Single molecule: setup and run¶
Note: This example requires AMS2024 or later.
To follow along, either
Download
sal_single_molecule_setup_run.py(run as$AMSBIN/amspython sal_single_molecule_setup_run.py).Download
sal_single_molecule_setup_run.ipynb(see also: how to install Jupyterlab in AMS)
Initial imports¶
from scm.simple_active_learning import SimpleActiveLearningJob
import scm.plams as plams
from scm.external_engines.core import interface_is_installed
assert interface_is_installed(
"m3gnet"
), "You must first install m3gnet with the AMS package manager"
Initialize PLAMS¶
plams.init()
PLAMS working folder: /path/plams_workdir
Input system¶
mol = plams.from_smiles("OCC=O")
for at in mol:
at.properties = {}
plams.plot_molecule(mol)
Reference engine settings¶
For time reasons we use the UFF force field as the reference method. Typically you would instead train to DFT using ADF, BAND, or Quantum ESPRESSO.
ref_s = plams.Settings()
ref_s.input.ForceField.Type = "UFF"
ref_s.runscript.nproc = 1
print(plams.AMSJob(settings=ref_s).get_input())
Engine ForceField
Type UFF
EndEngine
Molecular dynamics settings¶
Here, we use the convenient AMSNVTJob recipe to easily initialize
sone MD settings.
md_s = plams.AMSNVTJob(temperature=300, timestep=0.5, nsteps=10000).settings
print(plams.AMSJob(settings=md_s).get_input())
MolecularDynamics
BinLog
DipoleMoment False
PressureTensor False
Time False
End
CalcPressure False
Checkpoint
Frequency 1000000
End
InitialVelocities
Temperature 300
Type Random
End
NSteps 10000
Thermostat
Tau 200.0
Temperature 300
Type NHC
End
TimeStep 0.5
Trajectory
SamplingFreq 100
WriteBonds True
WriteCharges True
WriteEngineGradients False
WriteMolecules True
WriteVelocities True
End
End
Task MolecularDynamics
ParAMS ML Training settings¶
(Technical note: When using SimpleActiveLearningJob the ParAMS
settings go under input.ams. When using ParAMSJob the settings
instead simply go under input. See the ParAMS Python tutorials.)
ml_s = plams.Settings()
ml_s.input.ams.MachineLearning.Backend = "M3GNet"
ml_s.input.ams.MachineLearning.CommitteeSize = 1
ml_s.input.ams.MachineLearning.M3GNet.Model = "UniversalPotential"
ml_s.input.ams.MachineLearning.MaxEpochs = 200
print(SimpleActiveLearningJob(settings=ml_s).get_input())
MachineLearning
Backend M3GNet
CommitteeSize 1
M3GNet
Model UniversalPotential
End
MaxEpochs 200
End
Active learning settings¶
al_s = plams.Settings()
al_s.input.ams.ActiveLearning.Steps.Type = "Geometric"
al_s.input.ams.ActiveLearning.Steps.Geometric.Start = 10 # 10 MD frames
al_s.input.ams.ActiveLearning.Steps.Geometric.NumSteps = 5 # 10 AL steps
print(SimpleActiveLearningJob(settings=al_s).get_input())
ActiveLearning
Steps
Geometric
NumSteps 5
Start 10
End
Type Geometric
End
End
Simple Active Learning Job¶
settings = ref_s + md_s + ml_s + al_s
job = SimpleActiveLearningJob(settings=settings, molecule=mol, name="sal")
print(job.get_input())
ActiveLearning
Steps
Geometric
NumSteps 5
Start 10
End
Type Geometric
End
End
MachineLearning
Backend M3GNet
CommitteeSize 1
M3GNet
Model UniversalPotential
End
MaxEpochs 200
End
MolecularDynamics
BinLog
DipoleMoment False
PressureTensor False
Time False
End
CalcPressure False
Checkpoint
Frequency 1000000
End
InitialVelocities
Temperature 300.0
Type Random
End
NSteps 10000
Thermostat
Tau 200.0
Temperature 300.0
Type NHC
End
TimeStep 0.5
Trajectory
SamplingFreq 100
WriteBonds True
WriteCharges True
WriteEngineGradients False
WriteMolecules True
WriteVelocities True
End
End
Task MolecularDynamics
Engine ForceField
Type UFF
EndEngine
System
Atoms
O 1.5185424677 1.2528427606 -0.3382346351
C 1.0167107700 0.2231953999 0.1069866215
C -0.3341736669 -0.1931701775 -0.3199821682
O -1.3067693409 0.4447398660 0.4319048572
H 1.6797677292 -0.4284512471 0.6676972986
H -0.4715138766 0.0855120883 -1.3968084763
H -0.4208687404 -1.2969328351 -0.3134514232
H -1.6816953421 -0.0877358551 1.1618879254
End
BondOrders
1 2 2.0
2 3 1.0
3 4 1.0
2 5 1.0
3 6 1.0
3 7 1.0
4 8 1.0
End
Charge 0
End
Run the job¶
job.run(watch=True)
[31.01|17:43:33] JOB sal STARTED
[31.01|17:43:33] JOB sal RUNNING
[31.01|17:43:34] Simple Active Learning 2023.205, Nodes: 1, Procs: 1
[31.01|17:43:36] Composition of main system: C2H4O2
[31.01|17:43:36] All REFERENCE calculations will be performed with the following ForceField engine:
[31.01|17:43:36]
Engine forcefield
type UFF
EndEngine
[31.01|17:43:36] The following are the settings for the to-be-trained MACHINE LEARNING model:
[31.01|17:43:36]
MachineLearning
Backend M3GNet
CommitteeSize 1
M3GNet
Model UniversalPotential
End
MaxEpochs 200
End
[31.01|17:43:36] A single model will be trained (no committee).
[31.01|17:43:36] The ACTIVE LEARNING loop will contain 5 steps, using the following schema:
[31.01|17:43:36] Active Learning Step 1: 10 MD Steps (cumulative: 10)
[31.01|17:43:36] Active Learning Step 2: 46 MD Steps (cumulative: 56)
[31.01|17:43:36] Active Learning Step 3: 260 MD Steps (cumulative: 316)
[31.01|17:43:36] Active Learning Step 4: 1462 MD Steps (cumulative: 1778)
[31.01|17:43:36] Active Learning Step 5: 8222 MD Steps (cumulative: 10000)
[31.01|17:43:36] Total number of MD Steps: 10000
[31.01|17:43:36] Max attempts per active learning step: 15
[31.01|17:43:36]
[31.01|17:43:36] Tip: create a file called SKIP_STEP to skip an active learning step and move on to the next!
[31.01|17:43:36]
[31.01|17:43:36] Running initial reference MD for constructing initial training set...
[31.01|17:43:37] Expanding training and validation sets....
[31.01|17:43:37] Adding frames 2, 3, 4 from initial_reference_calculations/ams.rkf to training set.
[31.01|17:43:37] Adding frames 1 from initial_reference_calculations/ams.rkf to validation set.
[31.01|17:43:37] Current # training set entries: 3
[31.01|17:43:37] Current # validation set entries: 1
[31.01|17:43:37] Storing data in initial_reference_data
[31.01|17:43:37] Deleting initial_reference_calculations
[31.01|17:43:37] Initial reference data has been created. Storing data in folder: initial_reference_data
[31.01|17:43:37] Running initial ParAMS training in folder: initial_training
[31.01|17:43:39] JOB m3gnet STARTED
[31.01|17:43:39] Starting m3gnet.prerun()
[31.01|17:43:39] m3gnet.prerun() finished
[31.01|17:43:39] JOB m3gnet RUNNING
[31.01|17:43:39] Executing m3gnet.run
[31.01|17:44:04] training_set Epoch: 0 Loss: 0.001370
[31.01|17:44:04] validation_set Epoch: 0 Loss: 0.069857
[31.01|17:44:04] training_set Epoch: 10 Loss: 0.000821
[31.01|17:44:04] validation_set Epoch: 10 Loss: 0.003661
[31.01|17:44:05] training_set Epoch: 20 Loss: 0.000368
[31.01|17:44:05] validation_set Epoch: 20 Loss: 0.004117
[31.01|17:44:05] training_set Epoch: 30 Loss: 0.000160
[31.01|17:44:05] validation_set Epoch: 30 Loss: 0.004883
[31.01|17:44:06] training_set Epoch: 40 Loss: 0.000074
[31.01|17:44:06] validation_set Epoch: 40 Loss: 0.002111
[31.01|17:44:06] training_set Epoch: 50 Loss: 0.000040
[31.01|17:44:06] validation_set Epoch: 50 Loss: 0.001035
[31.01|17:44:07] training_set Epoch: 60 Loss: 0.000037
[31.01|17:44:07] validation_set Epoch: 60 Loss: 0.001459
[31.01|17:44:07] training_set Epoch: 70 Loss: 0.000030
[31.01|17:44:07] validation_set Epoch: 70 Loss: 0.000749
[31.01|17:44:08] training_set Epoch: 80 Loss: 0.000024
[31.01|17:44:08] validation_set Epoch: 80 Loss: 0.001935
[31.01|17:44:08] training_set Epoch: 90 Loss: 0.000021
[31.01|17:44:08] validation_set Epoch: 90 Loss: 0.001082
[31.01|17:44:09] training_set Epoch: 100 Loss: 0.000018
[31.01|17:44:09] validation_set Epoch: 100 Loss: 0.000597
[31.01|17:44:09] training_set Epoch: 110 Loss: 0.000015
[31.01|17:44:09] validation_set Epoch: 110 Loss: 0.000949
[31.01|17:44:10] training_set Epoch: 120 Loss: 0.000013
[31.01|17:44:10] validation_set Epoch: 120 Loss: 0.000559
[31.01|17:44:10] training_set Epoch: 130 Loss: 0.000012
[31.01|17:44:10] validation_set Epoch: 130 Loss: 0.000625
[31.01|17:44:11] training_set Epoch: 140 Loss: 0.000010
[31.01|17:44:11] validation_set Epoch: 140 Loss: 0.000463
[31.01|17:44:11] training_set Epoch: 150 Loss: 0.000009
[31.01|17:44:11] validation_set Epoch: 150 Loss: 0.000463
[31.01|17:44:12] training_set Epoch: 160 Loss: 0.000008
[31.01|17:44:12] validation_set Epoch: 160 Loss: 0.000370
[31.01|17:44:12] training_set Epoch: 170 Loss: 0.000007
[31.01|17:44:12] validation_set Epoch: 170 Loss: 0.000337
[31.01|17:44:13] training_set Epoch: 180 Loss: 0.000006
[31.01|17:44:13] validation_set Epoch: 180 Loss: 0.000305
[31.01|17:44:13] training_set Epoch: 190 Loss: 0.000006
[31.01|17:44:13] validation_set Epoch: 190 Loss: 0.000270
[31.01|17:44:15] Execution of m3gnet.run finished with returncode 0
[31.01|17:44:15] JOB m3gnet FINISHED
[31.01|17:44:15] Starting m3gnet.postrun()
[31.01|17:44:15] m3gnet.postrun() finished
[31.01|17:44:15] JOB m3gnet SUCCESSFUL
[31.01|17:44:26] Running all jobs through AMS....
