Postmodern Local Correlation Theory: Is there life after DLPNO theory?
Colloquia
- Speaker
-
Martin Head-Gordon
Pitzer Center for Theoretical Chemistry and Department of Chemistry, University of California Berkeley, USA - When
-
2026/09/03
12:00 - Place
- DIPC Josebe-Olarra Lecture Hall
- Add to calendar
-
iCal
Many-electron theory provides a route to systematically improvable polynomial-scaling approximations to the formally exponential-scaling Schrodinger equation for electrons in molecules, as exemplified at lowest order by second order perturbation theory (MP2), or coupled cluster theory through perturbative triples (CCSD(T)) for high accuracy. Yet the high polynomial scaling of compute effort with molecular size is still a widely recognized barrier to simulating large, complex molecules. The main effort in our field to address this challenge has been about two decades of development of “local correlation” using domain localized pair natural orbital (DL-PNO) theory. The result has been practical linear scaling algorithms for standard quantum chemistry methods for electron correlation ranging from MP2 to CCSD(T), and even beyond. However, the DL-PNO approach does have drawbacks: algebraic complications from using non-orthogonal localized virtual orbitals, and practical problems that manifest as slightly inadequate numerical precision, that is difficult to improve in current implementations. In this talk I will describe progress in developing a new alternative local correlation approach that is fully numerical. Usability is promoted by use of just a single master threshold to determine the compute cost versus numerical precision tradeoff. Simplicity is promoted by the use of a single set of orthonormal localized virtual orbitals. Several new ideas are employed to achieve high efficiency, with a form of electronic embedding proving to be particularly important. Results will be presented that assess performance versus precision as a function of threshold for a range of standard quantum chemical methods beginning with MP2 and progressing to higher order methods. If time permits, I will also discuss synergistic developments that improve MP2 theory itself, and touch on double hybrid density functional theory.
About the speaker
Professor Martin Head-Gordon is the Kenneth S. Pitzer Distinguished Professor of Chemistry at the University of California, Berkeley, and a Senior Faculty Scientist in the Chemical Sciences Division of Lawrence Berkeley National Laboratory. A leading figure in theoretical and computational chemistry, he has been a member of the Berkeley faculty since 1992. His research has made major contributions to molecular electronic structure theory, including linear-scaling quantum chemistry, density functional theory, energy decomposition analysis, and computational approaches to catalysis. He is a Fellow of the Royal Society and the American Chemical Society, a member of the U.S. National Academy of Sciences and the American Academy of Arts and Sciences, and President of the World Association of Theoretical and Computational Chemists (WATOC). He is also a driving force behind the Q-Chem quantum chemistry program.