Seperating a vector based on another vector

I have two arrays A & B.vector A has to be separated & processed based on B's vector value.For the length that B's value remains same,A has to be taken till that value & processed.

Example:-

A = 1:10;
B = [5 5 5 5 5 10 10 10 10 10];
%so B's first 5 values are same.so A's first 5 values should be assigned to a different array on which i'll do processing.
%Again B's next 5 values are same,so A's next 5 values are to be assigned to that another array on which i can process.
%My approach:-
C = unique(B);
for i = 1:length(C)
for j = 1:length(A)
if A(j) == unique(i)
D(j) = A(j) 
else
D(j) = [ ]
end
end
% do processing on D
end

Can this be done in any better way ?

[Guillaume]

I assume the line

if A(j) == unique(i)

is meant to read

if B(j) == C(i)

Otherwise the code makes no sense.

[Luffy]

yes,that's right

Answer 1

[Guillaume]

For a start your inner j loop is completely unnecessary, use logical vector indexing:

C = unique(B);
for idx = 1:numel(C)  %avoid using i, it's a matlab function
   D{idx} = A(B == C(idx)); %B == C(idx) is a logical vector, used to index A
end

Another way of achieving the same result is to use accumarray:

[~, ~, locations] = unique(B);
D = accumarray(locations, A, [], @(v) {v});

or arrayfun:

D = arrayfun(@(c) A(B == c), unique(B), 'UniformOutput', false);