[31.01|17:44:26] Storing results/optimization/training_set_results/latest
[31.01|17:44:26] Storing results/optimization/validation_set_results/latest
[31.01|17:44:26] PLAMS environment cleaned up successfully
[31.01|17:44:26] PLAMS run finished. Goodbye
[31.01|17:44:26] Initial model has been trained!
[31.01|17:44:26] ParAMSResults training_set validation_set
[31.01|17:44:26] energy MAE 0.0006 0.0006 eV
[31.01|17:44:26] forces MAE 0.0313 0.0527 eV/angstrom
[31.01|17:44:26]
[31.01|17:44:26] Starting active learning loop...
[31.01|17:44:26] ##########################
[31.01|17:44:26] ### Step 1 / Attempt 1 ###
[31.01|17:44:26] ##########################
[31.01|17:44:26] MD Steps: 10 (cumulative: 10)
[31.01|17:44:26] Current engine settings:
[31.01|17:44:26]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/initial_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:44:26] Running step1_attempt1_simulation...
[31.01|17:44:37] Job step1_attempt1_simulation finished
[31.01|17:44:37] Deleting files that are no longer needed...
[31.01|17:44:37] Launching reference calculation
[31.01|17:44:38] Reference calculation finished!
[31.01|17:44:38] Checking success for step1_attempt1
[31.01|17:44:38] CheckEnergy: Checking energy for MDStep10, n_atoms = 8
[31.01|17:44:38] CheckEnergy: normalization coefficient = 8
[31.01|17:44:38] CheckEnergy: Actual Threshold
[31.01|17:44:38] CheckEnergy: ΔE/8 -0.0180 0.2000 OK!
[31.01|17:44:38]
[31.01|17:44:38] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:44:38] CheckForces: ------------
[31.01|17:44:38] CheckForces: Reference job from step1_attempt1_reference_calc1
[31.01|17:44:38] CheckForces: Prediction job from final frame (MDStep10) of step1_attempt1_simulation
[31.01|17:44:38] CheckForces: ------------
[31.01|17:44:38] CheckForces: Histogram of forces
[31.01|17:44:38] CheckForces: eV/Ang Ref Pred
[31.01|17:44:38] CheckForces: -3 1 0
[31.01|17:44:38] CheckForces: -2 3 1
[31.01|17:44:38] CheckForces: -1 9 11
[31.01|17:44:38] CheckForces: 0 8 11
[31.01|17:44:38] CheckForces: 1 2 1
[31.01|17:44:38] CheckForces: 2 1 0
[31.01|17:44:38] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:44:38] CheckForces: Force components with an error exceeding the threshold:
[31.01|17:44:38] CheckForces: Ref Pred Delta Threshold
[31.01|17:44:38] CheckForces: -1.21 -0.50 0.71 0.57
[31.01|17:44:38] CheckForces: 1.93 0.54 1.39 0.63
[31.01|17:44:38] CheckForces: 0.98 0.20 0.78 0.55
[31.01|17:44:38] CheckForces: -1.85 -0.10 1.74 0.62
[31.01|17:44:38] CheckForces: -2.38 -0.78 1.60 0.68
[31.01|17:44:38] CheckForces: -0.84 0.09 0.93 0.54
[31.01|17:44:38] CheckForces: 2.27 0.67 1.60 0.67
[31.01|17:44:38] CheckForces: 0.13 -0.51 0.65 0.51
[31.01|17:44:38] CheckForces: Maximum deviation: 1.742 eV/angstrom
[31.01|17:44:38] CheckForces: Actual Threshold
[31.01|17:44:38] CheckForces: # > thr. 8 0 Not OK!
[31.01|17:44:38] CheckForces: MAE 0.543 0.30 Not OK!
[31.01|17:44:38] CheckForces: R^2 0.592 0.80 Not OK!
[31.01|17:44:38] CheckForces: --------------------
[31.01|17:44:38]
[31.01|17:44:38] Adding results from step1_attempt1_reference_calc1 to training set
[31.01|17:44:38] Current # training set entries: 4
[31.01|17:44:38] Current # validation set entries: 1
[31.01|17:44:38] Storing data in step1_attempt1_reference_data
[31.01|17:44:38] Deleting initial_reference_data
[31.01|17:44:38] Deleting step1_attempt1_reference_calc1
[31.01|17:44:38]
[31.01|17:44:38] Current (cumulative) timings:
[31.01|17:44:38] Time (s) Fraction
[31.01|17:44:38] Ref. calcs 2.04 0.033
[31.01|17:44:38] ML training 48.63 0.794
[31.01|17:44:38] Simulations 10.58 0.173
[31.01|17:44:38]
[31.01|17:44:38]
[31.01|17:44:38]
[31.01|17:44:38] --- Begin summary ---
[31.01|17:44:38] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:44:38] 1 1 FAILED Inaccurate 1.7423
[31.01|17:44:38] --- End summary ---
[31.01|17:44:38]
[31.01|17:44:38] Running more reference calculations....
[31.01|17:44:38] Running reference calculations on frames [6] from step1_attempt1_simulation/ams.rkf
[31.01|17:44:38] Calculating 1 frames in total
[31.01|17:44:38] Running step1_attempt1_reference_calc2
[31.01|17:44:39] Reference calculations finished!
[31.01|17:44:39] Adding results from step1_attempt1_reference_calc2 to validation set
[31.01|17:44:39] Current # training set entries: 4
[31.01|17:44:39] Current # validation set entries: 2
[31.01|17:44:39] Storing data in step1_attempt1_reference_data
[31.01|17:44:39] Deleting step1_attempt1_reference_calc2
[31.01|17:44:39] Launching reparametrization job: step1_attempt1_training
[31.01|17:44:40] JOB m3gnet STARTED
[31.01|17:44:40] Starting m3gnet.prerun()
[31.01|17:44:40] m3gnet.prerun() finished
[31.01|17:44:40] JOB m3gnet RUNNING
[31.01|17:44:40] Executing m3gnet.run
[31.01|17:45:05] training_set Epoch: 0 Loss: 0.000565
[31.01|17:45:05] validation_set Epoch: 0 Loss: 0.077548
[31.01|17:45:05] training_set Epoch: 10 Loss: 0.001188
[31.01|17:45:05] validation_set Epoch: 10 Loss: 0.012764
[31.01|17:45:06] training_set Epoch: 20 Loss: 0.000858
[31.01|17:45:06] validation_set Epoch: 20 Loss: 0.020567
[31.01|17:45:06] training_set Epoch: 30 Loss: 0.000445
[31.01|17:45:06] validation_set Epoch: 30 Loss: 0.014934
[31.01|17:45:07] training_set Epoch: 40 Loss: 0.000296
[31.01|17:45:07] validation_set Epoch: 40 Loss: 0.009303
[31.01|17:45:07] training_set Epoch: 50 Loss: 0.000215
[31.01|17:45:07] validation_set Epoch: 50 Loss: 0.007049
[31.01|17:45:08] training_set Epoch: 60 Loss: 0.000145
[31.01|17:45:08] validation_set Epoch: 60 Loss: 0.004677
[31.01|17:45:08] training_set Epoch: 70 Loss: 0.000082
[31.01|17:45:08] validation_set Epoch: 70 Loss: 0.001991
[31.01|17:45:09] training_set Epoch: 80 Loss: 0.000051
[31.01|17:45:09] validation_set Epoch: 80 Loss: 0.001933
[31.01|17:45:09] training_set Epoch: 90 Loss: 0.000037
[31.01|17:45:09] validation_set Epoch: 90 Loss: 0.001878
[31.01|17:45:10] training_set Epoch: 100 Loss: 0.000031
[31.01|17:45:10] validation_set Epoch: 100 Loss: 0.001710
[31.01|17:45:10] training_set Epoch: 110 Loss: 0.000027
[31.01|17:45:10] validation_set Epoch: 110 Loss: 0.001554
[31.01|17:45:11] training_set Epoch: 120 Loss: 0.000025
[31.01|17:45:11] validation_set Epoch: 120 Loss: 0.001411
[31.01|17:45:11] training_set Epoch: 130 Loss: 0.000023
[31.01|17:45:11] validation_set Epoch: 130 Loss: 0.001322
[31.01|17:45:12] training_set Epoch: 140 Loss: 0.000022
[31.01|17:45:12] validation_set Epoch: 140 Loss: 0.001259
[31.01|17:45:12] training_set Epoch: 150 Loss: 0.000020
[31.01|17:45:12] validation_set Epoch: 150 Loss: 0.001211
[31.01|17:45:13] training_set Epoch: 160 Loss: 0.000019
[31.01|17:45:13] validation_set Epoch: 160 Loss: 0.001169
[31.01|17:45:13] training_set Epoch: 170 Loss: 0.000018
[31.01|17:45:13] validation_set Epoch: 170 Loss: 0.001135
[31.01|17:45:14] training_set Epoch: 180 Loss: 0.000017
[31.01|17:45:14] validation_set Epoch: 180 Loss: 0.001105
[31.01|17:45:14] training_set Epoch: 190 Loss: 0.000016
[31.01|17:45:14] validation_set Epoch: 190 Loss: 0.001078
[31.01|17:45:16] Execution of m3gnet.run finished with returncode 0
[31.01|17:45:16] JOB m3gnet FINISHED
[31.01|17:45:16] Starting m3gnet.postrun()
[31.01|17:45:16] m3gnet.postrun() finished
[31.01|17:45:16] JOB m3gnet SUCCESSFUL
[31.01|17:45:27] Running all jobs through AMS....
[31.01|17:45:28] Storing results/optimization/training_set_results/latest
[31.01|17:45:28] Storing results/optimization/validation_set_results/latest
[31.01|17:45:28] PLAMS environment cleaned up successfully
[31.01|17:45:28] PLAMS run finished. Goodbye
[31.01|17:45:28] ParAMSResults training_set validation_set
[31.01|17:45:28] energy MAE 0.0042 0.0063 eV
[31.01|17:45:28] forces MAE 0.0504 0.0809 eV/angstrom
[31.01|17:45:28] Newly created parameter file/dir: step1_attempt1_training/results/optimization/m3gnet/m3gnet
[31.01|17:45:28] Done!
