[Rd] Re: convolve() design bug

Martin Maechler Martin Maechler <maechler@stat.math.ethz.ch>
Fri, 17 Nov 2000 16:35:05 +0100

>>>>> "Bill" == Bill Simpson <wsi@gcal.ac.uk> writes   --- on R-help :


  Bill> Use convolve().  Here is an example:
  Bill> #Bracewell The Fourier transform and its applications, chap 3 (p.32 in 2nd
  Bill> #edition): {2 2 3 3 4} * {1 1 2} = {2 4 9 10 13 10 8}
  Bill> x<-c(2,2,3,3,4)
  Bill> h<-c(1,1,2)
  Bill> out<-convolve(x,rev(h),type="o") #need to do rev due to fault in R code

  Bill> I pointed it out several times to the R-developers that
  Bill> convolve() should do convolution by default! However they prefer
  Bill> it to do correlation by default (which makes no sense to
  Bill> me). Anyway the workaround is to use rev(h).

Dear Bill,
as one of those not too responsive way back, here are some 
non-convincing reasons: 

  At that time, we had density rely on convolve's behavior back then,
  and the it was old legacy R code on which others might have relied, 
  etc, etc

But then, I must admit I had forgotten for a while and later never felt
urged anymore.

Can we have something like a ``survey'' ?
Does anybody heavily rely on convolve()'s current behavior ?
{teaching / packages / ...}

Bill, can you (again, I know, exuse us!) detail your suggested changes,
ideally by sending patches against the current *.R and *.Rd files ?
Thank you.

Martin Maechler <maechler@stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO D10	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch