At the end, apply sum_array(permute(n)). Or you could use actual random numbers using this same technique, but there maybe repeated values.

lucky shot

well done guillaume, thanks for trying random numbers as well. great effort!

well done alfonso, thanks for trying random numbers as well. im learning new techniques here as well.

true randomness is a bit involved to test (particularly without the stats toolbox), plus the sum constrain adds some additional complexity to the randomness of the resulting values (see for example this Cody problem: http://www.mathworks.com/matlabcentral/cody/problems/331-compute-area-from-fixed-sum-cumulative-probability). In any way, the simplest solution to discourage this sort of extremely-non-random solutions would probably be running your current testsuite twice (with identical inputs) and then checking that the returned vectors are different...

Another possibility, would be to test that there is no repetition in the returned values, which can easily be tested with: assert(all(diff(sort(sum_array))));

you're right guillaume, i apppreciate that.

well done! :) quick solution. however i was expecting coder to generate random values each epoch :)

I've now provided a solution that does generate random values.

i've changed the test suit following ur approach, thanks.