ML NMDINT: Difference between revisions
No edit summary |
|||
| Line 3: | Line 3: | ||
Description: Tag to control the minimum interval to get training samples in the machine learning force field method. | Description: Tag to control the minimum interval to get training samples in the machine learning force field method. | ||
---- | ---- | ||
The usage of this tag in combination with the learning algorithms is described here: [[Machine learning force field calculations: | The usage of this tag in combination with the learning algorithms is described here: [[Machine learning force field calculations: Basics#Sampling of training data and local reference configurations|here]] | ||
This tag defines a lower threshold for taking new configurations from the MD, so that as long as the upper threshold for the Bayesian error (e.g. {{TAG|ML_CDOUB}} times {{TAG|ML_CTIFOR}}) is not exceeded, at least {{TAG|ML_NMDINT}} MD steps are preformed using the MLFF (i.e. no first principles calculation is performed). This avoids that many nearly identical structures are added. | This tag defines a lower threshold for taking new configurations from the MD, so that as long as the upper threshold for the Bayesian error (e.g. {{TAG|ML_CDOUB}} times {{TAG|ML_CTIFOR}}) is not exceeded, at least {{TAG|ML_NMDINT}} MD steps are preformed using the MLFF (i.e. no first principles calculation is performed). This avoids that many nearly identical structures are added. | ||
Revision as of 10:09, 2 November 2021
ML_NMDINT = [integer]
Default: ML_NMDINT = 10
Description: Tag to control the minimum interval to get training samples in the machine learning force field method.
The usage of this tag in combination with the learning algorithms is described here: here
This tag defines a lower threshold for taking new configurations from the MD, so that as long as the upper threshold for the Bayesian error (e.g. ML_CDOUB times ML_CTIFOR) is not exceeded, at least ML_NMDINT MD steps are preformed using the MLFF (i.e. no first principles calculation is performed). This avoids that many nearly identical structures are added.
Related Tags and Sections
ML_LMLFF, ML_MCONF_NEW, ML_CDOUB, ML_CTIFOR, ML_MHIS