def test_db_to_dict(self):
tmpdir = tempfile.mkdtemp()
dfile = os.path.join(tmpdir, 'junk.db')
recorder = DBCaseRecorder(dfile)
# create some Cases where some are missing a variable
outputs = ['comp1.z', 'comp2.z']
inputs = ['comp1.x', 'comp1.y', 'comp1.y2']
recorder.register(self, inputs, outputs)
for i in range(10):
inputs = [i, i*2, i*3]
outputs = [i*i, float('NaN')]
recorder.record(self, inputs, outputs, None, '', '')
varnames = ['comp1.x', 'comp1.y', 'comp1.y2']
varinfo = case_db_to_dict(dfile, varnames)
self.assertEqual(len(varinfo), 3)
# each var list should have 10 data values in it
for lst in varinfo.values():
self.assertEqual(len(lst), 10)
# now use caseiter_to_dict to grab the same data
varinfo = caseiter_to_dict(recorder.get_iterator(), varnames)
# each var list should have 10 data values in it
for lst in varinfo.values():
self.assertEqual(len(lst), 10)
try:
shutil.rmtree(tmpdir, onerror=onerror)
except OSError:
logging.error("problem removing directory %s", tmpdir)
def test_db_to_dict(self):
tmpdir = tempfile.mkdtemp()
dfile = os.path.join(tmpdir, 'junk.db')
recorder = DBCaseRecorder(dfile)
# create some Cases where some are missing a variable
outputs = ['comp1.z', 'comp2.z']
inputs = ['comp1.x', 'comp1.y', 'comp1.y2']
recorder.register(self, inputs, outputs)
for i in range(10):
inputs = [i, i * 2, i * 3]
outputs = [i * i, float('NaN')]
recorder.record(self, inputs, outputs, None, '', '')
varnames = ['comp1.x', 'comp1.y', 'comp1.y2']
varinfo = case_db_to_dict(dfile, varnames)
self.assertEqual(len(varinfo), 3)
# each var list should have 10 data values in it
for lst in varinfo.values():
self.assertEqual(len(lst), 10)
# now use caseiter_to_dict to grab the same data
varinfo = caseiter_to_dict(recorder.get_iterator(), varnames)
# each var list should have 10 data values in it
for lst in varinfo.values():
self.assertEqual(len(lst), 10)
try:
shutil.rmtree(tmpdir, onerror=onerror)
except OSError:
logging.error("problem removing directory %s", tmpdir)
def test_db_to_dict(self):
tmpdir = tempfile.mkdtemp()
dfile = os.path.join(tmpdir, 'junk.db')
recorder = DBCaseRecorder(dfile)
# create some Cases where some are missing a variable
outputs = ['comp1.z', 'comp2.z']
cases = []
for i in range(10):
if i > 1:
msg = ''
else:
msg = 'an error occurred'
if i < 5:
inputs = [('comp1.x', i), ('comp1.y', i * 2),
('comp1.y2', i * 3)]
else:
inputs = [('comp1.x', i), ('comp1.y', i * 2)]
recorder.record(Case(inputs=inputs, outputs=outputs, msg=msg))
varnames = ['comp1.x', 'comp1.y', 'comp1.y2']
varinfo = case_db_to_dict(dfile, varnames)
self.assertEqual(len(varinfo), 3)
# each var list should have 3 data values in it (5 with the required variables minus
# 2 with errors
for name, lst in varinfo.items():
self.assertEqual(len(lst), 3)
# now use caseiter_to_dict to grab the same data
varinfo = caseiter_to_dict(recorder.get_iterator(), varnames)
# each var list should have 3 data values in it (5 with the required variables minus
# 2 with errors
for name, lst in varinfo.items():
