[Maxima-bugs] [ maxima-Bugs-1487703 ] integrate((sqrt(x^4-6*x^2+1)-x^2+1)/(2*x), x) fails

[SourceForge.net <no...@so...>]

Bugs item #1487703, was opened at 2006-05-12 17:36
Message generated for change (Settings changed) made by robert_dodier
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104933&aid=1487703&group_id=4933

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
>Category: Lisp Core - Integration
Group: None
>Status: Closed
>Resolution: Fixed
Priority: 5
Submitted By: David Billinghurst (billingd)
Assigned to: Nobody/Anonymous (nobody)
Summary: integrate((sqrt(x^4-6*x^2+1)-x^2+1)/(2*x),x) fails 

Initial Comment:
With CVS maxima (2006-05-13)

e:(sqrt(x^4-6*x^2+1)-x^2+1)/(2*x);
integrate(e,x);

*** - Lisp stack overflow. RESET


The change is recent - since Feb 06. With maxima 5.9.3

(%i2) integrate(e,x);
                   2                   2
              log(x  + 2 x - 1)   log(x  - 2 x - 1)
              ----------------- + -----------------
          /           2                   
2                        2
          [ %
e                                                    x
          I --------------------------------------- 
dx + log(x) - --
          ]                    
x                                  2
          /
(%o2)     --------------------------------------------
--------------
                                      2

When fixed, re-enable DE 105 in 
share/contrib/diffequations/tests/rtestode_murphy_2_2.
mac


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

>Comment By: Robert Dodier (robert_dodier)
Date: 2006-06-04 23:18

Message:
Logged In: YES 
user_id=501686

Testing with a recent build (built 2006/05/?? with sbcl
0.9.9) I find that all 7 examples (the one in the initial
report and the other 6 in the followups) yield the same
result as with 5.9.3 official release.

Closing this report as fixed. Feel free to reopen it if
appropriate.

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

Comment By: Raymond Toy (rtoy)
Date: 2006-05-14 11:33

Message:
Logged In: YES 
user_id=28849

I tried the checked-in version of sin.lisp.  I think the 6
test integrals now produce the same answers now as in 5.9.3.


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

Comment By: David Billinghurst (billingd)
Date: 2006-05-13 18:21

Message:
Logged In: YES 
user_id=365569

I tried this  (diff attached).  There isn't any change to 
the testsuite but there are a number of changes to the 
ode_contrib testsuite.  I have reduced the first few to 
find the changes in integrate() results.  There are both 
good and bad changes.

integrate(2*(1-tan(x)^2)*csc(2*x)*(csc(2*x)+cot(2*x)),x);
integrate(1/(u*((sqrt(u^2+1)*abs(x)+u*x)/(u*x)-1)),u);
integrate(1/(u*((sqrt(1-u^2)*abs(x)+u*x)/(u*x)-1)),u);
integrate(-(x*sqrt(y^2+x^2)+y)/(x*sqrt(y^2+x^2)),x);
integrate(1/(u*((a*sqrt(u^2+b)*abs(x)+u*x)/(u*x)-1)),u);
integrate((a*x^3-x*sqrt(a^2*x^4-b))/sqrt(a^2*x^4-b),x);

I haven't attached the results - it looks ugly - but can 
do if required.


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

Comment By: Raymond Toy (rtoy)
Date: 2006-05-12 19:55

Message:
Logged In: YES 
user_id=28849

There are two causes for this, I think.  First, intform no
longer sets $radexand to $all because that causes sqrt(x^2)
to be x, which is not always right.

Second, the line cond test

    ((not (alike1 exp (setq y ($expand exp))))

near the end of integrator is the actual source of the loop.

I don't really understand why it wants to do this, but
commenting out test and body allows this integral to work as
it used to.  Plus there are no failures in the testsuite. 
Additionally, some commented out integration tests no longer
loop either.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=104933&aid=1487703&group_id=4933