How use correctly Rcpp::pt( )

Question

I am reading the chapter 4 of "Seamless R and C++ Integration with Rcpp" and I had a little problem. In the "listing 4.13" the book given a example about how to use a function of R. I tried use other functions (different of the example) and I had success. My code is here: ``` #include // [[Rcpp::export]] Rcpp::DataFrame myrandom(Rcpp::NumericVector x) { int n = x.size(); Rcpp::NumericVector y1(n), y2(n), y3(n); y1 = Rcpp::pexp(x,1.0,1,0); y2 = Rcpp::pnorm(x,0.0,1.0,1,0); y3 = Rcpp::ppois(x,3.0,1,0); return Rcpp::DataFrame::create(Rcpp::Named("Exp") = y1,Rcpp::Named("Norm") = y2, Rcpp::Named("Pois") = y3); } sourceCpp("random.cpp") myrandom(c(0.5,1)) ``` In this case is OK, but when I try use Rcpp::pt I don't have success. My code is here. ``` #include // [[Rcpp::export]] Rcpp::DataFrame myrandom2(Rcpp::NumericVector x) { int n = x.size(); Rcpp::NumericVector y1(n); y1 = Rcpp::pt(x,3.0,0,1,0); return Rcpp::DataFrame::create(Rcpp::Named("T") = y1); } sourceCpp("random2…

Accepted Answer

Aren't C++ compiler messages lovely? ;-)

I think you just helped squash a really old bug. For both t and F, R actually has two functions group (r|p|q|d)t and (r|p|q|d)nt (ditto for f and nf) where the second form allows for the non-centrality parameter. And I think out header for nt.h was wrong.

If you make this change in file `include/Rcpp/stats/nt.h`

```
@@ -22,7 +22,7 @@
 #ifndef Rcpp__stats__nt_h
 #define Rcpp__stats__nt_h

-RCPP_DPQ_2(t,::Rf_dt,::Rf_pt,::Rf_qt)
+RCPP_DPQ_2(nt,::Rf_dnt,::Rf_pnt,::Rf_qnt)

#endif
```

(which you can just edit in the header, no reinstallation needed) then things seem to work:

```
#include <Rcpp.h>

// [[Rcpp::export]]
Rcpp::DataFrame myrandom(Rcpp::NumericVector x) {
  Rcpp::NumericVector y1, y2, y3, y4, y5, y6, y7, y8;

y1 = Rcpp::pexp(x,1.0,1,0);
  y2 = Rcpp::pnorm(x,0.0,1.0,1,0);
  y3 = Rcpp::ppois(x,3.0,1,0);
  y4 = Rcpp::pt(x,3.0,1,0);
  y5 = Rcpp::pnt(x,3.0,2,TRUE,FALSE);
  y6 = Rcpp::pnt(x,3.0,2,TRUE,TRUE);
  y7 = Rcpp::pnt(x,3.0,2,FALSE,TRUE);
  y8 = Rcpp::pnt(x,3.0,2,FALSE,FALSE);

return Rcpp::DataFrame::create(Rcpp::Named("Exp") = y1,
                                 Rcpp::Named("Norm") = y2, 
                                 Rcpp::Named("Pois") = y3,
                                 Rcpp::Named("t") = y4,
                                 Rcpp::Named("nt1") = y5,
                                 Rcpp::Named("nt2") = y6,
                                 Rcpp::Named("nt3") = y7,
                                 Rcpp::Named("nt4") = y8);
}
```

I still need to check the numerics against R, but at least it builds now. (And TRUE/FALSE versus 1/0 does not matter; that gets cast correctly).

Edit: And here is some output:

```
> sourceCpp("/tmp/so1.cpp")  
R> myrandom(seq(0.2, 0.5, by=0.1))  
       Exp     Norm      Pois        t       nt1      nt2        nt3      nt4   
1 0.181269 0.579260 0.0497871 0.572865 0.0351335 -3.34860 -0.0357655 0.964866   
2 0.259182 0.617911 0.0497871 0.608118 0.0434710 -3.13566 -0.0444441 0.956529  
3 0.329680 0.655422 0.0497871 0.642032 0.0535219 -2.92767 -0.0550074 0.946478  
4 0.393469 0.691462 0.0497871 0.674276 0.0654790 -2.72603 -0.0677212 0.934521   
R>   
```

How use correctly Rcpp::pt( )

Solution

Problem