- Notifications
You must be signed in to change notification settings - Fork4
General Purpose Optimization in R using C++: provides a unified C++ wrapper to call functions of the algorithms underlying the optim() solver
NotificationsYou must be signed in to change notification settings
ypan1988/roptim
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
- Perform general purpose optimization in
Rusing theArmadilloC++library for numerical linear algebra. - A unified wrapper interface is used to call
Ccode of the five optimization algorithms (namelyNelder-Mead,BFGS,CG,L-BFGS-BandSANN) underlying functionoptim()(packagestats) provided by defaultRinstallation.
Get the development version from github:
install.packages("devtools")library(devtools)devtools::install_github("ypan1988/roptim",dependencies=TRUE)
Or the released version from CRAN:
install.packages("roptim")An example of usingstats::optim() inR environment:
fr<-function(x) {## Rosenbrock Banana functionx1<-x[1]x2<-x[2]100* (x2-x1*x1)^2+ (1-x1)^2}grr<-function(x) {## Gradient of 'fr'x1<-x[1]x2<-x[2] c(-400*x1* (x2-x1*x1)-2* (1-x1),200* (x2-x1*x1))}res<- optim(c(-1.2,1),fr,grr,method="BFGS",control=list(trace=T),hessian=TRUE)res
Corresponding code written inC++ using packageroptim (filedemo.cpp):
#include<cmath>// std::pow#include<RcppArmadillo.h>// [[Rcpp::depends(RcppArmadillo)]]#include<roptim.h>// [[Rcpp::depends(roptim)]]usingnamespaceroptim;classRosen :publicFunctor {public:doubleoperator()(const arma::vec &x)override {double x1 =x(0);double x2 =x(1);return100 *std::pow((x2 - x1 * x1),2) +std::pow(1 - x1,2); }voidGradient(const arma::vec &x, arma::vec &gr)override { gr = arma::zeros<arma::vec>(2);double x1 =x(0);double x2 =x(1);gr(0) = -400 * x1 * (x2 - x1 * x1) -2 * (1 - x1);gr(1) =200 * (x2 - x1 * x1); }};// [[Rcpp::export]]voidrosen_bfgs() { Rosen rb; Roptim<Rosen>opt("BFGS"); opt.control.trace =1; opt.set_hessian(true); arma::vec x = {-1.2,1}; opt.minimize(rb, x); Rcpp::Rcout <<"-------------------------" << std::endl; opt.print();}
Compile and run the function inR:
library(Rcpp)sourceCpp("~/demo.cpp")# you may need to change the directoryrosen_bfgs()
Then you will get expected output as follows:
initial value 24.200000 iter 10 value 1.367383iter 20 value 0.134560iter 30 value 0.001978iter 40 value 0.000000final value 0.000000 converged-------------------------.par() 1.0000 1.0000.value()9.59496e-018.fncount()110.grcount()43.convergence()0.message()NULL.hessian() 8.0200e+002 -4.0000e+002 -4.0000e+002 2.0000e+002Note: more examples are provided insrc/roptim_examples.cpp.