I am trying to use the Kuiper and Watson U^2 tests available in the
CircStats or circular packages to test for uniformity on a circle.
However, the implementation of these tests gives ranges of p-values that
are not dependent on n, the size of sample.
Are there any methods (including/excluding implementation) for calculating
the p-values (with respect to n) of these tests with arbitrary precision?
The closest thing that I found was an asymptotic approximation for the
Kuiper p-value implemented in MatLab.