R help - Oct 2004 - Beginners problem

Hi,

I'm new to R and have a problem with a little test program (see below). 
Why doesn't <<- in function rk4
assign the new value to y so that it is seen in rktest. I thought that 
<<- does exactly this. But it seems that I
didn't get it right. I would be very appreciative for an explanation of 
that behaviour of <<-. I know how to
write the whole thing so that it works (return the updated y to rktest) 
but I would like to learn how variable scope
in R works.

Thanks in advance

Rolf

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

rk4 <- function(x,y,f,h) {
  n  <- length(y)
  hh <- h*0.5
  k1 <- f(x,y)
  k2 <- f(x,y+k1*hh)
  k3 <- f(x,y+k2*hh)
  k4 <- f(x,y+k3*h)
  y <<- y + h/6.0*(k1 + 2.0*k2 + 2.0*k3 + k4))
}

rkf <- function(x,y) {
  -y
}

rktest <- function(){
  y <- 1.0
  x <- 0.0
  h <- 0.1
  for(i in 1:10) {
    rk4(x,y,rkf,h)
    print(y)
  }
}

rktest()

[Please try to make sure the code you posted contains no syntax error.]

Try:
> rktest <- function(){
+     rk4 <- function(x,y,f,h) {
+         n  <- length(y)
+         hh <- h*0.5
+         k1 <- f(x,y)
+         k2 <- f(x,y+k1*hh)
+         k3 <- f(x,y+k2*hh)
+         k4 <- f(x,y+k3*h)
+         y <<- y + h/6.0*(k1 + 2.0*k2 + 2.0*k3 + k4)
+     }
+     y <- 1.0
+     x <- 0.0
+     h <- 0.1
+     for(i in 1:10) {
+         rk4(x,y,rkf,h)
+         print(y)
+     }
+ }
> 
> rktest()
[1] 0.9048375
[1] 0.8187309
[1] 0.7408184
[1] 0.6703203
[1] 0.6065309
[1] 0.5488119
[1] 0.4965856
[1] 0.4493293
[1] 0.40657
[1] 0.3678798

You expected `y' in rktest() to be modified by the super-assignment, which
is incorrect.  In your version, you'll find the super-assignment creating
`y' in the global environment, rather than modifying the copy in rktest().
The reason is that the order variables are looked up in rk4() is:

1. within rk4()
2. where rk4() is defined
3. if 2 above is a function, where that function is defined
...
eventually: the global environment

BTW, I'm not sure if you really meant to do 

   y <<- y + ...

because the `y' on the RHS will be the local copy (if exists).

Andy
> From: Rolf Wester
> 
> Hi,
> 
> I'm new to R and have a problem with a little test program 
> (see below). 
> Why doesn't <<- in function rk4
> assign the new value to y so that it is seen in rktest. I 
> thought that 
> <<- does exactly this. But it seems that I
> didn't get it right. I would be very appreciative for an 
> explanation of 
> that behaviour of <<-. I know how to
> write the whole thing so that it works (return the updated y 
> to rktest) 
> but I would like to learn how variable scope
> in R works.
> 
> Thanks in advance
> 
> Rolf
> 
> --------------------------------------------------------------
> --------------------------------------------------------
> 
> rk4 <- function(x,y,f,h) {
>   n  <- length(y)
>   hh <- h*0.5
>   k1 <- f(x,y)
>   k2 <- f(x,y+k1*hh)
>   k3 <- f(x,y+k2*hh)
>   k4 <- f(x,y+k3*h)
>   y <<- y + h/6.0*(k1 + 2.0*k2 + 2.0*k3 + k4))
> }
> 
> rkf <- function(x,y) {
>   -y
> }
> 
> rktest <- function(){
>   y <- 1.0
>   x <- 0.0
>   h <- 0.1
>   for(i in 1:10) {
>     rk4(x,y,rkf,h)
>     print(y)
>   }
> }
> 
> rktest()
> 
> ______________________________________________
> 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
> 
>

Rolf Wester <rolf.wester at ilt.fraunhofer.de> writes:
> Hi,
> 
> I'm new to R and have a problem with a little test program (see
> below). Why doesn't <<- in function rk4
> assign the new value to y so that it is seen in rktest. I thought that
> <<- does exactly this. But it seems that I
> didn't get it right. I would be very appreciative for an explanation
> of that behaviour of <<-. I know how to
> write the whole thing so that it works (return the updated y to
> rktest) but I would like to learn how variable scope
> in R works.
R has lexical scope and rk4() is not lexically scoped in rktest(). I
believe that there's a rather thorough discussion of these matters in
the FAQ.
> rk4 <- function(x,y,f,h) {
....
>   y <<- y + h/6.0*(k1 + 2.0*k2 + 2.0*k3 + k4))
> }
> 
> rkf <- function(x,y) {
>   -y
> }
> 
> rktest <- function(){
>   y <- 1.0
>   x <- 0.0
>   h <- 0.1
>   for(i in 1:10) {
>     rk4(x,y,rkf,h)
>     print(y)
>   }
> }
> 
> rktest()

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907

Rolf Wester wrote:
> Hi,
> 
> I'm new to R and have a problem with a little test program (see below).
> Why doesn't <<- in function rk4
The reason is lexical scoping (see FAQ), and I suggest to implement rk4 
as function with a return value (see # !!!). BTW there are already a rk4 
(and lsoda) functions in the odesolve package.

Thomas P.

Your example formally corrected:

rk4 <- function(x,y,f,h) {
  n  <- length(y)
  hh <- h*0.5
  k1 <- f(x,y)
  k2 <- f(x,y+k1*hh)
  k3 <- f(x,y+k2*hh)
  k4 <- f(x,y+k3*h)
  y + h/6.0*(k1 + 2.0*k2 + 2.0*k3 + k4) # !!!
}

rkf <- function(x,y) {
  -y
}

rktest <- function(){
  y <- 1.0
  x <- 0.0
  h <- 0.1
  for(i in 1:10) {
    y <- rk4(x,y,rkf,h) # !!!
    print(y)
  }
}

rktest()