[NumPy-Tickets] [NumPy] #1742: cholesky error ?

[NumPy Trac]

#1742: cholesky error ?
---------------------------+------------------------------------------------
  Reporter:  pcrs          |       Owner:  pv    
      Type:  defect        |      Status:  closed
  Priority:  normal        |   Milestone:  2.0.0 
 Component:  numpy.linalg  |     Version:  1.5.1 
Resolution:  invalid       |    Keywords:        
---------------------------+------------------------------------------------
Changes (by pv):

  * status:  new => closed
  * resolution:  => invalid


Old description:

> from numpy import *
> from numpy.linalg import cholesky
> a=array([[1.0, 0.0],[0.0, 1.0]])
> k=0.3;
> b=dot(a,array([[cos(k), sin(k)],[-sin(k), cos(k)]]))
> cholesky(b)
> array([[ 0.97741316,  0.        ],
>        [-0.30234932,  0.92947371]])
> dot(c,c.transpose())
> array([[ 0.95533649, -0.29552021],
>        [-0.29552021,  0.95533649]])
> b
> array([[ 0.95533649,  0.29552021],
>        [-0.29552021,  0.95533649]])
> Isn't dot(c,c.transpose()) not supposed to be equal to b? the sign of
> b[0,1] is sign inverted

New description:

 {{{
 from numpy import *
 from numpy.linalg import cholesky
 a=array([[1.0, 0.0],[0.0, 1.0]])
 k=0.3;
 b=dot(a,array([[cos(k), sin(k)],[-sin(k), cos(k)]]))
 cholesky(b)
 array([[ 0.97741316,  0.        ],
        [-0.30234932,  0.92947371]])
 dot(c,c.transpose())
 array([[ 0.95533649, -0.29552021],
        [-0.29552021,  0.95533649]])
 b
 array([[ 0.95533649,  0.29552021],
        [-0.29552021,  0.95533649]])
 }}}
 Isn't dot(c,c.transpose()) not supposed to be equal to b? the sign of
 b[0,1] is sign inverted

--

Comment:

 Cholesky decomposition is defined only for hermitian, positive-definite
 matrices.
 Your matrix `b` is not hermitian.

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/1742#comment:1>
NumPy <http://projects.scipy.org/numpy>
My example project