Vectorize np.linalg.qr #11741
Description
QR decomposition is not vectorized, i.e. expects only 2-dimensional arrays as input, while you can pass 3-dimensional arrays (a number of matrices along 0-axis) to SVD:
Reproducing code example:
This gives error:
import numpy as np
A = np.random.normal(size=(3,2,2))
[Q, R] = np.linalg.qr(A)while SVD returns proper arrays U, S and V
[U, S, V] = np.linalg.svd(A)
Error message:
---------------------------------------------------------------------------
LinAlgError Traceback (most recent call last)
<ipython-input-237-7b6d808ca118> in <module>()
----> 1 np.linalg.qr(A)
~/lib/python3.6/site-packages/numpy/linalg/linalg.py in qr(a, mode)
858
859 a, wrap = _makearray(a)
--> 860 _assertRank2(a)
861 _assertNoEmpty2d(a)
862 m, n = a.shape
~/lib/python3.6/site-packages/numpy/linalg/linalg.py in _assertRank2(*arrays)
196 if a.ndim != 2:
197 raise LinAlgError('%d-dimensional array given. Array must be '
--> 198 'two-dimensional' % a.ndim)
199
200 def _assertRankAtLeast2(*arrays):
LinAlgError: 3-dimensional array given. Array must be two-dimensional
Numpy/Python version information:
1.15.0 3.6.6 (v3.6.6:4cf1f54eb7, Jun 26 2018, 17:02:57)
[GCC 4.2.1 (Apple Inc. build 5666) (dot 3)]