[Numpy-discussion] [ANN] NumPy 0.9.6 released

Sven Schreiber svetosch at gmx.net
Wed Mar 15 07:04:05 EST 2006


mfmorss at aep.com schrieb:
>> matrix([[1, 0, 0],
>>       [0, 1, 0],
>>       [0, 0, 1]])
>>
>>
>>>>> n.linalg.inverse(a)
>>>>>
>>>>>
>> array([[ 1.,  0.,  0.],
>>       [ 0.,  1.,  0.],
>>       [ 0.,  0.,  1.]])
>>
>> I was asked to report if functions in linalg still don't honor the
>> matrix input type; voila.
>>
> 
> As someone who is just starting to experiment with numpy, may I ask, what
> in general is the expected result of
> numpy.linalg.inverse(integer_matrix)?  Since not all such inverses are
> integer matrices, what does it mean to "honor the matrix input type?"  Is
> there a matrix analogue of Python's integer scalar division?
> 

Sorry for this jargon, I was just referring to "matrix" in, "array" out,
which should not happen anymore for the linear algebra functions.
-sven




More information about the NumPy-Discussion mailing list