def test_model():
raw_trace = np.genfromtxt(TEST_DATA_PATH, delimiter=',')
dt = 0.0001
vcinfer = VCInfer(raw_trace, dt, config='config_testing.json')
vcinfer.momenta()
vcinfer.psd_calc()
vcinfer.estimate_taus()
vcinfer.reset_obs_unc()
vcinfer.set_obs_unc_mean()
vcinfer.set_obs_unc_std()
dist_list = ['LogNormal', 'TruncNormal'] #, 'Exponential' it takes ages
for dist_type in dist_list:
vcinfer.first_guess(distType=dist_type)
vcinfer.optimize_likelihood()
vcinfer.create_model()
vcinfer.run_sampler()
vcinfer.plot_mc_results(save_pdf=True)
assert 'map' in vcinfer.inference_results
assert 'mh' in vcinfer.inference_results
for el in ['weights', 'parameters', 'metrics']:
assert el in vcinfer.inference_results['map']
for el in ['weights', 'parameters', 'dic', 'mcmc_object']:
assert el in vcinfer.inference_results['mh']
def test_likelihood_est_f():
config_file = os.path.join(CONFIG_DIR, 'config.json')
with open(config_file, mode='r') as fp:
config = json.load(fp)
raw_trace = np.genfromtxt(TEST_DATA_PATH, delimiter=',')
dt = 0.0001
vcinfer = VCInfer(raw_trace, dt)
vcinfer.momenta()
vcinfer.psd_calc()
vcinfer.estimate_taus()
vcinfer.reset_obs_unc()
vcinfer.set_obs_unc_mean()
vcinfer.set_obs_unc_std()
vcinfer.first_guess(distType='LogNormal')
vcinfer.optimize_likelihood()
vcinfer.plot_likelihood(save_pdf=True)
assert len(vcinfer.likelihood_samples.index
) == config['likelihood_par']['nsamples']
assert 'kde_ll' in vcinfer.likelihood_estim
assert 'moments_sim' in vcinfer.likelihood_estim
assert len(
vcinfer.likelihood_estim['moments_sim']['mean']) == vcinfer.n_moments
def test_estimate_tau():
raw_trace = np.genfromtxt(TEST_DATA_PATH, delimiter=',')
dt = 0.0001
vcinfer = VCInfer(raw_trace, dt)
vcinfer.psd_calc()
vcinfer.estimate_taus()
vcinfer.plot_tau_estimation(save_pdf=True)
assert vcinfer.param_priors['tau1'] > 0.
assert vcinfer.param_priors['tau2'] > 0.
assert vcinfer.param_priors['constant'] > 0.
def test_set_obs_unc_std():
raw_trace = np.genfromtxt(TEST_DATA_PATH, delimiter=',')
dt = 0.0001
vcinfer = VCInfer(raw_trace, dt)
vcinfer.momenta()
vcinfer.psd_calc()
vcinfer.estimate_taus()
vcinfer.reset_obs_unc()
assert vcinfer.obs_bias['mean'][1] == 0
vcinfer.set_obs_unc_std()
assert vcinfer.obs_bias['mean'][1] > 0
def test_first_guess():
raw_trace = np.genfromtxt(TEST_DATA_PATH, delimiter=',')
dt = 0.0001
vcinfer = VCInfer(raw_trace, dt)
vcinfer.momenta()
vcinfer.psd_calc()
vcinfer.estimate_taus()
vcinfer.reset_obs_unc()
vcinfer.set_obs_unc_mean()
vcinfer.set_obs_unc_std()
dist_list = ['LogNormal', 'Exponential', 'TruncNormal']
for dist_type in dist_list:
logging.info('******************** Inference using {} weights'.format(
dist_type))
vcinfer.first_guess(distType=dist_type)
assert vcinfer.param_priors['pw0'][0] > 0.
assert vcinfer.param_priors['pw1'][0] > 0.
assert vcinfer.param_priors['freq'][0] > 0.