AI Paper Atlas

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

ICML 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Hamiltonian Monte Carlo · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add MSIS multi-state importance sampling for overlap estimation (ablation)Excited Pfaffian · 12-state C2 PES and 33-state Beoverlap-estimate variance / training stabilityImproved
use Excited Pfaffian parametrization instead of naive multi-state architecturenaive multi-state (Ndet=1) · 33-state Bestep time (s): 48 -> 7.2Improved
use Excited Pfaffian instead of naive architecture at equal capacitynaive Ndet=1 · 33-state Betotal energy per stepNo clear effect
increase determinants from Ndet=1 to 16Excited Pfaffian · 33-state Bestep time (s): 7.2 -> 8Mixed
train jointly across atoms rather than separatelyExcited Pfaffian vs NES · second-row atomstotal samples: 16B/64B -> 800MImproved
train jointly across molecules rather than separatelyExcited Pfaffian vs prior · 12 moleculestotal samples: 12B and 48B -> 800MImproved
use Excited Pfaffian instead of NES for per-step computeNES · C2 PESper-step speedup: > 200xImproved
use Excited Pfaffian instead of ground-state Neural PfaffianNeural Pfaffian · C2 near state crossing at 1.3 Aground-state energyImproved
use Excited Pfaffian for more states than prior neural VMCprior neural VMC · C2 PESstates modeled: 8 -> 12Improved
add bridge sampling for normalizing constant ratiosMSIS · 16 states, 8 moleculesstep-time overhead (ms): 0.83No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy