Regensburg 2025 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
TT: Fachverband Tiefe Temperaturen
TT 17: Correlated Electrons: Method Development
TT 17.4: Vortrag
Dienstag, 18. März 2025, 10:15–10:30, H33
Neural network supported Configuration Interaction calculations in quantum many-body clusters — •Louis Thirion1, Pavlo Bilous2, Yorick L. A. Schmerwitz3,4, Gianluca Levi3, Elvar Ö. Jónsson3, Henri Menke5, Maurits Haverkort6, Adriana Pálffy-Buß7, Hannes Jónsson3, and Philipp Hansmann1 — 1University of Erlangen-Nürnberg — 2Max Planck Institute for the Science of Light, Erlangen — 3University of Iceland, Reykjavik — 4Max Planck Institute for Coal Research, Mühlheim — 5Max Planck Institute for Solid State Research, Stuttgart — 6University of Heidelberg — 7University of Würzburg
A novel method is presented for computing the ground state in finite-size quantum many-body systems using configuration interaction (CI) enhanced by machine learning. Our recently developed Python library Solax [1] is used for this purpose. A neural network classifier is trained to select an efficient many-body basis in an iterative procedure. It addresses the exponential growth of the Hilbert space while maintaining accuracy. Validation with the Single Impurity Anderson Model shows a basis reduction by at least an order of magnitude compared to standard truncation schemes [2]. Application to the N2 molecule with ≤ 2× 105 Slater determinants, gives results comparable to full CI calculations with nearly 1010 determinants [3]. We aim to extend this method to multi-tier embedding schemes for predicting critical energy scales in heterogeneous catalysis.
[1] L.Thirion, P.Hansmann, P.Bilous, arXiv:2408.16915v1;
[2] P.Bilous, L.Thirion et al., arXiv:2406.00151;
[3] Y.L.A.Schmerwitz, L.Thirion et al., arXiv:2406.08154.
Keywords: configuration interaction; machine learning; neural network classifier