o Qi+@s&ddlmZddlmZdddZdS))check_random_state)randomized_svdNcCs\|dks t|tr t|}|jd}||krtd||ft||||d\}}}|||fS)a :param X: scipy.sparse.csr_matrix Input matrix :param n_components: int Size of embedding dimension :param n_iter: int Maximum number of iteration. Default is 5 :param random_state: random state Default is None Returns ---------- U : numpy.ndarray Representation matrix of rows. shape = (n_rows, n_components) Sigma : numpy.ndarray Eigenvalue of dimension. shape = (n_components, n_components) Diagonal value are in decreasing order VT : numpy.ndarray Representation matrix of columns. shape = (n_components, n_cols) Usage ----- >>> U, Sigma, VT = svd(X, n_components=100) Nz/n_components must be < n_features; got %d >= %d)n_iter random_state) isinstanceintrshape ValueErrorr)X n_componentsrr n_featuresUSigmaVTrE/home/ubuntu/.local/lib/python3.10/site-packages/soynlp/utils/math.pysvds  r)rN) sklearn.utilsrsklearn.utils.extmathrrrrrrs