spde.matrices {RGeostats}  R Documentation 
Build the Precision matrix from a set of meshes
spde.matrices(points=NA, meshes=NA, model=NA, db=NA, pattern=NA, nostat=NA, mode=0, size=NA, lp=NA, li=NA, lx=NA, d=NA, up=NA, uj=NA, ux=NA, zp=NA, zi=NA, verbose = FALSE)
points,meshes 
Arguments resulting from the triangulation step. See

model 
The 
db 
The 
pattern 
Definition of the SPDE pattern. It corresponds to the halfspace pattern. Its dimensions are the same along all space axes. This argument is used when mode=2. 
nostat 
List of nonstationary parameters.
For details see 
mode 
Type of construction for the SPDE matrices:
Depending on this argument, one or several arrays are returned (see Details). 
size 
Dimension of the matrices L, U, D and Z 
lp,li,lx 
Definition of the L matrix which must be sparse, lower triangular, stored by column. The row indices of L must be sorted. 
d 
Diagonal of D, of size n 
up,uj,ux 
Definition of the U matrix which must be sparse, upper triangular, stored by row. The row indices of U need not be sorted. If not provided, the elements from up, ui and ux are use instead (which corresponds to the case where the initial matrix is symmetric). 
zp,zi 
Z is sparse, stored by column. The tow indices of Z must be sorted 
verbose 
Verbose flag. 
Several matrices used in the SPDE framework are provided depending on the 'mode' argument, such as:
Q The precision matrix
G The G array
TildeC The diagonal matrix
InvTildeC The diagonal (inverse) matrix
Vars The diagonal of the inverse Q matrix
According to 'mode' the following arrays are returned:
0 Q, G, TildeC and InvTildeC
1 Vars
2 Q
All matrices are sparse. They are returned in the 'triplet' form. Each nonzero element of the sparse matrix is specified by its row, column and the item value.