Options (kinsol)

A struct with options field names and corresponding values. Specifying the options with a struct allows to use the same options set in different solver calls. Values are overriden by separately setting individual options.

OPT = struct();
OPT.display = "iter";
OPT.tol = 1e-6;
OPT.stepTol = 1e-8;
function F=fun(x)    
    F = [2*x(1)-x(2)-exp(-x(1))
          -x(1)+2*x(2)-exp(-2*x(2))];
end
x = kinsol(fun, [0 0], options = OPT)

The solver method given as a string. kinsol accepts "Newton" (inexact Newton method), "lineSearch" (inexact Newton method with linesearch globalization), "Picard" (Picard iterations), and "fixedPoint" (fixed-point iterations).

The maximum number of nonlinear iterations (default is 200).

stopping tolerance on the scaled maximum norm of the system function (default is %eps^(1/3))

stopping tolerance on the minimum scaled step length (default is %eps^(2/3)).

The maximum length of the Newton step (default is 1000 times the scaled norm of intial guess).

The typical value of system function argument when the latter is close to the solution. This vector is used for scaling the argument components (default is ones(n,1), where n is the number of equations).

The typical value of system function when argument is far from the solution. This vector is used for scaling the system function components (default is ones(n,1), where n is the number of equations).

The number of acceleration Anderson vectors in fixed-point or Picard iterations (default is 0).

The damping factor in fixed-point or Picard iterations, must be in ]0,1] (default is 1, which means no damping).

the maximum number of iterations before updating the Jacobian (default is 10) when using Newton method. Must be a multiple of resMonFreq (see below).

the maximum number of iterations before checking the residual (default is 5).

A vector with the indices of the solution to constrain. When a theoretical solution exists and some of its components have a given sign then any of the following four constraints can be imposed: y(i)<0, y(i)<=0, y(i)>=0, y(i)>0. This option is only available when using inexact Newton method or Newton method with linesearch.

These options allow to specify a user-supplied Jacobian or its approximation (not available with fixed-point method). Please see the dedicated Jacobian page. Note that when using Picard iterations the Jacobian is supposed to be constant.

The solver can call a user function after each successfull internal step. See the dedicated Callback page for explanations and use cases.

A string allowing to choose the verbosity of the solver output: "iter" (the default, giving the iteration details and termination message), "final" (termination message only) and "none".