def test_no_logging(self):
self.logger.set_level(0)
with dcs.capture_output() as out:
dcs.run_bpe(self.opti_opts)
output = out.getvalue().strip()
out.close()
self.assertEqual(output, '')
def test_to_convergence(self):
self.opti_opts.max_iters = 100
with dcs.capture_output() as out:
dcs.run_bpe(self.opti_opts)
lines = out.getvalue().strip().split('\n')
out.close()
for line in lines:
if line.startswith('Declare convergence'):
break
else:
self.assertTrue(False, "Didn't converge")
def test_two_sided(self):
with dcs.capture_output() as out:
self.opti_opts.slope_method = 'two_sided'
dcs.run_bpe(self.opti_opts)
lines = out.getvalue().strip().split('\n')
out.close()
for (ix, line) in enumerate(lines):
if line == 'Running iteration 1.':
self.assertTrue(lines[ix+1].startswith(' Running model with magnitude'))
self.assertTrue(lines[ix+2].startswith(' Running model with magnitude'))
break
else:
self.assertTrue(False, 'two sided had issues')
# rerun at log_level 0
self.logger.set_level(0)
dcs.run_bpe(self.opti_opts)
def test_nominal(self):
with dcs.capture_output() as out:
(bpe_results, results) = dcs.run_bpe(self.opti_opts)
output = out.getvalue().strip()
out.close()
self.assertTrue(output.startswith('******************************\nValidating optimization options.'))
self.assertTrue(isinstance(bpe_results, dcs.BpeResults))
self.assertTrue(isinstance(results, np.ndarray))
def test_saving(self):
self.logger.set_level(0)
self.opti_opts.max_iters = 0
self.opti_opts.output_folder = dcs.get_tests_dir()
self.opti_opts.output_results = 'temp_results.hdf5'
dcs.run_bpe(self.opti_opts)
def test_max_likelihood(self):
self.logger.set_level(0)
self.opti_opts.is_max_like = True
dcs.run_bpe(self.opti_opts)
best=20,
min_=1,
max_=1000,
typical=60,
minstep=0.01))
opti_opts.params.append(
dcs.OptiParam('phase',
best=180,
min_=0,
max_=360,
typical=100,
minstep=0.1))
# Run code
if rerun:
(bpe_results, results) = dcs.run_bpe(opti_opts)
else:
bpe_results = dcs.BpeResults.load(
os.path.join(opti_opts.output_folder, opti_opts.output_results))
results = sim_model(
sim_params) # just re-run, nothing is actually saved by this model
# Plot results
if make_plots:
# build opts
opts = dcs.Opts()
opts.case_name = 'Model Results'
opts.save_path = dcs.get_output_dir()
opts.save_plot = True
# make model plots
opti_opts.tol_delta_step = 1e-7
opti_opts.tol_delta_cost = 1e-8
opti_opts.step_limit = 5
opti_opts.x_bias = 0.8
opti_opts.grow_radius = 2
opti_opts.shrink_radius = 0.5
opti_opts.trust_radius = 1.0
# Parameters to estimate
opti_opts.params.append(dcs.OptiParam('magnitude', best=2.5, min_=-10, max_=10, typical=5, minstep=0.01))
opti_opts.params.append(dcs.OptiParam('frequency', best=20, min_=1, max_=1000, typical=60, minstep=0.01))
opti_opts.params.append(dcs.OptiParam('phase', best=180, min_=0, max_=360, typical=100, minstep=0.1))
# Run code
if rerun:
(bpe_results, results) = dcs.run_bpe(opti_opts)
else:
bpe_results = dcs.BpeResults.load(os.path.join(opti_opts.output_folder, opti_opts.output_results))
results = sim_model(sim_params) # just re-run, nothing is actually saved by this model
# Plot results
if make_plots:
# build opts
opts = dcs.Opts()
opts.case_name = 'Model Results'
opts.save_path = dcs.get_output_dir()
opts.save_plot = True
# make model plots
dcs.plot_time_history(time, results, description='Output vs. Time', opts=opts, truth=truth)