PGM Algorithms

PGM algorithm development is supported by a base class from which classes for specific optimisation problems may be derived. This base class provides the machinery for managing iterations and stopping conditions, etc. so that only methods providing the details of a specific problem need to be defined in a derived class.

Auxiliary Classes

The following classes provide auxiliary functionality for the PGM algorithms.

  • BacktrackBase an iterative process to find the optimal step size for the gradient descent (\(L^{-1}\)).

  • StepSizePolicyBase a process to adapt the step size for the gradient descent (\(L^{-1}\)).

  • MomentumBase a process to update the momentum coefficient \(t^{(j)}\).