[R] Finding local minima and maxima

[Gabor Grothendieck]

1. See ?turnpoints in package pastecs.

2. do it yourself with rapply in the zoo package:

   library(zoo)
   f <- function(x) x[2] > max(x[-2])
   rapply(zoo(rank(x, ties = "first")), 3, function(x) f(x) - f(-x))

   Note that if a run is an extrema then the last member of a
   run of maxima and the first member of a run of minima will
   be identified as the extremum.

3. There are also a number of other solutions on r-help you
   could find if you look through the result of the R command:
 
   RSiteSearch("peak")

On 6/15/05, Sean Davis <sdavis2 at mail.nih.gov> wrote:
> Thanks, Uwe.
> 
> Actually, I didn't mean to make this complicated--simple vector x and
> values at x in a vector y.  Just wanted to determine the local minima
> and maxima of y[x], but y is not continuous (but is a "smooth" function
> of x).  I will probably use hist or density to do it and then look at
> these curves or cuts.
> 
> Sean
> 
> On Jun 15, 2005, at 8:16 AM, Uwe Ligges wrote:
> 
> > Sean Davis wrote:
> >
> >> I have data in the form of (x,y) pairs and would like to find local
> >> minima and maxima (typically the zeros of the 2nd derivative) of the
> >> y values.  I looked at numericDeriv, but I don't have an "expression"
> >> per se.  I looked at optim, also, but it looks like it will find only
> >> one "global" max or min.  I can code up my own piecewise derivatives,
> >> but wondered if there is an existing function to do this.
> >
> > If you don't have derivatives, and you don't have an expression, and
> > you don't have a criteria how to minimize, please tell me of any known
> > method that can handle your problem well enough.
> >
> > You could look whether y[i-1] <y[i] and y[i+1]<y[i] if y is ordered
> > according to x, if this most trivial and very non-robust method fits
> > to your definition...
> >
> > Uwe Ligges
> >
> >
> >> Thanks,
> >> Sean
> >> ______________________________________________
> >> R-help at stat.math.ethz.ch mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-help
> >> PLEASE do read the posting guide!
> >> http://www.R-project.org/posting-guide.html
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
>