Max of vector in higher dimensional array subject to constraints

Hello all,

I currently have code that's working but inefficient due to nested for loops, and I would like to vectorize it as possible. My data set is a higher dimensional array (a,b,.....z), and I'd like to find the maximum of each vector of z subject to a constraint based on the values of the other vectors.

For example, take a 3 dimensional array (a,b,c). I will have calculated a matrix (a x b-sized) of a constraint. I'd like to find a (a x b-sized) matrix that gives the maximum of each c vector that is measured only on the domain of (constraint:end). Let me know if there is a way to do this without for loops.

(edited for clarity)

Answer 1

[Matt J]

N=ndims(A); %A is the given matrix

z=size(A,N);  

idx=reshape(1:z, [ones(1,N-1),z] ) < constraint; %requires R2016b or higher

B=A;
 B(idx)=-inf;

result=max(B,[],N);

[James Lee]

Hey Matt,

This worked. I had to modify it a bit because I realized I had made a small mistake in the prompt (the constraint was an index to the matrix as opposed to a value) but this approach helped a lot. I appreciate it!

[Matt J]

My solution did assume that constraint is a matrix of indices. So, I'm not sure why you had to change anything. Glad you got what you need, though.

Answer 2

[darova]

Use max()

B = max(A,[],3);