(Roulette Wheel) Sorting multiple elements in an array

I'm perfoming a Roulette Wheel selection (http://www.edc.ncl.ac.uk/assets/hilite_graphics/rhjan07g02.png), and I want to compute the selection of n elements. How can I do this avoiding the use of loops?

For example, I have the following prob vector:

prob = [0.1 0.3 0.4 0.15 0.05];

The selection of a single element=0.2 would be:

cumprob = cumsum(prob);
selected = find(element<=prob,1,'first')

selected = 2

But, what about computing the selection of n elements? The intuitive and slow way would be:

cumprob = cumsum(prob);
for id = 1:1:n 
   selected(id) = find(element(id)<=prob,1,'first');
end

Is there any way to implement this avoiding the use of the for loop?

Thanks in advance.

Method 1: use discretize (requires Matlab version R2015a or newer)

Method 2: use arrayfun which is way slower.

Test code:

n = 5e6;
element = rand(n,1);
prob = [0.1 0.3 0.4 0.15 0.05];

cumprob = cumsum(prob);

tic
selected1 = zeros(n,1);
for id = 1:1:n 
   selected1(id) = find(element(id)<=cumprob,1,'first');
end
toc

tic
selected2 = discretize(element,[0,cumprob]);
toc
isequal(selected1, selected2)

tic
selected3 = arrayfun(@(e) sum(e>=cumprob)+1, element);
toc
isequal(selected1, selected3)

Timing (inaccurate, but works) and accuracy comparison:

Elapsed time is 5.634721 seconds.
Elapsed time is 0.059813 seconds.

ans =

     1

Elapsed time is 18.838859 seconds.

ans =

     1

>>