Quantum chemistry plays a crucial role in several fields (physics, biology, engineering...), it is used to predict molecular properties and to help to interpret experimental findings.

In quantum chemistry, the state of a molecular system is determined by its wave function. Accurate calculations of the wavefunction require the computations and storage of numerous two-electron integrals, which impede the accuracy of the predictions and/or the size of the molecular systems that can be investigated. In order to overcome these shortcomings, we will use state-of-the-art Machine Learning algorithms and methods to compute efficiently these two-electronic integrals. Our approach is expected to accelerate quantum chemistry calculations by several orders of magnitude, thus, it will allow us to treat larger molecular systems with higher accuracy than what is currently possible with standard quantum chemistry computational methods.

This project consists of two main phases:

• Phase 1: Approximate the set of two-electronic integrals with a physical based model.
• Phase 2: Complete the simplistic model built in phase 1 by a component based on machine learning. This second component will correct the errors of the physical model.

Our thesis project brings together two teams: Quantum Chemistry (LCPMR) and Artificial Intelligence (ISIR), thus, it enables to ally two complementary communities and to promote AI to face new challenges whose results will contribute to a wide range of scientific applications.

 PhD student: Kaoutar EL HALOUI
 
PhD supervisors: Nicolas Sisourat, Nicolas Thome
 
Research laboratories:

• LCPMR - Laboratoire Chimie Physique Matière et Rayonnement
• ISIR - Institut des systèms intelligents et de robotique