def __call__(self):
if PY2:
with open(self.filename, 'rb') as f:
return load(f)
else:
try:
with open(self.filename, 'rb') as f:
return load(f)
except UnicodeDecodeError:
# this happens when loading numpy arrays which have been
# pickled with Python 2
with open(self.filename, 'rb') as f:
return load(f, encoding='latin-1')
def _load(self, i):
if i in self._cache:
return self._cache[i]
else:
with open(os.path.join(self.dir, str(i)), 'rb') as f:
v = load(f)
self._cache[i] = v
if len(self._cache) > self.cache_size:
self._cache.popitem(last=False)
return v
def _load(self, i):
if i in self._cache:
return self._cache[i]
else:
with open(os.path.join(self.dir, str(i)), 'rb') as f:
v = load(f)
self._cache[i] = v
if len(self._cache) > self.cache_size:
self._cache.popitem(last=False)
return v
def d(self):
if self._deleted:
raise ValueError('Dataset has been deleted')
if hasattr(self, '_d'):
return self._d
if not self.finished:
if self.failed:
logger.warn('Loading data of failed dataset {}.'.format(
self.name))
else:
if os.path.exists(os.path.join(self.path, 'DATA')):
logger.warn(
'Loading data of unfinished dataset {}.'.format(
self.name))
else:
raise ValueError(
'No data has been written to unfinished dataset {}.'.
format(self.name))
if PY2:
with open(os.path.join(self.path, 'DATA'), 'rb') as f:
self._data = load(f)
else:
try:
with open(os.path.join(self.path, 'DATA'), 'rb') as f:
self._data = load(f)
except UnicodeDecodeError:
# this happens when loading numpy arrays which have been
# pickled with Python 2
with open(os.path.join(self.path, 'DATA'), 'rb') as f:
self._data = load(f, encoding='latin-1')
for v in self._data.values():
if isinstance(v, DataLoader):
v.filename = os.path.join(self.path, v.filename)
elif isinstance(v, list):
for vv in v:
if isinstance(vv, DataLoader):
vv.filename = os.path.join(self.path, vv.filename)
self._d = self.DataDict()
self._d.__dict__ = self._data
return self._d
def get(self, key):
key = base64.b64encode(key)
response = self.server.get(self.secret, key)
assert len(response) == 2 and isinstance(response[0], bool) and isinstance(response[1], str)
if response[0]:
file_path = response[1]
with open(file_path) as f:
value = load(f)
return True, value
else:
return False, None
def get(self, key):
key = base64.b64encode(key)
response = self.server.get(self.secret, key)
assert len(response) == 2 and isinstance(response[0], bool) and isinstance(response[1], str)
if response[0]:
file_path = response[1]
with open(file_path, 'rb') as f:
value = load(f)
return True, value
else:
return False, None
def get(self, key):
c = self.conn.cursor()
t = (key,)
c.execute('SELECT filename FROM entries WHERE key=?', t)
result = c.fetchall()
if len(result) == 0:
return False, None
elif len(result) == 1:
file_path = os.path.join(self.path, result[0][0])
with open(file_path, 'rb') as f:
value = load(f)
return True, value
else:
raise RuntimeError('Cache is corrupt!')
def get(self, key):
c = self.conn.cursor()
t = (key, )
c.execute('SELECT filename FROM entries WHERE key=?', t)
result = c.fetchall()
if len(result) == 0:
return False, None
elif len(result) == 1:
file_path = os.path.join(self.path, result[0][0])
with open(file_path, 'rb') as f:
value = load(f)
return True, value
else:
raise RuntimeError('Cache is corrupt!')
def __init__(self, argument_dict):
"""If argument_dict contains a value for max_keys this maximum amount of cache values kept in the
internal cache file, otherwise its set to sys.maxlen.
If necessary values are deleted from the cache in FIFO order.
