def createPsuadeInFile(data, filename, includePDF=True):
outf = open(filename, 'w')
outf.write('PSUADE\n')
#Write inputs
outf.write('INPUT\n')
outf.write(' dimension = %d\n' % data.getNumInputs())
names = data.getInputNames()
mins = data.getInputMins()
maxs = data.getInputMaxs()
indices = list(range(data.getNumInputs()))
for i, name, minimum, maximum in zip(indices, names, mins, maxs):
outf.write(' variable %d %s = %e %e\n' %
(i + 1, name, minimum, maximum))
distributions = data.getInputDistributions()
for i, dist in zip(indices, distributions):
distType = dist.getDistributionType()
distParams = dist.getParameterValues()
if distType == Distribution.SAMPLE:
outf.write(' PDF %d %c' %
(i + 1, Distribution.getPsuadeName(distType)))
if distParams[0] is not None:
filename = distParams[0]
if platform.system() == 'Windows':
import win32api
filename = win32api.GetShortPathName(filename)
outf.write(' %s' % filename)
if distParams[1] is not None:
outf.write(' %d' % distParams[1])
outf.write('\n')
elif includePDF: # write out PDF info for all non-uniform PDF types
if distType != Distribution.UNIFORM:
outf.write(' PDF %d %c' %
(i + 1, Distribution.getPsuadeName(distType)))
if distParams[0] is not None:
outf.write(' %e' % distParams[0])
if distParams[1] is not None:
outf.write(' %e' % distParams[1])
outf.write('\n')
outf.write('END\n')
#Write outputs
outf.write('OUTPUT\n')
if data.getNumOutputs() == 0:
outf.write(' dimension = 1\n')
names = ['ghostOuput']
indices = list(range(1))
for i, name in zip(indices, names):
outf.write(' variable %d %s\n' % (i + 1, name))
else:
outf.write(' dimension = %d\n' % data.getNumOutputs())
names = data.getOutputNames()
indices = list(range(data.getNumOutputs()))
for i, name in zip(indices, names):
outf.write(' variable %d %s\n' % (i + 1, name))
outf.write('END\n')
#Write Method
outf.write('METHOD\n')
outf.write(' sampling = %s\n' %
SamplingMethods.getPsuadeName(data.getSampleMethod()))
outf.write(' num_samples = %d\n' % data.getNumSamples())
if data.getSampleMethod() != SamplingMethods.GMOAT:
outf.write(' randomize\n')
outf.write('END\n')
#Write Application
outf.write('APPLICATION\n')
driverString = data.getDriverName()
if driverString is None:
driverString = 'NONE'
elif platform.system() == 'Windows':
if os.path.exists(driverString):
import win32api
driverString = win32api.GetShortPathName(driverString)
else:
driverString = 'NONE'
driverString = 'NONE'
outf.write(' driver = %s\n' % driverString)
outf.write(' save_frequency = 1\n')
outf.write('END\n')
#Write Analysis
outf.write('ANALYSIS\n')
outf.write(' diagnostics 2\n')
outf.write('END\n')
outf.write('END\n')
outf.close()
def writeOUUdata(outfile, outputs, constraints, derivatives, data, xtable,
**kwargs):
# Charles TODO: Handle y is now a list of inputs
# Charles TODO: Handle derivatives
# defaults
rseed = None
driver = data.getDriverName()
if driver is None:
driver = 'NONE'
optdriver = data.getOptDriverName()
if optdriver is None:
optdriver = 'NONE'
ensoptdriver = data.getEnsembleOptDriverName()
if ensoptdriver is None:
ensoptdriver = 'NONE'
auxdriver = data.getAuxDriverName()
if auxdriver is None:
auxdriver = 'NONE'
inputLB = None
inputUB = None
inputDefaults = None
distributions = None
init_input = None
# process keyworded arguments
for key in kwargs:
k = key.lower()
if k == 'randseed':
rseed = kwargs[key]
elif k == 'driver':
driver = kwargs[key]
elif k == 'optdriver':
optdriver = kwargs[key]
elif k == 'ensoptdriver':
ensoptdriver = kwargs[key]
elif k == 'auxdriver':
auxdriver = kwargs[key]
elif k == 'inputlowerbounds':
inputLB = kwargs[key]
elif k == 'inputupperbounds':
inputUB = kwargs[key]
elif k == 'inputpdf':
distributions = kwargs[key]
elif k == 'init_input':
init_input = kwargs[key]
inputTypes = data.getInputTypes()
nInputs = data.getNumInputs()
inputNames = data.getInputNames()
variableInputIndices = []
for e in xtable:
if e['type'] != u'Fixed':
variableInputIndices.append(inputNames.index(e['name']))
nVariableInputs = len(variableInputIndices)
nOutputs = len(outputs)
nSamples = num_fmin = 1 # number of random restarts
nConstraints = constraints.count(True)
nDerivatives = derivatives.count(True)
totalOutputs = nOutputs + nConstraints + nDerivatives
f = open(outfile, 'w')
if init_input:
f.write('PSUADE_IO\n')
f.write('%d %d %d\n' % (nVariableInputs, totalOutputs, nSamples))
f.write("1 0\n") # assume initial point has not been run
for x in init_input:
f.write(' % .16e\n' % x)
for i in xrange(totalOutputs):
f.write(' 9.9999999999999997e+34\n')
f.write("PSUADE_IO\n")
# TO DO: merge with RSAnalyzer.writeRSdata()
f.write('PSUADE\n')
