MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: elliptic integrals

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg636] Re: [mg602] elliptic integrals
  • From: Richard Mercer <richard at seuss.math.wright.edu>
  • Date: Mon, 03 Apr 1995 14:11:37 -0400

>  first, load package <<Calculus`EllipticIntegrate`
>  

>  Then:  Integrate[1/Sqrt[(a + b x + x^2)(a - b x + x^2)],
>  x]
>  

>  I get an expression that includes the symbol
>  "Calculus`EllipticIntegrate`Private`a$24".  What is this?

The technical term is "bug". 


>  What do I do with it?  Things get even more complicated
>  when I put some limits on my integral (as I will need to
>  do).
>  


Open your package Calculus`EllipticIntegrate` and search for

{a, c} = -Map

There are two such locations. At the end of each such line (but before the  
semicolon) type 


//Union

and save the package. This seems to fix the problem. The problem was the code  
was written under the assumption that exactly two roots would result from  
solving certain equations. In your example each root was listed twice, which is  
fixed by the //Union. An even better fix might be 


Take[#,2]& @ -Map .... //Union

to guarantee more than two roots will not screw up the calculation.

Richard Mercer


  • Prev by Date: How to construct pure expressions
  • Next by Date: Re: SuperPrimes
  • Previous by thread: elliptic integrals
  • Next by thread: transparency in Mma graphics