C++ package with a list of common optimization algorithms. This package heavily relies on the C++ Boost library (version >= 1.41.0). To compile, you should also have installed CMake (version >= 1.26).

Latest sources can be checked out via:

git clone git://github.com:bruneli/optimize.git

or (if you have write privileges):

git clone git@github.com:bruneli/optimize.git

Then to compile

mkdir build
cd build
cmake ..
make all test

Every function which takes as an input argument a double * pointer and returns a double can be minimized. For example a C-like function should be something like:

double myfunc(double * const x)
{
  return (x[0] - 0.5)*(x[0] - 0.5) + (x[1] - 2.)*(x[1] - 2.);
}
double x0[2] = {0., 0.};
minimize_nm min_nm(&myfunc, 2, (double*)x0);
min_nm.minimize();

If you have python and boost-python properly compiled, it is possible to run the different minimization methods with a python function via a callback function as shown in the following example:

>>> import liboptimize as cppopt
>>> myf = lambda x : (x[0] - 2.)*(x[0] - 2.) + (x[1] - 2.)*(x[1] - 2.) 
>>> x0 = (0., 0.)
>>> cppopt.minimize(myf, x0)

The file python/ex_minimize.py provides a comparison in term of computation time between the C++ optimize and the python based scipy.optimize packages when running the Nelder-Mead (Downhill simplex) and the BFGS algorithms.

To run with the C++ optimize package, you should make sure that the PYTHONPATH environment variable can access the liboptimize.so library like:

export PYTHONPATH=/path/to/my/package/optimize/lib:${PYTHONPATH}