parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Create the spotted model
model = Fulda_lumped(datetime.datetime(begin,1,1),
datetime.datetime(end,12,31), with_valid_data = True,
shift_one_day = True)
if 'i' in sys.argv:
runs = 0
elif 'v' in sys.argv:
sys.argv.remove('v')
best = eval(open(prefix + '-best.dict').read())
best.pop('Eff')
model.setparameters(**best)
model.begin = datetime.datetime(1986,1,1)
model.end = datetime.datetime(1988,12,31)
resQ = np.array(model.runmodel())
model.plotvalidation(np.array(resQ))
runs = 0
elif len(sys.argv)>1:
runs = int(sys.argv[1])
if runs:
sampler = Sampler(model, parallel=parallel)
# sampler.datawriter = DataWriter(prefix,model.params, model.begin, model.end, 0.0)
# multi objective datawriter
sampler.datawriter = DataWriter(prefix, model.params, model.begin, model.end, simthreshold_NS = 0.50,
simthreshold_pbias = 25.0, simthreshold_rsr = 0.70,
with_valid_data = model.with_valid_data,
shift_one_day = model.shift_one_day)
# Now we can sample with the implemented Latin hypercube algorithm:
sampler.sample(runs)
runs = 2
# File names of the forcing data
fnQ = "Q_Kammerzell_1979_1999.txt"
fnT = "T_kammerzell_1979_1999_max_min_avg.txt"
fnP = "P_Krigavg_kammerzell_1979_1999.txt"
# import algorithm
from spotpy.algorithms import mc as Sampler
# Find out if the model should run parallel (for supercomputer)
parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Create the model
model = ComplexLumped(datetime.datetime(begin, 1, 1),
datetime.datetime(end, 12, 31))
print(cmf.describe(model.project))
# If there is an command line argument, take its value for the amount of
# runs
if len(sys.argv) > 1:
runs = int(sys.argv[1])
# run the model
if runs:
sampler = Sampler(model,
parallel=parallel,
dbname="complex_lumped",
dbformat="csv",
save_sim=True)
sampler.sample(runs) #, subsets = 30)
datetime.datetime(end, 12, 31),
with_valid_data=True,
shift_one_day=True)
if 'i' in sys.argv:
runs = 0
elif 'v' in sys.argv:
sys.argv.remove('v')
best = eval(open(prefix + '-best.dict').read())
best.pop('Eff')
model.setparameters(**best)
model.begin = datetime.datetime(1986, 1, 1)
model.end = datetime.datetime(1988, 12, 31)
resQ = np.array(model.runmodel())
model.plotvalidation(np.array(resQ))
runs = 0
elif len(sys.argv) > 1:
runs = int(sys.argv[1])
if runs:
sampler = Sampler(model,
parallel=parallel,
dbformat="csv",
dbname="test")
# sampler.datawriter = DataWriter(prefix,model.params, model.begin, model.end, 0.0)
# multi objective datawriter
# sampler.datawriter = DataWriter(prefix, model.params, model.begin, model.end, simthreshold_NS = 0.50,
# simthreshold_pbias = 25.0, simthreshold_rsr = 0.70,
# with_valid_data = model.with_valid_data,
# shift_one_day = model.shift_one_day)
# Now we can sample with the implemented Latin hypercube algorithm:
sampler.sample(runs)
if __name__ == '__main__':
# Get sampler
from spotpy.algorithms import lhs as Sampler
# Check if we are running on a supercomputer or local
parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Run the models
runs = 5
num_cells = [1, 2, 4, 8]
results = {}
for num in num_cells:
# Create the model
model = ScalingTester(num_cells=num)
print(cmf.describe(model.project))
# Create the sampler
sampler = Sampler(model,
parallel=parallel,
dbname=model.dbname,
dbformat='csv',
save_sim=False)
sampler.sample(runs)
results[str(num)] = sampler.status.objectivefunction
