RcppRoll (version 0.2.2)

rollit_raw: Generate your own Weighted C++ Roll Function

Description

Using this, you can write and wrap in your own C++ function.

Usage

rollit_raw(fun, depends = NULL, includes = NULL, inline = TRUE,
  name = NULL, additional = NULL, ...)

Arguments

fun

A character string defining the function call. See examples for usage.

depends

Other libraries to link to. Linking is done through Rcpp attributes.

includes

Other C++ libraries to include. For example, to include boost/math.hpp, you would pass c("<boost/math.hpp>").

inline

boolean; mark this function as inline? This may or may not increase execution speed.

name

string; a name to internally assign to your generated C++ functions.

additional

Other C++ code you want to include; e.g. helper functions. This code will be inserted as-is above the code in fun.

...

Optional arguments passed to sourceCpp.

Details

The signature of fun is fixed as:

double <name>( NumericVector& x, NumericVector& weights, const int& n, const int& N, const int& ind)

where

• X_SUB is a #define macro that expands to the sub-vector being rolled over,

• X(i) is a #define macro that expands to the current element of X_SUB in a loop being rolled over,

• x is a reference to the entire vector (not just the sub-vector being rolled over),

• weights are the weights,

• n is the window size,

• N is the number of non-zero weights passed,

• ind is the current position along vector x.

Because the variables are being passed by reference, you should not modify them, unless you're prepared for strange behavior. See examples for a potential usage.

Examples

# NOT RUN {
## implement a weighted rolling 'sum of squares'
fun <- "
double out = 0;
const double m = mean( X_SUB );
for( int i=0; i < n; i++ ) {
  out += weights[i] * ( (X(i)-m) * (X(i)-m) ) / (N-1);
  }
return out;
"

rolling_var <- rollit_raw( fun )
x <- 1:5
rolling_var( x, 5 ) == var(x)

## a (slow-ish) implementation of rolling kurtosis
fun <- "
double numerator = 0;
double denominator = 0;
const double m = mean( X_SUB );
for( int i=0; i < n; i++ ) {
  double tmp = ( X(i) - m ) * ( X(i) - m );
  numerator += tmp * tmp;
  denominator += tmp;
}
return N * numerator / ( denominator * denominator );
"

rolling_kurt <- rollit_raw( fun )
x <- rnorm(100)
rolling_kurt(x, 20)
# }