dplucenio/optimization

Repository to store all my studies and experiments from my masters Optimization discipline

optimization-teapm

Repository to store all my studies and experiments from my masters Optimization discipline Unconstrained optimization algorithms

Simple usage: Define objective function, define unconstrained problem and solve with any UnconstrainedOptimization solver available so far (SteepestDescent, ConjugateGradient and Newton Methods. Soon: ModifiedNewton and QuasiNewton).

Four different types of line search methods are available: equal intervals, golden section, quadratic interpolation and armijo method.

def rosenbrockFunction(x):
    x1, x2 = x[0], x[1] 
    return 10.0*x1**4.0 -20.0*x1**2.0*x2 + 10.0*x2**2.0 + x1**2.0 - 2.0*x1 + 5.0

p1 = UnconstrainedProblemSetup(
    f = rosenbrockFunction, 
    x0 = [- 1.0 , 3.0], 
    lineSearchMethod = goldenLineSearch, 
    absoluteEpsilon = 1e-6,
    maxIterations = 500,
    )

newtonOpt = NewtonOptimization(p1)
output = newtonOpt.solve()

Graphical optimization

Tools to ouput two variables (for proof of concept, teaching) objective functions, equality and inequality constraints. Shows feasible region and also optimization path for the existing algorithms implemented so far.