combinations = itertools.product(\
matched['ratios'].items(),\
pulses, \
range(settings.get('nTrials'))
)
simulationRoot = rootDirectory
print(f"Making {simulationRoot}")
os.makedirs(simulationRoot, exist_ok=1)
settings['equilibrium'] = matched
# make settings
print(settings['graph'])
IO.savePickle(os.path.join(simulationRoot, 'settings'),\
settings)
# load settings file
elif os.path.isdir(file):
print('reading settings')
# load old settings
# if root in settings
# read settings
oldSettings = None
for (root, dir, files) in os.walk(file):
for f in files:
fn = os.path.join(root, f)
if f.endswith('settings.pickle'):
oldSettings = fn
break
if not oldSettings:
settings = dict(
repeats = repeats,\
deltas = deltas,\
nSamples = nSamples,\
step = step,\
burninSamples = burninSamples,\
pulseSizes = pulseSizes,\
updateType = updateType,\
nNodes = graph.number_of_nodes(),\
nTrials = nTrials,\
# this is added
graph = graph,\
mapping = model.mapping,\
rmapping = model.rmapping,\
directory = targetDirectory,\
nudgeType = nudgeType,\
CHECK = CHECK,\
tempres = tempres, \
magSide = magSide,\
targetDirectory = targetDirectory,\
start = start, \
now = now,\
)
runFile = genDataFile(idx)
IO.savePickle(runFile, settings)
cmd = f"sbatch control_slurm.sh {runFile}".split()
subprocess.call(cmd)
corrTimeSettings = IO.loadResults(targetDirectory, 'corrTimeSettings')
burninSteps = mixingResults['burninSteps']
distSamples = mixingResults['distSamples']
# try to load neighbourhood shell data. If it doesn't exist yet, generate it
try:
if len(args.neighboursDir) > 0:
neighboursG = IO.loadPickle(args.neighboursDir, 'neighboursG')
else:
neighboursG = IO.loadPickle(targetDirectory, 'neighboursG')
except:
print(f'determining neighbours')
neighboursG = model.neighboursAtDistAllNodes(nodes, maxDist)
if len(args.neighboursDir) > 0:
os.makedirs(args.neighboursDir, exist_ok=True)
IO.savePickle(args.neighboursDir, 'neighboursG', neighboursG)
else:
IO.savePickle(targetDirectory, 'neighboursG', neighboursG)
snapshotSettings = dict( \
nSamples = args.numSamples, \
burninSteps = burninSteps, \
distSamples = distSamples, \
maxDist = maxDist, \
nBins = args.bins
)
# repetitive runs of the MI estimation procedure
for r in range(args.runs):
# m.equilibriate(magRatios, eqSettings)
combinations = itertools.product(\
m.matched['ratios'].items(),\
pulseSizes, \
range(settings.get('nTrials'))
)
settings['graph'] = g
# setup filepaths
now = datetime.datetime.now().isoformat()
simulationRoot = os.path.join(\
rootDirectory, now)
print(f"Making {simulationRoot}")
os.makedirs(simulationRoot, exist_ok=1)
settings['equilibrium'] = m.matched
IO.savePickle(os.path.join(simulationRoot, 'settings'),\
settings)
for (ratio, pulse, trial) in combinations:
# tmp added properties
settings['trial'] = trial
settings['ratio'] = ratio
if pulse:
for node in m.graph.nodes():
tmp = copy.deepcopy(m)
tmp.t = ratio[1]
intervention = {node: pulse}
tmp.nudges = intervention
settings['pulse'] = intervention
