Pseudo-Random-Number Generators

As part of the course work you have implemented a generator giving uniformly distributed pseudo random numbers and a generator giving normally distributed pseudo random numbers.

The first, a Combined Tausworthe generator suggested by Tezuka and L'ecuyer, is of random bit type with period length about 10¹⁸ and guaranteeing good quality of random points in boxes.

The latter generator based on an acceptance-rejection technique is suggested by Leva. It is using easily computable bounding curves to reduce the slow calculation of the exact boundary function (0.012 per random deviate). The generator calls the Tausworthe generator to generate the uniform deviates needed.
The reason why a generator for normally distributed random numbers was implemented is that normal deviates as coordiates give uniformly distributed directions in multidimensional space. Such directions are used in some stochastic optimization algorithms.

Both generators are fast and easy to implement.