twacube: bug fix in swarming
Swarming is the ability to explore successors of a state in random way. Since computing an array of indexes and shuffle it is costly, we opted to a semi-random approach where the "next" index is computed using
current * primes[tid] % succ.size()
(with current an index in [0..succ.size()-1] and primes[tid] a big prime number)
This method has already been tested and provide good results for swarming.
Nonetheless, running needham.3.dve lead to incorrect results. It appears that current * primes[tid] % succ.size()
does not fit in an int (or even an unsigned). We should move to unsigned long long in order to tackle this problem. (The problem is that some state have more than 35 successors)