CoCoA » CoCoALib

Here are some more minor points (after speaking to Anna on the phone):

• Anna was concerned about potential cost if there are many gens;
• if there are more (non-zero) gens than indets then the LTs cannot be coprime;
• when scanning through the list of LTs, if the number of unseen indets is less than the number of remaining (non-zero) gens then they cannot all be pairwise coprime.

---> moved to #1255

Having seen in the code that computing a GBasis over QQ implies creating a new ring (with coeffs in ZZ), encourages me to push for avoiding computing a GBasis if the gens are already "obviously" a GBasis (e.g. LPPs are coprime).

One could also use a heuristic: reduce the gens mod p (for several primes), and then check that all S-pairs reduce to 0 mod p (probably with a time limit). Maybe this is too complicated while also being uncertain...