ML CX: Difference between revisions

Latest revision as of 15:24, 17 September 2022

ML_CX = [real]
Default: ML_CX = 0.0

Description: The parameter determines to which value the threshold (ML_CTIFOR) is updated within the machine learning force field methods.

The use of this tag in combination with the learning algorithms is described here: here.

If ML_ICRITERIA>0, ML_CTIFOR is set to the average of the Bayesian errors of the forces stored in a history. Note that ML_ICRITERIA=1 and ML_ICRITERIA=2, average over different data. In the first case the average is performed over Bayesian errors after updates of the force fields, and in the second case over all recent Bayesian error estimates (see ML_ICRITERIA). In both cases, if ML_CTIFOR is updated, it is set to

ML_CTIFOR = (average of the stored Bayesian errors in the history) *(1.0 + ML_CX).

Obviously setting ML_CX to a positive value will result in fewer first principles calculations and fewer updates of the MLFF, whereas negative values result in more frequent first principles calculations (as well as updates of the MLFF). Typical values of ML_CX are between -0.2 and 0.0 for ML_ICRITERIA=1, and 0.0 and 0.3 for ML_ICRITERIA=2 (a good starting value is 0.2 for ML_ICRITERIA=2). For training runs using heating, the default usually results in very well balanced machine learned force fields. When the training is performed at a fixed temperature, it is often desirable to decrease to ML_CX=-0.1, in order to increase the number of first principle calculations and thus the size of the training set (the default can result in too few training data).

The number of entries in the history are controlled by ML_MHIS for ML_ICRITERIA=1, and it is currently fixed to 400 for ML_ICRITERIA=2 (in future releases 50 x ML_MHIS).

@@ Line 1: / Line 1: @@
-{{TAGDEF|ML_FF_XMIX|[integer]|0.5}}
+{{DISPLAYTITLE:ML_CX}}
+{{TAGDEF|ML_CX|[real]|0.0}}
-Description: The parameter determines how the  threshold  ({{TAG|ML_FF_CTIFOR}}) is updated in the machine learning force field methods.
+Description: The parameter determines to which value the threshold  ({{TAG|ML_CTIFOR}}) is updated within the machine learning force field methods.
 ----
+The use of this tag in combination with the learning algorithms is described here: [[Machine learning force field calculations: Basics#Threshold for error of forces|here]].
-If {{TAG|ML_FF_LCRITERIA}} is .TRUE.,
+If {{TAG|ML_ICRITERIA}}>0, {{TAG|ML_CTIFOR}} is set to the average of the Bayesian errors of the forces stored in a history. Note that {{TAG|ML_ICRITERIA}}=1 and {{TAG|ML_ICRITERIA}}=2, average over different data. In the first case the average is performed over Bayesian errors after updates of the force fields, and in the second case over all recent Bayesian error estimates (see {{TAG|ML_ICRITERIA}}). In both cases, if {{TAG|ML_CTIFOR}} is updated, it is set to
-{{TAG|ML_FF_CTIFOR}} is set to the average of the  Bayesian errors of the forces stored in a history (see {{TAG|ML_FF_LCRITERIA}}). The number of entries in the history are controlled by  {{TAG|ML_FF_MHIS}}. To avoid too abrupt changes, the average Bayesian error is mixed with the  old value of {{TAG|ML_FF_CTIFOR}}:
-{{TAG|ML_FF_CTIFOR}} = (1-{{TAG|ML_FF_XMIX}}) * {{TAG|ML_FF_CTIFOR}}  + {{TAG|ML_FF_XMIX}} * (average of the stored Bayesian errors).
+{{TAG|ML_CTIFOR}} = (average of the stored Bayesian errors in the history) *(1.0 + {{TAG|ML_CX}}).
-This implies that  for {{TAG|ML_FF_XMIX}}=1, the old value stored in {{TAG|ML_FF_CTIFOR}} is simply overwritten by the current average Bayesian error. This tag is particularly useful at the beginning to avoid that the value supplied in the INCAR file changes too quickly.
+Obviously setting  {{TAG|ML_CX}} to a positive value will result in fewer first principles calculations and fewer updates of the MLFF, whereas negative values result in more frequent first principles calculations (as well as updates of the MLFF). Typical values of  {{TAG|ML_CX}} are between -0.2 and 0.0 for {{TAG|ML_ICRITERIA}}=1, and 0.0 and 0.3 for {{TAG|ML_ICRITERIA}}=2 (a good starting value is 0.2 for {{TAG|ML_ICRITERIA}}=2). For training runs using heating, the default usually results in very well balanced machine learned force fields. When the training is performed at a fixed temperature, it is often desirable to decrease to {{TAG|ML_CX}}=-0.1, in order to increase the number of first principle calculations and thus the size of the training set (the default can result in too few training data).
-== Related Tags and Sections ==
+The number of entries in the history are controlled by  {{TAG|ML_MHIS}} for {{TAG|ML_ICRITERIA}}=1, and it is currently fixed to 400 for {{TAG|ML_ICRITERIA}}=2 (in future releases 50 x {{TAG|ML_MHIS}}).
-{{TAG|ML_FF_LMLFF}}, {{TAG|ML_FF_LCRITERIA}}, {{TAG|ML_FF_CTIFOR}}, {{TAG|ML_FF_MHIS}}
-{{sc|ML_FF_LCRITERIA|Examples|Examples that use this tag}}
+== Related tags and articles ==
+{{TAG|ML_LMLFF}}, {{TAG|ML_ICRITERIA}}, {{TAG|ML_CTIFOR}}, {{TAG|ML_MHIS}}, {{TAG|ML_CSIG}}, {{TAG|ML_CSLOPE}}
+{{sc|ML_LCRITERIA|Examples|Examples that use this tag}}
 ----
-[[Category:INCAR]][[Category:Machine Learning]][[Category:Machine Learned Force Fields]][[Category: Alpha]]
+[[Category:INCAR tag]][[Category:Machine-learned force fields]]