[Scipy-tickets] [SciPy] #1740: Large-argument Bessel functions are wrong

SciPy Trac scipy-tickets@scipy....
Tue Oct 2 16:47:55 CDT 2012


#1740: Large-argument Bessel functions are wrong
------------------------------+---------------------------------------------
 Reporter:  AndreasKloeckner  |       Owner:  somebody   
     Type:  defect            |      Status:  new        
 Priority:  normal            |   Milestone:  Unscheduled
Component:  Other             |     Version:  0.11.0     
 Keywords:                    |  
------------------------------+---------------------------------------------
 >>> import scipy.special as s
 >>> s.jv(489.5, 240)
 1.0331464177361223e-101

 But:

 >>> import mpmath as mp
 >>> mp.mp.dps = 100
 >>> mp.besselj(489.5, 240)
 mpf('6.360630292677643219422925387157638090642323742574897809454463193606633580841269279451987762707114621539e-102')

 Octave is happy, too:

 octave:1> besselj(489.5, 240)
 ans =  6.3606e-102

 Matlab gets this right, too.

 (And before you argue that these numbers are too small to be relevant
 anyhow: they're usually immediately multiplied by a factor that brings
 them back to O(1), at which point their being wrong does matter.)

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1740>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list