CODEWITHSUNDEEP
julia
Maciek Leks
Maciek Leks

Reputation: 1448

Column of the max for each row in a matrix

Consider the matrix:

M = [1.0 2.0 3.0; 5.0 4.0 3.0; 1.0 100.0 12.0]

I want to get column of the max value in each row. So it should be:

col = [3; 1; 2]

, since

M[1,3] -> 3.0;
M[2,1] -> 5.0;
M[3,2] -> 100.00;

In Octave it is easy to achieve:

[max, col] = max(M,[],2)

, where col=[3;2;1].

In Julia I can find only findmax function which returns an absolute index of the max element for each row. So it would be:

max, aindx = findmax(M,2)

, where aindx=[7,2,6]

M[7] = 3.0; M[2] = 5.0; M[6] = 100;

Where to find Julia equivelent for Octave max(M,[],2)?

My current workaround:

max, aindx = findmax(M, 2);
msize=size(M);
col = zeros(msize[1], 1);
for i=1:msize[1] 
  _, col[i] = ind2sub(msize,aindx[i]);
end

Upvotes: 4

Views: 2684

Answers (1)

tholy
tholy

Reputation: 12179

Julia's findmax is more flexible than Octave's max: you can find the maximum over multiple dimensions at once. As a consequence, it returns a linear index.

As you've noted, you can use ind2sub to compute whatever index(es) you want. If you use this a lot, you might want to define your "workaround" as a function to make it convenient to use. You can put that function in your .juliarc.jl if you want to make sure it's always available.

Upvotes: 6

Related Questions