1
General Questions and Answers / Interface MTP Training
« on: November 27, 2024, 23:03 »
Dear Admin,
Q1. I tried the code on "https://docs.quantumatk.com/manual/Types/CrystalInterfaceTrainingParameters/CrystalInterfaceTrainingParameters.html#NL.Study.MomentTensorPotential.CrystalInterfaceTraining.CrystalInterfaceTrainingParameters" for Iron/Copper interface but I get this error everytime. I tried changing random seed but still has same error. Kindly help resolve.
Master : Handling exception in delegator process.
Encountered TaskExecutionError: An exception was raised while executing task "a90b78e8ad0a11ef90b9801844e91328".
Traceback (most recent call last):
File "zipdir/NL/ComputerScienceUtilities/ParallelTools/DynamicTaskScheduler.py", line 1391, in __startParallelExecutionOnWorkerProcesses
File "zipdir/NL/ComputerScienceUtilities/Workflow/Workflow.py", line 1187, in _runTask
File "zipdir/NL/ComputerScienceUtilities/Workflow/Workflow.py", line 708, in run
File "zipdir/NL/Study/MomentTensorPotential/FitMomentTensorPotential.py", line 464, in _execute
NL.ComputerScienceUtilities.Exceptions.NLExecutionError: The training dataset contains no configurations. This error can probably be avoided by picking a different random seed.
Q2. Please answer the following. Step 1: I trained an MTP based with active learning of few configurations Step 2: I want to include more configurations so I combined reference data and active learning candidates from step 1 and passed onto Step 2. Step 3: I again want to include few more configurations. Now should I include all configurations from Step 1 and Step2 or only step 2? Also, in retaining should I change basis size as more configurations are being added or keep same for all Steps?
Q1. I tried the code on "https://docs.quantumatk.com/manual/Types/CrystalInterfaceTrainingParameters/CrystalInterfaceTrainingParameters.html#NL.Study.MomentTensorPotential.CrystalInterfaceTraining.CrystalInterfaceTrainingParameters" for Iron/Copper interface but I get this error everytime. I tried changing random seed but still has same error. Kindly help resolve.
Master : Handling exception in delegator process.
Encountered TaskExecutionError: An exception was raised while executing task "a90b78e8ad0a11ef90b9801844e91328".
Traceback (most recent call last):
File "zipdir/NL/ComputerScienceUtilities/ParallelTools/DynamicTaskScheduler.py", line 1391, in __startParallelExecutionOnWorkerProcesses
File "zipdir/NL/ComputerScienceUtilities/Workflow/Workflow.py", line 1187, in _runTask
File "zipdir/NL/ComputerScienceUtilities/Workflow/Workflow.py", line 708, in run
File "zipdir/NL/Study/MomentTensorPotential/FitMomentTensorPotential.py", line 464, in _execute
NL.ComputerScienceUtilities.Exceptions.NLExecutionError: The training dataset contains no configurations. This error can probably be avoided by picking a different random seed.
Q2. Please answer the following. Step 1: I trained an MTP based with active learning of few configurations Step 2: I want to include more configurations so I combined reference data and active learning candidates from step 1 and passed onto Step 2. Step 3: I again want to include few more configurations. Now should I include all configurations from Step 1 and Step2 or only step 2? Also, in retaining should I change basis size as more configurations are being added or keep same for all Steps?