"""
argument_dict['filename'] = argument_dict.get('filename', os.path.join(gettempdir(), 'pymor'))
super(LimitedFileBackend, self).__init__(argument_dict)
self.logger.debug('LimitedFileBackend args {}'.format(pformat(argument_dict)))
self._max_keys = argument_dict.get('max_keys', sys.maxsize)
self._keylist_fn = self.filename + '.keys'
self._max_size = argument_dict.get('max_size', None)
try:
self._keylist, self._size = load(open(self._keylist_fn, 'rb'))
except Exception:
self._keylist = deque()
self._size = 0
self._enforce_limits(None)
self.print_limit()
def analyze_pickle_histogram(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced discretization ...')
rb_discretization = load(open(args['REDUCED_DATA']))
mus = rb_discretization.parameter_space.sample_randomly(args['SAMPLES'])
us = []
for mu in mus:
print('Solving reduced for {} ... '.format(mu), end='')
sys.stdout.flush()
us.append(rb_discretization.solve(mu))
print('done')
print()
if hasattr(rb_discretization, 'estimate'):
ests = []
for u, mu in zip(us, mus):
print('Estimating error for {} ... '.format(mu), end='')
sys.stdout.flush()
ests.append(rb_discretization.estimate(u, mu=mu))
print('done')
if args['--detailed']:
print('Loading high-dimensional data ...')
discretization, reconstructor = load(open(args['--detailed']))
errs = []
for u, mu in zip(us, mus):
print('Calculating error for {} ... '.format(mu))
sys.stdout.flush()
err = discretization.solve(mu) - reconstructor.reconstruct(u)
if args['--error-norm']:
errs.append(np.max(getattr(discretization, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
print('done')
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
if hasattr(rb_discretization, 'estimate') and args['--detailed']:
# setup axes
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
bottom_h = left_h = left+width+0.02
rect_scatter = [left, bottom, width, height]
rect_histx = [left, bottom_h, width, 0.2]
rect_histy = [left_h, bottom, 0.2, height]
plt.figure(1, figsize=(8, 8))
axScatter = plt.axes(rect_scatter)
axHistx = plt.axes(rect_histx)
axHisty = plt.axes(rect_histy)
# scatter plot
total_min = min(min(ests), min(errs)) * 0.9
total_max = max(max(ests), max(errs)) * 1.1
axScatter.set_xscale('log')
axScatter.set_yscale('log')
axScatter.set_xlim([total_min, total_max])
axScatter.set_ylim([total_min, total_max])
axScatter.set_xlabel('errors')
axScatter.set_ylabel('estimates')
axScatter.plot([total_min, total_max], [total_min, total_max], 'r')
axScatter.scatter(errs, ests)
# plot histograms
x_hist, x_bin_edges = np.histogram(errs, bins=np.logspace(np.log10(total_min), np.log10(total_max), 100))
axHistx.bar(x_bin_edges[1:], x_hist, width=x_bin_edges[:-1] - x_bin_edges[1:], color='blue')
y_hist, y_bin_edges = np.histogram(ests, bins=np.logspace(np.log10(total_min), np.log10(total_max), 100))
axHisty.barh(y_bin_edges[1:], y_hist, height=y_bin_edges[:-1] - y_bin_edges[1:], color='blue')
axHistx.set_xscale('log')
axHisty.set_yscale('log')
axHistx.set_xticklabels([])
axHisty.set_yticklabels([])
axHistx.set_xlim(axScatter.get_xlim())
axHisty.set_ylim(axScatter.get_ylim())
axHistx.set_ylim([0, max(max(x_hist), max(y_hist))])
axHisty.set_xlim([0, max(max(x_hist), max(y_hist))])
plt.show()
elif hasattr(rb_discretization, 'estimate'):
total_min = min(ests) * 0.9
total_max = max(ests) * 1.1
hist, bin_edges = np.histogram(ests, bins=np.logspace(np.log10(total_min), np.log10(total_max), 100))
plt.bar(bin_edges[1:], hist, width=bin_edges[:-1] - bin_edges[1:], color='blue')
plt.xlim([total_min, total_max])
plt.xscale('log')
plt.xlabel('estimated error')
plt.show()
elif args['--detailed']:
total_min = min(ests) * 0.9
total_max = max(ests) * 1.1
hist, bin_edges = np.histogram(errs, bins=np.logspace(np.log10(total_min), np.log10(total_max), 100))
plt.bar(bin_edges[1:], hist, width=bin_edges[:-1] - bin_edges[1:], color='blue')
plt.xlim([total_min, total_max])
plt.xscale('log')
plt.xlabel('error')
plt.show()
else:
raise ValueError('Nothing to plot!')
