It really dependes on the nature of the computations that you are doing to
obtain these real vectors, and also on the objective of your problem. For
example, if the real vectors are the result of some numerical approximation,
then you can set the threshold to be equal to the error involved in the
approaximation, if you know that. If these thresholds are too high to be
appropriate for your objective, then you can increase the accuracy of your
numerical approximation and obtain a smaller threshold.
Ravi.
____________________________________________________________________
Ravi Varadhan, Ph.D.
Assistant Professor,
Division of Geriatric Medicine and Gerontology
School of Medicine
Johns Hopkins University
Ph. (410) 502-2619
email: rvaradhan at jhmi.edu
----- Original Message -----
From: Hesen Peng <hesen.peng at gmail.com>
Date: Thursday, October 2, 2008 9:55 pm
Subject: [R] When to set small values to 0?
To: R help forum <r-help at r-project.org>
> My dear R buddies,
>
> I've run into a problem when doing numerical computation recently. In
> a program that I've been working on, I usually get a vector of real
> values which are theoretically (and it's correct) supposed to decrease
> until reaching zero after a given value. However, most of the value
> just wander at 10^-8 scale and never shrink exactly to zero. So I
> guess I should manually add a threshold to set the values to zero. But
> does anyone know any choice of these thresholds? Thanks very much.
>
> --
> ??? Hesen Peng
>
> ______________________________________________
> R-help at r-project.org mailing list
>
> PLEASE do read the posting guide
> and provide commented, minimal, self-contained, reproducible code.