def get_all_results_by_param(self, mlmc_est, param_name):
"""
Get all result values by given param name, e.g. param_name = "temp" - return all temperatures...
:param mlmc_est: Estimate instance
:param param_name: Sample result param name
:return: moments means, moments vars -> two numpy arrays
"""
n_moments = 3
mlmc_est.mlmc.clean_select()
mlmc_est.mlmc.select_values({param_name: (0, ">=")})
domain = Estimate.estimate_domain(mlmc_est.mlmc)
moments_fn = Monomial(n_moments, domain, False, ref_domain=domain)
means, vars = mlmc_est.estimate_moments(moments_fn)
return means, vars
def process(self):
"""
Use collected data
:return: None
"""
assert os.path.isdir(self.work_dir)
mlmc_est_list = []
for nl in [1]: # high resolution fields
mlmc = self.setup_config(nl, clean=False)
# Use wrapper object for working with collected data
mlmc_est = Estimate(mlmc)
mlmc_est_list.append(mlmc_est)
#self.plot_density(mlmc)
self.plot_temp_power(mlmc_est)
def process(self):
"""
Use collected data
:return: None
"""
assert os.path.isdir(self.work_dir)
mlmc_est_list = []
# for nl in [ 1,3,5,7,9]:
import time
for nl in [5]: # high resolution fields
start = time.time()
mlmc = self.setup_config(nl, clean=False)
print("celkový čas ", time.time() - start)
# Use wrapper object for working with collected data
mlmc_est = Estimate(mlmc)
mlmc_est_list.append(mlmc_est)
def plot_param(self, mlmc_est, ax, X, col, param_name, legend):
"""
Get all result values by given param name, e.g. param_name = "temp" - return all temperatures...
:param mlmc_est: Estimate instance
:param param_name: Sample result param name
:return: moments means, moments vars -> two numpy arrays
"""
n_moments = 3
mlmc_est.mlmc.clean_select()
mlmc_est.mlmc.select_values(None, selected_param=param_name)
print("plot param:", param_name)
samples = mlmc_est.mlmc.levels[0].sample_values[:,0,:]
print(" shape: ", samples.shape)
#print(np.any(np.isnan(samples), axis=1))
domain = Estimate.estimate_domain(mlmc_est.mlmc)
domain_diff = domain[1] - domain[0]
print(" domain: ", domain)
moments_fn = Monomial(n_moments, domain, False, ref_domain=domain)
N = mlmc_est.mlmc.n_samples[0]
mom_means, mom_vars = mlmc_est.estimate_moments(moments_fn)
mom_means, mom_vars = mom_means[1:, :], mom_vars[1:, :] # omit time=0
q_mean = mom_means[:, 1]
q_mean_err = np.sqrt(mom_vars[:, 1])
q_var = (mom_means[:, 2] - q_mean ** 2) * N / (N-1)
# print(" means: ", mom_means[-1, :])
# print(" vars: ", mom_vars[-1, :])
# print(" alt var: ", mom_vars[-1, 1]*N)
# print(" qvar : ", q_var[-1])
# print(" qvar_err : ", q_var[-1])
#q_var = q_mean_err * N
q_std = np.sqrt(q_var)
q_std_err = np.sqrt(q_var + np.sqrt(mom_vars[:, 2] * N / (N-1)))
ax.fill_between(X, q_mean - q_mean_err, q_mean + q_mean_err,
color=col, alpha=1, label=legend + " - mean")
ax.fill_between(X, q_mean - q_std, q_mean + q_std,
color=col, alpha=0.2, label=legend + " - std")
ax.fill_between(X, q_mean - q_std_err, q_mean - q_std,
color=col, alpha=0.4, label = None)
ax.fill_between(X, q_mean + q_std, q_mean + q_std_err,
