Car.proper C[] matrix

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?