# [R] uniroot() problem

Albyn Jones jones at reed.edu
Wed Dec 31 07:18:01 CET 2008

```One can't tell for sure without seeing the function, but I'd guess
that you have   a numerical issue.  Here is an example to reflect upon:

> f=function(x) (exp(x)-exp(50))*(exp(x)+exp(50))
> uniroot(f,c(0,100))
\$root
[1] 49.99997

\$f.root
[1] -1.640646e+39

\$iter
[1] 4

\$estim.prec
[1] 6.103516e-05

> .Machine\$double.eps^0.25/2
[1] 6.103516e-05

uniroot thinks it has converged, at least in relative terms.  Note
that the estimated precision is related to the machine epsilon, used
in the default value for "tol".  try fiddling with the tol argument.

> uniroot(f,c(0,100),tol=1/10^12)
\$root
[1] 50

\$f.root
[1] 1.337393e+31

\$iter
[1] 4

\$estim.prec
[1] 5.186962e-13

albyn

Quoting megh <megh700004 at yahoo.com>:

>
> I have a strange problem with uniroot() function. Here is the result :
>
>> uniroot(th, c(-20, 20))
> \$root
> [1] 4.216521e-05
>
> \$f.root
> [1] 16.66423
>
> \$iter
> [1] 27
>
> \$estim.prec
> [1] 6.103516e-05
>
> Pls forgive for not reproducing whole code, here my question is how "f.root"
> can be 16.66423? As it is finding root of a function, it must be near Zero.
> Am I missing something?
>
> --
> View this message in context:
> http://www.nabble.com/uniroot%28%29-problem-tp21227702p21227702.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help