for key, value in results.items():
print("The model with {} cell(s) has a best NS of {}".format(
key, value))
# File names of the forcing data
fnQ = "Q_Kammerzell_1979_1999.txt"
fnT = "T_kammerzell_1979_1999_max_min_avg.txt"
fnP = "P_Krigavg_kammerzell_1979_1999.txt"
# import algorithm
from spotpy.algorithms import rope as Sampler
# Find out if the model should run parallel (for supercomputer)
parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Create the model
model = SimpleLumped(datetime.datetime(begin, 1, 1),
datetime.datetime(end, 12, 31))
# If there is an command line argument, take its value for the amount of
# runs
if len(sys.argv) > 1:
runs = int(sys.argv[1])
# run the model
if runs:
sampler = Sampler(model,
parallel=parallel,
dbname="simple_lumped_hargreaves",
dbformat="csv",
save_sim=True,
save_threshold=[0.0, 0.0])
sampler.sample(runs, subsets=30)
fnP = "P_Krigavg_kammerzell_1979_1999.txt"
fnSun = "sunshine_hours_mw_fulda_wasserkuppe_1979_1989.txt"
fnWind = "windspeed_m_s_mw_fulda_wasserkuppe_1979_1989.txt"
fnVapor = "vapor_pressure_kpa_mw_fulda_wasserkuppe_1979_1989.txt"
fnRelHum = "rel_hum_percent_mw_fulda_wasserkuppe_1979_1989.txt"
# import algorithm
from spotpy.algorithms import rope as Sampler
# Find out if the model should run parallel (for supercomputer)
parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Create the model
model = ComplexLumped(datetime.datetime(begin, 1, 1),
datetime.datetime(end, 12, 31))
# If there is an command line argument, take its value for the amount of
# runs
if len(sys.argv) > 1:
runs = int(sys.argv[1])
# run the model
if runs:
sampler = Sampler(model,
parallel=parallel,
dbname="complex_lumped_penman",
dbformat="csv",
save_sim=True,
save_threshold=[0.0, 0.0])
sampler.sample(runs, subsets=30)
# http://stackoverflow.com/questions/419163/what-does-if-name-main-do
if __name__ == '__main__':
# Get the Monte-Carlo sampler
from spotpy.algorithms import mc as Sampler
# Create the model
model = SingleStorage(datetime.datetime(1980, 1, 1),
datetime.datetime(1985, 12, 31))
runs = get_runs(default=1)
# Create the sampler
sampler = Sampler(model,
parallel=parallel(),
dbname=model.dbname, dbformat='csv',
save_sim=True)
# Print our configuration
print(spotpy.describe.describe(sampler))
# Do the sampling
if runs > 1:
# Now we can sample with the implemented Monte Carlo algorithm:
sampler.sample(runs)
else:
result = model.simulation(verbose=True)
for name, value in spotpy.objectivefunctions.calculate_all_functions(model.evaluation(), result):
try:
plt.plot(x, self.evaluation(), 'k')
plt.show()
#%%
# http://stackoverflow.com/questions/419163/what-does-if-name-main-do
if __name__ == '__main__':
# Importiere Algorithmus
from spotpy.algorithms import lhs as Sampler
# Finde heraus, ob das ganze parallel laufen soll (für Supercomputer)
parallel = 'mpi' if 'OMPI_COMM_WORLD_SIZE' in os.environ else 'seq'
# Create the spotted model
model = SingleStorage(datetime.datetime(begin, 1, 1),
datetime.datetime(end, 12, 31))
if runs:
sampler = Sampler(model,
parallel=parallel,
dbname='lhs-1stor',
dbformat='csv',
save_sim=True)
# sampler.datawriter = DataWriter(model.params, model.begin, model.end, 0.0)
# Now we can sample with the implemented Monte Carlo algortihm:
sampler.sample(runs)
# plottet das Ergebnis
# TODO: Threshold anpassen für die graue Fläche im Plot. Eure Modelle schaffen hoffentlich deutlich mehr...
# model.plotsimulation(0.3)
