R devel - Jul 1998 - S speedup in R

Venables and Ripley describe a speedup where you take a structure like

x<-NULL
for(i in sequence)
	y<-c(y,function(x,i))
	
and convert it to one like

x<-numeric(its length)
for(i in sequence)
	y[i]<-function(x,i)

I tried this speedup on some simple examples and it made them twice as
fast.

But now I am hitting a snag with some real code.

This original version works:
cif<-function(x, y, tau, h=tau[2]-tau[1])
{
# conditional cross-intensity function for two point process realizations.
# estimated at tau.
# x and y are vectors of event times (sorting is not necessary).
# tau is the lag (or vector of lags) 
# h is binwidth.
# see Brillinger, Bryant, & Segundo (1976) eq 13, mhatAB(u)
n <- length(x)
cif.numerator <- NULL
for(tt in tau)
	{
	point.counter <- sum(round(rank(c(x + tt + h/2, y))[1:n], 0) -
	round(rank(c(x + tt - h/2, y))[1:n], 0))
	cif.numerator <- c(cif.numerator,point.counter)
	}
cif.numerator/(n * h)
}

But the sped-up version just returns a vector of zeros:

cif<-function(x, y, tau, h=tau[2]-tau[1])
{
# conditional cross-intensity function for two point process realizations.
# estimated at tau.
# x and y are vectors of event times (sorting is not necessary).
# tau is the lag (or vector of lags) 
# h is binwidth.
# see Brillinger, Bryant, & Segundo (1976) eq 13, mhatAB(u)
n <- length(x)
cif.numerator <- numeric(length(tau))
for(tt in tau)
	{
	cif.numerator[tt] <- sum(round(rank(c(x + tt + h/2, y))[1:n], 0) -
	round(rank(c(x + tt - h/2, y))[1:n], 0))
	}
cif.numerator/(n * h)
}

I can't see why this isn't working. Is this hitting a peculiarity of 
R? Thanks very much for any help.

Bill Simpson

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Bill Simpson <wsimpson@uwinnipeg.ca> writes:

...
> But the sped-up version just returns a vector of zeros:
> 
> cif<-function(x, y, tau, h=tau[2]-tau[1])
> {
> # conditional cross-intensity function for two point process realizations.
> # estimated at tau.
> # x and y are vectors of event times (sorting is not necessary).
> # tau is the lag (or vector of lags) 
> # h is binwidth.
> # see Brillinger, Bryant, & Segundo (1976) eq 13, mhatAB(u)
> n <- length(x)
> cif.numerator <- numeric(length(tau))
> for(tt in tau)
> 	{
> 	cif.numerator[tt] <- sum(round(rank(c(x + tt + h/2, y))[1:n], 0) -
> 	round(rank(c(x + tt - h/2, y))[1:n], 0))
> 	}
> cif.numerator/(n * h)
> }
You are using tt for an index in addition to being the value of an
element of the tau vector.  Unless tau always happens to be
1:length(tau) that won't work.

You probably want to change the "for" statement to
  for (tt in seq( along = tau ) )
 	{
 	cif.numerator[tt] <- sum(round(rank(c(x + tau[tt] + h/2, y))[1:n], 0) -
 	round(rank(c(x + tau[tt] - h/2, y))[1:n], 0))
 	}
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Thanks very much Douglas for pointing out my dumb mistakes.

Bill

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._