[18099] in Athena Bugs
Your OLC question about "maple"
daemon@ATHENA.MIT.EDU (tarvizo@MIT.EDU)
Thu Aug 3 14:13:10 2000
From: tarvizo@MIT.EDU
Date: Thu, 3 Aug 2000 14:13:02 -0400
Message-Id: <200008031813.OAA17154@dewey.mit.edu>
To: bugs@MIT.EDU, 3partysw@MIT.EDU
Reply-to: olc@MIT.EDU
--------
Hello,
This is in response to a maple problem that one user had reported,
which our records say you have been previously notfied about.
(I've appended the original problem at the end anyway.)
Tresi Arvizo
Athena Consultant
*** Reply from user sahai@ZOOEY.MIT.EDU [0].
[Sat 29-Jul-00 10:33pm]
As far as the Maple bug is concerned, I posted it to USENET
and a helpful person pointed me to:
http://www.math.ubc.ca/~israel/advisor
Where this particular bug is documented and a little
workaround given. The workaround is to express the
integral so that it only has one fractional power in
it. I suggest that Athena place a link to this
particular page on its Maple page so that other users
can know where to check for possible bugs/workarounds
before they get confused.
*****
Topic: maple
Question:
Buggy???
This bug has been replicated in both Linux and on Sun systems. I've tried both Maple 6 and 5.1. Same bug
both systems.
The symbolic integration routine gives an inconsistent answer to:
plot([evalf(Int(1/(t^(5/2)*(s-t)^(5/2)),t = 1 .. (s-1))),
evalf(int(1/(t^(5/2)*(s-t)^(5/2)),t = 1 .. (s-1))),
(evalf(Int(1/(t^(5/2)*(s-t)^(5/2)),t = 1 .. (s-1))))
/
(evalf(int(1/(t^(5/2)*(s-t)^(5/2)),t = 1 .. (s-1))))] ,
s=2 .. 10);
The two plots should be identical, and the ratio should be 1 for all s.
But instead, things look different. There is no warning message or
any other external indication that something is wrong.
Please contact me if there is any known workaround.
-Anant Sahai
sahai@mit.edu
