format editing added local lobpcg temporarily

lobpcg · lobpcg · commit 539dfd48e77f · 2018-10-07T17:38:30.000-04:00
diff --git a/sklearn/decomposition/pca.py b/sklearn/decomposition/pca.py
@@ -185,16 +185,17 @@ class PCA(_BasePCA):
         .. versionadded:: 0.18.0
 
     iterated_power : int >= 0, or 'auto', (default 'auto')
-        Number of iterations of svd_solver = 'lobpcg' or for the power method computed by
-        svd_solver == 'randomized'.
+        Number of iterations of svd_solver = 'lobpcg' or for the power method
+        computed by svd_solver == 'randomized'.
 
         .. versionadded:: 0.18.0
 
     random_state : int, RandomState instance or None, optional (default None)
         If int, random_state is the seed used by the random number generator;
         If RandomState instance, random_state is the random number generator;
         If None, the random number generator is the RandomState instance used
-        by `np.random`. Used when ``svd_solver`` == 'arpack', 'lobpcg', or 'randomized'.
+        by `np.random` in
+        ``svd_solver`` == 'arpack', 'lobpcg', or 'randomized'.
 
         .. versionadded:: 0.18.0
 
diff --git a/sklearn/decomposition/truncated_svd.py b/sklearn/decomposition/truncated_svd.py
@@ -13,7 +13,8 @@
 
 from ..base import BaseEstimator, TransformerMixin
 from ..utils import check_array, check_random_state
-from ..utils.extmath import randomized_svd, lobpcg_svd, safe_sparse_dot, svd_flip
+from ..utils.extmath import randomized_svd, lobpcg_svd,\
+                            safe_sparse_dot, svd_flip
 from ..utils.sparsefuncs import mean_variance_axis
 
 __all__ = ["TruncatedSVD"]
@@ -33,8 +34,8 @@ class TruncatedSVD(BaseEstimator, TransformerMixin):
     context, it is known as latent semantic analysis (LSA).
 
     This estimator supports 3 algorithms: a fast randomized SVD solver, and
-    a "naive" algorithm that uses ARPACK or LOBPCG as an eigensolver on (X * X.T) or
-    (X.T * X), whichever is more efficient.
+    a "naive" algorithm that uses ARPACK or LOBPCG as an eigensolver 
+    on the normal matrix (X * X.T) or (X.T * X), whichever is more efficient.
 
     Read more in the :ref:`User Guide <LSA>`.
 
@@ -48,11 +49,13 @@ class TruncatedSVD(BaseEstimator, TransformerMixin):
 
     algorithm : string, default = "randomized"
         SVD solver to use. Either "arpack" for the ARPACK wrapper in SciPy
-        (scipy.sparse.linalg.svds), or "lobpcg" for LOBPCG (scipy.sparse.linalg.lobpcg),
-        or "randomized" for the randomized algorithm due to Halko (2009).
+        (scipy.sparse.linalg.svds), or
+        "lobpcg" for LOBPCG (scipy.sparse.linalg.lobpcg) Knyazev (2001), or 
+        "randomized" for the randomized algorithm due to Halko (2009).
 
     n_iter : int, optional (default 5)
-        Number of iterations for randomized or LOBPCG SVD solver. Not used by ARPACK.
+        Number of iterations for randomized or LOBPCG SVD solver.
+        Not used by ARPACK.
         The default is larger than the default in `randomized_svd` to handle
         sparse matrices that may have large slowly decaying spectrum.
 
diff --git a/sklearn/utils/extmath.py b/sklearn/utils/extmath.py
@@ -25,7 +25,8 @@
 from ..externals.six.moves import xrange
 from .sparsefuncs_fast import csr_row_norms
 from .validation import check_array
-from scipy.sparse.linalg import lobpcg
+from . import lobpcg
+# from scipy.sparse.linalg import lobpcg
 
 
 @deprecated("sklearn.utils.extmath.norm was deprecated in version 0.19 "
@@ -381,9 +382,9 @@ def randomized_svd(M, n_components, n_oversamples=10, n_iter='auto',
 
 
 def lobpcg_svd(M, n_components, n_oversamples=10, n_iter='auto',
-               transpose='auto', 
+               transpose='auto',
                flip_sign=True, random_state=0):
-    """Computes a truncated SVD using LOBPCG mimicking the randomized SVD setup
+    """Computes a truncated SVD using LOBPCG mimicking the randomized SVD
 
     Parameters
     ----------
@@ -475,12 +476,12 @@ def lobpcg_svd(M, n_components, n_oversamples=10, n_iter='auto',
         Q = Q.astype(M.dtype, copy=False)
 
     A = - safe_sparse_dot(M, M.T)
-    # 1. LOBPCG default option largest=True is currently broken, so we 
+    # 1. LOBPCG default option largest=True is currently broken, so we
     # go the smallest (negative) of the negative normal matrix A
     # 2. In contrast to randomised, lobpcg allows setting up useful
-    # tol=lobpcg_tol but adding it below currently results in 
+    # tol=lobpcg_tol but adding it below currently results in
     # Docstring Error:
-    # sklearn.cluster.spectral.discretize arg mismatch. 
+    # sklearn.cluster.spectral.discretize arg mismatch.
     _, Q = lobpcg(A, Q, maxiter=n_iter, largest=False)
 
     # project M to the (k + p) dimensional space using the basis vectors
diff --git a/sklearn/utils/lobpcg.py b/sklearn/utils/lobpcg.py
