R help - Mar 2010 - function odiag(): assigning values to off-diagonal

hi
I'm trying to use the function odiag(x) for matrix calculations. I would 
like to assign new values to an off-diagonal in a matrix.
When I use the diag (x) function I could write something like

p<-matrix(seq(1:12),ncol=4)
p.new<-matrix(rep(0,12),ncol=4)

diag(p.new)<-diag(p)
p.new

But this won't work with odiag. How can I turn odiag (x) into something 
like diag (x) in order to assign off-diagonals new values?
In the end I would like to do something like this:

h<-matrix(rep(c(0.2,0.3,0.3,1),4),ncol=4,byrow=T)     #harvest rates for 
the 3 diferent years
S<-matrix(rep(c(0.9,0.8,0.7),4),ncol=3,byrow=T)         #survival rates

for (i in 0:(L-2)){

    odiag(p,i) <- 
(odiag(h,i)*c(1,cumprod((1-odiag(h[1:R,1:(L-1)],i))*odiag(S[1:R,1:(L-1)],i))))
   
     p[Rp-i,Lp] <- 1-sum(odiag(p[1:Rp,-Lp],i)) ## add one more value
      p[1,L]<-h[1,L]
    p[2,Lp]<-1-p[1,L]
   }

Does anybody know how I can do this?

Thank you very much for the help

Best wishes

Ben

On Mar 1, 2010, at 10:53 AM, Benedikt Gehr wrote:
> hi
> I'm trying to use the function odiag(x) for matrix calculations.
You assume that everyone is using the package that contains odiag?

> I would like to assign new values to an off-diagonal in a matrix.
> When I use the diag (x) function I could write something like
>
> p<-matrix(seq(1:12),ncol=4)
> p.new<-matrix(rep(0,12),ncol=4)
>
> diag(p.new)<-diag(p)
> p.new
>
> But this won't work with odiag. How can I turn odiag (x) into  
> something like diag (x) in order to assign off-diagonals new values?
Perhaps something like:

p.new[row(p.new) == col(p.new) - 1] <- diag(p)
 > p.new
      [,1] [,2] [,3] [,4]
[1,]    1    1    0    0
[2,]    0    5    5    0
[3,]    0    0    9    9
> In the end I would like to do something like this:
>
> h<-matrix(rep(c(0.2,0.3,0.3,1),4),ncol=4,byrow=T)     #harvest rates  
> for the 3 diferent years
> S<-matrix(rep(c(0.9,0.8,0.7),4),ncol=3,byrow=T)         #survival  
> rates
>
> for (i in 0:(L-2)){
It's not clear if this is correct code on you machine, but on our  
machines we don't have "L".
>
>   odiag(p,i) <- (odiag(h,i)*c(1,cumprod((1-odiag(h[1:R,1: 
> (L-1)],i))*odiag(S[1:R,1:(L-1)],i))))
>      p[Rp-i,Lp] <- 1-sum(odiag(p[1:Rp,-Lp],i)) ## add one more value
>     p[1,L]<-h[1,L]
>   p[2,Lp]<-1-p[1,L]
>  }
>
> Does anybody know how I can do this?
Do what? You have given non-executable code but no verbal description  
of what you are attempting.
>
> Thank you very much for the help
>
> Best wishes
>

David Winsemius, MD
Heritage Laboratories
West Hartford, CT