def analyze_pickle_convergence(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced discretization ...')
rb_discretization = load(open(args['REDUCED_DATA']))
if args['--detailed']:
print('Loading high-dimensional data ...')
discretization, reconstructor = load(open(args['--detailed']))
if not hasattr(rb_discretization, 'estimate') and not args['--detailed']:
raise ValueError('Nothing to do! (Neither estimates nor true error can be computed.)')
dim = rb_discretization.solution_space.dim
if args['--ndim']:
dims = np.linspace(0, dim, args['--ndim'], dtype=np.int)
else:
dims = np.arange(dim + 1)
mus = rb_discretization.parameter_space.sample_randomly(args['SAMPLES'])
ESTS = []
ERRS = []
T_SOLVES = []
T_ESTS = []
for N in dims:
rd, rc, _ = reduce_to_subbasis(rb_discretization, N)
print('N = {:3} '.format(N), end='')
us = []
print('solve ', end='')
sys.stdout.flush()
start = time.time()
for mu in mus:
us.append(rd.solve(mu))
T_SOLVES.append((time.time() - start) * 1000. / len(mus))
print('estimate ', end='')
sys.stdout.flush()
if hasattr(rb_discretization, 'estimate'):
ests = []
start = time.time()
for u, mu in zip(us, mus):
# print('e', end='')
# sys.stdout.flush()
ests.append(rd.estimate(u, mu=mu))
ESTS.append(max(ests))
T_ESTS.append((time.time() - start) * 1000. / len(mus))
if args['--detailed']:
print('errors', end='')
sys.stdout.flush()
errs = []
for u, mu in zip(us, mus):
err = discretization.solve(mu) - reconstructor.reconstruct(rc.reconstruct(u))
if args['--error-norm']:
errs.append(np.max(getattr(discretization, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
ERRS.append(max(errs))
print()
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
plt.subplot(1, 2, 1)
if hasattr(rb_discretization, 'estimate'):
plt.semilogy(dims, ESTS, label='max. estimate')
if args['--detailed']:
plt.semilogy(dims, ERRS, label='max. error')
plt.xlabel('dimension')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(dims, T_SOLVES, label='avg. solve time')
if hasattr(rb_discretization, 'estimate'):
plt.plot(dims, T_ESTS, label='avg. estimate time')
plt.xlabel('dimension')
plt.ylabel('milliseconds')
plt.legend()
plt.show()
def analyze_pickle_histogram(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced model ...')
rom, parameter_space = load(open(args['REDUCED_DATA'], 'rb'))
mus = parameter_space.sample_randomly(args['SAMPLES'])
us = []
for mu in mus:
print(f'Solving reduced for {mu} ... ', end='')
sys.stdout.flush()
us.append(rom.solve(mu))
print('done')
print()
if hasattr(rom, 'estimate'):
ests = []
for u, mu in zip(us, mus):
print(f'Estimating error for {mu} ... ', end='')
sys.stdout.flush()
ests.append(rom.estimate(u, mu=mu))
print('done')
if args['--detailed']:
print('Loading high-dimensional data ...')
