I am writing a scientific paper and i would like to mention which of the various models of error/uncertainty propagation for functions operating on RIF parameters sage is using.
If you had a pointer to a URL or reference to literature or so, i would be thrilled.
Thanks
There's the source:
https://git.sagemath.org/sage.git/tre...
Quickly browsing the file shows RIF elements are pairs of mprf floats, one with "round up" and the other with "round down". It's just relying on mpfr doing its rounding properly.
For more advanced functionality, however, it's using a library, MPFI:
http://perso.ens-lyon.fr/nathalie.rev...
your should probably reference that library, and the documentation there hopefully has the information you're looking for.
Is that a valid way of doing error propagation? does it not consider the kind of function operating on the floats for the effect on the resulting error? --- checked that out, and yes it is.
You might also consider using Arb.
See SageMath documentation for arbitrary-precision computation.
I agree that Arb is probably under more active development, but I'd expect that MPFI is also certified. The only point of doing interval or ball arithmetic is to get certified results.
Okay, I updated my answer.
Asked: 5 years ago
Seen: 369 times
Last updated: Apr 14 '19
Copyright Sage, 2010. Some rights reserved under creative commons license. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
