Equivariant representations for molecular Hamiltonians and N-center atomic-scale properties

Considerations of symmetry underpin the mathematical representations of the atomic configurations that are used by machine learning models to predict properties of various molecular structures. Though these models generally rely on a description of atom-centered environments, many of the quantities that are relevant for quantum mechanical calculations – notably the single-particle Hamiltonian Ĥ matrix when written in an atomic-orbital basis – aren’t associated with a single center, but rather with two or more atoms in the structure. In the paper “Equivariant representations for molecular Hamiltonians and N-center atomic-scale properties,” recently published in the Journal of Chemical Physics, researchers discuss a family of structural descriptors that generalize atom-centered density correlation features to the N-centers case and show how it can be applied to efficiently learn the matrix elements of Ĥ. These N-centers features are fully equivariant in terms of both translations and rotations as well as in terms of permutations of the indices associated with the atoms and are therefore suitable for use in constructing symmetry-adapted machine-learning models of new classes of properties of molecules and materials.

Manifolds in commonly used atomic fingerprints lead to failure in machine-learning four-body interactions

The existence of manifolds in two atomic environment fingerprints commonly used to characterize the local environments of atoms in machine learning and other contexts causes a failure to machine learn four-body interactions such as torsional energies, which are an important part of standard force fields. No such manifolds can be found for the Overlap Matrix (OM) fingerprint due to its intrinsic many-body character, making it an appealing alternative, NCCR MARVEL researchers at the University of Basel found in a paper recently published in the Journal of Chemical Physics.

MARVEL Distinguished Lecture — Sharon Glotzer

The 29th NCCR MARVEL Distinguished Lecture will be given by Prof. Sharon Glotzer, department chair of chemical engineering at the University of Michigan, entitled "A theory of entropic bonding".

MARVEL Distinguished Lecture — Garnet Chan

The 30th NCCR MARVEL Distinguished Lecture will be given by Prof. Garnet Chan, professor of chemistry at the California Institute of Technology. He will be discussing the possibility of exponential quantum advantage in quantum chemistry.

MARVEL Distinguished Lecture — Alán Aspuru-Guzik

The 31st NCCR MARVEL Distinguished Lecture will be given by Prof. Alán Aspuru-Guzik, professor of chemistry and computer science at the University of Toronto, and entitled: "There is no time for science as usual: Materials Acceleration Platforms".

MARVEL Junior Seminar — February 2022

The MARVEL Junior Seminars aim to intensify interactions between the MARVEL Junior scientists belonging to different research groups. The MARVEL Junior Seminar series is now held in videoconferencing mode. We are pleased to propose the 48th MARVEL Junior Seminar: Francesco Libbi (THEOS, EPFL) and Anna Hehn (University of Zurich) will present their research. The seminar will be chaired by Michele Kotiuga (EPFL).

MADICES workshop 2022

This CECAM workshop is devoted to machine-actionable data interoperability for chemical sciences (MADICES): bridging experiments, simulations, and machine learning for spectral data. Developers, scientists, and data specialists will be brought together to discuss the hurdles and opportunities of data interoperability in the context of the chemical and materials sciences. We will strive for general technical recommendations, with X-ray absorption spectroscopy as the prototype use case.

From women's eyes

"In order to achieve full and equal access to and participation in science for women and girls, the United Nations General Assembly declared February 11 as the International day of Women and Girls in Science in 2015” (from the UN website). CECAM will mark this important occasion with a special afternoon dedicated to women who contributed to our field… and by proposing an unusual tool to encourage (also) young girls to consider a scientific path.

Psi-k conference

Psi-k 2022 is the 6th general conference for the worldwide Psi-k community. Typically held every 5 years, this event is the largest worldwide in first-principles simulations. With more than 1000 participants attending the past 2010 and 2015 editions, it promises to be the most exciting, defining event in the field, offering an intense but enjoyable atmosphere, in addition to a chance to explore the beautiful region around Lausanne and Lake Geneva.  

MARVEL Junior Seminar — December 2024

The MARVEL Junior Seminars aim to intensify interactions between the MARVEL Junior scientists belonging to different research groups – this is held in hybrid mode, in order to maintain in-person contacts and allow off-campus attendees to follow the seminars remotely! We are pleased to propose the 63rd MARVEL Junior Seminar: Mauro Dossena (Integrated systems laboratory, ETHZ) and Stefan Riemelmoser (Chair of atomic scale simulation - CSEA, EPFL) will present their research. 

MARVEL Review and Retreat January 2025

The eleventh MARVEL Review and Retreat will take place on January 13-15, 2025. The event will gather all MARVEL members, group leaders as well as postdocs and students. Presentations will show to all the community where the projects stand and what are the next steps. 

MARVEL Distinguished Lecture — Alexandre Tkatchenko

The 39th NCCR MARVEL Distinguished Lecture will be given by Prof. Alexandre Tkatchenko, University of Luxembourg. He will be presenting a lecture entitled: "AI-Driven Fully Quantum Biomolecular Simulations".