This is now ticket #33245 <https://trac.sagemath.org/ticket/33245>
El domingo, 30 de enero de 2022 a las 13:38:38 UTC+1, emanuel.c...@gmail.com escribió: > FWIW, after rebuilding Sage with system’s giac : > > sage: integrate(e^x/(x^2+1), x, -pi, pi) > 1/2*(imag_part(Ei(pi + I))*tan(1/2)^2 - imag_part(Ei(pi - I))*tan(1/2)^2 + > 2*real_part(Ei(pi + I))*tan(1/2) + 2*real_part(Ei(pi - I))*tan(1/2) - > imag_part(Ei(pi + I)) + imag_part(Ei(pi - I)))/(tan(1/2)^2 + 1) - > 1/2*(imag_part(Ei(-pi + I))*tan(1/2)^2 - imag_part(Ei(-pi - I))*tan(1/2)^2 + > 2*real_part(Ei(-pi + I))*tan(1/2) + 2*real_part(Ei(-pi - I))*tan(1/2) - > imag_part(Ei(-pi + I)) + imag_part(Ei(-pi - I)))/(tan(1/2)^2 + 1) > sage: giac.version() > "giac 1.7.0, (c) B. Parisse and R. De Graeve, Institut Fourier, Universite de > Grenoble I" > > But now, ptestlong exhibits new bugs > <https://groups.google.com/g/sage-release/c/aOpjpfOXgro/m/03WJYuNTFwAJ>. > Sigh (again)… > > HTH, > > Le dimanche 30 janvier 2022 à 07:25:48 UTC+1, Emmanuel Charpentier a > écrit : > >> To add insult to injury, trying to re-enable system’s giac (make >> giac-clean && configure failed. Somehow,./configurefinds no *suitable* >> systemgiacversion, and re-builds Sage's giac. It seems that I should >> also cleanpari` and consorts and re-re-build… >> >> Sigh… >> >> Le vendredi 28 janvier 2022 à 23:10:23 UTC+1, dmo...@deductivepress.ca a >> écrit : >> >>> I confirm the problem so please open a ticket. >>> >>> I tried 9.5rc4 on MacOS 11.5.2, I tried 9.4 and 9.5rc4 on Ubuntu 20.04 >>> (CoCalc), and I tried 9.5rc3 on a 32-bit Debian virtual machine. Maybe >>> something would have happened eventually, but none of them gave an answer >>> within 20 minutes. The giac version seems to be 1.6.0 in all cases. >>> >>> On Friday, January 28, 2022 at 1:23:08 PM UTC-7 emanuel.c...@gmail.com >>> wrote: >>> >>>> Can’t reproduce on Sage 9.5.rc1 running in Debian testing. First thing >>>> after session start : >>>> >>>> sage: integrate(e^x/(x^2+1), x, -pi, pi) >>>> // Giac share root-directory:/usr/share/giac/ >>>> // Giac share root-directory:/usr/share/giac/ >>>> Added 0 synonyms >>>> 1/2*(imag_part(Ei(pi + I))*tan(1/2)^2 - imag_part(Ei(pi - I))*tan(1/2)^2 + >>>> 2*real_part(Ei(pi + I))*tan(1/2) + 2*real_part(Ei(pi - I))*tan(1/2) - >>>> imag_part(Ei(pi + I)) + imag_part(Ei(pi - I)))/(tan(1/2)^2 + 1) - >>>> 1/2*(imag_part(Ei(-pi + I))*tan(1/2)^2 - imag_part(Ei(-pi - I))*tan(1/2)^2 >>>> + 2*real_part(Ei(-pi + I))*tan(1/2) + 2*real_part(Ei(-pi - I))*tan(1/2) - >>>> imag_part(Ei(-pi + I)) + imag_part(Ei(-pi - I)))/(tan(1/2)^2 + 1) >>>> >>>> BTW: >>>> >>>> sage: integrate(e^x/(x^2+1), x).simplify_full() >>>> 1/2*(2*I*cos(1/2)^2 + 2*cos(1/2)*sin(1/2) - I)*Ei(x + I) + >>>> 1/2*(-2*I*cos(1/2)^2 + 2*cos(1/2)*sin(1/2) + I)*Ei(x - I) >>>> sage: integrate(e^x/(x^2+1), x, algorithm="fricas").simplify_full() >>>> 1/2*(-I*Ei(x - I)*e^(2*I) + I*Ei(x + I))*e^(-I) >>>> >>>> And these expressions are equal : >>>> >>>> sage: integrate(e^x/(x^2+1), x).trig_reduce().exponentialize().factor() >>>> -1/2*I*(Ei(x - I)*e^(2*I) - Ei(x + I))*e^(-I) >>>> sage: integrate(e^x/(x^2+1), x, algorithm="fricas") >>>> -1/2*I*(Ei(x - I)*e^(2*I) - Ei(x + I))*e^(-I) >>>> >>>> Mathematica gives the same (mathematical) result (but this can’t be >>>> backtranslated to Sage now, for lack of ExpIntegralEi in the >>>> translation dictionary) : >>>> >>>> sage: mathematica.Integrate(e^x/(x^2+1), x) >>>> ((-I/2)*(E^(2*I)*ExpIntegralEi[-I + x] - ExpIntegralEi[I + x]))/E^I >>>> >>>> HTH, >>>> >>>> Le vendredi 28 janvier 2022 à 11:56:14 UTC+1, Emmanuel Briand a écrit : >>>> >>>>> Using Sage 9.4 installed with conda on Mac OS X 10.13.6, I observe an >>>>> erratic behaviour of "integrate" for a very specific integral. >>>>> >>>>> The following command runs forever: >>>>> >>>>> sage: integrate(e^x/(x^2+1), x, -pi, pi) >>>>> >>>>> (The problem does not show up with another interval of integration, >>>>> e.g. 0, pi or -pi, 0 instead of -pi, pi). >>>>> >>>>> After interrupting, the following message is shown: >>>>> >>>>> ^CInterrupting Giac... >>>>> >>>>> integrate(e^x/(x^2 + 1), x, -pi, pi) >>>>> >>>>> Running a second time the command returns the question, which is ok. >>>>> >>>>> sage: integrate(e^x/(x^2+1), x, -pi, pi) >>>>> integrate(e^x/(x^2 + 1), x, -pi, pi) >>>>> >>>>> But running a third time (or fourth o more) the command returns a >>>>> wrong result: >>>>> >>>>> sage: integrate(e^x/(x^2+1), x, -pi, pi) >>>>> -pi*e^x/(x^2 + 1) - x*e^x/(x^2 + 1) >>>>> >>>>> Actually, after that, no integral of th same function can be >>>>> calculated: >>>>> >>>>> sage: integrate(e^x/(x^2+1), x,0, 1) >>>>> -x*e^x/(x^2 + 1) >>>>> >>>>> Intgrls of other functions are ok. >>>>> >>>>> >>>>> >>>>> The problem does not show up when avoiding giac: >>>>> >>>>> sage: integrate(e^x/(x^2+1), x, -pi, pi, algorithm='maxima') >>>>> integrate(e^x/(x^2 + 1), x, -pi, pi) >>>>> >>>>> Asking for the giac version gives no clue: >>>>> >>>>> sage: giac.version() >>>>> "Done" >>>>> >>>>> Should I open a ticket for this? >>>>> >>>>> Emmanuel Briand >>>>> >>>>> >>>>> >>>>> -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/a870341f-e35b-4c60-b8b9-1a29a68bc630n%40googlegroups.com.