color=col, alpha=0.4, label = None)
def process(self):
"""
Use collected data
:return: None
"""
assert os.path.isdir(self.work_dir)
mlmc_est_list = []
for nl in [1]: # high resolution fields
mlmc = self.setup_config(nl, clean=False)
# Use wrapper object for working with collected data
mlmc_est = Estimate(mlmc)
mlmc_est_list.append(mlmc_est)
# self.result_text(mlmc)
# self.plot_density(mlmc)
mlmc_est.mlmc.clean_select()
th_02_model_params = ["power", "temp", "temp_min", "temp_max"]
th_03_model_params = [n + "_ref" for n in th_02_model_params]
print("02_th - stimulated EGS")
good_mask = self.select_samples_with_good_values(mlmc_est, th_02_model_params)
print("03_th - reference EGS")
good_mask = self.select_samples_with_good_values(mlmc_est, th_03_model_params, good_mask=good_mask)
#good_mask[np.random.choice(len(good_mask), int(len(good_mask)*0.93))] = False
#print("N Random choice: ", sum(good_mask))
mlmc_est.mlmc.subsample_by_indices(good_mask)
# self.plot_temp_power(mlmc_est, th_02_model_params)
# self.plot_temp_power(mlmc_est, th_03_model_params)
#self.plot_histogram_const(mlmc_est, 'fr_cs_med', "fr_cs_median", "Median of $\delta_f$ [mm]", bins=30, scale=1000)
#self.plot_histogram_const(mlmc_est, 'fr_cond_med', "fr_cond_median", "Median of $k_f$ [m.s$^{-1}$]", bins=30, scale=1)
#self.plot_histogram_flux(mlmc_est, 'bc_flux', "total_flux_histogram", bins=30)
self.plot_histogram(mlmc_est, 'temp', "temp_histogram", "Temperature [$^\circ$C]", bins=30)
self.plot_histogram(mlmc_est, 'power', "power_histogram", "Power [MW]", bins=30)
self.plot_temp_ref_comparison(mlmc_est)
self.plot_power_ref_comparison(mlmc_est)
# n_samples = int(mlmc_est.mlmc.n_samples)
# print("N samples: ", n_samples)
# temp_v_ele = np.zeros((2 * n_samples, 2 * n_samples), dtype=int)
# print("good temperature interval: <{},{}>".format(MIN_T, MAX_T))
# bad_temp_flag = np.zeros((len(n_bad_els),))
# bad_ele_flag = np.zeros((len(n_bad_els),))
# for i in range(0, n_samples):
# bad_temp_flag[i] = min(temp_min[i][0]) < MIN_T or max(temp_max[i][0]) > MAX_T
# bad_ele_flag[i] = n_bad_els[i][0][0] > 0
# ty = i if bad_temp_flag[i] else n_samples + i
# tx = i if bad_ele_flag[i] else n_samples + i
# temp_v_ele[ty, tx] = 1
# [np.sum(temp_v_ele[:n_samples, :n_samples]),
# np.sum(temp_v_ele[:n_samples, n_samples:2 * n_samples])],
# [np.sum(temp_v_ele[n_samples:2 * n_samples, :n_samples]),
# np.sum(temp_v_ele[n_samples:2 * n_samples, n_samples:2 * n_samples])]]
#
# print("temp[BAD, GOOD]' x mesh[BAD, GOOD]]\n{}\n{}".format(temp_v_ele_sum[0], temp_v_ele_sum[1]))
#
# # print(n_bad_els.shape)
# n_bad_elements = np.zeros((len(n_bad_els),))
# for i in range(len(n_bad_els)):
# n_bad_elements[i] = n_bad_els[i][0][0]
#
# print("min n_bad_element = ", min(n_bad_elements))
# print("max n_bad_element = ", max(n_bad_elements))
#
# # compute EX and varX
# bad_temp_ele_avg = 0 # EX of bad elements for bad temperature
# good_temp_ele_avg = 0 # EX of bad elements for good temperature
# for i in range(len(n_bad_els)):
