How to vectorize this code with logical idexing

Hi all,

I have the following code, how can I avoid the for to make it faster?

a=[10:5:50];
b=[10:20:50];
c=rand(size(a));

for i=length(b)-1
   d=c(a>=b(i) & a<b(i+1));
end

thanks

cheers

[Jos (10584)]

Since you're only storing the last values of d in each iteration, you can skip the for-loop completely:

i = length(b)-1
d = c(a>=b(i)) & a<b(i+1))

but I am pretty sure you intend to do something else.

What should happen with the values of d obtained in the iterations 1 to length(b)-2?

Answer 1

[Andrei Bobrov]

a=[10:5:50];
b=[10:20:50];
c=rand(size(a));

[~,ii] = histc(a,b);
d = accumarray(ii(:),c(:),size(b(:)),@(x){x})

[pietro]

Hi Andrei,

thanks. I tried with my code but the empty bins at the end of the array are skipped. Here an axample:

a=[10:5:20, 55:5:90];
b=[10:20:170];
c=rand(size(a));

[er,ii] = histc(a,b);
d = accumarray(ii(:),c(:),[],@(x){x});

d should be a 8 elements array instead it is 4 elements array. How could I do to take into account the empty bins at the end of the array?

[Andrei Bobrov]

corrected

[pietro]

It works. Thanks