On Saturday, 9 February 2013 at 01:25:58 UTC, John Colvin wrote:
This is a fair point. Being able to obtain the kth permutation
of a large set would indeed be useful, even if iteration is not
feasible. For example, you might want to examine the k=2^n
perturbations, as you have some a priori knowledge that they
contain the solution you're looking for.
In that case we'd want to be able to index with an effectively
arbitrary size index. I don't have any experience with bigint
but I presume it's the correct tool for the job.
BigInt would seem the easiest to implement as things presently
stand, as it should only require you to modify the type(s) for
the index while the remainder is the same.
Using cent would need require the 128bit type implemented and
the array one would take a lot of work and maybe change the whole
algorithm to try and compensate for it.
Regardless I'd like to see cent implemented. Hmmm although
there's a predetermined definition for the 128bit type, was there
for a 256bit type? Although that might be getting ahead of what
we may need them for.
