[scilab-Users] Re: intg: results differ substantially from those from Wolfram Alpha, which are correct?

[Bassam Awad]

Any one knows how to unsubscribe to the list?



--- On Fri, 5/20/11, Jean-Yves Baudais <jean-yves.baudais at insa-rennes.fr> wrote:

From: Jean-Yves Baudais <jean-yves.baudais at insa-rennes.fr>
Subject: Re: [scilab-Users] Re: intg: results differ substantially from those from Wolfram Alpha, which are correct?
To: users at lists.scilab.org
Date: Friday, May 20, 2011, 1:05 PM

Hello,

> ok, for normal distribution I found a cure - folded normal dist (
> http://en.wikipedia.org/wiki/Folded_normal_distribution) that avoids the
> abs() in the integral; but what about other cases?

  But in the folded normal dist there is no log function as in your function. Your problem is the same as

--> function y=f(x),y=exp(-x), endfunction
--> intg(0,1e10,f)
ans=
0

The integral calculation is performed with Monte-Carlo method (I think), so the probability to obtain non zero contribution, when the upper limit increases, decreases. With lower upper limit the result is good

--> intg(0,1e2,f)
ans=
1

When lim(x->0) f(x)=0, you must pay attention to the upper limit. It is your case.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/users/attachments/20110520/7766447f/attachment.htm>