I don’t think your program can be used for this; it only works for small integers.

It seems unfair to call this a problem with JavaFan's solution. As you point out, the bottleneck is mathematical—there are so many partitions—rather than programmatic, so any program that does what the poster seems to want will be slow on large numbers. Maybe the poster only needs a few partitions, or knows something about the data-sets, so that some optimisations can be applied?