Suppose I have the following c++ code in a file named test.cpp
#include <Rcpp.h>
//[[Rcpp::export]]
Rcpp::NumericMatrix MyAbar (const Rcpp::NumericMatrix & x, int T){
unsigned int outrows = x.nrow(), i = 0, j = 0;
double d;
Rcpp::NumericMatrix out(outrows,outrows);
// Rcpp::LogicalVector comp;
for (i = 0; i < outrows - 1; i++){
Rcpp::NumericVector v1 = x.row(i);
Rcpp::NumericVector ans(outrows);
for (j = i + 1; j < outrows ; j ++){
d = mean(Rcpp::runif( T ) < x(i,j));
out(j,i)=d;
out(i,j)=d;
}
}
return out;
}
I know with the following command, I can have my own package
Rcpp.package.skeleton("test",cpp_files = "~/Desktop/test.cpp")
However, what if I want to combine the following R function which call the Rcpp-function into the package
random = function(A, T){
if (!is.matrix(A)){
A = Reduce("+",A)/T
}
# global constant and threshold
n = nrow(A)
B_0 = 3
w = min(sqrt(n),sqrt(T * log(n)))
q = B_0 * log(n) / (sqrt(n) * w)
A2 = MyAbar(A)
diag(A2) <- NA
K = A2 <= rowQuantiles(A2, probs=q, na.rm =TRUE)
diag(K) = FALSE
P = K %*% A * ( 1/(rowSums(K) + 1e-10))
return( (P + t(P))*0.5 )
}
How can i make it?
So you are asking how to make an R package? There are many good tutorials.
To a first approximation:
R/random.R
man/random.Rd
or by learning package roxygen2
NAMESPACE
is for and that DESCRIPTION
is right