A number of standard optimization functions for R along with sampling methods.
There is a unified interface to sampling the functions. One can run
sample.func("rosenbrok", n=250, k=5, method="lhs.sampling")
to get a 250
sample Latin hypercube in 5D of the rosenbrock function.
The following multi-D scalar functions are implemented. They are all defined on an arbitrary number of inputs.
- ackley
- rosenbrock
- schwefel
- sinc
- spherical
- zakharov
The following sampling methods are supported with their internal names in
parenthesis. With the exception of hexagonal.sample
these are all defined
on an arbitrary number of dimensions.
- Latin hypercube (
lh.sample
) - Uniform random (
random.sample
) - Cartesian lattice (
cartesian.sample
) - Hexagonal lattice (
hexagonal.sample
) - Toroidal sampling (
torus.sample
) - Sobol sequence (
sobol.sample
) - Halton sequence (
halton.sample
)