OK. Thanks you. Found it. In base to the code is feasible in a couple of weeks. I-ll implement a second order kahan algorithm. Third order would be overkill ( I don't think anyone is gonna use it for high precision calculus ). But it's not only sum and average, variance (stats) and any other formula with sums is affected. That's why I need time to track it. And by the way, what about using long double for intermediate steps, to improve precision. For what I know glibc or whatever uses gcc on fedora uses it, or at least exp((double)10) returns the same as exp((long double)10.
El mié, 9 dic 2020 a las 23:00, Eike Rathke (<er...@redhat.com>) escribió: > Hi, > > On Wednesday, 2020-12-09 20:25:17 +0100, Dante Doménech wrote: > > > Does anyone know where are implemented the calc formulas? > > Regina already gave pointers. Ask if you need more. > > > I'd like to check out if it uses the kahan algorithm, pairwise sum or > > something else. > > Kahan is not used, but there's > https://bugs.documentfoundation.org/show_bug.cgi?id=137679 > to implement it. I thought Roman would start on that but maybe > I misunderstood. So if you'd like to tackle that you're welcome. > > The current approach is quite simple, it just remembers one value to add > that later. > > Eike > > -- > GPG key 0x6A6CD5B765632D3A - 2265 D7F3 A7B0 95CC 3918 630B 6A6C D5B7 6563 > 2D3A >
_______________________________________________ LibreOffice mailing list LibreOffice@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/libreoffice
