This is a function interface that is designed to allow efficient implementations of minimisation algorithms by recognising the fact that the function and its gradient may be repeatedly evaluated at the same point in different parts of the code - hence we can save ourselves some computation.

Constrained minimiser.

This is a base class for running multidimensional optimisation It requires the following objects to exist in object space (they are used by the State base class default constructor) Functionx "functionx": the function to be minimised VVector "vector": the vector to work with ConstrainedMinimiser also requires "constraintClass" to exist.

Cubic line search with positivity constraints dimensions become inactive when they go to zero and the gradient > GEPS

This is a function interface that is designed to allow efficient implementations of minimisation algorithms by recognising the fact that the function and its gradient may be repeatedly evaluated at the same point in different parts of the code - hence we can save ourselves some computation.

unconstrained minimiser for smooth functions: - ConjugateGradient - OR Quasi-newton (using GillMurray uydates) - Safeguarded cubic interpolation line search using gradients