[Numpy-discussion] dot/inner with kwargs?

Hans Meine meine@informatik.uni-hamburg...
Wed Aug 6 10:47:51 CDT 2008


I am trying to do something like the following as efficiently as possible:
result[...,0] = (dat * cos(arange(100))).sum(axis = -1)
result[...,1] = (dat * sin(arange(100))).sum(axis = -1)

Thus, I was looking for dot / inner with an 'output' arg.

Q1: What is the difference between dot and inner (apart from the fact that dot 
does not seem to have docs)?

In http://www.scipy.org/Numpy_Example_List_With_Doc#inner, I read that:
> Like the generic NumPy equivalent the product sum is over
> the last dimension of a and b.
(BTW: "innerproduct" should be "inner")

Q2: What is "the generic NumPy equivalent" here?

Ciao, /  /
    /  / ANS

More information about the Numpy-discussion mailing list