Skip to content

alas, no vecnorm

3 messages · Faheem Mitha, Douglas Bates, Brian Ripley

Faheem Mitha
# 
I wanted a function that would give the euclidean distance of a vector. 

Then I was happy, because I found vecnorm listed on pg 55 of V&R (3rd
edn) which I had just bought today.

Then I was sad, because R did not have vecnorm.

Then I was happy again, because I bethought myself that I could copy the
function vecnorm from splus to my code.

Then I was sad again because R complained
Error in .Fortran("d2norm", as.integer(length(x)), as.double(x), value =
double(1)) : 
	C/Fortran function name not in load table

So, after this tempest of emotion, the question still remains: does R have
this function under another name, or was it simply overlooked?

                              Best regards, Faheem Mitha.


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Douglas Bates
# 
Faheem Mitha <faheem at email.unc.edu> writes:
Check in src/appl/blas.f and you will find

*
*  dnrm2() returns the euclidean norm of a vector via the function
*  name, so that
*
*     dnrm2 := sqrt( x'*x )
*
*
*  -- This version written on 25-October-1982.
*     Modified on 14-October-1993 to inline the call to DLASSQ.
*     Sven Hammarling, Nag Ltd.
*
*
      double precision function dnrm2 ( n, x, incx )

Because this is a Fortran function and not a subroutine you can't call
it directly from R.  However, it would be an informative exercise to
take the manual "Writing R Extensions" and decide how to write a
Fortran or C wrapper function that can be called from R.

In C, for the .Call interface it could be written like 

#include <R.h>
#include <Rdefines.h>

extern double F77_NAME(dnrm2)(int *, double *, int *);

SEXP dnorm2(SEXP x)
{
    int one = 1, n;
    SEXP val;

    PROTECT(x = AS_NUMERIC(x)); n = LENGTH(x);
    PROTECT(val = NEW_NUMERIC(1));
    NUMERIC_POINTER(val)[0] =
        F77_CALL(dnrm2)(&n, NUMERIC_POINTER(x), &one);
    UNPROTECT(2);
    return(val);
}

In fact, for Intel machines and any others that use 80 bit floating
point registers, all the dancing around with scaling that is done in
dnrm2 is unnecessary.  The brute force approach of accumulating the
sum of the squares of the entries and taking the square root would
work perfectly fine.  However, since dnrm2 is available, we might as
well use it.
.Call("dnorm2", x)

(Warning - I didn't test that code although I did compile it.)
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Brian Ripley
#
On Thu, 4 May 2000, Faheem Mitha wrote:

            

      
      
      
    

        
Funny, it is on page 55 of the _second_ edition, and not in the third.
It is the third edition that is R-aware.  Have you really just bought a
second edition?

            

      
      
      
    

        
Well, it was not overlooked.  R is not an S-PLUS clone, it is similar to S,
and AFAIK vecnorm is an S-PLUS extension.

vecnorm <- function(x, p=2) sum(x^p)^(1/p)

will do almost all of what vecnorm does, if not as fast as the internal
function, but do you need the speed?  I at least thought vecnorm was not a
priority for R.