fom, reductor = load(open(args['--detailed'], 'rb'))
errs = []
for u, mu in zip(us, mus):
print(f'Calculating error for {mu} ... ')
sys.stdout.flush()
err = fom.solve(mu) - reductor.reconstruct(u)
if args['--error-norm']:
errs.append(
np.max(getattr(fom, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
print('done')
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
if hasattr(rom, 'estimate') and args['--detailed']:
# setup axes
left, width = 0.1, 0.65
bottom, height = 0.1, 0.65
bottom_h = left_h = left + width + 0.02
rect_scatter = [left, bottom, width, height]
rect_histx = [left, bottom_h, width, 0.2]
rect_histy = [left_h, bottom, 0.2, height]
plt.figure(1, figsize=(8, 8))
axScatter = plt.axes(rect_scatter)
axHistx = plt.axes(rect_histx)
axHisty = plt.axes(rect_histy)
# scatter plot
total_min = min(np.min(ests), np.min(errs)) * 0.9
total_max = max(np.max(ests), np.max(errs)) * 1.1
axScatter.set_xscale('log')
axScatter.set_yscale('log')
axScatter.set_xlim([total_min, total_max])
axScatter.set_ylim([total_min, total_max])
axScatter.set_xlabel('errors')
axScatter.set_ylabel('estimates')
axScatter.plot([total_min, total_max], [total_min, total_max], 'r')
axScatter.scatter(errs, ests)
# plot histograms
x_hist, x_bin_edges = np.histogram(errs,
bins=_bins(total_min, total_max))
axHistx.bar(x_bin_edges[1:],
x_hist,
width=x_bin_edges[:-1] - x_bin_edges[1:],
color='blue')
y_hist, y_bin_edges = np.histogram(ests,
bins=_bins(total_min, total_max))
axHisty.barh(y_bin_edges[1:],
y_hist,
height=y_bin_edges[:-1] - y_bin_edges[1:],
color='blue')
axHistx.set_xscale('log')
axHisty.set_yscale('log')
axHistx.set_xticklabels([])
axHisty.set_yticklabels([])
axHistx.set_xlim(axScatter.get_xlim())
axHisty.set_ylim(axScatter.get_ylim())
axHistx.set_ylim([0, max(np.max(x_hist), np.max(y_hist))])
axHisty.set_xlim([0, max(np.max(x_hist), np.max(y_hist))])
plt.show()
elif hasattr(rom, 'estimate'):
total_min = np.min(ests) * 0.9
total_max = np.max(ests) * 1.1
hist, bin_edges = np.histogram(ests, bins=_bins(total_min, total_max))
plt.bar(bin_edges[1:],
hist,
width=bin_edges[:-1] - bin_edges[1:],
color='blue')
plt.xlim([total_min, total_max])
plt.xscale('log')
plt.xlabel('estimated error')
plt.show()
elif args['--detailed']:
total_min = np.min(ests) * 0.9
total_max = np.max(ests) * 1.1
hist, bin_edges = np.histogram(errs, bins=_bins(total_min, total_max))
plt.bar(bin_edges[1:],
hist,
width=bin_edges[:-1] - bin_edges[1:],
color='blue')
plt.xlim([total_min, total_max])
plt.xscale('log')
plt.xlabel('error')
plt.show()
else:
raise ValueError('Nothing to plot!')
def analyze_pickle_convergence(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced model ...')
rom, parameter_space = load(open(args['REDUCED_DATA'], 'rb'))
if not args['--detailed']:
raise ValueError('High-dimensional data file must be specified.')
print('Loading high-dimensional data ...')
fom, reductor = load(open(args['--detailed'], 'rb'))
fom.enable_caching('disk')
dim = rom.solution_space.dim
if args['--ndim']:
dims = np.linspace(0, dim, args['--ndim'], dtype=np.int)
else:
dims = np.arange(dim + 1)
mus = parameter_space.sample_randomly(args['SAMPLES'])
ESTS = []
ERRS = []
T_SOLVES = []
T_ESTS = []
for N in dims:
rom = reductor.reduce(N)
print(f'N = {N:3} ', end='')
us = []
print('solve ', end='')
sys.stdout.flush()
start = time.time()
for mu in mus:
us.append(rom.solve(mu))
T_SOLVES.append((time.time() - start) * 1000. / len(mus))
print('estimate ', end='')
sys.stdout.flush()
if hasattr(rom, 'estimate'):
ests = []
start = time.time()
for u, mu in zip(us, mus):
# print('e', end='')
# sys.stdout.flush()
ests.append(rom.estimate(u, mu=mu))
ESTS.append(max(ests))
T_ESTS.append((time.time() - start) * 1000. / len(mus))
print('errors', end='')
sys.stdout.flush()
errs = []
for u, mu in zip(us, mus):
err = fom.solve(mu) - reductor.reconstruct(u)
if args['--error-norm']:
errs.append(
np.max(getattr(fom, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
ERRS.append(max(errs))
print()
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
plt.subplot(1, 2, 1)
if hasattr(rom, 'estimate'):
plt.semilogy(dims, ESTS, label='max. estimate')
plt.semilogy(dims, ERRS, label='max. error')
plt.xlabel('dimension')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(dims, T_SOLVES, label='avg. solve time')
if hasattr(rom, 'estimate'):
plt.plot(dims, T_ESTS, label='avg. estimate time')
plt.xlabel('dimension')
plt.ylabel('milliseconds')
plt.legend()
plt.show()
def analyze_pickle_convergence(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced discretization ...')
