Thema der Dissertation:
Development of a practical deep-learning quantum Monte Carlo method Thema der Disputation:
Advantages of a practical deep-learning quantum Monte Carlo method
Development of a practical deep-learning quantum Monte Carlo method Thema der Disputation:
Advantages of a practical deep-learning quantum Monte Carlo method
Abstract: This lecture focuses on a deep-learning-based variational quantum Monte Carlo (VMC) approach to molecular electronic structure. In contrast to conventional second-quantized quantum chemical methods, which rely on Gaussian basis sets and large determinant expansions, deep-learning VMC approach operates directly in real space with flexible many-body wave functions parameterized by neural networks. This eliminates basis set errors, and provides an intuitive, variationally controlled framework with direct access to a compact wave function representation. Building on this core idea, I highlight how deep-learning VMC enables straightforward and unified treatments of different electronic states, nuclear geometries, and even chemical species, offering a conceptually simple yet powerful alternative for accurately describing strongly correlated electronic systems.
Zeit & Ort
05.05.2026 | 13:00
Seminarraum 010
(Fachbereich Mathematik und Informatik, Arnimallee 12, 14195 Berlin)
&
WebEx