vectorizing a function

["Leeds, Mark" <mleeds@mlp.com>]

i sent out a question

about vectorizing a function

a few days ago. the solution

that was posted ( using pmin instead of min ) 

is appreciated and

gets close to what i want but it doesn't

quite work because of the

recursive nature of what

i want to do. the cumprod part

of the solution just accumulates the product

blindly regardless of the condition

so that what makes it not quite work.

 

if anyone knows the filter function

well enough to use it to avoid the loop,

or any other way, the knowledge would be

appreciated.

 

the original function is deceptively recursive

and may not be as simple as it

looks when trying to avoid looping.

 

 

                                    thanks

 

                                    mark

 

 

the looping version of the function

and the almost solution

are given below for people ( like me )

who deleted the original email.

 

--------------------------------------------------------------

 

originalfunction<-function(invector)

 

outvector<-invector

 

for ( i in 2:length(invector) ) {

 

if  ( invector[i] < 1 ) {

 

outvector[i]_invector[i]*outvector[i-1]

 

}

 

 

return(outvector)

 

-------------------------------------------------------------------

 

almostworks<-function(invector) {

 

outvector<-invector

part<-cumprod(pmin(invector,1))

outvector[invector < 1]<-part[invector < 1]

 

return(outvector)

 

-------------------------------------------------------------------