from sparse_ho.ho import grad_search from sparse_ho.utils import Monitor from scipy.sparse import csc_matrix n_samples = 10 n_features = 20 n_active = 5 tol = 1e-16 max_iter = 50000 SNR = 3 rho = 0.1 X_train, y_train, beta_star, noise, sigma_star = get_synt_data( dictionary_type="Gaussian", n_samples=n_samples, n_features=n_features, n_times=1, n_active=n_active, rho=rho, SNR=SNR, seed=0) X_train = csc_matrix(X_train) X_test, y_test, beta_star, noise, sigma = get_synt_data( dictionary_type="Gaussian", n_samples=n_samples, n_features=n_features, n_times=1, n_active=n_active, rho=rho, SNR=SNR, seed=1) X_test = csc_matrix(X_test)
from sparse_ho import Forward from sparse_ho import ImplicitForward from sparse_ho import Implicit from sparse_ho import Backward from sparse_ho.criterion import HeldOutMSE, SmoothedSURE n_samples = 100 n_features = 100 n_active = 5 SNR = 3 rho = 0.1 X, y, beta_star, noise, sigma_star = get_synt_data(dictionary_type="Toeplitz", n_samples=n_samples, n_features=n_features, n_times=1, n_active=n_active, rho=rho, SNR=SNR, seed=0) X_s = csc_matrix(X) idx_train = np.arange(0, 50) idx_val = np.arange(50, 100) alpha_max = (np.abs(X[idx_train, :].T @ y[idx_train])).max() / n_samples p_alpha = 0.9 alpha = p_alpha * alpha_max log_alpha = np.log(alpha) log_alphas = np.log(alpha_max * np.geomspace(1, 0.1)) tol = 1e-16