Unexpected result calculating the determinant of a singular matrix (42S)

[ttw]

I can't reproduce the FXND today. I will post the following when I get a "useful" version of convert fraction to partial quotients (not to hard, still debating on input style: two integers or fraction). At times I need statistics for the results; I'll probably just add some post-processing, things like sum of PCs or max or alternating sum and difference.

I also have some programs that convert a list of partial quotients into the quadratic irrational which that as the repeated part. Going the other to is fun. I still have a bit of work to do as this one is slow if using fractions but very complex if dealing with numerator and denominator separately. Some care is needed as the integer part (first PC) has to be handled separately, then there's a non-repeating part. Getting it in general is a bit tedious.

I may just post a "good" version with arbitrary choices and let anyone who wants to use these modify them to suit.

I've been using integers which are larger than 2^64 which can cause some problems. However keeping everything in integers is pretty helpful.

I did find out a few funny things: IDIV2 is very slow compared to separate parts (even IQUOT and IREMAINDER), the matrix form for continued fractions is really slow too.