Towards Accelerated SCF Workflows with Equivariant Density-Matrix Learning and Analytic Refinement
📰 ArXiv cs.AI
Learn to accelerate SCF workflows using equivariant density-matrix learning and analytic refinement with dm-PhiSNet
Action Steps
- Build a dm-PhiSNet model using the PhiSNet architecture and equivariant constraints
- Train the model in two stages with physically motivated objectives
- Predict one-electron reduced density matrices (1-RDMs) from molecular geometries using the trained model
- Refine the predicted 1-RDMs with a lightweight analytic method
- Apply the refined 1-RDMs to accelerate SCF workflows
Who Needs to Know This
Quantum chemists and materials scientists can benefit from this approach to speed up their SCF workflows, while machine learning engineers can apply the equivariant learning techniques to other physics-based problems
Key Insight
💡 Equivariant learning can be used to predict density matrices directly from molecular geometries, accelerating SCF workflows
Share This
Accelerate SCF workflows with equivariant density-matrix learning and analytic refinement using dm-PhiSNet! #AI #QuantumChemistry
Key Takeaways
Learn to accelerate SCF workflows using equivariant density-matrix learning and analytic refinement with dm-PhiSNet
Full Article
Title: Towards Accelerated SCF Workflows with Equivariant Density-Matrix Learning and Analytic Refinement
Abstract:
arXiv:2604.27256v1 Announce Type: cross Abstract: We present \textsc{dm-PhiSNet}, a physically constrained \textsc{PhiSNet}-based equivariant model that predicts one-electron reduced density matrices (1-RDMs) directly from molecular geometries in an atomic-orbital (AO) basis for accelerated self-consistent field (SCF) workflows. Training follows a two-stage schedule with progressively introduced physically motivated objectives, and the resulting predictions are refined by a lightweight analytic
Abstract:
arXiv:2604.27256v1 Announce Type: cross Abstract: We present \textsc{dm-PhiSNet}, a physically constrained \textsc{PhiSNet}-based equivariant model that predicts one-electron reduced density matrices (1-RDMs) directly from molecular geometries in an atomic-orbital (AO) basis for accelerated self-consistent field (SCF) workflows. Training follows a two-stage schedule with progressively introduced physically motivated objectives, and the resulting predictions are refined by a lightweight analytic
DeepCamp AI