# if bad_temp_flag[i]:
# bad_temp_ele_avg += n_bad_elements[i]
# else:
# good_temp_ele_avg += n_bad_elements[i]
# if temp_v_ele_sum[0][0] != 0:
# bad_temp_ele_avg /= temp_v_ele_sum[0][0]
# if temp_v_ele_sum[1][0] != 0:
# good_temp_ele_avg /= temp_v_ele_sum[1][0]
# print("bad_temp EX: ", bad_temp_ele_avg)
# print("good_temp EX: ", good_temp_ele_avg)
#
# bad_temp_ele_var = 0 # varX of bad elements for bad temperature
# good_temp_ele_var = 0 # varX of bad elements for good temperature
# for i in range(len(n_bad_els)):
# if bad_temp_flag[i]:
# bad_temp_ele_var += (n_bad_elements[i]-bad_temp_ele_avg)**2
# else:
# good_temp_ele_var += (n_bad_elements[i]-good_temp_ele_avg)**2
# if temp_v_ele_sum[0][0] != 0:
# bad_temp_ele_var /= temp_v_ele_sum[0][0]
# if temp_v_ele_sum[1][0] != 0:
# good_temp_ele_var /= temp_v_ele_sum[1][0]
# print("bad_temp varX: ", bad_temp_ele_var)
# print("good_temp varX: ", good_temp_ele_var)
# print("bad_temp s: ", np.sqrt(bad_temp_ele_var))
# print("good_temp s: ", np.sqrt(good_temp_ele_var))
#
# # print("temp_v_ele:\n", temp_v_ele)
# temp_v_ele_sum = [
# [np.sum(temp_v_ele[:n_samples, :n_samples]),
# np.sum(temp_v_ele[:n_samples, n_samples:2 * n_samples])],
# [np.sum(temp_v_ele[n_samples:2 * n_samples, :n_samples]),
# np.sum(temp_v_ele[n_samples:2 * n_samples, n_samples:2 * n_samples])]]
#
# print("temp[BAD, GOOD]' x mesh[BAD, GOOD]]\n{}\n{}".format(temp_v_ele_sum[0], temp_v_ele_sum[1]))
#
# # print(n_bad_els.shape)
# n_bad_elements = np.zeros((len(n_bad_els),))
# for i in range(len(n_bad_els)):
# n_bad_elements[i] = n_bad_els[i][0][0]
#
# print("min n_bad_element = ", min(n_bad_elements))
# print("max n_bad_element = ", max(n_bad_elements))
#
# # compute EX and varX
# bad_temp_ele_avg = 0 # EX of bad elements for bad temperature
# good_temp_ele_avg = 0 # EX of bad elements for good temperature
# for i in range(len(n_bad_els)):
# if bad_temp_flag[i]:
# bad_temp_ele_avg += n_bad_elements[i]
# else:
# good_temp_ele_avg += n_bad_elements[i]
# bad_temp_ele_avg /= temp_v_ele_sum[0][0]
# good_temp_ele_avg /= temp_v_ele_sum[1][0]
# print("bad_temp EX: ", bad_temp_ele_avg)
# print("good_temp EX: ", good_temp_ele_avg)
#
# bad_temp_ele_var = 0 # varX of bad elements for bad temperature
# good_temp_ele_var = 0 # varX of bad elements for good temperature
# for i in range(len(n_bad_els)):
# if bad_temp_flag[i]:
# bad_temp_ele_var += (n_bad_elements[i]-bad_temp_ele_avg)**2
# else:
# good_temp_ele_var += (n_bad_elements[i]-good_temp_ele_avg)**2
# bad_temp_ele_var /= temp_v_ele_sum[0][0]
# good_temp_ele_var /= temp_v_ele_sum[1][0]
# print("bad_temp varX: ", bad_temp_ele_var)
# print("good_temp varX: ", good_temp_ele_var)
# print("bad_temp s: ", np.sqrt(bad_temp_ele_var))
# print("good_temp s: ", np.sqrt(good_temp_ele_var))
#
# # print("n_bad_els:\n", n_bad_els)
#
# # print("TEMP_MIN: ", temp_min)
# # print("TEMP_MAX: ", temp_max)
# #
# # print("TEMP_MIN: ", min(temp_min[0][0]))
# # print("TEMP_MAX: ", max(temp_max[1][0]))
print("PROCESS FINISHED :)")