, where identifying low-energy adsorption geometries is essential. By improving the speed and accuracy of adsorbate placement, AdsorbFlow could facilitate advancements in catalysis research and applications, potentially leading to more efficient chemical processes.
Key Takeaways
- AdsorbFlow achieves faster and more accurate adsorbate placement using only 5 iterative steps.
- The model outperforms previous methods like AdsorbDiff and AdsorbML in success rates and anomaly rates.
- Energy information is integrated through classifier-free guidance, enhancing the model's efficiency.
- AdsorbFlow retains high performance even on out-of-distribution systems, indicating robustness.
- This approach could significantly impact the field of computational heterogeneous catalysis.
Computer Science > Machine Learning arXiv:2602.19289 (cs) [Submitted on 22 Feb 2026] Title:AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement Authors:Jiangjie Qiu, Wentao Li, Honghao Chen, Leyi Zhao, Xiaonan Wang View a PDF of the paper titled AdsorbFlow: energy-conditioned flow matching enables fast and realistic adsorbate placement, by Jiangjie Qiu and 4 other authors View PDF HTML (experimental) Abstract:Identifying low-energy adsorption geometries on catalytic surfaces is a practical bottleneck for computational heterogeneous catalysis: the difficulty lies not only in the cost of density functional theory (DFT) but in proposing initial placements that relax into the correct energy basins. Conditional denoising diffusion has improved success rates, yet requires $\sim$100 iterative steps per sample. Here we introduce AdsorbFlow, a deterministic generative model that learns an energy-conditioned vector field on the rigid-body configuration space of adsorbate translation and rotation via conditional flow matching. Energy information enters through classifier-free guidance conditioning -- not energy-gradient guidance -- and sampling reduces to integrating an ODE in as few as 5 steps. On OC20-Dense with full DFT single-point verification, AdsorbFlow with an EquiformerV2 backbone achieves 61.4% SR@10 and 34.1% SR@1 -- surpassing AdsorbDiff (31.8% SR@1, 41.0% SR@10) at every evaluation level and AdsorbML (47.7% SR@10) -- while using 20 ...
