integrate



 > Hello
 > 
 > With maxima 5.6, the following command :
 > 
 > integrate(1/cos(x)^2,x,0,%PI/4);
 > give the answer :
 > lim(x->0) tan(x+%PI/4) 
 > 
 > I would expect that maxima returns the real 1, because it is a very 
 > classical function to integrate.
 > 
 > Best regards
 > 
 > ****************************
 > Michel Gosse 
 > michel.gosse@ac-poitiers.fr
 > michel.gosse@freesbee.fr
 > ****************************

There used to be a more severe bug in maxima-5.6 which appeared in the
above integral with slightly different limits:

(C9)  integrate((sec(t))^2,t,-%pi/4,0);

                                      %PI
(D9)                                  ---
                                       2   

This Bill Schelter fixed (see note below) and now the Maxima gives the
correct answer: 1

I don't understand his fix but those who do may get an idea from it.

Sincerely,
Moses

-----------------------------------------------------------------  

From: Bill Schelter <wfs at mail>
To: glasner@math.psu.edu
Subject: Re: bug in maxima
Date: Sun, 25 Mar 2001 01:11:05 -0600

Thank you for the report
   (C9)  integrate((sec(t))^2,t,-%pi/4,0);
 incorrect.

I have made the following change in the maxima source file sin.lisp
to correct the bug you reported.

(in-package "MAXIMA")

(DEFUN MONSTERTRIG (EXP VAR TRIGARG)
  (if (not (atom trigarg)) (return-from monstertrig (rischint exp
var)))
  (PROG (NOTSAME W A B Y D)