GPR model predicts record molecular dipole moments
Stony Brook researchers developed a Gaussian Process Regression (GPR) model that predicts molecular dipole moments with less than 5% error. The AI identifies record-high dipoles in seconds, replacing months of traditional quantum chemistry simulations.
AI is making brute-force quantum chemistry calculations obsolete for molecular screening by turning months of simulation into seconds of prediction. The GPR model replaces CCSD(T) quantum chemistry calculations that take weeks with predictions delivered in seconds, maintaining an impressive error rate under 5%. By identifying unexpected high dipoles in alkali-noble metal combinations, the AI is actively expanding the known library of molecules suitable for quantum interaction studies. Requiring only basic atomic properties like electron affinity and ionization potential makes the model highly accessible for rapid materials discovery.
DISCOVERED
22d ago
2026-03-21
PUBLISHED
22d ago
2026-03-21
RELEVANCE
AUTHOR
Secure-Technology-78