Skip to main content

SIESTA Y SU INTEROPERABILIDAD PARA LOS NUEVOS RETOS EN SIMULACIONES ATOMISTICAS

PRE2019-089784

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal de Promoción del Talento y su Empleabilidad en I+D+i
Subprogram Subprograma Estatal de Formación
Call Contratos Predoctorales
Call year 2019
State Plan 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre FUNDACIÓ INSTITUT CATALÀ DE NANOCIÈNCIA I NANOTECNOLOGIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 1
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Graph2Mat: universal graph to matrix conversion for electron density prediction

  • Febrer, Pol
  • Jørgensen, Peter Bjørn
  • Pruneda, Miguel
  • García Arribas, Alberto
  • Ordejón, Pablo
  • Bhowmik, Arghya
The electron density is a fundamental observable of an atomic system from which all ground-state properties can be computed. As a prediction target for machine learning (ML) models, electron density is often represented on a dense real space grid, which is data heavy, or through density fitting approximations. In this work, we show the power of targeting the density matrix (DM), a linear-scaling sparse SE(3) equivariant matrix that encodes the exact density. We introduce Graph2Mat, a universal function for converting molecular graphs into equivariant matrices. We demonstrate how a ML model that combines this Graph2Mat approach with state-of-the-art molecular graph representations can accurately predict the DM of molecular systems. The models achieve state-of-the-art performance on electron density prediction by matching the accuracy of grid-based methods, while using datasets that are at least one order of magnitude smaller. Accurately predicted electron densities can also accelerate density functional theory (DFT) calculations by reducing the number of self-consistent field (SCF) iterations needed to converge. In this work, we get an average 40% reduction on the number of SCF steps in DFT calculations of QM9 molecules with SIESTA. The novel prediction model also allows for two new and promising measures of uncertainty (total charge error and self-consistency error) that will facilitate its practical usage, e.g. within active learning workflows. These results open the door for many applications using hybrid ML-accelerated DFT methodologies, and uncertainty aware single iteration ab initio molecular dynamics., PF has received funding from the Project PID2022-139776NB-C62 funded by MCIN/AEI https://doi.org/10.13039/501100011033 ERDF, EU and the Grant PRE2019-089784 funded by MCIN/AEI https://doi.org/10.13039/501100011033 and by ESF Investing in your future. This work has been carried in the framework of the Materials Science PhD Program at Universitat Autònoma de Barcelona (UAB). AB was supported by Independent Research Fund Denmark with project DELIGHT (Grant No. 0217-00326B) TeraBatt project (Grant No. 2035-00232B) and ADANA project (Grant No. 3164-00297B). AG and MP were supported by Grant PID2022-139776NB-C61 of the Spanish MCIN/AEI https://doi.org/10.13039/501100011033, by the ERDF, A way of making Europe, and by Generalitat de Catalunya (Grant 2021SGR01519). AG acknowledges AEI through the Severo Ochoa MaTrans42 (CEX2023-0001263-S) Excellence Centre distinction, and the European High Performance Computing Joint Undertaking under grant Agreement no. 101093374 (Materials design at the exascale), co-funded by AEI PCI2022-134978-2. This project has received funding from the European Union's project Battery Interface Genome—Materials Acceleration Platform: EU-H2020 Grant Agreement ID: 957189. PF, MP, AG and PO acknowledge support from EU MaX CoE (Grant 101093374) and Grants PCI2022-134972-2, PCI2022-134978-2, PID2022-139776NB-C61 and PID2022-139776NB-C62 funded by the Spanish MCIN/AEI https://doi.org/10.13039/501100011033 and by the ERDF, A way of making Europe, and Grant 2021SGR01519 from Generalitat de Catalunya. The ICN2 is supported by the Severo Ochoa Centres of Excellence programme, Grant No. CEX2021-001214-S, funded by MCIN/AEI https://doi.org/10.13039/501100011033., With funding from the Spanish government through the "Severo Ochoa Centre of Excelence" accreditation (CEX2023-001263-S), With funding from the Spanish government through the "Severo Ochoa Centre of Excelence" accreditation (CEX2021-001214-S), Peer reviewed