Remove unused support for sparse matrices

kmeans/_kmeans.pyx

@@ -19,7 +19,6 @@ from libc.stdlib cimport malloc, calloc, free
 from libc.string cimport memset
 from libc.math cimport sqrt
 from scipy.linalg.cython_blas cimport sgemm, dgemm
-from scipy import sparse
 
 np.import_array()
 
@@ -244,38 +243,12 @@ cpdef void _relocate_empty_clusters_dense(
         weight_in_clusters[new_cluster_id] = weight
         weight_in_clusters[old_cluster_id] -= weight
 
-def _csr_row_norms(X):
-    """L2 norm of each row in CSR matrix X."""
-    if X.dtype not in [np.float32, np.float64]:
-        X = X.astype(np.float64)
-    return _csr_row_norms(X.data, X.shape, X.indices, X.indptr)
-
-def _csr_row_norms(np.ndarray[floating, ndim=1, mode="c"] X_data,
-                   shape,
-                   np.ndarray[integral, ndim=1, mode="c"] X_indices,
-                   np.ndarray[integral, ndim=1, mode="c"] X_indptr):
-    cdef:
-        unsigned long long n_samples = shape[0]
-        unsigned long long i
-        integral j
-        double sum_
-
-    norms = np.empty(n_samples, dtype=X_data.dtype)
-    cdef floating[::1] norms_view = norms
-
-    for i in range(n_samples):
-        sum_ = 0.0
-        for j in range(X_indptr[i], X_indptr[i + 1]):
-            sum_ += X_data[j] * X_data[j]
-        norms_view[i] = sum_
-
-    return norms
 
 def row_norms(X, squared=False):
     """Row-wise (squared) Euclidean norm of X.
 
-    Equivalent to np.sqrt((X * X).sum(axis=1)), but also supports sparse
-    matrices and does not create an X.shape-sized temporary.
+    Equivalent to np.sqrt((X * X).sum(axis=1)), but does not create
+    an X.shape-sized temporary.
 
     Performs no input validation.
 
@@ -291,12 +264,7 @@ def row_norms(X, squared=False):
     array-like
         The row-wise (squared) Euclidean norm of X.
     """
-    if sparse.issparse(X):
-        if not isinstance(X, sparse.csr_matrix):
-            X = sparse.csr_matrix(X)
-        norms = _csr_row_norms(X)
-    else:
-        norms = np.einsum('ij,ij->i', X, X)
+    norms = np.einsum('ij,ij->i', X, X)
 
     if not squared:
         np.sqrt(norms, norms)