nonlin_solve Module Reference

nonlin_solve More...

Data Types
type	brent_solver
	Defines a solver based upon Brent's method for solving an equation of one variable without using derivatives. More...
type	line_search_solver
	A class describing nonlinear solvers that use a line search algorithm to improve convergence behavior. More...
type	newton_1var_solver
	Defines a solver based upon Newtons's method for solving an equation of one variable. The algorithm uses a bisection method in conjunction with Newton's method in order to keep bounds upon the Newton iterations. More...
type	newton_solver
	Defines a Newton solver. More...
type	quasi_newton_solver
	Defines a quasi-Newton type solver based upon Broyden's method. More...

Functions/Subroutines
subroutine, public	test_convergence (x, xo, f, g, lg, xtol, ftol, gtol, c, cx, cf, cg, xnorm, fnorm)
	Tests for convergence.

Detailed Description

nonlin_solve

Purpose

To provide various routines capapble of solving systems of nonlinear equations.

Function/Subroutine Documentation

test_convergence()

subroutine, public nonlin_solve::test_convergence	(	real(real64), dimension(:), intent(in)	x,
		real(real64), dimension(:), intent(in)	xo,
		real(real64), dimension(:), intent(in)	f,
		real(real64), dimension(:), intent(in)	g,
		logical, intent(in)	lg,
		real(real64), intent(in)	xtol,
		real(real64), intent(in)	ftol,
		real(real64), intent(in)	gtol,
		logical, intent(out)	c,
		logical, intent(out)	cx,
		logical, intent(out)	cf,
		logical, intent(out)	cg,
		real(real64), intent(out)	xnorm,
		real(real64), intent(out)	fnorm )

Tests for convergence.

Parameters

[in]	x	The current solution estimate.
[in]	xo	The previous solution estimate.
[in]	f	The current residual based upon x.
[in]	g	The current estimate of the gradient vector at x.
[in]	lg	Set to true if the gradient slope check should be performed; else, false.
[in]	xtol	The tolerance on the change in variable.
[in]	ftol	The tolerance on the residual.
[in]	gtol	The tolerance on the slope of the gradient.
[out]	c	True if the solution converged on either the residual or change in variable.
[out]	cx	True if convergence occurred due to change in variable.
[out]	cf	True if convergence occurred due to residual.
[out]	cg	True if convergence occured due to slope of the gradient.
[out]	xnorm	The largest magnitude component of the scaled change in variable vector.
[out]	fnorm	The largest magnitude residual component

Definition at line 629 of file nonlin_solve.f90.