MathGroup Archive 1992

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

Search the Archive

Mathematica 1.2 bug in BesselI[nu,x] with nu > 20 and nu ~ 1.5x

  • To: mathgroup at yoda.physics.unc.edu
  • Subject: Mathematica 1.2 bug in BesselI[nu,x] with nu > 20 and nu ~ 1.5x
  • From: "NELSON M. BLACHMAN" <blachman at gtewd.mtv.gsc.gte.com>
  • Date: Thu, 24 Sep 1992 16:15:50 PDT

  I've just noticed that Mma 1.2 gives BesselI[44,30.] = 0. and 	
BesselI[60,40.] = 0., with even more peculiar (frighteningly large |values|)  
for indexes and arguemnts in these neighborhoods.  For larger arguments, 
Mma 1.2 apparently satisfactorily uses the asymptotic series and for indexes 
under 21 it seems OK.  It would be good to know just when it can be trusted.  .  

  In Mma 2.0 this BesselI bug has apparently been corrected, but those who 
may have used or may use Mma 1.2 programs involving numerical values of 
BesselI[nu,x] with nu > 20 should be warned not to trust the results without 
thorough checking.

   ////\\\	
  /  _ _  \			Nelson M. Blachman   	
(|---o-o---|)			Office of the Chief Engineer 	
 |    L    |			GTE Government Systems Corporation  
  \ (---) /			Mountain View, California 94039-7188  
   \\\|///			(415)-966-2247  blachman at gtewd.mtv.gsc.gte.com  
    \\|//                                       blachman%gtewd.dnet at gte.com





  • Prev by Date: Re: setting tolerance for small answers
  • Next by Date: Time Stamps in Notebooks
  • Previous by thread: Request for User Info
  • Next by thread: Time Stamps in Notebooks