openswmm::findroot Namespace Reference

Typedefs
using	NewtonFunc = std::function<void(double x, double* f, double* df)>
	Synchronous function and derivative evaluator for newton().
using	RidderFunc = std::function<double(double x)>
	Synchronous scalar function evaluator for ridder().

Functions
int	newton (double x1, double x2, double *rts, double xacc, const NewtonFunc &func)
	Find a root with Newton steps and a bisection fallback.
double	ridder (double x1, double x2, double xacc, const RidderFunc &func)
	Estimate a bracketed root with derivative-free Ridder iterations.

Variables
constexpr int	MAXIT = 60
	Maximum iterations.

Typedef Documentation

NewtonFunc

using openswmm::findroot::NewtonFunc = std::function<void(double x, double* f, double* df)>

Synchronous function and derivative evaluator for newton().

Parameters

[in]	x	Candidate root, in the caller's independent-variable units.
[out]	f	Non-null solver-owned slot for f(x).
[out]	df	Non-null solver-owned slot for df/dx.

Note

The callback must fill both slots and must not retain their addresses.

RidderFunc

using openswmm::findroot::RidderFunc = std::function<double(double x)>

Synchronous scalar function evaluator for ridder().

Parameters

[in]

x

Candidate root in caller-selected units.

Returns

f(x); no derivative is required.

Function Documentation

newton()

int openswmm::findroot::newton	(	double	x1,
		double	x2,
		double *	rts,
		double	xacc,
		const NewtonFunc &	func )

Find a root with Newton steps and a bisection fallback.

Parameters

[in]	x1	Bracket endpoint where f(x1) < 0, regardless of coordinate order.
[in]	x2	Bracket endpoint where f(x2) > 0.
[in,out]	rts	Non-null caller-owned initial guess inside the bracket; overwritten with the final estimate on both convergence and iteration exhaustion.
[in]	xacc	Positive absolute step tolerance in the units of x1 and x2.
[in]	func	Nonempty synchronous callback producing f and df/dx for each candidate. It is not retained; callback exceptions propagate.

Returns

Number of function evaluations on convergence or a stationary floating-point estimate; zero when more than MAXIT evaluations are needed.

The caller must orient and validate the bracket. This function does not pre-evaluate endpoints; preserving callback sequence matches legacy findroot_Newton(). Newton steps outside the bracket, or steps that converge too slowly, are replaced by bisection.

Here is the caller graph for this function:

ridder()

double openswmm::findroot::ridder	(	double	x1,
		double	x2,
		double	xacc,
		const RidderFunc &	func )

Estimate a bracketed root with derivative-free Ridder iterations.

Parameters

[in]	x1	First finite bracket coordinate.
[in]	x2	Second finite bracket coordinate; endpoint function values must have opposite signs unless an endpoint is already a root.
[in]	xacc	Positive absolute tolerance in the units of x1 and x2, applied to successive estimates and bracket width.
[in]	func	Nonempty synchronous evaluator, not retained. Exceptions propagate; callback invocation order follows legacy findroot_Ridder().

Returns

An exact endpoint root, an approximate root, or -1.0e20 if the endpoints do not straddle a root. On iteration exhaustion the last estimate is returned without a distinct failure code.

Note

When a new estimate differs by at most xacc, the previous estimate is returned. The callback's most recent input need not equal the result.

Variable Documentation

MAXIT

int openswmm::findroot::MAXIT = 60

constexpr

Maximum iterations.