Profile picture
Semantic Scholar

About

I develop mathematically grounded computational methods for problems in molecular science, materials science, and machine learning. My work lies at the intersection of applied mathematics, computational physics, and scientific machine learning, with an emphasis on turning theoretical ideas into robust numerical methods and working software.

I am currently a postdoctoral researcher at the DAVID laboratory, Université de Versailles Saint-Quentin-en-Yvelines (UVSQ), and collaborating with CEA Paris-Saclay. I work on structure elucidation for molecules and crystalline solids with Jérémie Cabessa (UVSQ), Marie-Pierre Gaigeot (UEVE), Thibault Charpentier (CEA), and Mihai-Cosmin Marinica (CEA).

Previously, I completed my PhD in the MATHERIALS team at Inria Paris and at the CERMICS laboratory, École Nationale des Ponts et Chaussées, under the supervision of Éric Cancès and Antoine Levitt.

Research Interests

I enjoy working closely with researchers and developers to formalize open scientific problems, derive computational methods, and deliver proof-of-concept implementations—from mathematical analysis to working code. My core expertise is in the numerical analysis of electronic-structure methods, with applications to quantum chemistry and condensed-matter physics.

More recently, I have been developing deep generative models for molecular structure elucidation. I built a PyTorch research framework implementing discrete diffusion models and stochastic interpolants with Transformer architectures. This work includes the derivation and implementation of efficient sampling strategies and guidance mechanisms. A paper and an open-source implementation are currently in preparation.

Further details are available in the Research and Code sections.

Topics: scientific machine learning; generative modeling; inverse problems and inverse design; diffusion and post-diffusion models; electronic structure; quantum chemistry; density-functional theory; numerical analysis; numerical and Riemannian optimization.

Email: laurent(dot)vidal(at)uvsq(dot)fr