R help - Aug 2004 - Problem (bug?) with vector indices

Hi all,

I found a problem on R that looks like a bug to me. I am working with R
for windows verson 1.9.0, but the problem also happens on the linux
version 1.9.0. I hope that the problem is on my reasoning, and I would be
grateful if someone could explain me why this happens.
There goes the example:
> x<-seq(-2,2,0.001)
> y<-rep(0,length(x))
> for (i in x){y[1000*i+2001]<-i^2}
> plot(x,y)    # Here I found something wrong
> which(y==0)
 [1]  244  353  357  361  365  369  482  486  490  494  579  583  587  591
 595
[16]  599  603  607  611  615  619  708  712  716  720  724  728  732  736
 740
[31]  744  837  841  845  849  853  857  861  865  869  962  966  970  974
 978
[46]  982  986  990  994  998 2001
> x[244]
[1] -1.757
> i<-x[244]
> 1000*i+2001
[1] 244
> y[1000*i+2001]
[1] 3.087049
> y[244]
[1] 0
> y[243]
[1] 3.087049


What means that the expression 1000*i+2001, which is equal to 244, inside
the brackets is evaluated as 243! The same occurs with every value of x
that results in a zero value of y, the right value is stored in y one
position before it should be, overwriting the previous value at this
position.
The bug is on R or on my mind?

Thanks a lot.
Rodrigo Drummond

____________________________________________
Rodrigo D. Drummond
Laboratorio Genoma Funcional
Centro de Biologia Molecular e Eng. Genetica
Universidade Estadual de Campinas
Caixa Postal 6010
13083-875 - Campinas - SP - Brasil
Tel: xx-19-3788-1119 Fax: xx-19-3788-1089

"Rodrigo Drummond" <rduarte at unicamp.br> writes:
> Hi all,
> 
> I found a problem on R that looks like a bug to me. I am working with R
> for windows verson 1.9.0, but the problem also happens on the linux
> version 1.9.0. I hope that the problem is on my reasoning, and I would be
> grateful if someone could explain me why this happens.
> There goes the example:
> 
> > x<-seq(-2,2,0.001)
> > y<-rep(0,length(x))
> > for (i in x){y[1000*i+2001]<-i^2}
...
> > x[244]
> [1] -1.757
> > i<-x[244]
> > 1000*i+2001
> [1] 244
> > y[1000*i+2001]
> [1] 3.087049
> > y[244]
> [1] 0
> > y[243]
> [1] 3.087049
> 
> 
> What means that the expression 1000*i+2001, which is equal to 244, inside
> the brackets is evaluated as 243! The same occurs with every value of x
> that results in a zero value of y, the right value is stored in y one
> position before it should be, overwriting the previous value at this
> position.
> The bug is on R or on my mind?
The latter: 

Have a look at (1000*x[244]+2001)-244

Beware floating point arithmetic!

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907