Machine learning enables long time scale molecular photodynamics simulations.


Journal

Chemical science
ISSN: 2041-6520
Titre abrégé: Chem Sci
Pays: England
ID NLM: 101545951

Informations de publication

Date de publication:
21 Sep 2019
Historique:
received: 09 04 2019
accepted: 02 08 2019
entrez: 21 12 2019
pubmed: 21 12 2019
medline: 21 12 2019
Statut: epublish

Résumé

Photo-induced processes are fundamental in nature but accurate simulations of their dynamics are seriously limited by the cost of the underlying quantum chemical calculations, hampering their application for long time scales. Here we introduce a method based on machine learning to overcome this bottleneck and enable accurate photodynamics on nanosecond time scales, which are otherwise out of reach with contemporary approaches. Instead of expensive quantum chemistry during molecular dynamics simulations, we use deep neural networks to learn the relationship between a molecular geometry and its high-dimensional electronic properties. As an example, the time evolution of the methylenimmonium cation for one nanosecond is used to demonstrate that machine learning algorithms can outperform standard excited-state molecular dynamics approaches in their computational efficiency while delivering the same accuracy.

Identifiants

pubmed: 31857878
doi: 10.1039/c9sc01742a
pii: c9sc01742a
pmc: PMC6849489
doi:

Types de publication

Journal Article

Langues

eng

Pagination

8100-8107

Subventions

Organisme : Austrian Science Fund FWF
ID : W 1232
Pays : Austria

Informations de copyright

This journal is © The Royal Society of Chemistry 2019.

Références

J Chem Phys. 2006 Apr 14;124(14):144110
pubmed: 16626183
Angew Chem Int Ed Engl. 2017 Oct 9;56(42):12828-12840
pubmed: 28520235
J Chem Phys. 2012 Nov 28;137(20):204305
pubmed: 23206001
J Phys Chem Lett. 2018 Oct 4;9(19):5660-5663
pubmed: 30200766
Chem Sci. 2017 Apr 1;8(4):3192-3203
pubmed: 28507695
J Chem Phys. 2018 Nov 28;149(20):204106
pubmed: 30501255
Nat Commun. 2018 Oct 10;9(1):4195
pubmed: 30305624
Chem Sci. 2018 Dec 21;10(8):2298-2307
pubmed: 30881655
Chem Sci. 2017 Oct 1;8(10):6924-6935
pubmed: 29147518
J Chem Theory Comput. 2011 May 10;7(5):1253-8
pubmed: 26610121
J Chem Phys. 2018 Oct 14;149(14):144107
pubmed: 30316273
Wiley Interdiscip Rev Comput Mol Sci. 2018 Nov-Dec;8(6):e1370
pubmed: 30450129
Acc Chem Res. 2017 Jul 18;50(7):1785-1793
pubmed: 28665584
Annu Rev Phys Chem. 2011;62:621-43
pubmed: 21219147
J Chem Theory Comput. 2016 Mar 8;12(3):1207-19
pubmed: 26854874
J Chem Phys. 2019 Feb 14;150(6):064105
pubmed: 30769998
Sci Adv. 2017 Dec 13;3(12):e1701816
pubmed: 29242828
Phys Chem Chem Phys. 2019 Jul 14;21(26):14205-14213
pubmed: 30523350
Science. 2002 Aug 2;297(5582):822-5
pubmed: 12161649
Science. 2018 Jul 27;361(6400):360-365
pubmed: 30049875
J Am Chem Soc. 2016 Dec 14;138(49):15911-15916
pubmed: 27682199
Nature. 2016 Jan 28;529(7587):484-9
pubmed: 26819042
Science. 2018 Jan 19;359(6373):325-329
pubmed: 29348237
J Phys Chem B. 2008 Jan 17;112(2):405-13
pubmed: 18081339
J Phys Chem Lett. 2018 Dec 6;9(23):6702-6708
pubmed: 30403870
J Phys Chem Lett. 2018 Oct 18;9(20):6096-6102
pubmed: 30286602
J Phys Chem Lett. 2018 Jun 7;9(11):2725-2732
pubmed: 29732893
Phys Chem Chem Phys. 2014 Dec 21;16(47):25883-95
pubmed: 25354307
Nature. 2017 Mar 15;543(7645):355-365
pubmed: 28300093
Phys Rev Lett. 2015 Mar 6;114(9):096405
pubmed: 25793835
J Chem Phys. 2006 Apr 21;124(15):154104
pubmed: 16674215
J Chem Theory Comput. 2015 May 12;11(5):2187-98
pubmed: 26574419
J Chem Phys. 2016 Nov 7;145(17):170901
pubmed: 27825224
Angew Chem Int Ed Engl. 2018 Apr 9;57(16):4164-4169
pubmed: 29216413
J Chem Phys. 2004 Apr 1;120(13):5871-4
pubmed: 15267468
Chem Sci. 2016 Aug 1;7(8):5139-5147
pubmed: 30155164
J Phys Chem Lett. 2018 Aug 2;9(15):4355-4359
pubmed: 30024765
Science. 2004 Dec 3;306(5702):1765-8
pubmed: 15576616
J Phys Chem A. 2007 Aug 30;111(34):8500-8
pubmed: 17685594
Nat Commun. 2018 Sep 24;9(1):3887
pubmed: 30250077
Science. 2007 Feb 2;315(5812):625-9
pubmed: 17272716
J Chem Phys. 2018 Jun 28;148(24):241733
pubmed: 29960353
Nat Chem. 2014 Mar;6(3):242-7
pubmed: 24557140
J Chem Phys. 2005 Nov 22;123(20):204101
pubmed: 16351234
Nature. 2018 Jul;559(7715):547-555
pubmed: 30046072
Phys Rev Lett. 2010 Apr 2;104(13):136403
pubmed: 20481899
Science. 2002 Dec 20;298(5602):2395-8
pubmed: 12493917
J Phys Chem A. 2009 Apr 23;113(16):4356-66
pubmed: 19298071

Auteurs

Julia Westermayr (J)

Institute of Theoretical Chemistry , Faculty of Chemistry , University of Vienna , 1090 Vienna , Austria . Email: philipp.marquetand@univie.ac.at.

Michael Gastegger (M)

Machine Learning Group , Technical University of Berlin , 10587 Berlin , Germany.

Maximilian F S J Menger (MFSJ)

Institute of Theoretical Chemistry , Faculty of Chemistry , University of Vienna , 1090 Vienna , Austria . Email: philipp.marquetand@univie.ac.at.
Dipartimento di Chimica e Chimica Industriale , University of Pisa , Via G. Moruzzi 13 , 56124 Pisa , Italy.

Sebastian Mai (S)

Institute of Theoretical Chemistry , Faculty of Chemistry , University of Vienna , 1090 Vienna , Austria . Email: philipp.marquetand@univie.ac.at.

Leticia González (L)

Institute of Theoretical Chemistry , Faculty of Chemistry , University of Vienna , 1090 Vienna , Austria . Email: philipp.marquetand@univie.ac.at.

Philipp Marquetand (P)

Institute of Theoretical Chemistry , Faculty of Chemistry , University of Vienna , 1090 Vienna , Austria . Email: philipp.marquetand@univie.ac.at.

Classifications MeSH