self.assertEqual(len(lst), 3)
try:
shutil.rmtree(tmpdir)
except OSError:
logging.error("problem removing directory %s" % tmpdir)
def test_db_to_dict(self):
tmpdir = tempfile.mkdtemp()
dfile = os.path.join(tmpdir, 'junk.db')
recorder = DBCaseRecorder(dfile)
# create some Cases where some are missing a variable
outputs = ['comp1.z', 'comp2.z']
cases = []
for i in range(10):
if i>1:
msg = ''
else:
msg = 'an error occurred'
if i<5:
inputs = [('comp1.x', i), ('comp1.y', i*2), ('comp1.y2', i*3)]
else:
inputs = [('comp1.x', i), ('comp1.y', i*2)]
recorder.record(Case(inputs=inputs, outputs=outputs, msg=msg))
varnames = ['comp1.x','comp1.y','comp1.y2']
varinfo = case_db_to_dict(dfile, varnames)
self.assertEqual(len(varinfo), 3)
# each var list should have 3 data values in it (5 with the required variables minus
# 2 with errors
for name,lst in varinfo.items():
self.assertEqual(len(lst), 3)
# now use caseiter_to_dict to grab the same data
varinfo = caseiter_to_dict(recorder.get_iterator(), varnames)
# each var list should have 3 data values in it (5 with the required variables minus
# 2 with errors
for name,lst in varinfo.items():
self.assertEqual(len(lst), 3)
try:
shutil.rmtree(tmpdir)
except OSError:
logging.error("problem removing directory %s" % tmpdir)
X_range = arange(-5,10.2,.25)
Y_range = arange(0,15.2,.25)
X,Y = meshgrid(X_range,Y_range)
Z = branin(X,Y)
iterator = analysis.branin_meta_model.recorder.get_iterator()
plt.contour(X,Y,Z,arange(1,200,2),zorder=1)
cb = plt.colorbar(shrink=.45)
#plot the initial training data
data_train = case_db_to_dict(os.path.join(analysis._tdir,'trainer.db'),
['branin_meta_model.y',
'branin_meta_model.x',
'branin_meta_model.f_xy'])
plt.scatter(data_train['branin_meta_model.x'],
data_train['branin_meta_model.y'],s=30,c='#572E07',zorder=10)
data_EI = case_db_to_dict(os.path.join(analysis._tdir,'retrain.db'),
['branin_meta_model.y',
'branin_meta_model.x',
'branin_meta_model.f_xy'])
count = len(data_EI['branin_meta_model.x'])
colors = arange(0,count)/float(count)
color_map = get_cmap('spring')
X_range = arange(-5, 10.2, .25)
Y_range = arange(0, 15.2, .25)
X, Y = meshgrid(X_range, Y_range)
Z = branin(X, Y)
iterator = analysis.branin_meta_model.recorder.get_iterator()
plt.contour(X, Y, Z, arange(1, 200, 2), zorder=1)
cb = plt.colorbar(shrink=.45)
#plot the initial training data
data_train = case_db_to_dict(os.path.join(analysis._tdir, 'trainer.db'), [
'branin_meta_model.y', 'branin_meta_model.x', 'branin_meta_model.f_xy'
])
plt.scatter(data_train['branin_meta_model.x'],
data_train['branin_meta_model.y'],
s=30,
c='#572E07',
zorder=10)
data_EI = case_db_to_dict(os.path.join(analysis._tdir, 'retrain.db'), [
'branin_meta_model.y', 'branin_meta_model.x', 'branin_meta_model.f_xy'
])
count = len(data_EI['branin_meta_model.x'])
colors = arange(0, count) / float(count)
return (y - (5.1 / (4. * pi ** 2.)) * x ** 2. + 5. * x / pi - 6.) ** 2. + \
10. * (1. - 1. / (8. * pi)) * cos(x) + 10.
