[Scipy-tickets] [SciPy] #1218: Very poor relative accuracy in discrete survival functions

SciPy Trac scipy-tickets@scipy....
Sat Jul 3 16:25:54 CDT 2010


#1218: Very poor relative accuracy in discrete survival functions
-------------------------+--------------------------------------------------
 Reporter:  dsimcha      |       Owner:  somebody
     Type:  defect       |      Status:  new     
 Priority:  normal       |   Milestone:  0.8.0   
Component:  scipy.stats  |     Version:  0.7.0   
 Keywords:               |  
-------------------------+--------------------------------------------------
Changes (by josefpktd):

  * component:  Other => scipy.stats


Comment:

 do we need this precision ?
 1e-10 and 1e-66 are in my opinion essentially zero.

 I think the current generic cdf function only works with finite lower
 bound in the support.

 For bounded support we could improve the tail precision of the sf by
 starting at the upper boundary.

 cdf and a sf would need a rewrite to work correctly with open support on
 the real line. In that case, we need a stop criterion for desired
 precision, which would mean we have to decide on the required precision.

 (A similar question of a stop criterion is currently in the generic
 integration in the moment calculation, but maybe less serious because the
 integration is over the entire (effective) range. And it's the same
 problem for the new expect function. )

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1218#comment:1>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list