ML W1: Difference between revisions

From VASP Wiki
No edit summary
 
(13 intermediate revisions by 2 users not shown)
Line 1: Line 1:
{{TAGDEF|ML_FF_W1_MB|[real]|0.0}}
{{DISPLAYTITLE:ML_W1}}
{{TAGDEF|ML_W1|[real]|0.1}}


Description: This tag defines the weight for the angular descriptor within the machine learning force field method.
Description: This tag defines the weight <math>\beta</math> for the radial (and angular) descriptor within the machine learning force field method (see [[Machine learning force field: Theory#Potential energy fitting|this section]]).
----
----
The weight for the angular descriptor <math>W_{2}</math> is internally computed from the weight of the radial descriptor <math>W_{1}</math> as:


== Related Tags and Sections ==
<math>W_{2}=1.0-W_{1}.</math>
{{TAG|ML_FF_LMLFF}}, {{TAG|ML_FF_W2_MB}}, {{TAG|ML_FF_ISOAP1_MB}}


{{sc|ML_FF_W1_MB|Examples|Examples that use this tag}}
The value for {{TAG|ML_W1}} must be chosen in the interval <math>[0, 1]</math>.
 
By default, the angular and radial descriptors are both used although the latter is weighed less. In principle a weight of 0 for one of them is selectable which allows the code to internally skip the respective computation. However, it is generally recommended to use both descriptors to achieve satisfying training results.
 
== Related tags and articles ==
{{TAG|ML_LMLFF}}, {{TAG|ML_RCUT1}}, {{TAG|ML_RCUT2}}, {{TAG|ML_SION1}}, {{TAG|ML_SION2}}
 
{{sc|ML_W1|Examples|Examples that use this tag}}
----
----


[[Category:INCAR]][[Category:Machine Learning]][[Category:Machine Learned Force Fields]][[Category:VASP6]]
[[Category:INCAR tag]][[Category:Machine-learned force fields]]

Latest revision as of 13:31, 8 April 2022

ML_W1 = [real]
Default: ML_W1 = 0.1 

Description: This tag defines the weight for the radial (and angular) descriptor within the machine learning force field method (see this section).


The weight for the angular descriptor is internally computed from the weight of the radial descriptor as:

The value for ML_W1 must be chosen in the interval .

By default, the angular and radial descriptors are both used although the latter is weighed less. In principle a weight of 0 for one of them is selectable which allows the code to internally skip the respective computation. However, it is generally recommended to use both descriptors to achieve satisfying training results.

Related tags and articles

ML_LMLFF, ML_RCUT1, ML_RCUT2, ML_SION1, ML_SION2

Examples that use this tag