[31.01|17:45:28] Deleting initial_training
[31.01|17:45:28] ##########################
[31.01|17:45:28] ### Step 1 / Attempt 2 ###
[31.01|17:45:28] ##########################
[31.01|17:45:28] MD Steps: 10 (cumulative: 10)
[31.01|17:45:28] Current engine settings:
[31.01|17:45:28]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step1_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:45:28] Running step1_attempt2_simulation...
[31.01|17:45:39] Job step1_attempt2_simulation finished
[31.01|17:45:39] Deleting files that are no longer needed...
[31.01|17:45:39] Launching reference calculation
[31.01|17:45:40] Reference calculation finished!
[31.01|17:45:40] Checking success for step1_attempt2
[31.01|17:45:50] CheckEnergy: Checking energy for MDStep10, n_atoms = 8
[31.01|17:45:50] CheckEnergy: normalization coefficient = 8
[31.01|17:45:50] CheckEnergy: Actual Threshold
[31.01|17:45:50] CheckEnergy: ΔE/8 -0.0034 0.2000 OK!
[31.01|17:45:50] CheckEnergy: ΔΔE/8 -0.0045 0.0050 OK! (relative to step1_attempt1_simulation:MDStep10)
[31.01|17:45:50]
[31.01|17:45:50] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:45:50] CheckForces: ------------
[31.01|17:45:50] CheckForces: Reference job from step1_attempt2_reference_calc1
[31.01|17:45:50] CheckForces: Prediction job from final frame (MDStep10) of step1_attempt2_simulation
[31.01|17:45:50] CheckForces: ------------
[31.01|17:45:50] CheckForces: Histogram of forces
[31.01|17:45:50] CheckForces: eV/Ang Ref Pred
[31.01|17:45:50] CheckForces: -2 1 0
[31.01|17:45:50] CheckForces: -1 13 15
[31.01|17:45:50] CheckForces: 0 8 7
[31.01|17:45:50] CheckForces: 1 2 2
[31.01|17:45:50] CheckForces: 2 0 0
[31.01|17:45:50] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:45:50] CheckForces: All force components are within the acceptable error!
[31.01|17:45:50] CheckForces: Maximum deviation: 0.573 eV/angstrom
[31.01|17:45:50] CheckForces: Actual Threshold
[31.01|17:45:50] CheckForces: # > thr. 0 0 OK!
[31.01|17:45:50] CheckForces: MAE 0.157 0.30 OK!
[31.01|17:45:50] CheckForces: R^2 0.929 0.80 OK!
[31.01|17:45:50] CheckForces: --------------------
[31.01|17:45:50]
[31.01|17:45:50] Adding results from step1_attempt2_reference_calc1 to validation set
[31.01|17:45:50] Current # training set entries: 4
[31.01|17:45:50] Current # validation set entries: 3
[31.01|17:45:50] Storing data in step1_attempt2_reference_data
[31.01|17:45:50] Deleting step1_attempt1_reference_data
[31.01|17:45:50] Deleting step1_attempt2_reference_calc1
[31.01|17:45:50]
[31.01|17:45:50] Current (cumulative) timings:
[31.01|17:45:50] Time (s) Fraction
[31.01|17:45:50] Ref. calcs 3.86 0.031
[31.01|17:45:50] ML training 97.72 0.794
[31.01|17:45:50] Simulations 21.56 0.175
[31.01|17:45:50]
[31.01|17:45:50]
[31.01|17:45:50] Step 1 finished successfully!
[31.01|17:45:50]
[31.01|17:45:50] --- Begin summary ---
[31.01|17:45:50] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:45:50] 1 1 FAILED Inaccurate 1.7423
[31.01|17:45:50] 1 2 SUCCESS Accurate 0.5734
[31.01|17:45:50] --- End summary ---
[31.01|17:45:50]
[31.01|17:45:50] ##########################
[31.01|17:45:50] ### Step 2 / Attempt 1 ###
[31.01|17:45:50] ##########################
[31.01|17:45:50] MD Steps: 46 (cumulative: 56)
[31.01|17:45:50] Current engine settings:
[31.01|17:45:50]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step1_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:45:50] Running step2_attempt1_simulation...
[31.01|17:46:01] Job step2_attempt1_simulation finished
[31.01|17:46:01] Deleting files that are no longer needed...
[31.01|17:46:01] Deleting step1_attempt1_simulation
[31.01|17:46:01] Launching reference calculation
[31.01|17:46:02] Reference calculation finished!
[31.01|17:46:02] Checking success for step2_attempt1
[31.01|17:46:13] CheckEnergy: Checking energy for MDStep56, n_atoms = 8
[31.01|17:46:13] CheckEnergy: normalization coefficient = 8
[31.01|17:46:13] CheckEnergy: Actual Threshold
[31.01|17:46:13] CheckEnergy: ΔE/8 -0.0163 0.2000 OK!
[31.01|17:46:13] CheckEnergy: ΔΔE/8 -0.0129 0.0050 Not OK! (relative to step1_attempt2_simulation:MDStep10)
[31.01|17:46:13]
[31.01|17:46:13] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:46:13] CheckForces: ------------
[31.01|17:46:13] CheckForces: Reference job from step2_attempt1_reference_calc1
[31.01|17:46:13] CheckForces: Prediction job from final frame (MDStep56) of step2_attempt1_simulation
[31.01|17:46:13] CheckForces: ------------
[31.01|17:46:13] CheckForces: Histogram of forces
[31.01|17:46:13] CheckForces: eV/Ang Ref Pred
[31.01|17:46:13] CheckForces: -4 1 0
[31.01|17:46:13] CheckForces: -3 0 1
[31.01|17:46:13] CheckForces: -2 2 1
[31.01|17:46:13] CheckForces: -1 10 9
[31.01|17:46:13] CheckForces: 0 7 10
[31.01|17:46:13] CheckForces: 1 3 3
[31.01|17:46:13] CheckForces: 2 1 0
[31.01|17:46:13] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:46:13] CheckForces: Force components with an error exceeding the threshold:
[31.01|17:46:13] CheckForces: Ref Pred Delta Threshold
[31.01|17:46:13] CheckForces: 2.25 1.18 1.07 0.67
[31.01|17:46:13] CheckForces: -0.94 -0.37 0.57 0.55
[31.01|17:46:13] CheckForces: Maximum deviation: 1.068 eV/angstrom
[31.01|17:46:13] CheckForces: Actual Threshold
[31.01|17:46:13] CheckForces: # > thr. 2 0 Not OK!
[31.01|17:46:13] CheckForces: MAE 0.279 0.30 OK!
[31.01|17:46:13] CheckForces: R^2 0.933 0.80 OK!
[31.01|17:46:13] CheckForces: --------------------
[31.01|17:46:13]
[31.01|17:46:13] Adding results from step2_attempt1_reference_calc1 to training set
[31.01|17:46:13] Current # training set entries: 5
[31.01|17:46:13] Current # validation set entries: 3
[31.01|17:46:13] Storing data in step2_attempt1_reference_data
[31.01|17:46:13] Deleting step1_attempt2_reference_data
[31.01|17:46:13] Deleting step2_attempt1_reference_calc1
[31.01|17:46:13]
[31.01|17:46:13] Current (cumulative) timings:
[31.01|17:46:13] Time (s) Fraction
[31.01|17:46:13] Ref. calcs 4.64 0.034
[31.01|17:46:13] ML training 97.72 0.725
[31.01|17:46:13] Simulations 32.51 0.241
[31.01|17:46:13]
[31.01|17:46:13]
[31.01|17:46:13]
[31.01|17:46:13] --- Begin summary ---
[31.01|17:46:13] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:46:13] 1 1 FAILED Inaccurate 1.7423
[31.01|17:46:13] 1 2 SUCCESS Accurate 0.5734
[31.01|17:46:13] 2 1 FAILED Inaccurate 1.0679
[31.01|17:46:13] --- End summary ---
[31.01|17:46:13]
[31.01|17:46:13] Running more reference calculations....
[31.01|17:46:13] Running reference calculations on frames [17] from step2_attempt1_simulation/ams.rkf
[31.01|17:46:13] Calculating 1 frames in total
[31.01|17:46:13] Running step2_attempt1_reference_calc2
[31.01|17:46:13] Reference calculations finished!
[31.01|17:46:14] Adding results from step2_attempt1_reference_calc2 to validation set
[31.01|17:46:14] Current # training set entries: 5
[31.01|17:46:14] Current # validation set entries: 4
[31.01|17:46:14] Storing data in step2_attempt1_reference_data
[31.01|17:46:14] Deleting step2_attempt1_reference_calc2
[31.01|17:46:14] Launching reparametrization job: step2_attempt1_training
[31.01|17:46:15] JOB m3gnet STARTED
[31.01|17:46:15] Starting m3gnet.prerun()
[31.01|17:46:15] m3gnet.prerun() finished
[31.01|17:46:15] JOB m3gnet RUNNING
[31.01|17:46:15] Executing m3gnet.run
[31.01|17:46:39] training_set Epoch: 0 Loss: 0.000139
[31.01|17:46:39] validation_set Epoch: 0 Loss: 0.070191
[31.01|17:46:40] training_set Epoch: 10 Loss: 0.000339
[31.01|17:46:40] validation_set Epoch: 10 Loss: 0.011495
[31.01|17:46:40] training_set Epoch: 20 Loss: 0.000375
[31.01|17:46:40] validation_set Epoch: 20 Loss: 0.009220
[31.01|17:46:41] training_set Epoch: 30 Loss: 0.000277
[31.01|17:46:41] validation_set Epoch: 30 Loss: 0.010914
[31.01|17:46:41] training_set Epoch: 40 Loss: 0.000153
[31.01|17:46:41] validation_set Epoch: 40 Loss: 0.009465
[31.01|17:46:42] training_set Epoch: 50 Loss: 0.000098
[31.01|17:46:42] validation_set Epoch: 50 Loss: 0.007721
[31.01|17:46:42] training_set Epoch: 60 Loss: 0.000070
[31.01|17:46:42] validation_set Epoch: 60 Loss: 0.006123
[31.01|17:46:43] training_set Epoch: 70 Loss: 0.000056
[31.01|17:46:43] validation_set Epoch: 70 Loss: 0.006239
[31.01|17:46:43] training_set Epoch: 80 Loss: 0.000051
[31.01|17:46:43] validation_set Epoch: 80 Loss: 0.006012
[31.01|17:46:44] training_set Epoch: 90 Loss: 0.000043
[31.01|17:46:44] validation_set Epoch: 90 Loss: 0.005937
[31.01|17:46:44] training_set Epoch: 100 Loss: 0.000038
[31.01|17:46:44] validation_set Epoch: 100 Loss: 0.005861
[31.01|17:46:45] training_set Epoch: 110 Loss: 0.000034
[31.01|17:46:45] validation_set Epoch: 110 Loss: 0.005772
[31.01|17:46:45] training_set Epoch: 120 Loss: 0.000030
[31.01|17:46:45] validation_set Epoch: 120 Loss: 0.005683
[31.01|17:46:46] training_set Epoch: 130 Loss: 0.000026
[31.01|17:46:46] validation_set Epoch: 130 Loss: 0.005601
[31.01|17:46:46] training_set Epoch: 140 Loss: 0.000023
[31.01|17:46:46] validation_set Epoch: 140 Loss: 0.005525
[31.01|17:46:47] training_set Epoch: 150 Loss: 0.000021
[31.01|17:46:47] validation_set Epoch: 150 Loss: 0.005458
[31.01|17:46:47] training_set Epoch: 160 Loss: 0.000019
[31.01|17:46:47] validation_set Epoch: 160 Loss: 0.005396
[31.01|17:46:48] training_set Epoch: 170 Loss: 0.000017
[31.01|17:46:48] validation_set Epoch: 170 Loss: 0.005336
[31.01|17:46:48] training_set Epoch: 180 Loss: 0.000015
[31.01|17:46:48] validation_set Epoch: 180 Loss: 0.005280
[31.01|17:46:49] training_set Epoch: 190 Loss: 0.000014
[31.01|17:46:49] validation_set Epoch: 190 Loss: 0.005225
[31.01|17:46:50] Execution of m3gnet.run finished with returncode 0
[31.01|17:46:50] JOB m3gnet FINISHED
[31.01|17:46:50] Starting m3gnet.postrun()
[31.01|17:46:50] m3gnet.postrun() finished
[31.01|17:46:50] JOB m3gnet SUCCESSFUL
[31.01|17:47:01] Running all jobs through AMS....
