R help - Jan 2006 - numericDeriv() giving a vector when multiple variables input

R Help List --

I have defined two time-series-vector-valued-functions, let them be f and g,
and want to find the numeric derivative of f with respect to the variable x
where f depends on x through g:
(d/dx)(f (g(x) )

Moreover, x is a vector

I tried this out the long way (naming every element of the x vector and then
making the 'theta' argument in numericDeriv() the character vector of
all
these names) and the result is just one time series vector; I was hoping for
a matrix. Also weirdly, if I instead make theta equal to just one of the
named elements of x, I get the same time series vector; the same happens for
any subset of the named elements

My call to numericDeriv looks like this (vphi acts as x,
swz.kalman.vectoracts as g,
decision.ts.vector acts as f):

***
numericDeriv(
expr = decision.ts.vector(
  a = swz.kalman.vector(
    zeta1=zeta[1], u = Phi$u, phi = Phi$Phi,
    vphi
c(varphi1,varphi2,varphi3,varphi4,varphi5,varphi6,varphi7,varphi8,varphi9,varphi10,

varphi11,varphi12,varphi13,varphi14,varphi15,varphi16,varphi17,varphi18,varphi19,

varphi20,varphi21,varphi22,varphi23,varphi24,varphi25,varphi26,varphi27,varphi28,

varphi29,varphi30,varphi31,varphi32,varphi33,varphi34,varphi35,varphi36,varphi37,
    varphi38,varphi39,varphi40,varphi41,varphi42),
    alpha.prior = specs$alpha.prior
    ),
  phi = Phi$Phi, lambda = specs$lambda, delta = specs$delta, pi.star
specs$pi.star, u.2star = specs$u.2star
  ),
theta
c("varphi1","varphi2","varphi3","varphi4","varphi5","varphi6","varphi7","varphi8","varphi9","varphi10",

"varphi11","varphi12","varphi13","varphi14","varphi15","varphi16","varphi17","varphi18","varphi19",

"varphi20","varphi21","varphi22","varphi23","varphi24","varphi25","varphi26","varphi27","varphi28",

"varphi29","varphi30","varphi31","varphi32","varphi33","varphi34","varphi35","varphi36","varphi37",
   
"varphi38","varphi39","varphi40","varphi41","varphi42")
);
***

As you can see, it includes some calls to other objects that are lying
around. Maybe from this email someone can tell me my mistake with how I've
called things; otherwise, I'm happy to send along the .R files to whomever
is so kind as offer guidance.

I appreciate your time,
Seth

--
Seth Pruitt
Department of Economics
University of California, San Diego
sjpruitt@ucsd.edu
http://dss.ucsd.edu/~sjpruitt

	[[alternative HTML version deleted]]

I've never used "numericDeriv" before, but from reading the 
documentation and working throught the example, a little thought led me 
to the following example:

f <- function(g., A){
  A %*% g.
}
g <- function(x1, x2, x3, x4, G){
   G %*%c(x1, x2, x3, x4)
}
env234 <- new.env()
A <- array(1:6, dim=c(2,3))
G <- array(1:12, dim=c(3,4))
assign("A", A, env=env234)
assign("G", G, env=env234)
assign("x1", 1., env=env234)
assign("x2", 2., env=env234)
assign("x3", 3., env=env234)
assign("x4", 4., env=env234)
(AG <- numericDeriv(quote(f(g(x1, x2, x3, x4, G), A)),
            c("x1", "x2", "x3", "x4"),
env234))
(AG. <- A%*%G)
(AG.%*%1:4)

	  This seemed to work for me and give me the correct answer.  If this 
is not enough, RSiteSearch("numericDeriv") gave me 115 hits;  many of 
these are not relevant to your question, but I would expect that some 
might be.

	  hope this helps.
	  spencer graves

Seth Pruitt wrote:
> R Help List --
> 
> I have defined two time-series-vector-valued-functions, let them be f and
g,
> and want to find the numeric derivative of f with respect to the variable x
> where f depends on x through g:
> (d/dx)(f (g(x) )
> 
> Moreover, x is a vector
> 
> I tried this out the long way (naming every element of the x vector and
then
> making the 'theta' argument in numericDeriv() the character vector
of all
> these names) and the result is just one time series vector; I was hoping
for
> a matrix. Also weirdly, if I instead make theta equal to just one of the
> named elements of x, I get the same time series vector; the same happens
for
> any subset of the named elements
> 
> My call to numericDeriv looks like this (vphi acts as x,
> swz.kalman.vectoracts as g,
> decision.ts.vector acts as f):
> 
> ***
> numericDeriv(
> expr = decision.ts.vector(
>   a = swz.kalman.vector(
>     zeta1=zeta[1], u = Phi$u, phi = Phi$Phi,
>     vphi >
c(varphi1,varphi2,varphi3,varphi4,varphi5,varphi6,varphi7,varphi8,varphi9,varphi10,
> 
>
varphi11,varphi12,varphi13,varphi14,varphi15,varphi16,varphi17,varphi18,varphi19,
> 
>
varphi20,varphi21,varphi22,varphi23,varphi24,varphi25,varphi26,varphi27,varphi28,
> 
>
varphi29,varphi30,varphi31,varphi32,varphi33,varphi34,varphi35,varphi36,varphi37,
>     varphi38,varphi39,varphi40,varphi41,varphi42),
>     alpha.prior = specs$alpha.prior
>     ),
>   phi = Phi$Phi, lambda = specs$lambda, delta = specs$delta, pi.star >
specs$pi.star, u.2star = specs$u.2star
>   ),
> theta >
c("varphi1","varphi2","varphi3","varphi4","varphi5","varphi6","varphi7","varphi8","varphi9","varphi10",
> 
>
"varphi11","varphi12","varphi13","varphi14","varphi15","varphi16","varphi17","varphi18","varphi19",
> 
>
"varphi20","varphi21","varphi22","varphi23","varphi24","varphi25","varphi26","varphi27","varphi28",
> 
>
"varphi29","varphi30","varphi31","varphi32","varphi33","varphi34","varphi35","varphi36","varphi37",
>    
"varphi38","varphi39","varphi40","varphi41","varphi42")
> );
> ***
> 
> As you can see, it includes some calls to other objects that are lying
> around. Maybe from this email someone can tell me my mistake with how
I've
> called things; otherwise, I'm happy to send along the .R files to
whomever
> is so kind as offer guidance.
> 
> I appreciate your time,
> Seth
> 
> --
> Seth Pruitt
> Department of Economics
> University of California, San Diego
> sjpruitt at ucsd.edu
> http://dss.ucsd.edu/~sjpruitt
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> 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