o

scalation.minima

AugLagrangian

object AugLagrangian

The AugLagrangian class implements the Augmented Lagrangian Method for solving equality constrained optimization problems. Minimize objective function 'f' subject to constraint 'h' to find an optimal solution for 'x'.

min f(x) s.t. h(x) = 0

f = objective function h = equality contraint x = solution vector

Note: the hyper-parameters 'eta' and 'p0' will need to be tuned per problem.

See also

AugLagrangianTest for how to set up 'f', 'h' and 'grad' functions

Linear Supertypes
AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. AugLagrangian
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. Protected

Type Members

  1. type Gradient = (VectoD, Double, Double) => VectoD

Value Members

  1. def solve(x: VectoD, f: FunctionV_2S, h: FunctionV_2S, grad: Gradient): (VectoD, MatriD)

    Solve for an optimal solution to the equality constrained optimization problem.

    Solve for an optimal solution to the equality constrained optimization problem.

    x

    initial guess for solution vector

    f

    the objective function to be minimized

    h

    the equality constraint

    grad

    the gradient of Lagranian (must be specified by caller)