def _read_bwo_file(bwo_file_name):
"""Reads backwards-optimization results from file.
:param bwo_file_name: Path to input file (will be read by
`backwards_optimization.read_file`).
:return: bwo_dictionary: Dictionary returned by
`backwards_optimization.read_file`.
:return: model_metadata_dict: Dictionary returned by
`cnn.read_model_metadata`.
"""
print('Reading data from: "{0:s}"...'.format(bwo_file_name))
bwo_dictionary = backwards_opt.read_file(bwo_file_name)[0]
mean_before_matrices = [
numpy.expand_dims(a, axis=0)
for a in bwo_dictionary[backwards_opt.MEAN_INPUT_MATRICES_KEY]
]
mean_after_matrices = [
numpy.expand_dims(a, axis=0)
for a in bwo_dictionary[backwards_opt.MEAN_OUTPUT_MATRICES_KEY]
]
model_file_name = bwo_dictionary[backwards_opt.MODEL_FILE_KEY]
model_metafile_name = cnn.find_metafile(model_file_name)
print('Reading CNN metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
good_indices = numpy.array([
numpy.where(
training_option_dict[trainval_io.RADAR_HEIGHTS_KEY] == h)[0][0]
for h in RADAR_HEIGHTS_M_AGL
],
dtype=int)
mean_before_matrices[0] = mean_before_matrices[0][..., good_indices, :]
mean_after_matrices[0] = mean_after_matrices[0][..., good_indices, :]
good_indices = numpy.array([
training_option_dict[trainval_io.RADAR_FIELDS_KEY].index(f)
for f in RADAR_FIELD_NAMES
],
dtype=int)
mean_before_matrices[0] = mean_before_matrices[0][..., good_indices]
mean_after_matrices[0] = mean_after_matrices[0][..., good_indices]
bwo_dictionary[
backwards_opt.MEAN_INPUT_MATRICES_KEY] = mean_before_matrices
bwo_dictionary[
backwards_opt.MEAN_OUTPUT_MATRICES_KEY] = mean_after_matrices
training_option_dict[trainval_io.RADAR_HEIGHTS_KEY] = RADAR_HEIGHTS_M_AGL
training_option_dict[trainval_io.RADAR_FIELDS_KEY] = RADAR_FIELD_NAMES
model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY] = training_option_dict
return bwo_dictionary, model_metadata_dict
def _run():
"""Analyzes backwards-optimization experiment.
This is effectively the main method.
"""
num_l2_weights = len(L2_WEIGHTS)
num_minmax_weights = len(MINMAX_WEIGHTS)
mean_final_activation_matrix = numpy.full(
(num_l2_weights, num_minmax_weights), numpy.nan)
for i in range(num_l2_weights):
for j in range(num_minmax_weights):
this_file_name = (
'{0:s}/bwo_pmm_l2-weight={1:.10f}_minmax-weight={2:.10f}.p'
).format(TOP_EXPERIMENT_DIR_NAME, L2_WEIGHTS[i], MINMAX_WEIGHTS[j])
print('Reading data from: "{0:s}"...'.format(this_file_name))
this_bwo_dict = backwards_opt.read_file(this_file_name)[0]
mean_final_activation_matrix[i, j] = this_bwo_dict[
backwards_opt.MEAN_FINAL_ACTIVATION_KEY]
x_tick_labels = ['{0:.1f}'.format(r) for r in numpy.log10(MINMAX_WEIGHTS)]
y_tick_labels = ['{0:.1f}'.format(w) for w in numpy.log10(L2_WEIGHTS)]
axes_object = model_evaluation.plot_hyperparam_grid(
score_matrix=mean_final_activation_matrix,
colour_map_object=COLOUR_MAP_OBJECT,
min_colour_value=0.,
max_colour_value=1.)
axes_object.set_xticklabels(x_tick_labels, rotation=90.)
axes_object.set_yticklabels(y_tick_labels)
axes_object.set_xlabel(r'Min-max weight (log$_{10}$)')
axes_object.set_ylabel(r'L$_2$ weight (log$_{10}$)')
plotting_utils.plot_linear_colour_bar(
axes_object_or_matrix=axes_object,
data_matrix=mean_final_activation_matrix,
colour_map_object=COLOUR_MAP_OBJECT,
min_value=0.,
max_value=1.,
orientation_string='vertical',
extend_min=False,
extend_max=False,
font_size=FONT_SIZE)
output_file_name = '{0:s}/mean_final_activations.jpg'.format(
TOP_EXPERIMENT_DIR_NAME)
print('Saving figure to: "{0:s}"...'.format(output_file_name))
pyplot.savefig(output_file_name,
dpi=FIGURE_RESOLUTION_DPI,
pad_inches=0,
bbox_inches='tight')
pyplot.close()
def _composite_backwards_opt(
input_file_name, max_percentile_level, output_file_name):
"""Composites inputs and outputs for backwards optimization.
