hessprep.m

From Spinach Documentation Wiki
Jump to: navigation, search


Test and prepare a Hessian matrix to be definite and well-conditioned.

Syntax

    [grad,hess,data]=hessprep(grad,hess,optim,data)

Description

This function is used as a communication between the numerical optimisers (fminnewton.m, fminsimplex.m, fminkrotov.m) and regularisation/conditioning of hessreg.m.

Arguments

    grad      - gradient vector array
    
    hess      - Hessian matrix
    
    optim     - numerical optimisation options, needed to perform regularisation.
    
    data      - data structure inherited from optimiser.


Returns

    grad      - gradient vector array - may be modified by 'RFO' regularisation.
    
    hess      - regularised and conditioned Hessian matrix.
    
    data      - modified data structure inherited from optimiser.


See also

hessreg.m, fminnewton.m


Version 1.9, authors: David Goodwin