
A bug in Rubi 2
Posted:
Oct 7, 2011 1:12 PM


I was looking into elementary antiderivatives for the eight algebraic integrands listed in a recent post and noticed that the following rule in Rubi 2 needs a factor 1/a in the evaluation:
Int[1/((a_.+b_.*x_^2)*Sqrt[c_+d_.*x_^2]),x_Symbol] := x/Sqrt[c+d*x^2] /; FreeQ[{a,b,c,d},x] && ZeroQ[a*db*c]
For a=c=0, the evaluation should be 1/(2*b*x*Sqrt[d*x^2]).
I suggest to replace this rule by a conversion of the exponents (1,n) into (0,n1), which produces a factor d/b, and to finish with rules for a single quadratic binomial.
Martin.

