You should be able to use arma::sort_index
for this, which you mention in your post:
#include <RcppArmadillo.h>
// [[Rcpp::depends(RcppArmadillo)]]
// [[Rcpp::export]]
arma::vec arma_sort(arma::vec x, arma::vec y) {
return x(arma::sort_index(y));
}
/*** R
A <- c(0.5, 0.4, 0.2, 0.9)
B <- c(9, 1, 3, 5)
arma_sort(A, B)
*/
Result:
> arma_sort(A, B)
[,1]
[1,] 0.4
[2,] 0.2
[3,] 0.9
[4,] 0.5
Of course, there are other ways as well. Variations on this question have been asked a few times on Stack Overflow in the context of plain C++. Below I've adapted the answer here for Rcpp:
#include <Rcpp.h>
using namespace Rcpp;
// [[Rcpp::export]]
NumericVector Rcpp_sort(NumericVector x, NumericVector y) {
// Order the elements of x by sorting y
// First create a vector of indices
IntegerVector idx = seq_along(x) - 1;
// Then sort that vector by the values of y
std::sort(idx.begin(), idx.end(), [&](int i, int j){return y[i] < y[j];});
// And return x in that order
return x[idx];
}
/*** R
A <- c(0.5, 0.4, 0.2, 0.9)
B <- c(9, 1, 3, 5)
Rcpp_sort(A, B)
*/
Result:
> Rcpp_sort(A, B)
[1] 0.4 0.2 0.9 0.5