MathGroup Archive: December 2007 [00317]

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

Re: Hi presision problem

To: mathgroup at smc.vnet.net
Subject: [mg84116] Re: [mg84098] Hi presision problem
From: Daniel Lichtblau <danl at wolfram.com>
Date: Mon, 10 Dec 2007 20:41:35 -0500 (EST)
References: <200712100938.EAA16396@smc.vnet.net>

ArtÅ«ras Acus wrote:
> Dear Group,
> 
> How to calculate the expression with higher than machine precision?
> Mathematica 6 seems runs into serious problems (gives Inderetrminate)
> 
> N[1 - Pi^(-InverseGammaRegularized[1, 0, 1 - 1/Sqrt[E]]) , 20]
> 
> 
> Sincerely, Arturas Acus

This is a bug which came about due to subtle changes in N[] during 
development of version 5. It will be fixed for a future release. At 
present a workaround would be to Do the numerical evaluation on an 
argument of InverseGammaRegularized, e.g.

In[2]:= InputForm[1 - Pi^(-InverseGammaRegularized[1, 0, 1 - 
1/Sqrt[N[E,20]]])]

Out[2]//InputForm= 
0.43581041645224371305192054846675547818`19.857529934737872

Daniel Lichtblau
Wolfram Research

References:
- Hi presision problem
  - From: ArtÅras Acus <acus@itpa.lt>

Prev by Date: Re: Adding description to plots

Next by Date: Mathematical expectation

Previous by thread: Hi presision problem

Next by thread: Re: Hi presision problem