Skip to content

data.frame within a function (PR#9294) (cont'd)

2 messages · Riyan Cheng, Brian Ripley

Original
Riyan Cheng
# 
This continues the message "data.frame within a function (PR#9294)" that 
was posted on 2006/10/12. Duncan Murdoch kindly replied. I'm using the 
current version R 2.4.0, but the same issue exists. Just copy and paste 
the following code under R, and compare the output of f1() and f2() and 
the output of f3() and f4(). Does anybody have any idea? Thanks.

###################################################
# R code for demonstration only #
###########################

rmvnorm<- function (n, mean = rep(0, nrow(sigma)), sigma = 
diag(length(mean))){
  if (nrow(sigma) != ncol(sigma)) {
    stop("sigma must be a square matrix")
  }
  if (length(mean) != nrow(sigma)) {
    stop("mean and sigma have non-conforming size")
  }
  ev <- eigen(sigma, sym = TRUE)$values
  if (!all(ev >= -sqrt(.Machine$double.eps) * abs(ev[1])))
    warning("sigma is numerically not positive definite")
  sigsvd <- svd(sigma)
  retval <- t(sigsvd$v %*% (t(sigsvd$u) * sqrt(sigsvd$d)))
  retval <- matrix(rnorm(n * ncol(sigma)), nrow = n) %*% retval
  retval <- sweep(retval, 2, mean, "+")
  retval
}

f<- function(obj){
     update(obj,~ .+x,evaluate=T); cat("also ok\n")
}

#########################
# compare f1() and f2() #
#########################
f1<- function(){
     x<- rnorm(10)
     y<- rmvnorm(10,mean=c(1,2)); colnames(y)<- paste("y",1:2,sep="") 
#c("y1","y2")
     dtf<- data.frame(y,x)
     lm1<- lm(cbind(y1,y2)~1,data=dtf); cat("ok\n")
    
     update(lm1,~ .+x,evaluate=T); cat("also ok\n") # only this line is 
different
}
f2<- function(){
     x<- rnorm(10)
     y<- rmvnorm(10,mean=c(1,2)); colnames(y)<- paste("y",1:2,sep="") 
#c("y1","y2")
     dtf<- data.frame(y,x)
     lm1<- lm(cbind(y1,y2)~1,data=dtf); cat("ok\n")
    
     f(lm1) # only this line is different
}

f1()
f2()

#########################
# compare f3() and f4() #
#########################
f3<- function(){
     x<- rnorm(10)
     y<- rmvnorm(10,mean=c(1,2)); colnames(y)<- paste("y",1:2,sep="") 
#c("y1","y2")
     lm1<- lm(cbind(y1,y2)~1,data=data.frame(y,x)); cat("ok\n")
    
     update(lm1,~ .+x,evaluate=T); cat("also ok\n") # only this line is 
different
}
f4<- function(){
     x<- rnorm(10)
     y<- rmvnorm(10,mean=c(1,2)); colnames(y)<- paste("y",1:2,sep="") 
#c("y1","y2")
     lm1<- lm(cbind(y1,y2)~1,data=data.frame(y,x)); cat("ok\n")
    
     f(lm1) # only this line is different
}

f3()
f4()

#########
# the end #
#########
Brian Ripley
# 
This is not to do with your subject line, and not a bug (and PR#9294 as 
been closed).  You don't even say what you think the 'issue' is: it seems 
to be your lack of understanding of the scope rules.

I get
ok
Error in inherits(x, "data.frame") : object "dtf" not found

which is quite correct: dtf is not visible from the body of f() where you 
call update(evaluate=T) (the default, and use TRUE please).  This is why 
update has an 'evaluate' argument, so you can manage the scope used.
On Thu, 30 Nov 2006, Riyan Cheng wrote: