Skip to content

LR refactor #57

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Gromwud merged 2 commits into from
Dec 22, 2025
Merged

LR refactor #57

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b0cc885
lr refactor
Gromwud Dec 22, 2025
3366403
Merge branch 'ITMO-NSS-team:main' into main
Gromwud Dec 22, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
229 changes: 39 additions & 190 deletions epde/operators/common/fitness.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,206 +161,55 @@ def apply(self, objective: Equation, arguments: dict, force_out_of_place: bool =
objective.aic = None
objective.aic_calculated = True

# assert objective.simplified, 'Trying to evaluate not simplified equation.'

# Calculate r-loss
target_vals = target.reshape(*data_shape)
if not features is None:
features_vals = features.reshape(*data_shape, -1)
slices = [slice(None) for _ in range(target_vals.ndim)]
# if not features is None:
features_vals = features.reshape(*data_shape, -1)
sample_weights_vals = self.g_fun_vals.reshape(*data_shape)

if target_vals.ndim == 1:
lr = 0
for dim in range(target_vals.ndim):
horizons_default = 30
window_size = len(target_vals) // 2
num_horizons = len(target_vals) - window_size + 1
window_size = target_vals.shape[dim] // 2
num_horizons = window_size + 1
if num_horizons < horizons_default:
step_size = 1
else:
step_size = num_horizons // horizons_default
eq_window_weights = []
# Compute coefficients and collect statistics over horizons
if features is None:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
target_window = target_vals[start_idx:end_idx]
eq_window_weights.append(target_window.mean())
# lr = np.std(eq_window_weights, ddof=1) / np.sqrt(np.mean(np.pow(eq_window_weights, 2)))
# lr = np.abs(np.var(eq_window_weights, ddof=1) / np.mean(eq_window_weights))
lr = np.sqrt(np.std(eq_window_weights, ddof=1) ** 2 / (np.std(eq_window_weights, ddof=1) ** 2 + np.mean(eq_window_weights) ** 2))
else:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
target_window = target_vals[start_idx:end_idx]
feature_window = features_vals[start_idx:end_idx, :].reshape(-1, features.shape[-1])
# estimator = LinearRegression(fit_intercept=True)
estimator = Ridge(alpha=0, copy_X=True, fit_intercept=True, max_iter=20,
positive=False, random_state=None, tol=0.0001, solver='sparse_cg')
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals[start_idx:end_idx])
valuable_weights = estimator.coef_
eq_window_weights.append(valuable_weights)
eq_cv = np.array([
# np.std(_, ddof=1) / np.sqrt(np.mean(np.pow(_, 2)))
# np.abs(np.var(_, ddof=1) / np.mean(_))
# np.abs(np.std(_, ddof=1) / np.mean(_))
np.sqrt(np.std(_, ddof=1) ** 2 / (np.std(_, ddof=1) ** 2 + np.mean(_) ** 2))
for _ in zip(*np.array(eq_window_weights))
])
lr = np.nan_to_num(eq_cv).sum()

elif target_vals.ndim == 2:
lr = 0
for dim in range(target_vals.ndim):
horizons_default = 30
eq_window_weights = []
window_size = target_vals.shape[dim] // 2
num_horizons = target_vals.shape[dim] - window_size + 1
if num_horizons < horizons_default:
step_size = 1
else:
step_size = num_horizons // horizons_default
# Compute coefficients and collect statistics over horizons
if features is None:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
if dim == 0:
target_window = target_vals[start_idx:end_idx, :].reshape(-1)
else:
target_window = target_vals[:, start_idx:end_idx].reshape(-1)
eq_window_weights.append(target_window.mean())
# lr += np.std(eq_window_weights, ddof=1) / np.sqrt(np.mean(np.pow(eq_window_weights, 2)))
# lr += np.abs(np.std(eq_window_weights, ddof=1) / np.mean(eq_window_weights))
lr += np.sqrt(np.std(eq_window_weights, ddof=1) ** 2 / (np.std(eq_window_weights, ddof=1) ** 2 + np.mean(eq_window_weights) ** 2))
else:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
# estimator = LinearRegression(fit_intercept=True)
estimator = Ridge(alpha=0, copy_X=True, fit_intercept=True, max_iter=20,
positive=False, random_state=None, tol=0.0001, solver='sparse_cg')
if dim == 0:
target_window = target_vals[start_idx:end_idx, :].reshape(-1)
feature_window = features_vals[start_idx:end_idx, :].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[start_idx:end_idx, :].reshape(-1))
else:
target_window = target_vals[:, start_idx:end_idx].reshape(-1)
feature_window = features_vals[:, start_idx:end_idx].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, start_idx:end_idx].reshape(-1))
valuable_weights = estimator.coef_
eq_window_weights.append(valuable_weights)
eq_cv = np.array([
# np.std(_, ddof=1) / np.sqrt(np.mean(np.pow(_, 2)))
# np.abs(np.std(_, ddof=1) / np.mean(_))
np.sqrt(np.std(_, ddof=1) ** 2 / (np.std(_, ddof=1) ** 2 + np.mean(_) ** 2))
for _ in zip(*np.array(eq_window_weights))
])
lr += np.nan_to_num(eq_cv).sum()