# ... input ...
numFixed = nInputs - nVariableInputs
f.write('INPUT\n')
if numFixed > 0:
f.write(' num_fixed %d\n' % numFixed)
f.write(' dimension = %d\n' % nVariableInputs)
if inputLB is None:
inputLB = data.getInputMins()
if inputUB is None:
inputUB = data.getInputMaxs()
if inputDefaults is None:
inputDefaults = data.getInputDefaults()
indices = range(nInputs)
variableIndex = 1
fixedIndex = 1
for i, name, inType, lb, ub, default in zip(indices, inputNames, \
inputTypes, inputLB, inputUB, inputDefaults):
if i in variableInputIndices: #inType == Model.VARIABLE:
f.write(' variable %d %s = % .16e % .16e\n' % \
(variableIndex, name, lb, ub))
variableIndex = variableIndex + 1
else:
f.write(' fixed %d %s = % .16e\n' %
(fixedIndex, name, default))
fixedIndex = fixedIndex + 1
# inject discrete variables in psuade
opttypes = []
cnt = 0
for e in xtable:
cnt = cnt + 1
t = e['type']
if t == u'Opt: Primary Discrete (Z1d)':
opttypes.append(cnt)
nn = len(opttypes)
for ii in range(nn):
jj = opttypes[ii]
f.write(' discrete %d\n' % (jj))
if distributions is None:
distributions = SampleData.getInputDistributions(data)
for i, inType, dist in zip(indices, inputTypes, distributions):
if i in variableInputIndices: #inType == Model.VARIABLE:
distType = dist.getDistributionType()
distParams = dist.getParameterValues()
if distType != Distribution.UNIFORM:
f.write(' PDF %d %c' % (i+1, \
Distribution.getPsuadeName(distType)))
if distType == Distribution.SAMPLE:
error = 'OUU: In function writeOUUdata(), '
error = error + 'SAMPLE distribution is not supported.'
Common.showError(error)
return None
else:
if distParams[0] is not None:
f.write(' % .16e' % distParams[0])
if distParams[1] is not None:
f.write(' % .16e' % distParams[1])
f.write('\n')
f.write('END\n')
# ... output ...
outActive = nOutputs
nConstrs = 0
nDerivs = 0
for ii in range(len(constraints)):
if constraints[ii]:
outActive = outActive + 1
nConstrs = nConstrs + 1
for ii in range(len(derivatives)):
if derivatives[ii]:
outActive = outActive + 1
nDerivs = nDerivs + 1
if (nOutputs != 1):
error = 'OUU: In function writeOUUdata(), '
error = error + 'multi-objective optimization not supported.'
Common.showError(error)
return None
else:
if ((nConstrs > 0) and (nDerivs > 0)):
error = 'OUU: In function writeOUUdata(), '
error = error + 'LBFGS does not support inequality constraints.'
Common.showError(error)
return None
elif ((nDerivs > 0) and (nDerivs != nVariableInputs)):
error = 'OUU: In function writeOUUdata(), '
error = error + 'Number of derivatives not correct'
Common.showError(error)
return None
f.write('OUTPUT\n')
f.write(' dimension = %d\n' % (outActive))
outputNames = SampleData.getOutputNames(data)
for ii in range(nOutputs):
ind = outputs[ii]
f.write(' variable %d %s\n' % (ii + 1, outputNames[ind - 1]))
print(' variable %d %s\n' % (ii + 1, outputNames[ind - 1]))
outActive = nOutputs + 1
for ii in range(len(constraints)):
if constraints[ii]:
f.write(' variable %d %s\n' % (outActive, outputNames[ii]))
print(' variable %d %s\n' % (outActive, outputNames[ii]))
outActive = outActive + 1
for ii in range(len(derivatives)):
if derivatives[ii]:
f.write(' variable %d %s\n' % (outActive, outputNames[ii]))
print(' variable %d %s\n' % (outActive, outputNames[ii]))
outActive = outActive + 1
f.write('END\n')
# ... method ...
f.write('METHOD\n')
f.write(' sampling = MC\n') # OUU uses this to create
f.write(' num_samples = 1\n') # initial guess
if rseed is not None:
f.write('random_seed = %d\n' % rseed) # random seed
f.write('END\n')
# ... application ...
f.write('APPLICATION\n')
if platform.system() == 'Windows':
import win32api
if driver != 'NONE' and driver != 'PSUADE_LOCAL':
driver = win32api.GetShortPathName(driver)
if optdriver != 'NONE' and optdriver != 'PSUADE_LOCAL':
optdriver = win32api.GetShortPathName(optdriver)
if ensoptdriver != 'NONE' and ensoptdriver != 'PSUADE_LOCAL':
ensoptdriver = win32api.GetShortPathName(ensoptdriver)
if auxdriver != 'NONE' and auxdriver != 'PSUADE_LOCAL':
auxdriver = win32api.GetShortPathName(auxdriver)
f.write(' driver = %s\n' % driver)
f.write(' opt_driver = %s\n' % optdriver)
f.write(' ensemble_opt_driver = %s\n' % ensoptdriver)
f.write(' aux_opt_driver = %s\n' % auxdriver)
f.write(' launch_interval = 0\n')
f.write('END\n')
# ... analysis ...
f.write('ANALYSIS\n')
if (nDerivs > 0):
f.write(' optimization method = ouu_lbfgs\n')
else:
f.write(' optimization method = ouu\n')
f.write(' optimization num_local_minima = 1\n')
f.write(' optimization max_feval = 1000000\n')
f.write(' optimization fmin = 0.0\n')
f.write(' optimization tolerance = 1.000000e-06\n')
f.write(' optimization num_fmin = %d\n' % num_fmin)
f.write(' optimization print_level = 3\n')
#f.write(' analyzer output_id = %d\n' % y)
f.write(' analyzer output_id = 1\n')
f.write(' opt_expert\n')
f.write(' printlevel 0\n')
f.write('END\n')
f.write('END\n')
f.close()
return outfile