Improvements
Make the random engines compatible with the RandomGenerator API of Java 17.
All random generators can be created with new or via the RandomGenerator.of("<name>") or RandomGeneratorFactory.of("<name>") factory method.
final LCG64ShiftRandom random1 = new LCG64ShiftRandom();
final RandomGenerator random2 = RandomGenerator.of("LCG64ShiftRandom");
final RandomGenerator random3 = RandomGeneratorFactory.of("LCG64ShiftRandom").create();The following PRNGs are implemented:
KISS32Random: Implementation of an simple PRNG as proposed in Good Practice in (Pseudo) Random Number Generation for Bioinformatics Applications (JKISS32, page 3) David Jones, UCL Bioinformatics Group.KISS64Random: Implementation of an simple PRNG as proposed in Good Practice in (Pseudo) Random Number Generation for Bioinformatics Applications (JKISS64, page 10) David Jones, UCL Bioinformatics Group.LCG64ShiftRandom: This class implements a linear congruential PRNG with additional bit-shift transition. It is a port of the trng::lcg64_shift PRNG class of the TRNG library created by Heiko Bauke.MT19937_32Random: This is a 32-bit version of Mersenne Twister pseudorandom number generator.MT19937_64Random: This is a 64-bit version of Mersenne Twister pseudorandom number generator.XOR32ShiftRandom: This generator was discovered and characterized by George Marsaglia [Xorshift RNGs]. In just three XORs and three shifts (generally fast operations) it produces a full period of 232 - 1 on 32 bits. (The missing value is zero, which perpetuates itself and must be avoided.) High and low bits pass Diehard.XOR64ShiftRandom: This generator was discovered and characterized by George Marsaglia [Xorshift RNGs]. In just three XORs and three shifts (generally fast operations) it produces a full period of 264 - 1 on 64 bits. (The missing value is zero, which perpetuates itself and must be avoided.) High and low bits pass Diehard.