>> X_{ijk} = \sum_{l} A_{il}*B_{jl}*C_{kl}
>
> (A[:,newaxis,newaxis]*B[newaxis,:,newaxis]*C[newaxis,newaxis,:]).sum(axis=-1)
Thanks for the quick solution and practical exercise in broadcasting!
:-) However, this creates a temporary 4-array, right? Is there a way of
avoiding this memory requireme
Mon, 20 Sep 2010 23:34:58 +0200, Hagen Fürstenau wrote:
> I don't know if I'm overlooking something obvious, but is there a
> compact way of computing the 3-array
>
> X_{ijk} = \sum_{l} A_{il}*B_{jl}*C_{kl}
>
> out of the 2-arrays A, B, and C?
(A[:,newaxis,newaxis]*B[newaxis,:,newaxis]*C[newaxis
Hi,
I don't know if I'm overlooking something obvious, but is there a
compact way of computing the 3-array
X_{ijk} = \sum_{l} A_{il}*B_{jl}*C_{kl}
out of the 2-arrays A, B, and C?
- Hagen
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
h