Styjun

I am trying to use Rcpp for the first time. I have made a script in C++ that works fine independently of R, but when trying to implement it in Rcpp I am getting an error message about the return type. Here is the gist of what I've done:

cppSim <- '

#include <iostream>
#include <random>
#include <math.h>
//[[Rcpp::plugins(cpp11)]]
#include <Rcpp.h>
#include <cmath>
#include <iomanip>
#include <boost/math/distributions/students_t.hpp>
using boost::math::students_t;


using namespace std;
using namespace Rcpp;

// function to calculate p-value from a t-test

double ttestPValue(vector<double> obs1, vector<double> obs2, int nSamples) 

 ...

 return(q); \ q is a double, this bit works fine



NumericVector powerSimulation(int nSamples, int meanDiff, int nPerm, double pValueCutOff)

 ...

 // create vectors to hold data
 double sim1T[totalRD][nPerm]; 

 ...

 sim1T[rdmu - minRD][j] = ttestPValue(obs1, obs2, nSamples);

 double power[totalRD];
 for (int j = 0; j < 100; j++)
 power[j] = 0;
 

 for (int i = 0; i < totalRD; i++)
 for (int sum = 0; sum < nPerm; sum ++) 
 if (sim1T[i][sum] < pValueCutOff)
 power[i] = power[i] + 1;
 
 
 power[i] = (power[i] / nPerm)*100;
 

 return power; 
'

To run it I am using:

settings=getPlugin("Rcpp")
settings$env$PKG_CXXFLAGS=paste('-std=c++11',settings$env$PKG_CXXFLAGS,sep=' ')

simRcpp <- cxxfunction(signature(nSamplesR = "int", meanDiffR = "int", nPermR = "int", pValueCutOffR = "double"),
 plugin="Rcpp", 
 settings = settings,
 includes = cppSim,
 body='
 int nSamples = Rcpp::as<int>(nSamplesR);
 int meanDiff = Rcpp::as<int>(meanDiffR);
 int nPerm = Rcpp::as<int>(nPermR);
 double pValueCutOff = Rcpp::as<double>(pValueCutOffR);
 return Rcpp::wrap( cppSim(nSamples, meanDiff, nPerm, pValueCutOff));')

The error that I'm getting is:

file928498473e2.cpp: In function ‘Rcpp::NumericVector powerSimulation(int, int, int, double)’:
file928498473e2.cpp:165:8: error: invalid conversion from ‘double*’ to ‘const int&’ [-fpermissive]
 return power;
 ^

I get that my return type for power is wrong but how do I make it right? I've tried wrapping it as a NumericVector but that doesn't seem to work and I can't find an example with the same issue, as not everyone uses cxxfunction().

Please change

double power[totalRD];

into

std::vector<double> power(totalRD);