On:
math.sci.hiroshima-u.ac.jp/~m-mat/MT/MT2002/emt19937ar.html
The RNG author mentions a bug with the most significant bit not being utilized properly. (Probably irrelevant on our scale, but) the patch is to replace the first term of the SetRandomSeed's loop's assignment:
<<<
mt[mti] = (69069 * mt[mti-1]) & 0xffffffff;
---
mt[mti] = (1812433253UL * (mt[mti-1] ^ (mt[mti-1] >> 30)) + mti) & 0xffffffff;
>>>
math.sci.hiroshima-u.ac.jp/~m-mat/MT/MT2002/emt19937ar.html
The RNG author mentions a bug with the most significant bit not being utilized properly. (Probably irrelevant on our scale, but) the patch is to replace the first term of the SetRandomSeed's loop's assignment:
<<<
mt[mti] = (69069 * mt[mti-1]) & 0xffffffff;
---
mt[mti] = (1812433253UL * (mt[mti-1] ^ (mt[mti-1] >> 30)) + mti) & 0xffffffff;
>>>