elif target_vals.ndim == 3:
lr = 0
for dim in range(target_vals.ndim):
horizons_default = 30
eq_window_weights = []
window_size = target_vals.shape[dim] // 2
num_horizons = target_vals.shape[dim] - window_size + 1
if num_horizons < horizons_default:
step_size = 1
else:
step_size = num_horizons // horizons_default
# Compute coefficients and collect statistics over horizons
if features is None:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
if dim == 0:
target_window = target_vals[start_idx:end_idx, :, :].reshape(-1)
elif dim == 1:
target_window = target_vals[:, start_idx:end_idx, :].reshape(-1)
else:
target_window = target_vals[:, :, start_idx:end_idx].reshape(-1)
eq_window_weights.append(target_window.mean())
lr += np.sqrt(np.std(eq_window_weights, ddof=1) ** 2 / (np.std(eq_window_weights, ddof=1) ** 2 + np.mean(eq_window_weights) ** 2))
else:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
# estimator = LinearRegression(fit_intercept=True)
estimator = Ridge(alpha=0, copy_X=True, fit_intercept=True, max_iter=20,
positive=False, random_state=None, tol=0.0001, solver='sparse_cg')
if dim == 0:
target_window = target_vals[start_idx:end_idx, :, :].reshape(-1)
feature_window = features_vals[start_idx:end_idx, :, :].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[start_idx:end_idx, :, :].reshape(-1))
elif dim == 1:
target_window = target_vals[:, start_idx:end_idx, :].reshape(-1)
feature_window = features_vals[:, start_idx:end_idx, :].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, start_idx:end_idx, :].reshape(-1))
elif dim == 2:
target_window = target_vals[:, :, start_idx:end_idx].reshape(-1)
feature_window = features_vals[:, :, start_idx:end_idx].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, :, start_idx:end_idx].reshape(-1))
valuable_weights = estimator.coef_
eq_window_weights.append(valuable_weights)
eq_cv = np.array([
np.sqrt(np.std(_, ddof=1) ** 2 / (np.std(_, ddof=1) ** 2 + np.mean(_) ** 2))
for _ in zip(*np.array(eq_window_weights))
])
lr += np.nan_to_num(eq_cv).sum()

