Active Learning: Single-Molecule Setup and Run¶

Set up a minimal Simple Active Learning workflow for a small organic molecule, combining a reference method, molecular dynamics, and M3GNet retraining in Python.

Downloads: Notebook | Script ?

Requires: AMS2026 or later

Required AMS Packages: m3gnet

../_images/active_learning_single_molecule_setup_run_5_0_bcdacefc.png

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/to/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
... output trimmed ....
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
... output trimmed ....
     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:
... output trimmed ....
[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>

Python Script¶

#!/usr/bin/env python
# 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)