:param input_file_name: Path to input file. Will be read by
`backwards_optimization.read_file`.
:param max_percentile_level: See documentation at top of file.
:param output_file_name: Path to output file. Will be written by
`backwards_optimization.write_pmm_file`.
"""
print('Reading data from: "{0:s}"...'.format(input_file_name))
bwo_dictionary = backwards_opt.read_file(input_file_name)[0]
input_matrices = bwo_dictionary[backwards_opt.INPUT_MATRICES_KEY]
output_matrices = bwo_dictionary[backwards_opt.OUTPUT_MATRICES_KEY]
sounding_pressure_matrix_pa = bwo_dictionary[
backwards_opt.SOUNDING_PRESSURES_KEY]
print('Compositing non-optimized predictors...')
mean_input_matrices, mean_sounding_pressures_pa = _composite_predictors(
predictor_matrices=input_matrices,
max_percentile_level=max_percentile_level,
sounding_pressure_matrix_pa=sounding_pressure_matrix_pa)
print('Compositing optimized predictors...')
mean_output_matrices = _composite_predictors(
predictor_matrices=output_matrices,
max_percentile_level=max_percentile_level
)[0]
mean_initial_activation = numpy.mean(
bwo_dictionary[backwards_opt.INITIAL_ACTIVATIONS_KEY]
)
mean_final_activation = numpy.mean(
bwo_dictionary[backwards_opt.FINAL_ACTIVATIONS_KEY]
)
print('Writing output to: "{0:s}"...'.format(output_file_name))
backwards_opt.write_pmm_file(
pickle_file_name=output_file_name,
mean_denorm_input_matrices=mean_input_matrices,
mean_denorm_output_matrices=mean_output_matrices,
mean_initial_activation=mean_initial_activation,
mean_final_activation=mean_final_activation,
model_file_name=bwo_dictionary[backwards_opt.MODEL_FILE_KEY],
non_pmm_file_name=input_file_name,
pmm_max_percentile_level=max_percentile_level,
mean_sounding_pressures_pa=mean_sounding_pressures_pa)
def _read_bwo_file(bwo_file_name):
"""Reads backwards-optimization results from file.
:param bwo_file_name: Path to input file (will be read by
`backwards_optimization.read_file`).
:return: bwo_dictionary: Dictionary returned by
`backwards_optimization.read_file`.
:return: model_metadata_dict: Dictionary returned by
`cnn.read_model_metadata`.
"""
print('Reading data from: "{0:s}"...'.format(bwo_file_name))
bwo_dictionary = backwards_opt.read_file(bwo_file_name)[0]
mean_before_matrices = [
numpy.expand_dims(a, axis=0) for a in
bwo_dictionary[MEAN_INPUT_MATRICES_KEY]
]
mean_after_matrices = [
numpy.expand_dims(a, axis=0) for a in
bwo_dictionary[MEAN_OUTPUT_MATRICES_KEY]
]
model_file_name = bwo_dictionary[MODEL_FILE_KEY]
model_metafile_name = cnn.find_metafile(model_file_name)
print('Reading CNN metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
training_option_dict = model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY]
training_option_dict[trainval_io.UPSAMPLE_REFLECTIVITY_KEY] = False
all_refl_heights_m_agl = training_option_dict[trainval_io.RADAR_HEIGHTS_KEY]
good_flags = numpy.array(
[h in REFL_HEIGHTS_M_AGL for h in all_refl_heights_m_agl], dtype=bool
)
good_indices = numpy.where(good_flags)[0]
mean_before_matrices[0] = mean_before_matrices[0][..., good_indices, :]
mean_after_matrices[0] = mean_after_matrices[0][..., good_indices, :]
bwo_dictionary[MEAN_INPUT_MATRICES_KEY] = mean_before_matrices
bwo_dictionary[MEAN_OUTPUT_MATRICES_KEY] = mean_after_matrices
training_option_dict[trainval_io.RADAR_HEIGHTS_KEY] = REFL_HEIGHTS_M_AGL
model_metadata_dict[cnn.TRAINING_OPTION_DICT_KEY] = training_option_dict
return bwo_dictionary, model_metadata_dict
def _run():
"""Analyzes backwards-optimization experiment.
This is effectively the main method.
"""
num_weights = len(MINMAX_WEIGHTS)
for i in range(num_weights):
this_file_name = '{0:s}/bwo_minmax-weight={1:014.10f}_pmm.p'.format(
TOP_EXPERIMENT_DIR_NAME, MINMAX_WEIGHTS[i])
this_bwo_dict = backwards_opt.read_file(this_file_name)[0]
this_mean_final_activation = this_bwo_dict[
backwards_opt.MEAN_FINAL_ACTIVATION_KEY]
print(
('Min-max weight = 10^{0:.1f} ... mean final activation = {1:.4f}'
).format(numpy.log10(MINMAX_WEIGHTS[i]),
this_mean_final_activation))
def _run(input_file_name, diff_colour_map_name, max_diff_percentile,
plot_soundings, allow_whitespace, plot_panel_names, add_titles,
label_colour_bars, colour_bar_length, top_output_dir_name):
"""Plots results of backwards optimization.
This is effectively the main method.
:param input_file_name: See documentation at top of file.
:param diff_colour_map_name: Same.
:param max_diff_percentile: Same.
:param plot_soundings: Same.
:param allow_whitespace: Same.
:param plot_panel_names: Same.
:param add_titles: Same.
:param label_colour_bars: Same.
:param colour_bar_length: Same.
:param top_output_dir_name: Same.
"""
diff_colour_map_object = pyplot.cm.get_cmap(diff_colour_map_name)
print('Reading data from: "{0:s}"...'.format(input_file_name))
bwo_dictionary, pmm_flag = backwards_opt.read_file(input_file_name)
if pmm_flag:
input_matrices = bwo_dictionary.pop(
backwards_opt.MEAN_INPUT_MATRICES_KEY)
output_matrices = bwo_dictionary.pop(
backwards_opt.MEAN_OUTPUT_MATRICES_KEY)
full_storm_id_strings = [None]
storm_times_unix_sec = [None]
mean_sounding_pressures_pa = bwo_dictionary[
backwards_opt.MEAN_SOUNDING_PRESSURES_KEY]
if mean_sounding_pressures_pa is None:
sounding_pressure_matrix_pa = None
else:
sounding_pressure_matrix_pa = numpy.reshape(
mean_sounding_pressures_pa,
(1, len(mean_sounding_pressures_pa)))
for i in range(len(input_matrices)):
input_matrices[i] = numpy.expand_dims(input_matrices[i], axis=0)
output_matrices[i] = numpy.expand_dims(output_matrices[i], axis=0)
else:
input_matrices = bwo_dictionary.pop(backwards_opt.INPUT_MATRICES_KEY)
output_matrices = bwo_dictionary.pop(backwards_opt.OUTPUT_MATRICES_KEY)
full_storm_id_strings = bwo_dictionary[
backwards_opt.FULL_STORM_IDS_KEY]
storm_times_unix_sec = bwo_dictionary[backwards_opt.STORM_TIMES_KEY]
sounding_pressure_matrix_pa = bwo_dictionary[
backwards_opt.SOUNDING_PRESSURES_KEY]
model_file_name = bwo_dictionary[backwards_opt.MODEL_FILE_KEY]
model_metafile_name = '{0:s}/model_metadata.p'.format(
os.path.split(model_file_name)[0])
print('Reading metadata from: "{0:s}"...'.format(model_metafile_name))
model_metadata_dict = cnn.read_model_metadata(model_metafile_name)
print(SEPARATOR_STRING)
plot_examples.plot_examples(
list_of_predictor_matrices=input_matrices,
model_metadata_dict=model_metadata_dict,
pmm_flag=pmm_flag,
output_dir_name='{0:s}/before_optimization'.format(
top_output_dir_name),
plot_soundings=plot_soundings,
sounding_pressure_matrix_pascals=sounding_pressure_matrix_pa,
allow_whitespace=allow_whitespace,
plot_panel_names=plot_panel_names,
add_titles=add_titles,
label_colour_bars=label_colour_bars,
colour_bar_length=colour_bar_length,
plot_radar_diffs=False,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec)
print(SEPARATOR_STRING)
plot_examples.plot_examples(
list_of_predictor_matrices=output_matrices,
model_metadata_dict=model_metadata_dict,
pmm_flag=pmm_flag,
output_dir_name='{0:s}/after_optimization'.format(top_output_dir_name),
plot_soundings=plot_soundings,
sounding_pressure_matrix_pascals=sounding_pressure_matrix_pa,
allow_whitespace=allow_whitespace,
plot_panel_names=plot_panel_names,
add_titles=add_titles,
label_colour_bars=label_colour_bars,
colour_bar_length=colour_bar_length,
plot_radar_diffs=False,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec)
print(SEPARATOR_STRING)
difference_matrices = [
b - a for b, a in zip(output_matrices, input_matrices)
]
plot_examples.plot_examples(
list_of_predictor_matrices=difference_matrices,
model_metadata_dict=model_metadata_dict,
pmm_flag=pmm_flag,
output_dir_name='{0:s}/difference'.format(top_output_dir_name),
plot_soundings=False,
allow_whitespace=allow_whitespace,
plot_panel_names=plot_panel_names,
add_titles=add_titles,
label_colour_bars=label_colour_bars,
colour_bar_length=colour_bar_length,
plot_radar_diffs=True,
diff_colour_map_object=diff_colour_map_object,
max_diff_percentile=max_diff_percentile,
full_storm_id_strings=full_storm_id_strings,
storm_times_unix_sec=storm_times_unix_sec)