outfile_name = 'energies_output.out'
if not job.isfile(outfile_name):
return False
else:
return has_libatoms_exit_message(job.fn(outfile_name))
def has_force_output(job):
outfile_name = 'forces_output.out'
if not job.isfile(outfile_name):
return False
else:
return has_libatoms_exit_message(job.fn(outfile_name))
timings = FlowProject.make_group(name='timings')
@timings
@FlowProject.operation
@FlowProject.pre(gap_fit_success)
@FlowProject.post(has_energy_output)
@directives(omp_num_threads=1)
def time_quip_energies(job):
old_dir = os.getcwd()
quip_cmdline = build_quip_command_line(job)
print(quip_cmdline)
try:
os.chdir(job.workspace())
with open('energies_output.out', 'w') as outfile:
subprocess.run(quip_cmdline, stdout=outfile)
import sklearn
from sklearn.neural_network import MLPRegressor
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error, mean_absolute_error
TARGETS = ['COF', 'intercept']
IDENTIFIERS = ['terminal_group_1', 'terminal_group_2', 'terminal_group_3',
'backbone', 'frac-1', 'frac-2']
# get project root directory
proj = signac.get_project()
root_dir = proj.root_directory()
# define group that combines the different operations
neural_network = FlowProject.make_group(name='neural_network')
@FlowProject.label
def models_trained(job):
# post condition for train_evaluate
# checks if pickle files for models and scalers are in job workspace
# check external hard drive for models
for target in TARGETS:
if not os.path.isfile('/mnt/d/neural-networks-with-signac/workspace/' + job.id + '/{}_trained.pickle'.format(target)):
return False
if not os.path.isfile(job.fn('{}_scaler.pickle'.format(target))):
return False
return True
import signac
from flow import FlowProject
import os
from simtk import unit
from cg_openmm.cg_model.cgmodel import CGModel
from cg_openmm.parameters.reweight import get_temperature_list
from cg_openmm.simulation.rep_exch import *
from openmmtools.cache import global_context_cache
import numpy as np
import simtk.openmm as openmm
import pickle
from cg_openmm.thermo.calc import *
global_context_cache.platform = openmm.Platform.getPlatformByName("CUDA")
replica_exchange_group = FlowProject.make_group(name='replica_exchange')
analysis_group = FlowProject.make_group(name='analysis')
proj_directory = os.getcwd()
@FlowProject.label
def run_replica_exchange_done(job):
output_directory = os.path.join(job.workspace(), "output")
output_data = os.path.join(output_directory, "output.nc")
rep_exch_completed = 0
if os.path.isfile(output_data):
rep_exch_status = ReplicaExchangeSampler.read_status(output_data)
rep_exch_completed = rep_exch_status.is_completed
return rep_exch_completed
# A Signac-Flow project for one-component VLE simulations
import json
import os
import pickle
import shutil
import subprocess
import numpy as np
import signac
from flow import FlowProject, directive
# Melt, cool, volume, swap
equilibrate = FlowProject.make_group(name="equilibrate")
project = signac.get_project()
@FlowProject.label
def is_equil(job):
"""
For job tasks in the 'equilibrate' group, we only
want to run the task if it
"""
return job.sp.run == "equil"
@FlowProject.label
def job_marked_complete(job):
"""
For jobs that may have multiple steps, mark completion