An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20120731/bcb249fd/attachment.pl>
protential rounding error concern
3 messages · Jie, R. Michael Weylandt, Petr Savicky
On Tue, Jul 31, 2012 at 2:45 PM, Jie <jimmycloud at gmail.com> wrote:
Dear All, I am running a r code on 32bit win, involving absolutely small numbers. Although I tried sth like the ratio of numers like 10^(-100) and did not have issue to get the correct answer, but still a little concerned about it. Could anyone give some suggestion or have any experience?
This would be a canonical reference: http://www.validlab.com/goldberg/paper.pdf. Hope this helps, M
Best wishes,
Jie
[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
On Tue, Jul 31, 2012 at 03:45:14PM -0400, Jie wrote:
Dear All, I am running a r code on 32bit win, involving absolutely small numbers. Although I tried sth like the ratio of numers like 10^(-100) and did not have issue to get the correct answer, but still a little concerned about it. Could anyone give some suggestion or have any experience?
Hi. Numbers of the order 10^(-100) are still relatively far from the lower limit and are represented with full precision, which is 53 bits. The smallest number representable in full precision is 2^-1022 = 2.225074e-308 and the smallest representable number (in limited precision) is 2^-1074 = 4.940656e-324. See http://en.wikipedia.org/wiki/Double_precision_floating-point_format for more detail. Hope this helps. Petr Savicky.