Research map / Probabilistic and scientific modeling / Variational and generative sampling
Excited Pfaffians: Generalized Neural Wave Functions Across Structure and State
Nicholas Gao, Till Grutschus, Frank Noe, Stephan Günnemann
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add MSIS multi-state importance sampling for overlap estimation (ablation) | Excited Pfaffian · 12-state C2 PES and 33-state Be | overlap-estimate variance / training stability | Improved |
| use Excited Pfaffian parametrization instead of naive multi-state architecture | naive multi-state (Ndet=1) · 33-state Be | step time (s): 48 -> 7.2 | Improved |
| use Excited Pfaffian instead of naive architecture at equal capacity | naive Ndet=1 · 33-state Be | total energy per step | No clear effect |
| increase determinants from Ndet=1 to 16 | Excited Pfaffian · 33-state Be | step time (s): 7.2 -> 8 | Mixed |
| train jointly across atoms rather than separately | Excited Pfaffian vs NES · second-row atoms | total samples: 16B/64B -> 800M | Improved |
| train jointly across molecules rather than separately | Excited Pfaffian vs prior · 12 molecules | total samples: 12B and 48B -> 800M | Improved |
| use Excited Pfaffian instead of NES for per-step compute | NES · C2 PES | per-step speedup: > 200x | Improved |
| use Excited Pfaffian instead of ground-state Neural Pfaffian | Neural Pfaffian · C2 near state crossing at 1.3 A | ground-state energy | Improved |
| use Excited Pfaffian for more states than prior neural VMC | prior neural VMC · C2 PES | states modeled: 8 -> 12 | Improved |
| add bridge sampling for normalizing constant ratios | MSIS · 16 states, 8 molecules | step-time overhead (ms): 0.83 | No clear effect |
Closest papers by meaning
- Neural Pfaffians: Solving Many Many-Electron Schrödinger Equations (NeurIPS 2024 · similarity 0.92)
- Learning the Electronic Hamiltonian of Large Atomic Structures (ICML 2025 · similarity 0.86)
- Enhancing the Scalability and Applicability of Kohn-Sham Hamiltonians for Molecular Systems (ICLR 2025 · similarity 0.85)
- Gaussian Plane-Wave Neural Operator for Electron Density Estimation (ICML 2024 · similarity 0.85)
- A Theoretical Framework for an Efficient Normalizing Flow-Based Solution to the Electronic Schrödinger Equation (AAAI 2025 · similarity 0.84)
- The Importance of Being Scalable: Improving the Speed and Accuracy of Neural Network Interatomic Potentials Across Chemical Domains (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
