Difference between revisions of "Hauptseminar Porous Media SS 2021/Machine learning in physics"
(2 intermediate revisions by the same user not shown)  
Line 1:  Line 1:  
{{Seminartopic  {{Seminartopic  
number=3  number=3  
−  topic=Machine  +  topic=Machine learning for the construction of force fields 
date=TBA  date=TBA  
time=TBA  time=TBA  
Line 11:  Line 11:  
== Contents ==  == Contents ==  
−  +  Description: In recent years machine learning methods have been employed to develop nonparametric models for the approximation of the potential energy surface.  
+  These models, trained on abinitio simulation data, deliver the accuracy of these abinitio methods at the speed and system size of classical approaches.  
+  Development of these models requires several important steps, all of which are under active study and for which many approaches exist. Such steps include selecting  
+  training data, representing this data, and the choosing machine learning model to use.  
+  This talk will address current approaches for developing these models as well as well as some of the results that have been found by using the machine learning methods.  
+  The presenter will build upon the foundation work of the previous talks on DFT methods and atomistic methods in order to demonstrate the position of the machine learning  
+  approaches as a bridge between them.  
+  
+  === Important Points ===  
+  
+  * Gaussian process regression and Neural network potentials  
+  * Sampling configuration space for training data  
+  * Representation of molecular environments  
+  * Failures in ML potentials  
== Literature ==  == Literature ==  
−  +  <bibentry pdflink="yes">  
+  bartok15a  
+  ko21a  
+  langer20a  
+  chmiela18a  
+  chmiela19a  
+  rupp15b  
+  </bibentry> 
Revision as of 19:28, 5 February 2021
 Date
 TBA"TBA" contains an extrinsic dash or other characters that are invalid for a date interpretation.
 Time
 TBA
 Topic
 Machine learning for the construction of force fields
 Speaker
 TBA
 Tutor
 Samuel Tovey
Contents
Description: In recent years machine learning methods have been employed to develop nonparametric models for the approximation of the potential energy surface. These models, trained on abinitio simulation data, deliver the accuracy of these abinitio methods at the speed and system size of classical approaches. Development of these models requires several important steps, all of which are under active study and for which many approaches exist. Such steps include selecting training data, representing this data, and the choosing machine learning model to use. This talk will address current approaches for developing these models as well as well as some of the results that have been found by using the machine learning methods. The presenter will build upon the foundation work of the previous talks on DFT methods and atomistic methods in order to demonstrate the position of the machine learning approaches as a bridge between them.
Important Points
 Gaussian process regression and Neural network potentials
 Sampling configuration space for training data
 Representation of molecular environments
 Failures in ML potentials
Literature

Albert P. Bartók and Gábor Csányi.
"Gaussian Approximation Potentials: a brief tutorial introduction".
International Journal of Quantum Chemistry , 2015.
[PDF] (308 KB) [DOI] 
Ko, Tsz Wai and Finkler, Jonas A. and Goedecker, Stefan and Behler, Jörg.
"A fourthgeneration highdimensional neural network potential with accurate electrostatics including nonlocal charge transfer".
Nature Communications 12(1)(398), 2021.
[DOI] 
Marcel F. Langer and Alex Goe\ssmann and Matthias Rupp.
"Representations of molecules and materials for interpolation of quantummechanical simulations via machine learning".
2020.

Chmiela, Stefan and Sauceda, Huziel E. and Müller, KlausRobert and Tkatchenko, Alexandre.
"Towards exact molecular dynamics simulations with machinelearned force fields".
Nature Communications 9(1)(3887), 2018.
[DOI] 
Stefan Chmiela and Huziel E. Sauceda and Igor Poltavsky and KlausRobert Müller and Alexandre Tkatchenko.
"sGDML: Constructing accurate and data efficient molecular force fields using machine learning".
Computer Physics Communications 240(3845), 2019.
[URL] 
Rupp, Matthias.
"Machine learning for quantum mechanics in a nutshell".
International Journal of Quantum Chemistry 115(16)(10581073), 2015.
[URL]