|
|
last edited 10 years ago by test1 |
1 2 | ||
Editor: test1
Time: 2014/05/06 19:03:37 GMT+0 |
||
Note: |
added:
From test1 Tue May 6 19:03:37 +0000 2014
From: test1
Date: Tue, 06 May 2014 19:03:37 +0000
Subject:
Message-ID: <20140506190337+0000@axiom-wiki.newsynthesis.org>
Name: '#417 integration failure' => '#417 slow integration'
Axiom seems to take forever on:
integrate(sqrt(1-z)*(sqrt(1-sqrt(z))+sqrt(sqrt(z)+1))*(2*sqrt(1-_ sqrt(z))*sqrt(sqrt(z)+1)*(2*z-1)-2*z-sqrt(z)+1)/z^(3/4), z=0..1, "noPole")
same for the indefinite version. (Vladimir 49)
Martin
This is quite heavy computation and in current implementation it is done in rather inefficient
way. So very long compute time is expected. After doing setSimplifyDenomsFlag(true)
FriCAS
handles the last example below reasonably fast. The first below still needs too much time for
MathAction. Also, FriCAS thinks that the first below is nonelementary,
so it is left unevaluated). The case above is more complicated.
integrate(sqrt(1 - z)*sqrt(sqrt(1 + z^2)), z)
It seems that friCAS finally hangs calling reduce from FiniteDivisor?, which in turn is called by torsionIfCan in PFO. Similar::
(1) -> setSimplifyDenomsFlag(true)
(1) |
integrate(2*z*sqrt(sqrt(2*z*z+2*sqrt(2)*z+2)+z+sqrt(2)),z)
(2) |
(Vladimir 48)
Name:#417 integration failure
=> #417 slow integration