CODEWITHSUNDEEP
vectorizationoctave
smcmi
smcmi

Reputation: 98

Exponent vectorization in Octave

I'm attempting to take a general array A, raise it to the (element-wise) power of each member of a vector p, and sum the result, preferably in a vector operation, such that the result is the same size as A. Ideally, arrays of any size/dimension for A should be allowed. For example, if

A = [0 1 ; 2 3]
p = [2 3]

I want the result A.^p(1) + A.^p(2), which is [0 2 ; 12 36], just more elegantly, and for any size of A and length of p, and avoiding a loop.

I came up with the following, which expands into the next higher dimension of A, then sums along that dimension:

sum(repmat(A,[ones(1,ndims(A)) length(p)]) .^ repmat(reshape(p,[ones(1,ndims(A)) length(p)]),size(A)),ndims(A)+1)

which technically seems to work, but....ugh. Is there a cleaner way to do this?

Upvotes: 0

Views: 204

Answers (1)

rahnema1
rahnema1

Reputation: 15837

Assuming p to be a row vector:

result = reshape(sum(A(:) .^ p, 2), size(A));

Upvotes: 3

Related Questions