elif target_vals.ndim == 4:
lr = 0
for dim in range(target_vals.ndim):
horizons_default = 30
eq_window_weights = []
window_size = target_vals.shape[dim] // 2
num_horizons = target_vals.shape[dim] - window_size + 1
if num_horizons < horizons_default:
step_size = 1
else:
step_size = num_horizons // horizons_default
# Compute coefficients and collect statistics over horizons
if features is None:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
if dim == 0:
target_window = target_vals[start_idx:end_idx, :, :, :].reshape(-1)
elif dim == 1:
target_window = target_vals[:, start_idx:end_idx, :, :].reshape(-1)
elif dim == 2:
target_window = target_vals[:, :, start_idx:end_idx, :].reshape(-1)
else:
target_window = target_vals[:, :, :, start_idx:end_idx].reshape(-1)
eq_window_weights.append(target_window.mean())
lr += np.sqrt(np.std(eq_window_weights, ddof=1) ** 2 / (np.std(eq_window_weights, ddof=1) ** 2 + np.mean(eq_window_weights) ** 2))
else:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
# estimator = LinearRegression(fit_intercept=True)
estimator = Ridge(alpha=0, copy_X=True, fit_intercept=True, max_iter=20,
positive=False, random_state=None, tol=0.0001, solver='sparse_cg')
if dim == 0:
target_window = target_vals[start_idx:end_idx, :, :, :].reshape(-1)
feature_window = features_vals[start_idx:end_idx, :, :, :].reshape(-1, features.shape[-1])
estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[start_idx:end_idx, :, :, :].reshape(-1))
# elif dim == 1:
# target_window = target_vals[:, start_idx:end_idx, :, :].reshape(-1)
# feature_window = features_vals[:, start_idx:end_idx, :, :].reshape(-1, features.shape[-1])
# estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, start_idx:end_idx, :, :].reshape(-1))
# elif dim == 2:
# target_window = target_vals[:, :, start_idx:end_idx, :].reshape(-1)
# feature_window = features_vals[:, :, start_idx:end_idx, :].reshape(-1, features.shape[-1])
# estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, :, start_idx:end_idx, :].reshape(-1))
# elif dim == 3:
# target_window = target_vals[:, :, :, start_idx:end_idx].reshape(-1)
# feature_window = features_vals[:, :, :, start_idx:end_idx].reshape(-1, features.shape[-1])
# estimator.fit(feature_window, target_window, sample_weight=self.g_fun_vals.reshape(*data_shape, -1)[:, :, :, start_idx:end_idx].reshape(-1))
else:
continue
valuable_weights = estimator.coef_
eq_window_weights.append(valuable_weights)
eq_cv = np.array([
np.sqrt(np.std(_, ddof=1) ** 2 / (np.std(_, ddof=1) ** 2 + np.mean(_) ** 2))
for _ in zip(*np.array(eq_window_weights))
])
lr += np.nan_to_num(eq_cv).sum()

lr = lr / target_vals.ndim / (len(objective.structure) - 1)

slices_window = slices.copy()
# if features is None:
# for start_idx in range(0, num_horizons, step_size):
# end_idx = start_idx + window_size
# slices_window[dim] = slice(start_idx, end_idx)
# target_window = target_vals[*slices_window]
# eq_window_weights.append(target_window.mean())
# lr += np.sqrt(np.std(eq_window_weights, ddof=1) ** 2 / (np.std(eq_window_weights, ddof=1) ** 2 + np.mean(eq_window_weights) ** 2))
# else:
for start_idx in range(0, num_horizons, step_size):
end_idx = start_idx + window_size
slices_window[dim] = slice(start_idx, end_idx)
target_window = target_vals[*slices_window].reshape(-1)
feature_window = features_vals[*slices_window, :].reshape(-1, features.shape[-1])
sample_weights_window = sample_weights_vals[*slices_window].reshape(-1)
estimator = LinearRegression(fit_intercept=True)
# estimator = Ridge(alpha=0, copy_X=True, fit_intercept=True, max_iter=20,
# positive=False, random_state=None, tol=0.0001, solver='sparse_cg')
estimator.fit(feature_window, target_window, sample_weight=sample_weights_window)
valuable_weights = estimator.coef_
eq_window_weights.append(valuable_weights)
std = np.array(eq_window_weights).std(axis=0, ddof=1)
mu = np.array(eq_window_weights).mean(axis=0)
scale = []
for feature in range(features.shape[-1]):
scale.append(feature_window[:, feature] ** 2 / (feature_window[:, feature] ** 2 + target_window ** 2))
scale = np.array(scale).mean(axis=1)
eq_cv = np.sqrt(std ** 2 / (std ** 2 + mu ** 2) * scale)
lr += np.nan_to_num(eq_cv).sum()

lr = lr / (len(objective.structure) - 1) / target_vals.ndim
# if force_out_of_place:
# return lr

Expand Down
2 changes: 1 addition & 1 deletion epde/operators/common/sparsity.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ def get_cv(self, weights):
std = np.array(weights).std(axis=0, ddof=1)
mu = np.array(weights).mean(axis=0)
# cv = std ** 2 / (std ** 2 + mu ** 2)
cv = std ** 2 / (std ** 2 + mu ** 2)
cv = np.sqrt(std ** 2 / (std ** 2 + mu ** 2))
return cv

def calculate_weights(self, X, y):
Expand Down