[31.01|17:47:01] Storing results/optimization/training_set_results/latest
[31.01|17:47:01] Storing results/optimization/validation_set_results/latest
[31.01|17:47:01] PLAMS environment cleaned up successfully
[31.01|17:47:01] PLAMS run finished. Goodbye
[31.01|17:47:01] ParAMSResults training_set validation_set
[31.01|17:47:01] energy MAE 0.0067 0.0376 eV
[31.01|17:47:01] forces MAE 0.0468 0.1445 eV/angstrom
[31.01|17:47:01] Newly created parameter file/dir: step2_attempt1_training/results/optimization/m3gnet/m3gnet
[31.01|17:47:01] Done!
[31.01|17:47:01] Deleting step1_attempt1_training
[31.01|17:47:01] ##########################
[31.01|17:47:01] ### Step 2 / Attempt 2 ###
[31.01|17:47:01] ##########################
[31.01|17:47:01] MD Steps: 46 (cumulative: 56)
[31.01|17:47:01] Current engine settings:
[31.01|17:47:01]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step2_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:47:01] Running step2_attempt2_simulation...
[31.01|17:47:13] Job step2_attempt2_simulation finished
[31.01|17:47:13] Deleting files that are no longer needed...
[31.01|17:47:13] Launching reference calculation
[31.01|17:47:14] Reference calculation finished!
[31.01|17:47:14] Checking success for step2_attempt2
[31.01|17:47:25] CheckEnergy: Checking energy for MDStep56, n_atoms = 8
[31.01|17:47:25] CheckEnergy: normalization coefficient = 8
[31.01|17:47:25] CheckEnergy: Actual Threshold
[31.01|17:47:25] CheckEnergy: ΔE/8 0.0012 0.2000 OK!
[31.01|17:47:25] CheckEnergy: ΔΔE/8 0.0017 0.0050 OK! (relative to step2_attempt1_simulation:MDStep56)
[31.01|17:47:25]
[31.01|17:47:25] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:47:25] CheckForces: ------------
[31.01|17:47:25] CheckForces: Reference job from step2_attempt2_reference_calc1
[31.01|17:47:25] CheckForces: Prediction job from final frame (MDStep56) of step2_attempt2_simulation
[31.01|17:47:25] CheckForces: ------------
[31.01|17:47:25] CheckForces: Histogram of forces
[31.01|17:47:25] CheckForces: eV/Ang Ref Pred
[31.01|17:47:25] CheckForces: -4 0 0
[31.01|17:47:25] CheckForces: -3 1 1
[31.01|17:47:25] CheckForces: -2 1 1
[31.01|17:47:25] CheckForces: -1 10 11
[31.01|17:47:25] CheckForces: 0 9 8
[31.01|17:47:25] CheckForces: 1 2 3
[31.01|17:47:25] CheckForces: 2 1 0
[31.01|17:47:25] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:47:25] CheckForces: All force components are within the acceptable error!
[31.01|17:47:25] CheckForces: Maximum deviation: 0.392 eV/angstrom
[31.01|17:47:25] CheckForces: Actual Threshold
[31.01|17:47:25] CheckForces: # > thr. 0 0 OK!
[31.01|17:47:25] CheckForces: MAE 0.104 0.30 OK!
[31.01|17:47:25] CheckForces: R^2 0.980 0.80 OK!
[31.01|17:47:25] CheckForces: --------------------
[31.01|17:47:25]
[31.01|17:47:25] Adding results from step2_attempt2_reference_calc1 to training set
[31.01|17:47:25] Current # training set entries: 6
[31.01|17:47:25] Current # validation set entries: 4
[31.01|17:47:25] Storing data in step2_attempt2_reference_data
[31.01|17:47:25] Deleting step2_attempt1_reference_data
[31.01|17:47:25] Deleting step2_attempt2_reference_calc1
[31.01|17:47:25]
[31.01|17:47:25] Current (cumulative) timings:
[31.01|17:47:25] Time (s) Fraction
[31.01|17:47:25] Ref. calcs 7.06 0.036
[31.01|17:47:25] ML training 145.41 0.740
[31.01|17:47:25] Simulations 44.08 0.224
[31.01|17:47:25]
[31.01|17:47:25]
[31.01|17:47:25] Step 2 finished successfully!
[31.01|17:47:25]
[31.01|17:47:25] --- Begin summary ---
[31.01|17:47:25] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:47:25] 1 1 FAILED Inaccurate 1.7423
[31.01|17:47:25] 1 2 SUCCESS Accurate 0.5734
[31.01|17:47:25] 2 1 FAILED Inaccurate 1.0679
[31.01|17:47:25] 2 2 SUCCESS Accurate 0.3920
[31.01|17:47:25] --- End summary ---
[31.01|17:47:25]
[31.01|17:47:25] ##########################
[31.01|17:47:25] ### Step 3 / Attempt 1 ###
[31.01|17:47:25] ##########################
[31.01|17:47:25] MD Steps: 260 (cumulative: 316)
[31.01|17:47:25] Current engine settings:
[31.01|17:47:25]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step2_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:47:25] Running step3_attempt1_simulation...
[31.01|17:47:39] Job step3_attempt1_simulation finished
[31.01|17:47:39] Deleting files that are no longer needed...
[31.01|17:47:39] Deleting step1_attempt2_simulation
[31.01|17:47:39] Deleting step2_attempt1_simulation
[31.01|17:47:39] Launching reference calculation
[31.01|17:47:40] Reference calculation finished!
[31.01|17:47:40] Checking success for step3_attempt1
[31.01|17:47:50] CheckEnergy: Checking energy for MDStep316, n_atoms = 8
[31.01|17:47:50] CheckEnergy: normalization coefficient = 8
[31.01|17:47:50] CheckEnergy: Actual Threshold
[31.01|17:47:50] CheckEnergy: ΔE/8 0.0014 0.2000 OK!
[31.01|17:47:50] CheckEnergy: ΔΔE/8 0.0002 0.0050 OK! (relative to step2_attempt2_simulation:MDStep56)
[31.01|17:47:50]
[31.01|17:47:50] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:47:50] CheckForces: ------------
[31.01|17:47:50] CheckForces: Reference job from step3_attempt1_reference_calc1
[31.01|17:47:50] CheckForces: Prediction job from final frame (MDStep316) of step3_attempt1_simulation
[31.01|17:47:50] CheckForces: ------------
[31.01|17:47:50] CheckForces: Histogram of forces
[31.01|17:47:50] CheckForces: eV/Ang Ref Pred
[31.01|17:47:50] CheckForces: -3 0 0
[31.01|17:47:50] CheckForces: -2 2 1
[31.01|17:47:50] CheckForces: -1 8 10
[31.01|17:47:50] CheckForces: 0 13 11
[31.01|17:47:50] CheckForces: 1 1 2
[31.01|17:47:50] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:47:50] CheckForces: Force components with an error exceeding the threshold:
[31.01|17:47:50] CheckForces: Ref Pred Delta Threshold
[31.01|17:47:50] CheckForces: 0.14 1.11 0.98 0.51
[31.01|17:47:50] CheckForces: Maximum deviation: 0.976 eV/angstrom
[31.01|17:47:50] CheckForces: Actual Threshold
[31.01|17:47:50] CheckForces: # > thr. 1 0 Not OK!
[31.01|17:47:50] CheckForces: MAE 0.171 0.30 OK!
[31.01|17:47:50] CheckForces: R^2 0.856 0.80 OK!
[31.01|17:47:50] CheckForces: --------------------
[31.01|17:47:50]
[31.01|17:47:50] Adding results from step3_attempt1_reference_calc1 to training set
[31.01|17:47:50] Current # training set entries: 7
[31.01|17:47:50] Current # validation set entries: 4
[31.01|17:47:50] Storing data in step3_attempt1_reference_data
[31.01|17:47:50] Deleting step2_attempt2_reference_data
[31.01|17:47:50] Deleting step3_attempt1_reference_calc1
[31.01|17:47:50]
[31.01|17:47:50] Current (cumulative) timings:
[31.01|17:47:50] Time (s) Fraction
[31.01|17:47:50] Ref. calcs 7.84 0.037
[31.01|17:47:50] ML training 145.41 0.690
[31.01|17:47:50] Simulations 57.37 0.272
[31.01|17:47:50]
[31.01|17:47:50]
[31.01|17:47:51]
[31.01|17:47:51] --- Begin summary ---
[31.01|17:47:51] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:47:51] 1 1 FAILED Inaccurate 1.7423
[31.01|17:47:51] 1 2 SUCCESS Accurate 0.5734
[31.01|17:47:51] 2 1 FAILED Inaccurate 1.0679
[31.01|17:47:51] 2 2 SUCCESS Accurate 0.3920
[31.01|17:47:51] 3 1 FAILED Inaccurate 0.9762
[31.01|17:47:51] --- End summary ---
[31.01|17:47:51]
[31.01|17:47:51] Running more reference calculations....