X_range = arange(-5, 10.2, .25)
Y_range = arange(0, 15.2, .25)
X, Y = meshgrid(X_range, Y_range)
Z = branin(X, Y)
plt.contour(X, Y, Z, arange(1, 200, 2), zorder=1)
cb = plt.colorbar(shrink=.45)
# plot the initial training data
data_train = case_db_to_dict(
os.path.join(analysis.architecture._tdir, 'trainer.db'), [
'branin.y',
'branin.x',
])
plt.scatter(data_train['branin.x'],
data_train['branin.y'],
s=30,
c='#572E07',
zorder=10)
data_EI = {}
data_EI['branin.x'] = data_train['branin.x'][analysis.architecture.
initial_DOE_size:]
data_EI['branin.y'] = data_train['branin.y'][analysis.architecture.
initial_DOE_size:]
def branin(x,y):
return (y-(5.1/(4.*pi**2.))*x**2.+5.*x/pi-6.)**2.+\
10.*(1.-1./(8.*pi))*cos(x)+10.
X_range = arange(-5,10.2,.25)
Y_range = arange(0,15.2,.25)
X,Y = meshgrid(X_range,Y_range)
Z = branin(X,Y)
plt.contour(X,Y,Z,arange(1,200,2),zorder=1)
cb = plt.colorbar(shrink=.45)
#plot the initial training data
data_train = case_db_to_dict(os.path.join(analysis.architecture._tdir,'trainer.db'),
['branin.y',
'branin.x',])
plt.scatter(data_train['branin.x'],
data_train['branin.y'],s=30,c='#572E07',zorder=10)
data_EI = {}
data_EI['branin.x'] = data_train['branin.x'][analysis.architecture.initial_DOE_size:]
data_EI['branin.y'] = data_train['branin.y'][analysis.architecture.initial_DOE_size:]
count = len(data_EI['branin.x'])
print "%d adaptively sampled points"%count
colors = arange(0,count)/float(count)
color_map = get_cmap('spring')
row1 = []
row2 = []
for x, y in zip(x_row, y_row):
analysis.spiral_meta_model.x = x
analysis.spiral_meta_model.y = y
analysis.spiral_meta_model.execute()
row1.append(analysis.spiral_meta_model.f1_xy.mu)
row2.append(analysis.spiral_meta_model.f2_xy.mu)
Z1_pred.append(row1)
Z2_pred.append(row2)
Z1_pred = array(Z1_pred)
Z2_pred = array(Z2_pred)
#plot the initial training data
data_train = case_db_to_dict(os.path.join(analysis._tdir, 'trainer.db'), [
'spiral_meta_model.x', 'spiral_meta_model.y',
'spiral_meta_model.f1_xy', 'spiral_meta_model.f2_xy'
])
plt.scatter(data_train['spiral_meta_model.x'],
data_train['spiral_meta_model.y'],
s=30,
c='#572E07',
zorder=10)
data_EI = case_db_to_dict(os.path.join(analysis._tdir, 'retrain.db'), [
'spiral_meta_model.y', 'spiral_meta_model.x',
'spiral_meta_model.f1_xy', 'spiral_meta_model.f2_xy'
])
count = len(data_EI['spiral_meta_model.x'])
colors = arange(0, count) / float(count)
row2 = []
for x,y in zip(x_row,y_row):
analysis.spiral_meta_model.x = x
analysis.spiral_meta_model.y = y
analysis.spiral_meta_model.execute()
row1.append(analysis.spiral_meta_model.f1_xy.mu)
row2.append(analysis.spiral_meta_model.f2_xy.mu)
Z1_pred.append(row1)
Z2_pred.append(row2)
Z1_pred = array(Z1_pred)
Z2_pred = array(Z2_pred)
#plot the initial training data
data_train = case_db_to_dict(os.path.join(analysis._tdir,'trainer.db'),
['spiral_meta_model.x',
'spiral_meta_model.y',
'spiral_meta_model.f1_xy',
'spiral_meta_model.f2_xy'])
plt.scatter(data_train['spiral_meta_model.x'],
data_train['spiral_meta_model.y'],s=30,c='#572E07',zorder=10)
data_EI = case_db_to_dict(os.path.join(analysis._tdir,'retrain.db'),
['spiral_meta_model.y',
'spiral_meta_model.x',
'spiral_meta_model.f1_xy',
'spiral_meta_model.f2_xy'])
count = len(data_EI['spiral_meta_model.x'])
colors = arange(0,count)/float(count)
color_map = get_cmap('spring')