rb_discretization = load(open(args['REDUCED_DATA']))
if args['--detailed']:
print('Loading high-dimensional data ...')
discretization, reconstructor = load(open(args['--detailed']))
if not hasattr(rb_discretization, 'estimate') and not args['--detailed']:
raise ValueError('Nothing to do! (Neither estimates nor true error can be computed.)')
dim = rb_discretization.solution_space.dim
if args['--ndim']:
dims = np.linspace(0, dim, args['--ndim'], dtype=np.int)
else:
dims = np.arange(dim + 1)
mus = list(rb_discretization.parameter_space.sample_randomly(args['SAMPLES']))
ESTS = []
ERRS = []
T_SOLVES = []
T_ESTS = []
for N in dims:
rd, rc, _ = reduce_to_subbasis(rb_discretization, N)
print('N = {:3} '.format(N), end='')
us = []
print('solve ', end='')
sys.stdout.flush()
start = time.time()
for mu in mus:
us.append(rd.solve(mu))
T_SOLVES.append((time.time() - start) * 1000. / len(mus))
print('estimate ', end='')
sys.stdout.flush()
if hasattr(rb_discretization, 'estimate'):
ests = []
start = time.time()
for u, mu in zip(us, mus):
# print('e', end='')
# sys.stdout.flush()
ests.append(rd.estimate(u, mu=mu))
ESTS.append(max(ests))
T_ESTS.append((time.time() - start) * 1000. / len(mus))
if args['--detailed']:
print('errors', end='')
sys.stdout.flush()
errs = []
for u, mu in zip(us, mus):
err = discretization.solve(mu) - reconstructor.reconstruct(rc.reconstruct(u))
if args['--error-norm']:
errs.append(np.max(getattr(discretization, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
ERRS.append(max(errs))
print()
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
plt.subplot(1, 2, 1)
if hasattr(rb_discretization, 'estimate'):
plt.semilogy(dims, ESTS, label='max. estimate')
if args['--detailed']:
plt.semilogy(dims, ERRS, label='max. error')
plt.xlabel('dimension')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(dims, T_SOLVES, label='avg. solve time')
if hasattr(rb_discretization, 'estimate'):
plt.plot(dims, T_ESTS, label='avg. estimate time')
plt.xlabel('dimension')
plt.ylabel('milliseconds')
plt.legend()
plt.show()
def analyze_pickle_convergence(args):
args['SAMPLES'] = int(args['SAMPLES'])
print('Loading reduced model ...')
rom = load(open(args['REDUCED_DATA'], 'rb'))
if not args['--detailed']:
raise ValueError('High-dimensional data file must be specified.')
print('Loading high-dimensional data ...')
