[Numpy-discussion] Behaviour of vdot(array2d, array1d)

denis denis-bz-py@t-online...
Sun Jan 10 10:34:27 CST 2010


On 07/01/2010 18:51, Nikolas Tezak wrote:

> However when I do this, vdot raises a ValueError complaining that the
> "vectors have different lengths".

Nikolas,
   looks like a bug, in numpy 1.4 on mac ppc too. Use dot instead --

import numpy as np

x = 1j * np.ones(( 2, 3 ))
y = np.ones( 3 )
try:
     print "vdot:", np.vdot( x, y )
except ValueError, e:
     print "ValueError:", e
print "dot:", np.dot( x.conj(), y )


numpy core/numeric.py has
     from _dotblas import dot, vdot
but I don't know how to use testing/... for blas,
nor how to log a ticket -- experts please advise ?

cheers
   -- denis



More information about the NumPy-Discussion mailing list