Creates an ConjGradSolver object.

solver = cv.ConjGradSolver()
solver = cv.ConjGradSolver('OptionName', optionValue, ...)

Options

• Function Objective function that will be minimized, specified as a structure with the following fields (gradfun and gradeps are optional fields):
  • dims Number of dimensions
  • fun string, name of M-file that implements the calc() method. It should receive a vector of the specified dimension, and return a scalar value of the objective function evaluated at that point.
  • gradfun string, name of M-file that implements the getGradient() method. It should receive an ndims vector, and return a vector of partial derivatives. If an empty string is specified (default), the gradient is approximated using finite difference method as: F'(x) = (F(x+h) - F(x-h)) / 2*h (at the cost of exta function evaluations and less accuracy).
  • gradeps gradient step h used in finite difference method. default 1e-3
• TermCriteria Terminal criteria to the algorithm. default struct('type','Count+EPS', 'maxCount',5000, 'epsilon',1e-6)

All the parameters are optional, so this procedure can be called even without parameters at all. In this case, the default values will be used. As default value for terminal criteria are the only sensible ones, ObjectiveFunction should be set upon the obtained object, if the function was not given in the constructor. Otherwise, the two ways (submit it to ctor or miss it out and set the ObjectiveFunction property) are absolutely equivalent (and will drop the same errors in the same way, should invalid input be detected).