SecantNewtonNlpSolver

(class from pyomo.contrib.pynumero.algorithms.solvers.scipy_solvers)

class pyomo.contrib.pynumero.algorithms.solvers.scipy_solvers.SecantNewtonNlpSolver(nlp, timer=None, options=None)

Bases: NewtonNlpSolver

A wrapper around the SciPy scalar Newton solver for NLP objects that takes a specified number of secant iterations (default is 2) to try to converge a linear equation quickly then switches to Newton’s method if this is not successful. This strategy is inspired by calculate_variable_from_constraint in pyomo.util.calc_var_value.

__init__(nlp, timer=None, options=None)

Parameters:

nlp (ExtendedNLP) – An instance of ExtendedNLP that will be solved. ExtendedNLP is required to ensure that the NLP has equal numbers of primal variables and equality constraints.

Methods

__init__(nlp[, timer, options])
evaluate_function(x0)
evaluate_jacobian(x0)
solve([x0])

Attributes

OPTIONS

Member Documentation