[31.01|17:47:51] Running reference calculations on frames [26, 28, 30] from step3_attempt1_simulation/ams.rkf
[31.01|17:47:51] Calculating 3 frames in total
[31.01|17:47:51] Running step3_attempt1_reference_calc2
[31.01|17:47:51] Running step3_attempt1_reference_calc3
[31.01|17:47:52] Running step3_attempt1_reference_calc4
[31.01|17:47:53] Reference calculations finished!
[31.01|17:47:53] Adding results from step3_attempt1_reference_calc2 to validation set
[31.01|17:47:53] Adding results from step3_attempt1_reference_calc3 to training set
[31.01|17:47:53] Adding results from step3_attempt1_reference_calc4 to training set
[31.01|17:47:53] Current # training set entries: 9
[31.01|17:47:53] Current # validation set entries: 5
[31.01|17:47:53] Storing data in step3_attempt1_reference_data
[31.01|17:47:53] Deleting step3_attempt1_reference_calc2
[31.01|17:47:53] Deleting step3_attempt1_reference_calc3
[31.01|17:47:53] Deleting step3_attempt1_reference_calc4
[31.01|17:47:53] Launching reparametrization job: step3_attempt1_training
[31.01|17:47:54] JOB m3gnet STARTED
[31.01|17:47:54] Starting m3gnet.prerun()
[31.01|17:47:54] m3gnet.prerun() finished
[31.01|17:47:54] JOB m3gnet RUNNING
[31.01|17:47:54] Executing m3gnet.run
[31.01|17:48:39] training_set Epoch: 0 Loss: 0.003587
[31.01|17:48:39] validation_set Epoch: 0 Loss: 0.009850
[31.01|17:48:40] training_set Epoch: 10 Loss: 0.000570
[31.01|17:48:40] validation_set Epoch: 10 Loss: 0.010250
[31.01|17:48:41] training_set Epoch: 20 Loss: 0.000092
[31.01|17:48:41] validation_set Epoch: 20 Loss: 0.006772
[31.01|17:48:41] training_set Epoch: 30 Loss: 0.000063
[31.01|17:48:41] validation_set Epoch: 30 Loss: 0.005755
[31.01|17:48:42] training_set Epoch: 40 Loss: 0.000053
[31.01|17:48:42] validation_set Epoch: 40 Loss: 0.005245
[31.01|17:48:43] training_set Epoch: 50 Loss: 0.000044
[31.01|17:48:43] validation_set Epoch: 50 Loss: 0.005616
[31.01|17:48:44] training_set Epoch: 60 Loss: 0.000041
[31.01|17:48:44] validation_set Epoch: 60 Loss: 0.005354
[31.01|17:48:45] training_set Epoch: 70 Loss: 0.000035
[31.01|17:48:45] validation_set Epoch: 70 Loss: 0.005564
[31.01|17:48:45] training_set Epoch: 80 Loss: 0.000032
[31.01|17:48:45] validation_set Epoch: 80 Loss: 0.004750
[31.01|17:48:46] training_set Epoch: 90 Loss: 0.000030
[31.01|17:48:46] validation_set Epoch: 90 Loss: 0.004704
[31.01|17:48:47] training_set Epoch: 100 Loss: 0.000028
[31.01|17:48:47] validation_set Epoch: 100 Loss: 0.003894
[31.01|17:48:48] training_set Epoch: 110 Loss: 0.000025
[31.01|17:48:48] validation_set Epoch: 110 Loss: 0.004790
[31.01|17:48:49] training_set Epoch: 120 Loss: 0.000022
[31.01|17:48:49] validation_set Epoch: 120 Loss: 0.003588
[31.01|17:48:50] training_set Epoch: 130 Loss: 0.000022
[31.01|17:48:50] validation_set Epoch: 130 Loss: 0.004835
[31.01|17:48:50] training_set Epoch: 140 Loss: 0.000018
[31.01|17:48:50] validation_set Epoch: 140 Loss: 0.003450
[31.01|17:48:51] training_set Epoch: 150 Loss: 0.000018
[31.01|17:48:51] validation_set Epoch: 150 Loss: 0.003629
[31.01|17:48:52] training_set Epoch: 160 Loss: 0.000016
[31.01|17:48:52] validation_set Epoch: 160 Loss: 0.003157
[31.01|17:48:53] training_set Epoch: 170 Loss: 0.000016
[31.01|17:48:53] validation_set Epoch: 170 Loss: 0.003054
[31.01|17:48:54] training_set Epoch: 180 Loss: 0.000021
[31.01|17:48:54] validation_set Epoch: 180 Loss: 0.003291
[31.01|17:48:55] training_set Epoch: 190 Loss: 0.000015
[31.01|17:48:55] validation_set Epoch: 190 Loss: 0.004234
[31.01|17:48:57] Execution of m3gnet.run finished with returncode 0
[31.01|17:48:57] JOB m3gnet FINISHED
[31.01|17:48:57] Starting m3gnet.postrun()
[31.01|17:48:57] m3gnet.postrun() finished
[31.01|17:48:57] JOB m3gnet SUCCESSFUL
[31.01|17:49:08] Running all jobs through AMS....
[31.01|17:49:08] Storing results/optimization/training_set_results/latest
[31.01|17:49:08] Storing results/optimization/validation_set_results/latest
[31.01|17:49:08] PLAMS environment cleaned up successfully
[31.01|17:49:08] PLAMS run finished. Goodbye
[31.01|17:49:08] ParAMSResults training_set validation_set
[31.01|17:49:08] energy MAE 0.0069 0.0241 eV
[31.01|17:49:08] forces MAE 0.0433 0.1210 eV/angstrom
[31.01|17:49:08] Newly created parameter file/dir: step3_attempt1_training/results/optimization/m3gnet/m3gnet
[31.01|17:49:08] Done!
[31.01|17:49:08] Deleting step2_attempt1_training
[31.01|17:49:08] ##########################
[31.01|17:49:08] ### Step 3 / Attempt 2 ###
[31.01|17:49:08] ##########################
[31.01|17:49:08] MD Steps: 260 (cumulative: 316)
[31.01|17:49:08] Current engine settings:
[31.01|17:49:08]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step3_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:49:08] Running step3_attempt2_simulation...
[31.01|17:49:22] Job step3_attempt2_simulation finished
[31.01|17:49:22] Deleting files that are no longer needed...
[31.01|17:49:22] Launching reference calculation
[31.01|17:49:23] Reference calculation finished!
[31.01|17:49:23] Checking success for step3_attempt2
[31.01|17:49:34] CheckEnergy: Checking energy for MDStep316, n_atoms = 8
[31.01|17:49:34] CheckEnergy: normalization coefficient = 8
[31.01|17:49:34] CheckEnergy: Actual Threshold
[31.01|17:49:34] CheckEnergy: ΔE/8 -0.0065 0.2000 OK!
[31.01|17:49:34] CheckEnergy: ΔΔE/8 -0.0059 0.0050 Not OK! (relative to step3_attempt1_simulation:MDStep316)
[31.01|17:49:34]
[31.01|17:49:34] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:49:34] CheckForces: ------------
[31.01|17:49:34] CheckForces: Reference job from step3_attempt2_reference_calc1
[31.01|17:49:34] CheckForces: Prediction job from final frame (MDStep316) of step3_attempt2_simulation
[31.01|17:49:34] CheckForces: ------------
[31.01|17:49:34] CheckForces: Histogram of forces
[31.01|17:49:34] CheckForces: eV/Ang Ref Pred
[31.01|17:49:34] CheckForces: -4 0 0
[31.01|17:49:34] CheckForces: -3 1 1
[31.01|17:49:34] CheckForces: -2 2 2
[31.01|17:49:34] CheckForces: -1 9 9
[31.01|17:49:34] CheckForces: 0 10 9
[31.01|17:49:34] CheckForces: 1 1 3
[31.01|17:49:34] CheckForces: 2 1 0
[31.01|17:49:34] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:49:34] CheckForces: Force components with an error exceeding the threshold:
[31.01|17:49:34] CheckForces: Ref Pred Delta Threshold
[31.01|17:49:34] CheckForces: 2.31 1.56 0.76 0.67
[31.01|17:49:34] CheckForces: Maximum deviation: 0.756 eV/angstrom
[31.01|17:49:34] CheckForces: Actual Threshold
[31.01|17:49:34] CheckForces: # > thr. 1 0 Not OK!
[31.01|17:49:34] CheckForces: MAE 0.177 0.30 OK!
[31.01|17:49:34] CheckForces: R^2 0.969 0.80 OK!
[31.01|17:49:34] CheckForces: --------------------
[31.01|17:49:34]
[31.01|17:49:34] Adding results from step3_attempt2_reference_calc1 to training set
[31.01|17:49:34] Current # training set entries: 10
[31.01|17:49:34] Current # validation set entries: 5
[31.01|17:49:34] Storing data in step3_attempt2_reference_data
[31.01|17:49:34] Deleting step3_attempt1_reference_data
[31.01|17:49:34] Deleting step3_attempt2_reference_calc1
[31.01|17:49:34]
[31.01|17:49:34] Current (cumulative) timings:
[31.01|17:49:34] Time (s) Fraction
[31.01|17:49:34] Ref. calcs 11.64 0.038
[31.01|17:49:34] ML training 220.84 0.728
[31.01|17:49:34] Simulations 70.76 0.233
[31.01|17:49:34]
[31.01|17:49:34]
[31.01|17:49:34]
[31.01|17:49:34] --- Begin summary ---
[31.01|17:49:34] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:49:34] 1 1 FAILED Inaccurate 1.7423
[31.01|17:49:34] 1 2 SUCCESS Accurate 0.5734
[31.01|17:49:34] 2 1 FAILED Inaccurate 1.0679
[31.01|17:49:34] 2 2 SUCCESS Accurate 0.3920
[31.01|17:49:34] 3 1 FAILED Inaccurate 0.9762
[31.01|17:49:34] 3 2 FAILED Inaccurate 0.7560
[31.01|17:49:34] --- End summary ---
[31.01|17:49:34]
[31.01|17:49:34] Running more reference calculations....
[31.01|17:49:34] Running reference calculations on frames [26, 28, 30] from step3_attempt2_simulation/ams.rkf
[31.01|17:49:34] Calculating 3 frames in total
[31.01|17:49:34] Running step3_attempt2_reference_calc2
[31.01|17:49:35] Running step3_attempt2_reference_calc3
[31.01|17:49:36] Running step3_attempt2_reference_calc4
[31.01|17:49:36] Reference calculations finished!
