adding to a matrix with

I want to add one to the matrix A for each time the respective linear index appears in the vector Ex. lin_ind =

1     1     3     5     5     7     9     9    11    13    13
A = zeros(8, 8)
I want A(1) = 2 because 1 appears twice. A(2) = 0. A(3) = 1.

[Chaowei Chen]

just comment on "I want A(1) = 2 because 1 appears twice. A(2) = 0. A(3) = 1. "

This is histogram. A=hist(lin_ind,1:64);A=reshape(A,[8 8]);

[Walter Roberson]

Good point about it being a histogram.

I did some timing, and on my system,

A=reshape(accumarray(lin_ind(:),1,[64 1]),[8,8]);

is about twice as fast as

A=reshape(hist(lin_ind,1:64),[8,8]);

when tried with the lin_ind given above. I did not try with a larger lin_ind

Answer 1

[Walter Roberson]

You create an 8 x 8 array, and then you specify the additions in terms of 1-dimension indices, and from your question you clearly want gaps of 0 where there is no linear index with the corresponding value. The maximum value of your example lin_id is 13: does that mean you would want A(13) = 2 ? If so, then should that A(13) correspond to A(2,5), the 13th element of your 8 x 8 array? If so, then:

A = zeros(8,8);

t = accumarray(lin_ind(:),1);
A(1:length(t)) = A(1:length(t)) + t;

[Andrei Bobrov]

Hi Walter!

variant

[I,J] = ind2sub([8 8],lin_ind)

A = accumarray([I',J'] ,1,[8 8])

[ily]

Thanks, both are great solutions. I actually have the positions specified in row and column subscripts and I converted it to linear indices, so andrei's solution allows me to skip that step. I didn't know of the accumarray function before.