R help - Aug 2012 - Argument of a linear model

Hey

I have a little problem here:


I have an experimental space, lets say [-1,+1]^2, and I fit a second order
model above it. Regarding the whole experimental space the regression
function maps within [-3,+4], which means nothing else than
f^-1([-3,+4])=[-1,+1]
Now for example the question is: What is f^-1([-1,+2])=?

Is there any inverse function available in order to get the argument of a
regression function f?

cheers 
Martin



--
View this message in context:
http://r.789695.n4.nabble.com/Argument-of-a-linear-model-tp4641805.html
Sent from the R help mailing list archive at Nabble.com.

On Aug 30, 2012, at 12:08 AM, m4n14ccc wrote:
> Hey
>
> I have a little problem here:
>
>
> I have an experimental space, lets say [-1,+1]^2, and I fit a second  
> order
> model above it. Regarding the whole experimental space the regression
> function maps within [-3,+4], which means nothing else than
> f^-1([-3,+4])=[-1,+1]
> Now for example the question is: What is f^-1([-1,+2])=?
>
> Is there any inverse function available in order to get the argument  
> of a
> regression function f?
In a general sense such a problem is ill-defined mathematically  
because the inverse of a second order quadratic (my assumption  
regarding what you meant by "second order") will not necessarily be a
function in the mathematical sense of being one-to-one. You could  
first plot and then see if it makes sense to fit a surface to the  
point set generated by:

   sapply( seq(-1, 1, by=0.01) , f)

(Which could then be limited in your more restricted domain.) You  
could then predict those elements in seq(-1, 1, 0.01) which had  
corresponding images in f(seq(.)).

It would be more productive if you assembled a test case in R code.

-- 
David Winsemius, MD
Alameda, CA, USA