Albert Rich schrieb:   
   >   
   > On Friday, March 24, 2017 at 11:52:26 PM UTC-10, oldk...@gmail.com wrote:   
   > > On Saturday, March 25, 2017 at 5:25:59 PM UTC+8, Albert Rich wrote:   
   > > > On Friday, March 24, 2017 at 6:03:19 PM UTC-10, oldk...@gmail.com wrote:   
   > > > > Off topic a little, I find it interesting that Rubi fails Timofeev 493,   
   > > > >   
   > > > >      int(x^2/(x*cos(x)-sin(x))^2,x)  --> (cos(x) + x*sin(x))/(x*cos(x)   
   - sin(x))   
   > > > >   
   > > > > A pure transcendental function, and easy to solve by heuristic   
   > > > > methods.   
   > > >   
   > > > I'm a little slow.  What is the heuristic method used to integrate it?   
   > > >   
   > >   
   > > What I meant is:   
   > >   
   > > This integral can be solved if transcendental case of Risch algorithm   
   > > is implemented.   
   > >   
   > > Otherwise, a CAS system might try to solve it using heuristic method,   
   > > aka integration by parts, guess, etc.  For this integral, as Timofeev said,   
   > > it equals to int( x/sin(x) d(1/(x*cos(x)-sin(x)))), then integration by   
   part,   
   > > result is x/sin(x)/(x*cos(x)-sin(x)) + int(1/sin(x)^2,x) .   
   >   
   > The best I could come up with using integration by parts as you   
   > suggested was the following formula:   
   >   
   > int(x^m/(cos(x)^(m-2)*(cos(x)+x*sin(x))^2),x) -->   
   >   
   >     -x^(m-1)/(cos(x)^(m-1)*(cos(x)+x*sin(x))) +   
   >     (m-1)*int(x^(m-2)/cos(x)^m,x)   
   >   
   > If m>1 is an integer, x^(m-2)/cos(x)^m is integrable in terms of the   
   > polylogarithm function.  The Timofeev Chapter 7, problem 11 on page   
   > 344 is the special case when m=2.   
   >   
   > If anyone can derive a more general formula, please let me and Rubi   
   > know.   
   >   
      
   With respect to Timofeev 7.11, you have interchanged SIN(x) and COS(x),   
   but replacing x by x - pi/2 doesn't give Timofeev's case for m=2 back.   
      
   So this will become two new integration rules: One for the even   
   denominator COS(x) + x*SIN(x) and one for the odd denominator x*COS(x) -   
   SIN(x).   
      
   Presumably x should be replaced by a*x in the final Rubi rules.   
      
   Martin.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)