oclKronVectMult1col3Q: Function to call OpenCL program to premultiply a matrix by a...

In CARrampsOcl: Reparameterized and marginalized posterior sampling for conditional autoregressive models, OpenCL implementation

Description

Function to call OpenCL program to premultiply a matrix by a kronecker product of 3 matrices.

Usage

oclKronVectMult1col3Q(kernel, a, b, c, y, ...)

Arguments

kernel	OpenCL kernel to be run
a	first matrix in kronecker product
b	second matrix in kronecker product
c	third matrix in kronecker product
y	matrix to premultiply by kronecker product

Value

matrix; the product of the kronecker product of the first three arguments and the matrix y

Author(s)

Kate Cowles and Michael Seedorff

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function(a,b,b2,c)
{
   if(!is.numeric(a) | !is.numeric(b) | !is.numeric(c) )
          stop("a and b must be numeric matrices")

   na1 <- nrow(a)
   nb1 <- nrow(b)
   nb21 <- nrow(b2)
   nc1 <- ncol(c)
   a<- as.vector(t(a))
   b<- as.vector(t(b))
   b2<- as.vector(t(b2))
   c<- as.vector(c)     # change; not transposed before sent
   retvect <- rep(0, na1 * nb1 * nb21* nc1)

   out <- .C("doKronVectMult1col3QD", a=as.double(a), b = as.double(b),
              b2 = as.double(b2),
              c = as.double(c),
               retvect=as.double(retvect), na1 = as.integer(na1),
               nb1 = as.integer(nb1), nb21=as.integer(nb21), 
               nc1 = as.integer(nc1), PACKAGE="CARramps" )
   return(matrix(out$retvect, nrow=na1*nb1*nb21,byrow=F) ) # change byrow
  }

CARrampsOcl documentation built on May 2, 2019, 3:27 a.m.