Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: fyi, new build of CAS integration tests started. Mathematica11.2,Maple 17.2, Rubi 4.13
Posted:
Oct 3, 2017 4:40 PM


Interesting spectrum. Would your test suite, your test harness, give them automatically
different rating, or do you need to say, manually, which one is the best? Which one
would you choose?
Nasser M. Abbasi schrieb: > On 10/2/2017 6:39 PM, bursejan@gmail.com wrote: >> Hi, Nasser M. Abbasi, do you have: >> >> integ_0^(1/4) sqrt(xx^2) dx (*) >> >> as a test case? (something with complex >> numbers might go on, from looking at a >> wolfram alpha indefinite solution) >> > > Hello; > > I did wild card search on Rubi tests cases, and could not > find int( sqrt(xx^2) ). > > The closest is Timofeev problem 56 which is > 1/Sqrt[x  x^2] > > But Wolfram alpha gives same result as Mathematica 11.2 for this, which is > > In[4]:= Integrate[Sqrt[x  x^2], x] > Out[4]= (Sqrt[((1 + x))*x]*(Sqrt[1 + x]*Sqrt[x]*(1 + 2*x) >  Log[Sqrt[1 + x] + Sqrt[x]]))/(4*Sqrt[1 + x]*Sqrt[x]) > > Maple gives> > int(sqrt(xx^2),x); > > (1/4)*(2*x+1)*(x^2+x)^(1/2)+(1/8)*arcsin(2*x1) > > And Rubi gives > > Int[Sqrt[x  x^2], x] > Out[15]= (1/4) (12 x) Sqrt[xx^2]1/8 ArcSin[12 x] > > And Fricas gives > > int(sqrt(xx^2),x) > (1/4)*arctan(sqrt(x^2+x)/x) + (2*x1) * sqrt(x^2+x) > > Nasser > > >> (*) >> Issac Newton in 16651666 >> ?I am ashamed to tell you to how many figures I >> carried these computations, having not other >> business at the time.? >> http://www.math.tamu.edu/~dallen/masters/alg_numtheory/pi.pdf >> >> Am Freitag, 29. September 2017 18:12:44 UTC+2 schrieb Nasser M. Abbasi: >>> On 9/29/2017 11:05 AM, clicliclic@freenet.de wrote: >>> >>>>> If you mean run sympy on Rubi test cases? >>>> >>>> No, I had in mind the new rulebased integration module produced by two >>>> of this years's GSoC students. Their SymPy pull request is at: >>>> >>>> <https://github.com/sympy/sympy/pull/12978> >>>> >>>> Hopefully, this module would cut massively the number of timeouts that >>>> make the tests so difficult to run in the original SymPy integrator. >>>> But >>>> I suppose your HW will leave little room for such experiments. >>>> >>>> Martin. >>>> >>> >>> Sorry I misuderstood you. I never tried this new rubisympy package, >>> but will look at it. If I figure how to use it, will try to add >>> it to the test build. >>> >>> Best, >>> Nasser >> >