[31.01|17:49:36] Adding results from step3_attempt2_reference_calc2 to validation set
[31.01|17:49:36] Adding results from step3_attempt2_reference_calc3 to training set
[31.01|17:49:36] Adding results from step3_attempt2_reference_calc4 to training set
[31.01|17:49:36] Current # training set entries: 12
[31.01|17:49:36] Current # validation set entries: 6
[31.01|17:49:36] Storing data in step3_attempt2_reference_data
[31.01|17:49:36] Deleting step3_attempt2_reference_calc2
[31.01|17:49:36] Deleting step3_attempt2_reference_calc3
[31.01|17:49:36] Deleting step3_attempt2_reference_calc4
[31.01|17:49:36] Launching reparametrization job: step3_attempt2_training
[31.01|17:49:38] JOB m3gnet STARTED
[31.01|17:49:38] Starting m3gnet.prerun()
[31.01|17:49:38] m3gnet.prerun() finished
[31.01|17:49:38] JOB m3gnet RUNNING
[31.01|17:49:38] Executing m3gnet.run
[31.01|17:50:24] training_set Epoch: 0 Loss: 0.002272
[31.01|17:50:24] validation_set Epoch: 0 Loss: 0.056426
[31.01|17:50:25] training_set Epoch: 10 Loss: 0.000177
[31.01|17:50:25] validation_set Epoch: 10 Loss: 0.007855
[31.01|17:50:27] training_set Epoch: 20 Loss: 0.000050
[31.01|17:50:27] validation_set Epoch: 20 Loss: 0.002337
[31.01|17:50:28] training_set Epoch: 30 Loss: 0.000040
[31.01|17:50:28] validation_set Epoch: 30 Loss: 0.003837
[31.01|17:50:29] training_set Epoch: 40 Loss: 0.000031
[31.01|17:50:29] validation_set Epoch: 40 Loss: 0.003990
[31.01|17:50:30] training_set Epoch: 50 Loss: 0.000025
[31.01|17:50:30] validation_set Epoch: 50 Loss: 0.002706
[31.01|17:50:31] training_set Epoch: 60 Loss: 0.000031
[31.01|17:50:31] validation_set Epoch: 60 Loss: 0.002047
[31.01|17:50:32] training_set Epoch: 70 Loss: 0.000022
[31.01|17:50:32] validation_set Epoch: 70 Loss: 0.003074
[31.01|17:50:34] training_set Epoch: 80 Loss: 0.000023
[31.01|17:50:34] validation_set Epoch: 80 Loss: 0.001971
[31.01|17:50:35] training_set Epoch: 90 Loss: 0.000016
[31.01|17:50:35] validation_set Epoch: 90 Loss: 0.001933
[31.01|17:50:36] training_set Epoch: 100 Loss: 0.000018
[31.01|17:50:36] validation_set Epoch: 100 Loss: 0.002079
[31.01|17:50:37] training_set Epoch: 110 Loss: 0.000016
[31.01|17:50:37] validation_set Epoch: 110 Loss: 0.002864
[31.01|17:50:38] training_set Epoch: 120 Loss: 0.000016
[31.01|17:50:38] validation_set Epoch: 120 Loss: 0.001923
[31.01|17:50:39] training_set Epoch: 130 Loss: 0.000018
[31.01|17:50:39] validation_set Epoch: 130 Loss: 0.002417
[31.01|17:50:41] training_set Epoch: 140 Loss: 0.000014
[31.01|17:50:41] validation_set Epoch: 140 Loss: 0.002699
[31.01|17:50:42] training_set Epoch: 150 Loss: 0.000016
[31.01|17:50:42] validation_set Epoch: 150 Loss: 0.002029
[31.01|17:50:43] training_set Epoch: 160 Loss: 0.000012
[31.01|17:50:43] validation_set Epoch: 160 Loss: 0.001256
[31.01|17:50:44] training_set Epoch: 170 Loss: 0.000011
[31.01|17:50:44] validation_set Epoch: 170 Loss: 0.001377
[31.01|17:50:45] training_set Epoch: 180 Loss: 0.000016
[31.01|17:50:45] validation_set Epoch: 180 Loss: 0.003710
[31.01|17:50:46] training_set Epoch: 190 Loss: 0.000010
[31.01|17:50:46] validation_set Epoch: 190 Loss: 0.003736
[31.01|17:50:49] Execution of m3gnet.run finished with returncode 0
[31.01|17:50:49] JOB m3gnet FINISHED
[31.01|17:50:49] Starting m3gnet.postrun()
[31.01|17:50:49] m3gnet.postrun() finished
[31.01|17:50:49] JOB m3gnet SUCCESSFUL
[31.01|17:51:00] Running all jobs through AMS....
[31.01|17:51:00] Storing results/optimization/training_set_results/latest
[31.01|17:51:00] Storing results/optimization/validation_set_results/latest
[31.01|17:51:00] PLAMS environment cleaned up successfully
[31.01|17:51:00] PLAMS run finished. Goodbye
[31.01|17:51:01] ParAMSResults training_set validation_set
[31.01|17:51:01] energy MAE 0.0095 0.0107 eV
[31.01|17:51:01] forces MAE 0.0399 0.1005 eV/angstrom
[31.01|17:51:01] Newly created parameter file/dir: step3_attempt2_training/results/optimization/m3gnet/m3gnet
[31.01|17:51:01] Done!
[31.01|17:51:01] Deleting step3_attempt1_training
[31.01|17:51:01] ##########################
[31.01|17:51:01] ### Step 3 / Attempt 3 ###
[31.01|17:51:01] ##########################
[31.01|17:51:01] MD Steps: 260 (cumulative: 316)
[31.01|17:51:01] Current engine settings:
[31.01|17:51:01]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step3_attempt2_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:51:01] Running step3_attempt3_simulation...
[31.01|17:51:14] Job step3_attempt3_simulation finished
[31.01|17:51:14] Deleting files that are no longer needed...
[31.01|17:51:14] Launching reference calculation
[31.01|17:51:16] Reference calculation finished!
[31.01|17:51:16] Checking success for step3_attempt3
[31.01|17:51:28] CheckEnergy: Checking energy for MDStep316, n_atoms = 8
[31.01|17:51:28] CheckEnergy: normalization coefficient = 8
[31.01|17:51:28] CheckEnergy: Actual Threshold
[31.01|17:51:28] CheckEnergy: ΔE/8 -0.0014 0.2000 OK!
[31.01|17:51:28] CheckEnergy: ΔΔE/8 -0.0005 0.0050 OK! (relative to step3_attempt2_simulation:MDStep316)
[31.01|17:51:28]
[31.01|17:51:28] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:51:28] CheckForces: ------------
[31.01|17:51:28] CheckForces: Reference job from step3_attempt3_reference_calc1
[31.01|17:51:28] CheckForces: Prediction job from final frame (MDStep316) of step3_attempt3_simulation
[31.01|17:51:28] CheckForces: ------------
[31.01|17:51:28] CheckForces: Histogram of forces
[31.01|17:51:28] CheckForces: eV/Ang Ref Pred
[31.01|17:51:28] CheckForces: -3 0 0
[31.01|17:51:28] CheckForces: -2 5 5
[31.01|17:51:28] CheckForces: -1 8 8
[31.01|17:51:28] CheckForces: 0 8 8
[31.01|17:51:28] CheckForces: 1 1 1
[31.01|17:51:28] CheckForces: 2 2 2
[31.01|17:51:28] CheckForces: 3 0 0
[31.01|17:51:28] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:51:28] CheckForces: All force components are within the acceptable error!
[31.01|17:51:28] CheckForces: Maximum deviation: 0.178 eV/angstrom
[31.01|17:51:28] CheckForces: Actual Threshold
[31.01|17:51:28] CheckForces: # > thr. 0 0 OK!
[31.01|17:51:28] CheckForces: MAE 0.058 0.30 OK!
[31.01|17:51:28] CheckForces: R^2 0.995 0.80 OK!
[31.01|17:51:28] CheckForces: --------------------
[31.01|17:51:28]
[31.01|17:51:28] Adding results from step3_attempt3_reference_calc1 to validation set
[31.01|17:51:28] Current # training set entries: 12
[31.01|17:51:28] Current # validation set entries: 7
[31.01|17:51:28] Storing data in step3_attempt3_reference_data
[31.01|17:51:28] Deleting step3_attempt2_reference_data
[31.01|17:51:28] Deleting step3_attempt3_reference_calc1
[31.01|17:51:28]
[31.01|17:51:28] Current (cumulative) timings:
[31.01|17:51:28] Time (s) Fraction
[31.01|17:51:28] Ref. calcs 15.87 0.039
[31.01|17:51:28] ML training 305.05 0.753
[31.01|17:51:28] Simulations 84.44 0.208
[31.01|17:51:28]
[31.01|17:51:28]
[31.01|17:51:28] Step 3 finished successfully!
[31.01|17:51:28]
[31.01|17:51:28] --- Begin summary ---
[31.01|17:51:28] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:51:28] 1 1 FAILED Inaccurate 1.7423
[31.01|17:51:28] 1 2 SUCCESS Accurate 0.5734
[31.01|17:51:28] 2 1 FAILED Inaccurate 1.0679
[31.01|17:51:28] 2 2 SUCCESS Accurate 0.3920
[31.01|17:51:28] 3 1 FAILED Inaccurate 0.9762
[31.01|17:51:28] 3 2 FAILED Inaccurate 0.7560
[31.01|17:51:28] 3 3 SUCCESS Accurate 0.1781
[31.01|17:51:28] --- End summary ---
[31.01|17:51:28]
[31.01|17:51:28] ##########################
[31.01|17:51:28] ### Step 4 / Attempt 1 ###
[31.01|17:51:28] ##########################
[31.01|17:51:28] MD Steps: 1462 (cumulative: 1778)
[31.01|17:51:28] Current engine settings:
[31.01|17:51:28]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step3_attempt2_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:51:28] Running step4_attempt1_simulation...
[31.01|17:51:54] Job step4_attempt1_simulation finished
[31.01|17:51:54] Deleting files that are no longer needed...
[31.01|17:51:54] Deleting step2_attempt2_simulation
[31.01|17:51:54] Deleting step3_attempt1_simulation
[31.01|17:51:54] Deleting step3_attempt2_simulation
[31.01|17:51:55] Launching reference calculation
[31.01|17:51:56] Reference calculation finished!
[31.01|17:51:56] Checking success for step4_attempt1
[31.01|17:52:06] CheckEnergy: Checking energy for MDStep1778, n_atoms = 8
[31.01|17:52:06] CheckEnergy: normalization coefficient = 8
[31.01|17:52:06] CheckEnergy: Actual Threshold
[31.01|17:52:06] CheckEnergy: ΔE/8 0.0055 0.2000 OK!
