r.model.label_y = 'MLR_TOTAL'
r.model.header_line = 1
r.experiment.file_name = "Data.csv"
r.experiment.label_x = '# Time_{}_75'.format(iso)
r.experiment.label_y = 'SG_{}_75'.format(iso)
r.experiment.header_line = 0
r.experiment.yfactor = 1e-3
r.fitness_method = pr.FitnessMethodRMSE(n_points=100)
# use above model prototype (mps0) as template
mps = copy.deepcopy(mps0)
TEND = 600
# modify a single value of model parameter
mps[3].value = TEND
# create simulation setup object
template_file = "SimpleConeLaunchTest_{}_BestParaSet_Toast.fds".format(iso)
s = pr.SimulationSetup(name='cone_{}'.format(iso),
work_dir='cone_{}'.format(iso),
model_template=template_file,
model_parameter=mps,
model_executable='fds',
relations=r)
# append above object to simulation setup set
setups.append(s)
# use default values for optimiser
optimiser = pr.OptimiserProperties()
model_parameter=model_parameter_setups[i],
model_executable='fds',
relations=r[i])
ssetups.append(s)
# Initialise empty simulation setup sets.
setups = pr.SimulationSetupSet()
# Append above objects to simulation setup set.
for i in range(len(HeatingRatesTGA)):
setups.append(ssetups[i])
print('** setups generated')
# Provide values for optimiser.
optimiser = pr.OptimiserProperties(algorithm='sceua',
repetitions=150,
ngs=4,
# Sub-processes would be used for
# repetitions of an experiment.
num_subprocesses=1,
mpi=False)
print('** input file processed')
# define variable 'ops': optimisation parameter set
# define variable 'setups': simulation setup set
# define variable 'optimiser': properties for the optimiser
# import just for IDE convenience
import propti as pr
# fix the chid
CHID = 'TGA_analysis_01'
TEND = 9360
# use default values for optimiser
optimiser = pr.OptimiserProperties(algorithm='sceua', repetitions=10)
#ngs=4,
# define the optimisation parameter
op1 = pr.Parameter(name='ref_temp_comp_01',
place_holder='rtc01',
min_value=200,
max_value=400)
op2 = pr.Parameter(name='ref_rate_comp_01',
place_holder='rrc01',
min_value=0.001,
max_value=0.01)
op3 = pr.Parameter(name='ref_temp_comp_02',
place_holder='rtc02',
min_value=300,
max_value=600)
op4 = pr.Parameter(name='ref_rate_comp_02',
place_holder='rrc02',
min_value=0.001,
# Optimiser to perform FAST sensitivity analysis.
# Use the same data from optimiser for sensitivity analysis
num_subproc = optimiser.num_subprocesses
mpi_bool = optimiser.mpi
backup = optimiser.backup_every
# Compute number of repetitions required
k = len(ops) # total number of optimization params
(M, d) = 3, 2 # M = inference factor, d = freq. step <spotpy defaults>
rep = (1 + 4 * (M**2) * (1 + (k - 2) * d)) * k
sensitivity = pr.OptimiserProperties('fast',
repetitions=rep,
backup_every=backup,
db_name="propti_sensitivity_db",
db_type="csv",
num_subprocesses=num_subproc,
mpi=mpi_bool)
print(ver, setups, ops, sensitivity)
'''
Not writing any data to pickle file since there is only
one method to analyse sensitivity. It makes sense to Wiki this instead.
'''
# Write sensitivity pickle file
# out_file = open('propti_sensitivity.pickle.init', 'wb')
# pickle.dump((setups, ops, optimiser), out_file)
# out_file.close()
# Run optimization.