QS question
Currently in the quadratic sieve we sieve until we have a number of relations slightly above the number of primes in the factorbase. In nfs, however, we try postprocessing at regular intervals hoping that enough of the primes aren't used. Would it make sense to try some method like the one we use in nfs for qs? Do we have any knowledge of how many primes are not used in an average matrix from qs?