[31.01|17:52:06] CheckEnergy: ΔΔE/8 0.0069 0.0050 Not OK! (relative to step3_attempt3_simulation:MDStep316)
[31.01|17:52:06]
[31.01|17:52:06] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:52:06] CheckForces: ------------
[31.01|17:52:06] CheckForces: Reference job from step4_attempt1_reference_calc1
[31.01|17:52:06] CheckForces: Prediction job from final frame (MDStep1778) of step4_attempt1_simulation
[31.01|17:52:06] CheckForces: ------------
[31.01|17:52:06] CheckForces: Histogram of forces
[31.01|17:52:06] CheckForces: eV/Ang Ref Pred
[31.01|17:52:06] CheckForces: -4 1 1
[31.01|17:52:06] CheckForces: -3 0 0
[31.01|17:52:06] CheckForces: -2 0 2
[31.01|17:52:06] CheckForces: -1 13 10
[31.01|17:52:06] CheckForces: 0 5 6
[31.01|17:52:06] CheckForces: 1 5 4
[31.01|17:52:06] CheckForces: 2 0 1
[31.01|17:52:06] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:52:06] CheckForces: All force components are within the acceptable error!
[31.01|17:52:06] CheckForces: Maximum deviation: 0.339 eV/angstrom
[31.01|17:52:06] CheckForces: Actual Threshold
[31.01|17:52:06] CheckForces: # > thr. 0 0 OK!
[31.01|17:52:06] CheckForces: MAE 0.169 0.30 OK!
[31.01|17:52:06] CheckForces: R^2 0.983 0.80 OK!
[31.01|17:52:06] CheckForces: --------------------
[31.01|17:52:06]
[31.01|17:52:06] Adding results from step4_attempt1_reference_calc1 to training set
[31.01|17:52:06] Current # training set entries: 13
[31.01|17:52:06] Current # validation set entries: 7
[31.01|17:52:06] Storing data in step4_attempt1_reference_data
[31.01|17:52:06] Deleting step3_attempt3_reference_data
[31.01|17:52:06] Deleting step4_attempt1_reference_calc1
[31.01|17:52:06]
[31.01|17:52:06] Current (cumulative) timings:
[31.01|17:52:06] Time (s) Fraction
[31.01|17:52:06] Ref. calcs 16.71 0.039
[31.01|17:52:06] ML training 305.05 0.705
[31.01|17:52:06] Simulations 111.11 0.257
[31.01|17:52:06]
[31.01|17:52:06]
[31.01|17:52:06]
[31.01|17:52:06] --- Begin summary ---
[31.01|17:52:06] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:52:06] 1 1 FAILED Inaccurate 1.7423
[31.01|17:52:06] 1 2 SUCCESS Accurate 0.5734
[31.01|17:52:06] 2 1 FAILED Inaccurate 1.0679
[31.01|17:52:06] 2 2 SUCCESS Accurate 0.3920
[31.01|17:52:06] 3 1 FAILED Inaccurate 0.9762
[31.01|17:52:06] 3 2 FAILED Inaccurate 0.7560
[31.01|17:52:06] 3 3 SUCCESS Accurate 0.1781
[31.01|17:52:06] 4 1 FAILED Inaccurate 0.3389
[31.01|17:52:06] --- End summary ---
[31.01|17:52:06]
[31.01|17:52:06] Running more reference calculations....
[31.01|17:52:06] Running reference calculations on frames [37, 40, 43] from step4_attempt1_simulation/ams.rkf
[31.01|17:52:06] Calculating 3 frames in total
[31.01|17:52:06] Running step4_attempt1_reference_calc2
[31.01|17:52:07] Running step4_attempt1_reference_calc3
[31.01|17:52:08] Running step4_attempt1_reference_calc4
[31.01|17:52:08] Reference calculations finished!
[31.01|17:52:08] Adding results from step4_attempt1_reference_calc2 to validation set
[31.01|17:52:08] Adding results from step4_attempt1_reference_calc3 to training set
[31.01|17:52:08] Adding results from step4_attempt1_reference_calc4 to training set
[31.01|17:52:08] Current # training set entries: 15
[31.01|17:52:08] Current # validation set entries: 8
[31.01|17:52:08] Storing data in step4_attempt1_reference_data
[31.01|17:52:08] Deleting step4_attempt1_reference_calc2
[31.01|17:52:08] Deleting step4_attempt1_reference_calc3
[31.01|17:52:08] Deleting step4_attempt1_reference_calc4
[31.01|17:52:08] Launching reparametrization job: step4_attempt1_training
[31.01|17:52:10] JOB m3gnet STARTED
[31.01|17:52:10] Starting m3gnet.prerun()
[31.01|17:52:10] m3gnet.prerun() finished
[31.01|17:52:10] JOB m3gnet RUNNING
[31.01|17:52:10] Executing m3gnet.run
[31.01|17:52:35] training_set Epoch: 0 Loss: 0.002138
[31.01|17:52:35] validation_set Epoch: 0 Loss: 0.050272
[31.01|17:52:36] training_set Epoch: 10 Loss: 0.000159
[31.01|17:52:36] validation_set Epoch: 10 Loss: 0.012682
[31.01|17:52:37] training_set Epoch: 20 Loss: 0.000028
[31.01|17:52:37] validation_set Epoch: 20 Loss: 0.004910
[31.01|17:52:38] training_set Epoch: 30 Loss: 0.000023
[31.01|17:52:38] validation_set Epoch: 30 Loss: 0.002643
[31.01|17:52:39] training_set Epoch: 40 Loss: 0.000019
[31.01|17:52:39] validation_set Epoch: 40 Loss: 0.002616
[31.01|17:52:40] training_set Epoch: 50 Loss: 0.000017
[31.01|17:52:40] validation_set Epoch: 50 Loss: 0.002715
[31.01|17:52:41] training_set Epoch: 60 Loss: 0.000016
[31.01|17:52:41] validation_set Epoch: 60 Loss: 0.002266
[31.01|17:52:42] training_set Epoch: 70 Loss: 0.000018
[31.01|17:52:42] validation_set Epoch: 70 Loss: 0.002155
[31.01|17:52:43] training_set Epoch: 80 Loss: 0.000014
[31.01|17:52:43] validation_set Epoch: 80 Loss: 0.003127
[31.01|17:52:45] training_set Epoch: 90 Loss: 0.000013
[31.01|17:52:45] validation_set Epoch: 90 Loss: 0.002371
[31.01|17:52:46] training_set Epoch: 100 Loss: 0.000013
[31.01|17:52:46] validation_set Epoch: 100 Loss: 0.002530
[31.01|17:52:47] training_set Epoch: 110 Loss: 0.000013
[31.01|17:52:47] validation_set Epoch: 110 Loss: 0.002451
[31.01|17:52:48] training_set Epoch: 120 Loss: 0.000011
[31.01|17:52:48] validation_set Epoch: 120 Loss: 0.001932
[31.01|17:52:49] training_set Epoch: 130 Loss: 0.000010
[31.01|17:52:49] validation_set Epoch: 130 Loss: 0.001957
[31.01|17:52:50] training_set Epoch: 140 Loss: 0.000011
[31.01|17:52:50] validation_set Epoch: 140 Loss: 0.003718
[31.01|17:52:51] training_set Epoch: 150 Loss: 0.000009
[31.01|17:52:51] validation_set Epoch: 150 Loss: 0.002412
[31.01|17:52:52] training_set Epoch: 160 Loss: 0.000009
[31.01|17:52:52] validation_set Epoch: 160 Loss: 0.001933
[31.01|17:52:53] training_set Epoch: 170 Loss: 0.000009
[31.01|17:52:53] validation_set Epoch: 170 Loss: 0.002816
[31.01|17:52:54] training_set Epoch: 180 Loss: 0.000008
[31.01|17:52:54] validation_set Epoch: 180 Loss: 0.001849
[31.01|17:52:55] training_set Epoch: 190 Loss: 0.000008
[31.01|17:52:55] validation_set Epoch: 190 Loss: 0.002113
[31.01|17:52:57] Execution of m3gnet.run finished with returncode 0
[31.01|17:52:57] JOB m3gnet FINISHED
[31.01|17:52:57] Starting m3gnet.postrun()
[31.01|17:52:57] m3gnet.postrun() finished
[31.01|17:52:57] JOB m3gnet SUCCESSFUL
[31.01|17:53:09] Running all jobs through AMS....
[31.01|17:53:09] Storing results/optimization/training_set_results/latest
[31.01|17:53:09] Storing results/optimization/validation_set_results/latest
[31.01|17:53:09] PLAMS environment cleaned up successfully
[31.01|17:53:09] PLAMS run finished. Goodbye
[31.01|17:53:09] ParAMSResults training_set validation_set
[31.01|17:53:09] energy MAE 0.0132 0.0119 eV
[31.01|17:53:09] forces MAE 0.0337 0.0857 eV/angstrom
[31.01|17:53:09] Newly created parameter file/dir: step4_attempt1_training/results/optimization/m3gnet/m3gnet
[31.01|17:53:09] Done!
[31.01|17:53:09] Deleting step3_attempt2_training
[31.01|17:53:09] ##########################
[31.01|17:53:09] ### Step 4 / Attempt 2 ###
[31.01|17:53:09] ##########################
[31.01|17:53:09] MD Steps: 1462 (cumulative: 1778)
[31.01|17:53:09] Current engine settings:
[31.01|17:53:09]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step4_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:53:09] Running step4_attempt2_simulation...
[31.01|17:53:32] Job step4_attempt2_simulation finished
[31.01|17:53:32] Deleting files that are no longer needed...
[31.01|17:53:32] Launching reference calculation
[31.01|17:53:33] Reference calculation finished!
[31.01|17:53:33] Checking success for step4_attempt2
[31.01|17:53:43] CheckEnergy: Checking energy for MDStep1778, n_atoms = 8
[31.01|17:53:43] CheckEnergy: normalization coefficient = 8
[31.01|17:53:43] CheckEnergy: Actual Threshold
[31.01|17:53:43] CheckEnergy: ΔE/8 0.0006 0.2000 OK!
[31.01|17:53:43] CheckEnergy: ΔΔE/8 -0.0015 0.0050 OK! (relative to step4_attempt1_simulation:MDStep1778)
[31.01|17:53:43]
[31.01|17:53:43] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:53:43] CheckForces: ------------
[31.01|17:53:43] CheckForces: Reference job from step4_attempt2_reference_calc1
[31.01|17:53:43] CheckForces: Prediction job from final frame (MDStep1778) of step4_attempt2_simulation
[31.01|17:53:43] CheckForces: ------------
[31.01|17:53:43] CheckForces: Histogram of forces
[31.01|17:53:43] CheckForces: eV/Ang Ref Pred
[31.01|17:53:43] CheckForces: -5 1 1
[31.01|17:53:43] CheckForces: -4 0 0
[31.01|17:53:43] CheckForces: -3 1 1
[31.01|17:53:43] CheckForces: -2 2 2
[31.01|17:53:43] CheckForces: -1 6 6
[31.01|17:53:43] CheckForces: 0 9 9
[31.01|17:53:43] CheckForces: 1 3 3
[31.01|17:53:43] CheckForces: 2 2 2
[31.01|17:53:43] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:53:43] CheckForces: All force components are within the acceptable error!
