#19811: Upgrade arb to version 2.8.1
-------------------------------------+-------------------------------------
Reporter: jdemeyer | Owner:
Type: enhancement | Status: needs_work
Priority: major | Milestone: sage-7.0
Component: packages: | Resolution:
standard | Merged in:
Keywords: | Reviewers: Fredrik Johansson
Authors: Jeroen Demeyer | Work issues:
Report Upstream: N/A | Commit:
Branch: | d13d6a8deb50f723bacb0a6900925e6977211707
u/jdemeyer/upgrade_arb | Stopgaps:
Dependencies: |
-------------------------------------+-------------------------------------
Changes (by fredrik.johansson):
* status: positive_review => needs_work
Comment:
Well, there are four precision-related doctest failures when rebased on
the latest develop branch.
Two tests got better.
One test - rising factorial - got slightly worse. This is expected as I
changed the algorithm to use a gamma function quotient when n is large,
which gives a speedup but slightly worse error bounds.
One test - a hypergeometric function - got a lot worse (all digits are
lost). You need some 110 bits of precision now to get a similar result.
The reason is that Arb uses some heuristics when choosing where to
truncate asymptotic series, and these heuristics don't fully take the
parameters into account. With large parameters (in this case a = 100) some
precision is lost when z is not larger by orders of magnitude (in this
case z = 1000+1000i). The old code in Arb had a bug that made it choose
far too many terms in some cases, fixed in 2.8. This happened to result in
a very precise answer for this test case, purely by accident. The current
code in Arb is more correct, but errs on the side of stopping early, which
is worse for this particular test case. I think I see a reasonable way to
fix this in Arb, but that will be in the next version.
--
Ticket URL: <http://trac.sagemath.org/ticket/19811#comment:12>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
