rvgtest - Testing Non-Uniform Random Variate Generator

Non-uniform random variate generators are of fundamental importance in Monte-Carlo methods and stochastic simulation. They are based on the assumption that a source of uniformly distributed random numbers is available that produces real and truely random numbers. In practice, however, we have to use pseudo-random numbers which are generated by means of floating point numbers. Thus there are deviations from the theoretical framework caused by round-off errors which are due to the limitations of floating point arithmetic. In addition algorithm have to developed and implemented. As human can err coding errors or mistakes in the design of algorithm may cause defects in a generated random sample. It is of paramount importance to detect such errors using statistical tools and categorize which tests are sensitive to which errors. rvgtest is a collection of routines for testing non-uniform random variate generators. They hopefully should help to detect all kinds of error. Thus it provides routines for goodness-of-fit tests that are based on histograms and routines to estimate approximation errors for numerical inversion.