Interesting idea for encoding small numbers efficiently.
I'm trying to think of a way to use a set of prime numbers to assist in efficiently representing a series of arbitrary integers.
My first thoughts about this involve defining a set of n primes P in conjuction with the set of n values V such that Pi is the smallest prime greater than Vi for all i from 1..n. Basically I was thinking of using a set of primes as a sort of basis vector. But now I'm not sure I can make this idea work.
Anyway, the idea is to encode several small values, each of which may take from 2-8 bits to represent in less than a byte or two if the total bits of all the values are less than 12 (bonus points for using only a byte of the total bits are less than 7) and less than 3 bytes if the total bits are less than 20.