[31.01|17:53:43] CheckForces: Maximum deviation: 0.283 eV/angstrom
[31.01|17:53:43] CheckForces: Actual Threshold
[31.01|17:53:43] CheckForces: # > thr. 0 0 OK!
[31.01|17:53:43] CheckForces: MAE 0.077 0.30 OK!
[31.01|17:53:43] CheckForces: R^2 0.995 0.80 OK!
[31.01|17:53:43] CheckForces: --------------------
[31.01|17:53:43]
[31.01|17:53:43] Adding results from step4_attempt2_reference_calc1 to training set
[31.01|17:53:43] Current # training set entries: 16
[31.01|17:53:43] Current # validation set entries: 8
[31.01|17:53:43] Storing data in step4_attempt2_reference_data
[31.01|17:53:43] Deleting step4_attempt1_reference_data
[31.01|17:53:43] Deleting step4_attempt2_reference_calc1
[31.01|17:53:43]
[31.01|17:53:43] Current (cumulative) timings:
[31.01|17:53:43] Time (s) Fraction
[31.01|17:53:43] Ref. calcs 19.71 0.038
[31.01|17:53:43] ML training 365.82 0.704
[31.01|17:53:43] Simulations 134.03 0.258
[31.01|17:53:43]
[31.01|17:53:43]
[31.01|17:53:44] Step 4 finished successfully!
[31.01|17:53:44]
[31.01|17:53:44] --- Begin summary ---
[31.01|17:53:44] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:53:44] 1 1 FAILED Inaccurate 1.7423
[31.01|17:53:44] 1 2 SUCCESS Accurate 0.5734
[31.01|17:53:44] 2 1 FAILED Inaccurate 1.0679
[31.01|17:53:44] 2 2 SUCCESS Accurate 0.3920
[31.01|17:53:44] 3 1 FAILED Inaccurate 0.9762
[31.01|17:53:44] 3 2 FAILED Inaccurate 0.7560
[31.01|17:53:44] 3 3 SUCCESS Accurate 0.1781
[31.01|17:53:44] 4 1 FAILED Inaccurate 0.3389
[31.01|17:53:44] 4 2 SUCCESS Accurate 0.2827
[31.01|17:53:44] --- End summary ---
[31.01|17:53:44]
[31.01|17:53:44] ##########################
[31.01|17:53:44] ### Step 5 / Attempt 1 ###
[31.01|17:53:44] ##########################
[31.01|17:53:44] MD Steps: 8222 (cumulative: 10000)
[31.01|17:53:44] Current engine settings:
[31.01|17:53:44]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step4_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:53:44] Running step5_attempt1_simulation...
[31.01|17:55:05] Job step5_attempt1_simulation finished
[31.01|17:55:05] Deleting files that are no longer needed...
[31.01|17:55:05] Deleting step3_attempt3_simulation
[31.01|17:55:05] Deleting step4_attempt1_simulation
[31.01|17:55:05] Launching reference calculation
[31.01|17:55:06] Reference calculation finished!
[31.01|17:55:06] Checking success for step5_attempt1
[31.01|17:55:16] CheckEnergy: Checking energy for MDStep10000, n_atoms = 8
[31.01|17:55:16] CheckEnergy: normalization coefficient = 8
[31.01|17:55:16] CheckEnergy: Actual Threshold
[31.01|17:55:16] CheckEnergy: ΔE/8 -0.0008 0.2000 OK!
[31.01|17:55:16] CheckEnergy: ΔΔE/8 -0.0014 0.0050 OK! (relative to step4_attempt2_simulation:MDStep1778)
[31.01|17:55:16]
[31.01|17:55:16] CheckForces: Comparing predicted forces to reference forces (eV/angstrom) on MD snapshot
[31.01|17:55:16] CheckForces: ------------
[31.01|17:55:16] CheckForces: Reference job from step5_attempt1_reference_calc1
[31.01|17:55:16] CheckForces: Prediction job from final frame (MDStep10000) of step5_attempt1_simulation
[31.01|17:55:16] CheckForces: ------------
[31.01|17:55:16] CheckForces: Histogram of forces
[31.01|17:55:16] CheckForces: eV/Ang Ref Pred
[31.01|17:55:16] CheckForces: -4 0 0
[31.01|17:55:16] CheckForces: -3 1 1
[31.01|17:55:16] CheckForces: -2 1 1
[31.01|17:55:16] CheckForces: -1 12 11
[31.01|17:55:16] CheckForces: 0 7 8
[31.01|17:55:16] CheckForces: 1 2 2
[31.01|17:55:16] CheckForces: 2 1 1
[31.01|17:55:16] CheckForces: 3 0 0
[31.01|17:55:16] CheckForces: Threshold for 0 force: 0.50 eV/angstrom
[31.01|17:55:16] CheckForces: All force components are within the acceptable error!
[31.01|17:55:16] CheckForces: Maximum deviation: 0.530 eV/angstrom
[31.01|17:55:16] CheckForces: Actual Threshold
[31.01|17:55:16] CheckForces: # > thr. 0 0 OK!
[31.01|17:55:16] CheckForces: MAE 0.138 0.30 OK!
[31.01|17:55:16] CheckForces: R^2 0.975 0.80 OK!
[31.01|17:55:16] CheckForces: --------------------
[31.01|17:55:16]
[31.01|17:55:16] Adding results from step5_attempt1_reference_calc1 to training set
[31.01|17:55:16] Current # training set entries: 17
[31.01|17:55:16] Current # validation set entries: 8
[31.01|17:55:16] Storing data in step5_attempt1_reference_data
[31.01|17:55:16] Deleting step4_attempt2_reference_data
[31.01|17:55:16] Deleting step5_attempt1_reference_calc1
[31.01|17:55:16]
[31.01|17:55:16] Current (cumulative) timings:
[31.01|17:55:16] Time (s) Fraction
[31.01|17:55:16] Ref. calcs 20.44 0.034
[31.01|17:55:16] ML training 365.82 0.608
[31.01|17:55:16] Simulations 215.21 0.358
[31.01|17:55:16]
[31.01|17:55:16]
[31.01|17:55:16] Step 5 finished successfully!
[31.01|17:55:16]
[31.01|17:55:16] --- Begin summary ---
[31.01|17:55:16] Step Attempt Status Reason finalframe_forces_max_delta
[31.01|17:55:16] 1 1 FAILED Inaccurate 1.7423
[31.01|17:55:16] 1 2 SUCCESS Accurate 0.5734
[31.01|17:55:16] 2 1 FAILED Inaccurate 1.0679
[31.01|17:55:16] 2 2 SUCCESS Accurate 0.3920
[31.01|17:55:16] 3 1 FAILED Inaccurate 0.9762
[31.01|17:55:16] 3 2 FAILED Inaccurate 0.7560
[31.01|17:55:16] 3 3 SUCCESS Accurate 0.1781
[31.01|17:55:16] 4 1 FAILED Inaccurate 0.3389
[31.01|17:55:16] 4 2 SUCCESS Accurate 0.2827
[31.01|17:55:16] 5 1 SUCCESS Accurate 0.5300
[31.01|17:55:16] --- End summary ---
[31.01|17:55:16]
[31.01|17:55:16] The engine settings for the final trained ML engine are:
[31.01|17:55:16]
Engine MLPotential
Backend M3GNet
MLDistanceUnit angstrom
MLEnergyUnit eV
Model Custom
ParameterDir /path/plams_workdir/sal/step4_attempt1_training/results/optimization/m3gnet/m3gnet
EndEngine
[31.01|17:55:16] Active learning finished!
[31.01|17:55:16] Rerunning the simulation with the final parameters...
[31.01|17:56:52] Goodbye!
[31.01|17:56:52] JOB sal FINISHED
[31.01|17:56:52] JOB sal SUCCESSFUL
<scm.params.plams.simple_active_learning_job.SimpleActiveLearningResults at 0x7fd4176eaa90>
Complete Python code¶
#!/usr/bin/env amspython
# coding: utf-8
# ## Initial imports
from scm.simple_active_learning import SimpleActiveLearningJob
import scm.plams as plams
from scm.external_engines.core import interface_is_installed
assert interface_is_installed("m3gnet"), "You must first install m3gnet with the AMS package manager"
# ## Initialize PLAMS
plams.init()
# ## Input system
mol = plams.from_smiles("OCC=O")
for at in mol:
at.properties = {}
plams.plot_molecule(mol)
# ## Reference engine settings
# For time reasons we use the UFF force field as the reference method. Typically you would instead train to DFT using ADF, BAND, or Quantum ESPRESSO.
ref_s = plams.Settings()
ref_s.input.ForceField.Type = "UFF"
ref_s.runscript.nproc = 1
print(plams.AMSJob(settings=ref_s).get_input())
# ## Molecular dynamics settings
# Here, we use the convenient ``AMSNVTJob`` recipe to easily initialize sone MD settings.
md_s = plams.AMSNVTJob(temperature=300, timestep=0.5, nsteps=10000).settings
print(plams.AMSJob(settings=md_s).get_input())
# ## ParAMS ML Training settings
#
# (Technical note: When using ``SimpleActiveLearningJob`` the ParAMS settings go under ``input.ams``. When using ``ParAMSJob`` the settings instead simply go under ``input``. See the ParAMS Python tutorials.)
ml_s = plams.Settings()
ml_s.input.ams.MachineLearning.Backend = "M3GNet"
ml_s.input.ams.MachineLearning.CommitteeSize = 1
ml_s.input.ams.MachineLearning.M3GNet.Model = "UniversalPotential"
ml_s.input.ams.MachineLearning.MaxEpochs = 200
print(SimpleActiveLearningJob(settings=ml_s).get_input())
# ## Active learning settings
al_s = plams.Settings()
al_s.input.ams.ActiveLearning.Steps.Type = "Geometric"
al_s.input.ams.ActiveLearning.Steps.Geometric.Start = 10 # 10 MD frames
al_s.input.ams.ActiveLearning.Steps.Geometric.NumSteps = 5 # 10 AL steps
print(SimpleActiveLearningJob(settings=al_s).get_input())
# ## Simple Active Learning Job
settings = ref_s + md_s + ml_s + al_s
job = SimpleActiveLearningJob(settings=settings, molecule=mol, name="sal")
print(job.get_input())
# ## Run the job
job.run(watch=True)