[R] Car.proper C[] matrix

[Jason Gasper]

Here is the WinBUGS code

model {
for(i in 1:N) {m[i] <- 1/n[ind[i]] }
cumsum[1] <- 0
for(i in 2:(N+1)) {cumsum[i] <- sum(num[1:(i-1)]) }
for(k in 1:sumNumNeigh) {
  for(i in 1:N) {
# #pick[k,i] = 1 if cumsum[i] < k <= cumsum[i=1]; otherwise, pick[k,i] = 0
##step(e) 1 if e >= 0; 0 otherwise
    pick[k,i]<-step(k-cumsum[i]-epsilon)*step(cumsum[i+1]-k) }

  C[k]<-1/ inprod(num[], pick[k,]) }
epsilon <- 0.0001

Uwe Ligges wrote:
> If you want that people help to translate *code*, you have to specify 
> it ...
>
> Uwe Ligges
>
>
> Jason Gasper wrote:
>> I am hoping someone can help translate some WinBUGS code into R code. 
>> I would like to use R to create the C[] matrix required for a 
>> car.proper model in WinBUGS, but I am having a difficult time 
>> negotiating the coding. The C matrix provides normalized weights for 
>> each pair of spatial areas.   So the WinBUGS example is as follows:
>>
>> # of the weight matrix with elements Cij. The first J1 elements of 
>> the C[] vector contain the
>> # weights for the J1 neighbours of area i=1; the (J1+1) to J2 
>> elements of the C[] vector contain
>> # the weights for the J2 neighbours of area i=2; etc.
>> # To set up this vector, we need to define a variable cumsum, which 
>> gives the values of J1,
>> # J2, etc.; we then set up an index matrix pick[,] with N columns 
>> corresponding to the
>> # i=1,...,N areas, and with the same number of rows as there are 
>> elements in the C[] vector
>> # (i.e. sumNumNeigh). The elements C[ (cumsum[i]+1):cumsum[i+1] ] 
>> correspond to
>> # the set of weights Cij associated with area i, and so we set up ith 
>> column of the matrix pick[,]
>> # to have a 1 in all the rows k for which cumsum[i] < k <= 
>> cumsum[i+1], and 0's elsewhere.
>> # For example, let N=4 and cumsum=c(0,3,5,6,8), so area i=1 has 3 
>> neighbours, area i=2 has 2
>> # neighbours, area i=3 has 1 neighbour and area i=4 has 2 neighbours. 
>> The the matrix pick[,] is:
>> # pick
>> # 1, 0, 0, 0,
>> # 1, 0, 0, 0,
>> # 1, 0, 0, 0,
>> # 0, 1, 0, 0,
>> # 0, 1, 0, 0,
>> # 0, 0, 1, 0,
>> # 0, 0, 0, 1,
>> # 0, 0, 0, 1,
>> #
>> # We can then use the inner product (inprod(,)) function in WinBUGS 
>> and the kth row of pick to
>> # select which area corresponds to the kth element in the vector C[]; 
>> likewise, we can use inprod(,)
>> # and the ith column of pick to select the elements of C[] which 
>> correspond to area i.
>>
>> Basically I want to do this in R to speed things up a little. Has 
>> anyone written a function for this conversion?
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide 
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.

-- 
Jason Gasper
National Marine Fisheries Service
Alaska Region, Sustainable Fisheries Division
709 W. 9th St. Juneau, Alaska 99801 
Juneau, Alaska 99801

Phone  907-586-7237
Fax 907-586-7249