fom, reductor = load(open(args['--detailed'], 'rb'))
fom.enable_caching('disk')
dim = rom.solution_space.dim
if args['--ndim']:
dims = np.linspace(0, dim, args['--ndim'], dtype=np.int)
else:
dims = np.arange(dim + 1)
mus = rom.parameter_space.sample_randomly(args['SAMPLES'])
ESTS = []
ERRS = []
T_SOLVES = []
T_ESTS = []
for N in dims:
rom = reductor.reduce(N)
print(f'N = {N:3} ', end='')
us = []
print('solve ', end='')
sys.stdout.flush()
start = time.time()
for mu in mus:
us.append(rom.solve(mu))
T_SOLVES.append((time.time() - start) * 1000. / len(mus))
print('estimate ', end='')
sys.stdout.flush()
if hasattr(rom, 'estimate'):
ests = []
start = time.time()
for u, mu in zip(us, mus):
# print('e', end='')
# sys.stdout.flush()
ests.append(rom.estimate(u, mu=mu))
ESTS.append(max(ests))
T_ESTS.append((time.time() - start) * 1000. / len(mus))
print('errors', end='')
sys.stdout.flush()
errs = []
for u, mu in zip(us, mus):
err = fom.solve(mu) - reductor.reconstruct(u)
if args['--error-norm']:
errs.append(np.max(getattr(fom, args['--error-norm'] + '_norm')(err)))
else:
errs.append(np.max(err.l2_norm()))
ERRS.append(max(errs))
print()
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
plt.subplot(1, 2, 1)
if hasattr(rom, 'estimate'):
plt.semilogy(dims, ESTS, label='max. estimate')
plt.semilogy(dims, ERRS, label='max. error')
plt.xlabel('dimension')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(dims, T_SOLVES, label='avg. solve time')
if hasattr(rom, 'estimate'):
plt.plot(dims, T_ESTS, label='avg. estimate time')
plt.xlabel('dimension')
plt.ylabel('milliseconds')
plt.legend()
plt.show()
def convergence(
reduced_data: str = REDUCED_DATA,
detailed_data: str = Argument(..., help='File containing the high-dimensional model and the reductor.'),
samples: int = SAMPLES,
error_norm: str = ERROR_NORM,
ndim: int = Option(None, help='Number of reduced basis dimensions for which to estimate the error.')
):
print('Loading reduced model ...')
rom, parameter_space = load(open(reduced_data, 'rb'))
print('Loading high-dimensional data ...')
fom, reductor = load(open(detailed_data, 'rb'))
fom.enable_caching('disk')
dim = rom.solution_space.dim
if ndim:
dims = np.linspace(0, dim, ndim, dtype=np.int)
else:
dims = np.arange(dim + 1)
mus = parameter_space.sample_randomly(samples)
ESTS = []
ERRS = []
T_SOLVES = []
T_ESTS = []
for N in dims:
rom = reductor.reduce(N)
print(f'N = {N:3} ', end='')
us = []
print('solve ', end='')
sys.stdout.flush()
start = time.perf_counter()
for mu in mus:
us.append(rom.solve(mu))
T_SOLVES.append((time.perf_counter() - start) * 1000. / len(mus))
print('estimate ', end='')
sys.stdout.flush()
if hasattr(rom, 'estimate'):
ests = []
start = time.perf_counter()
for mu in mus:
# print('e', end='')
# sys.stdout.flush()
ests.append(rom.estimate_error(mu))
ESTS.append(max(ests))
T_ESTS.append((time.perf_counter() - start) * 1000. / len(mus))
print('errors', end='')
sys.stdout.flush()
errs = []
for u, mu in zip(us, mus):
err = fom.solve(mu) - reductor.reconstruct(u)
if error_norm:
errs.append(np.max(getattr(fom, error_norm + '_norm')(err)))
else:
errs.append(np.max(err.norm()))
ERRS.append(max(errs))
print()
print()
try:
plt.style.use('ggplot')
except AttributeError:
pass # plt.style is only available in newer matplotlib versions
plt.subplot(1, 2, 1)
if hasattr(rom, 'estimate'):
plt.semilogy(dims, ESTS, label='max. estimate')
plt.semilogy(dims, ERRS, label='max. error')
plt.xlabel('dimension')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(dims, T_SOLVES, label='avg. solve time')
if hasattr(rom, 'estimate'):
plt.plot(dims, T_ESTS, label='avg. estimate time')
plt.xlabel('dimension')
plt.ylabel('milliseconds')
plt.legend()
plt.show()
