solver Namespace Reference

Name space for solvers. More...

Classes
class	adaptive_rk4
	Adaptive Fourth Order Runge Kutta class. More...
class	rk2
	Second Order Runge Kutta class. More...
class	rk4
	Fourth Order Runge Kutta class. More...
class	solver_interface
	Class interface the solver. More...
class	split_simplextic
	Predictor corrector that trys to minimize the disperison residule. More...

Concepts
concept	method
	Solver method concept.

Functions
template<jit::float_scalar T, bool SAFE_MATH = false>
void	newton (workflow::manager< T, SAFE_MATH > &work, graph::output_nodes< T, SAFE_MATH > vars, graph::input_nodes< T, SAFE_MATH > inputs, graph::shared_leaf< T, SAFE_MATH > func, graph::shared_random_state< T, SAFE_MATH > state, const T tolarance=1.0E-30, const size_t max_iterations=1000, const T step=1.0)
	Determine the value of vars to minimze the loss function.

Detailed Description

Name space for solvers.

Function Documentation

newton()

template<jit::float_scalar T, bool SAFE_MATH = false>

void solver::newton	(	workflow::manager< T, SAFE_MATH > &	work,
		graph::output_nodes< T, SAFE_MATH >	vars,
		graph::input_nodes< T, SAFE_MATH >	inputs,
		graph::shared_leaf< T, SAFE_MATH >	func,
		graph::shared_random_state< T, SAFE_MATH >	state,
		const T	tolarance = 1.0E-30,
		const size_t	max_iterations = 1000,
		const T	step = 1.0
	)

Determine the value of vars to minimze the loss function.

This uses newtons methods to solver for D(x) = 0.

Template Parameters

T	Base type of the calculation.
SAFE_MATH	Use Safe Math operations.

Parameters

[in,out]	work	Workflow manager.
[in]	vars	The unknowns to solver for.
[in]	inputs	Inputs for jit compile.
[in]	func	Function to find the root of.
[in]	state	Random state node.
[in]	tolarance	Tolarance to solve the dispersion function to.
[in]	max_iterations	Maximum number of iterations before giving up.
[in]	step	Newton step size.