These two tests show a regression in the integrate function:
Current cvs:
integrate(-log(1-x)/x, x, 0, 1);
-> limit(li[2](%e^x)+x*log(1-%e^x)-x^2/2,x,inf)-%pi^2/6
Version 5.19.1:
integrate(-log(1-x)/x, x, 0, 1);
-> %pi^2/6
Current cvs:
integrate(-log(2-x)/(x-1), x, 0, 1);
-> -(4*%i*%pi*log(2)-%pi^2)/4+log(-1)*log(2)-%pi^2/6
rectform(%);
-> %pi^2/12
Version 5.19.1:
integrate(-log(2-x)/(x-1),x,0,1);
-> %pi^2/12
Andrej
On Sat, Mar 20, 2010 at 2:58 PM, Dieter Kaiser <drdieterkaiser at web.de> wrote:
> Revision 1.72 of defint.lisp causes two new errors in the
> share_testsuite:
>
> Running tests in rtest_simplify_sum:
>
> ********************** Problem 52 ***************
> Input:
> test_sum(sum(harmonic_number(n)/n/(1+n),n,1,inf),[],false,[],%pi^2/6)
>
>
> Result:
> (3*('limit(2*li[2](%e^?g1467538)+2*?g1467538*log(1-%
> e^?g1467538)-?g1467538^2,
> ? ? ? ? ? ?g1467538,inf))
> ?-2*%pi^2)
> ?/6
>
> This differed from the expected result:
> 0
>
> ********************** Problem 54 ***************
> Input:
> ev(test_sum(sum(harmonic_number(n)/n/2^n,n,1,inf),[],false,[],%pi^2/12),
> ? sum_by_integral_transforms = [logcontract])
>
>
> Result:
> (4*log(-1)*log(2)+4*li[2](2)-%pi^2)/4
>
> This differed from the expected result:
> 0
>
> 58/60 tests passed
>
> Dieter Kaiser
>
>
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>