Bootstrapping using Rcpp - rcpp

Question

I am implementing a bootstrap procedure using Rcpp and seem to have hit a performance bottleneck (?). Basically, I am creating a boot matrix using RcppArmadillo, calling the fit_power_law function from igraph and estimating the power law coefficient. The code is as follows:
#include <RcppArmadillo.h>
#include <RcppArmadilloExtensions/sample.h>
using namespace Rcpp;
// [[Rcpp::depends(RcppArmadillo)]]
double alphaPLFit(const arma::vec& x) {
Environment igraph("package:igraph");
Function fit_power_law = igraph["fit_power_law"];
List p = fit_power_law(Named("x", x));
return p["alpha"];
}
arma::mat bootstraps(const arma::vec& x, int times) {
arma::mat bootMatrix;
for (int t = 0; t < times; ++t) {
arma::vec resample = RcppArmadillo::sample(x, x.n_elem, true);
bootMatrix.insert_cols(t, resample);
}
return bootMatrix;
}
// [[Rcpp::export]]
arma::vec bootAlphaPLFit(const arma::vec& x, int times) {
arma::mat bootMatrix = bootstraps(x, times);
arma::vec alphas(times);
for (int i = 0; i < times; ++i) {
double alpha = alphaPLFit(bootMatrix.col(i));
alphas[i] = alpha;
}
return alphas;
}
/***R
library(igraph)
sampleData <- degree(sample_pa(n = 1e5, m = 3, directed = FALSE))
system.time(bootAlphaPLFit(sampleData, 500))
*/
This takes about a minute to run on my computer. As a comparison, the boot function from the boot package takes about 30 seconds, with a custom statistic that calls igraph::fit_power_law and extracts alpha.
Is there any way to speed this up?