[R] look for help on nlm in R

Martin Maechler maechler at stat.math.ethz.ch
Sat Jul 9 17:51:38 CEST 2005

>>>>> "MaggieZ" == Zhu, Maggie <maggie.zhu at pioneer.com>
>>>>>     on Thu, 7 Jul 2005 14:40:37 -0500 writes:

    MaggieZ> I had a hard time in learning nlm in R and appreciate any help. 

    MaggieZ> I encounted the following error message from time
    MaggieZ> to time when I tried different starting parameter
    MaggieZ> values (three parameter values in this case) in

     >>  nlm(f=SS.fun,p=c(0.1/40,0.1,2),hessian = FALSE,N.measure=object,h=20) 

     Error in f(x, ...) : only 0's may mix with negative subscripts 

    MaggieZ> Basically I know the three parameter values should all be positive numbers. So, how to select appropriate starting parameter values to prevent this kind of error message? 

Maggie, it's a bug in the 'SS.fun' function that you are asking
nlm() to minimize.  So it would be best to show us your

BTW:  "subscripts" are the `` i ''  in  x[i]
      in its the subscripts that become wrong in your SS.fun function.

Martin Maechler, ETH Zurich

More information about the R-help mailing list