I'm looking at speeding up some R code by re-writing in C++ & integrating via Rcpp. My Cpp is, to say the least, rusty: so would appreciate any suggestions. In particular, I'm looking for pointers on mapping a function onto all elements of an Rcpp NumericVector
. Here's an example.
I need to produce a new vector as follows:
tail
slice of an existing NumericVector
; I have this so far:
// [[Rcpp::export]]
NumericVector cppAdjustProbabilities( NumericVector& currentProbs,
const int index,
const double divisor ) {
//Note index <=0, e.g. -1 means remove first element
if(index == 0) {
return(currentProbs);
} else {
NumericVector newProbs = no_init(currentProbs.size()+index);
NumericVector::iterator i = currentProbs.begin() - index;
NumericVector::iterator j = newProbs.begin();
for(; i != currentProbs.end(); ++i, ++j) {
*j=*i/divisor;
}
return(newProbs);
}
}
This works, but I'd prefer to use a "map" approach. I looked at std::transform
, but it only supports a unary operation on vector elements - so I can't see how to pass the divisor in. This, for example, isn't valid:
std::transform(currentProbs.begin()-index, currentProbs.end(),
newProbs.begin(), [](double val) { return (val / divisor);} );
Is there a way to bring the divisor
into scope in the lambda? Or another way of doing it?
Thanks
With c++ lambda functions you can capture a value like this:
src1 <- 'NumericVector cppAdjustProbabilities( NumericVector& currentProbs,
const int index,
const double divisor ) {
//Note index <=0, e.g. -1 means remove first element
if(index == 0) {
return(currentProbs);
} else {
NumericVector newProbs = no_init(currentProbs.size()+index);
std::transform(currentProbs.begin()-index, currentProbs.end(),
newProbs.begin(), [&divisor](double val) { return (val / divisor);} );
// ^^^^^^^^
return(newProbs);
}
}'
Rcpp::cppFunction(src1)
currentProbs <- c(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9)
index <- -5L
divisor <- 2.0
cppAdjustProbabilities(currentProbs, index, divisor)
#> [1] 0.30 0.35 0.40 0.45
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.