R help - Sep 2011 - problem with function "Truncate" in package "distr"

Hello all,

Can someone tell me why the following mixture of two log-normal
distributions does not get truncated? What puzzles me is that the
function works almost always, but for certain combinations (like the
one below), it does not.

# R code example
library(distr)
mix<-UnivarMixingDistribution(Lnorm(3.2,0.5),Lnorm(5.4,0.6),mixCoeff=c(0.3,0.7))
mix.trunc<-Truncate(mix,lower=0.001,upper=3000)
distr.sample<-r(mix.trunc)(1000000)
range(distr.sample)

Why do I get values over 3000 (which was the defoned upper limit)?

Some help would be greatly appreciated.

Duarte Viana

On 21.09.2011 11:53, Duarte Viana wrote:
> Hello all,
>
> Can someone tell me why the following mixture of two log-normal
> distributions does not get truncated? What puzzles me is that the
> function works almost always, but for certain combinations (like the
> one below), it does not.
>
> # R code example
> library(distr)
>
mix<-UnivarMixingDistribution(Lnorm(3.2,0.5),Lnorm(5.4,0.6),mixCoeff=c(0.3,0.7))
> mix.trunc<-Truncate(mix,lower=0.001,upper=3000)
> distr.sample<-r(mix.trunc)(1000000)
> range(distr.sample)
>
> Why do I get values over 3000 (which was the defoned upper limit)?
>
> Some help would be greatly appreciated.

Some question for the author of package distr, I believe.

Uwe Ligges


> Duarte Viana
>
> ______________________________________________
> 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.

Am 21.09.2011 17:05, schrieb Uwe Ligges:
> 
> 
> On 21.09.2011 11:53, Duarte Viana wrote:
>> Hello all,
>>
>> Can someone tell me why the following mixture of two log-normal
>> distributions does not get truncated? What puzzles me is that the
>> function works almost always, but for certain combinations (like the
>> one below), it does not.
>>
>> # R code example
>> library(distr)
>>
mix<-UnivarMixingDistribution(Lnorm(3.2,0.5),Lnorm(5.4,0.6),mixCoeff=c(0.3,0.7))
>>
>> mix.trunc<-Truncate(mix,lower=0.001,upper=3000)
>> distr.sample<-r(mix.trunc)(1000000)
>> range(distr.sample)
>>
>> Why do I get values over 3000 (which was the defoned upper limit)?
>>
>> Some help would be greatly appreciated.
> 
> 
> Some question for the author of package distr, I believe.
> 
> Uwe Ligges
> 
seconded, Uwe, and settled off-line
[BTW was an issue of the globally set accuracy].

Best, Peter (one of the authors of distr)