Small host-side numerical, interpolation, fitting, and I/O utilities. More...

#include "util.hpp"

Public Types
using	HostVectorView = Kokkos::View< Real *, Kokkos::HostSpace >
	Unmanaged-compatible host view type passed to fit callbacks.

Static Public Member Functions
static void	A3deriv (const Kokkos::View< Real * > &y_in, const Kokkos::View< Real * > &x_in, Kokkos::View< Real * > &dydx_out, const Kokkos::View< Real * > &x_out)
	Evaluate derivatives of local four-point cubic interpolants.

static void	A3interp (const Kokkos::View< Real * > &y_in, const Kokkos::View< Real * > &x_in, Kokkos::View< Real * > &y_out, const Kokkos::View< Real * > &x_out)
	Evaluate local four-point cubic Lagrange interpolants.

static void	append_stat (const std::string &filename, const std::string &data)
	Append one newline-terminated record to a statistics file.

static int	choose_int (const Kokkos::View< int * > &arr, int val)
	Find the first occurrence of an integer in a Kokkos View.

static int	choose_int (const std::vector< int > &arr, int val)
	Find the first occurrence of an integer in a standard vector.

static void	creatd_stat (const std::string &filename)
	Create or truncate an empty statistics text file.

static int	execute_command_line_wrapper (const std::string &command)
	Execute a command through the host C runtime shell.

static void	gaussj (Kokkos::View< Real > &a, Kokkos::View< Real > &b, int n, int m)
	Solve dense linear systems by full-pivot Gauss–Jordan elimination.

static Real	interp1 (Real x, const Kokkos::View< Real * > &xp, const Kokkos::View< Real * > &yp)
	Linearly interpolate a monotone one-dimensional table.

static Real	log2real (bool lin)
	Convert a logical flag to an exact real indicator.

static void	mrqcof (const Kokkos::View< Real * > &x, const Kokkos::View< Real * > &y, const Kokkos::View< Real * > &sig, Kokkos::View< Real * > &a, const Kokkos::View< int * > &ia, Kokkos::View< Real ** > &alpha, Kokkos::View< Real * > &beta, Real &chisq, std::function< void(Real, const HostVectorView &, Real &, HostVectorView &)> funcs)
	Assemble Levenberg–Marquardt normal-equation coefficients.

template<typename ViewType >
static void	putarr_attime (const ViewType &arr, Real time, const std::string &name)
	Compatibility placeholder for time-tagged diagnostic array output.

static Real	zbrent (std::function< Real(Real)> func, Real x1, Real x2, Real tol)
	Find a bracketed scalar root with Brent's method.

Detailed Description

Small host-side numerical, interpolation, fitting, and I/O utilities.

Routines accept Kokkos Views for integration with ORB5X storage but currently execute through host mirrors. Coordinate and value units pass through unchanged; derivatives have y/x units, and tolerances use the units of their corresponding independent variable.

Member Typedef Documentation

◆ HostVectorView

using orb5x::Util::HostVectorView = Kokkos::View<Real*, Kokkos::HostSpace>

Unmanaged-compatible host view type passed to fit callbacks.

Member Function Documentation

◆ A3deriv()

static void orb5x::Util::A3deriv	(	const Kokkos::View< Real * > &	y_in,
		const Kokkos::View< Real * > &	x_in,
		Kokkos::View< Real * > &	dydx_out,
		const Kokkos::View< Real * > &	x_out
	)

static

Evaluate derivatives of local four-point cubic interpolants.

Parameters

[in]	y_in	Tabulated dependent values.
[in]	x_in	Increasing, distinct tabulation coordinates.
[in,out]	dydx_out	Derivatives in `y_in/x_in` units; unchanged when fewer than four input points are available.
[in]	x_out	Nondecreasing query coordinates.

◆ A3interp()

static void orb5x::Util::A3interp	(	const Kokkos::View< Real * > &	y_in,
		const Kokkos::View< Real * > &	x_in,
		Kokkos::View< Real * > &	y_out,
		const Kokkos::View< Real * > &	x_out
	)

static

Evaluate local four-point cubic Lagrange interpolants.

Parameters

[in]	y_in	Tabulated dependent values.
[in]	x_in	Increasing, distinct tabulation coordinates.
[in,out]	y_out	Output values; unchanged if fewer than four inputs.
[in]	x_out	Query coordinates, expected in nondecreasing order.

◆ append_stat()

static void orb5x::Util::append_stat	(	const std::string &	filename,
		const std::string &	data
	)

static

Append one newline-terminated record to a statistics file.

Parameters

[in]	filename	Output path.
[in]	data	Record text without the added newline.

◆ choose_int() [1/2]

static int orb5x::Util::choose_int	(	const Kokkos::View< int * > &	arr,
		int	val
	)

static

Find the first occurrence of an integer in a Kokkos View.

Parameters

[in]	arr	Values copied to host for the search.
[in]	val	Requested value.

Returns: Zero-based index, or -1 when absent.

◆ choose_int() [2/2]

static int orb5x::Util::choose_int	(	const std::vector< int > &	arr,
		int	val
	)

static

Find the first occurrence of an integer in a standard vector.

Parameters

[in]	arr	Values to search.
[in]	val	Requested value.

Returns: Zero-based index, or -1 when absent.

◆ creatd_stat()

static void orb5x::Util::creatd_stat ( const std::string & filename )

static

Create or truncate an empty statistics text file.

Parameters

[in] filename Output path.

Note: Failure is reported to standard error and is not thrown.

◆ execute_command_line_wrapper()

static int orb5x::Util::execute_command_line_wrapper ( const std::string & command )

static

Execute a command through the host C runtime shell.

Parameters

[in] command Shell command string.

Returns: Implementation-defined std::system status code.

Warning: Do not pass untrusted or unsanitized text.

◆ gaussj()

static void orb5x::Util::gaussj	(	Kokkos::View< Real ** > &	a,
		Kokkos::View< Real ** > &	b,
		int	n,
		int	m
	)

static

Solve dense linear systems by full-pivot Gauss–Jordan elimination.

Parameters

[in,out]	a	Leading `n x n` matrix, replaced by its inverse.
[in,out]	b	Leading `n x m` right-hand sides, replaced by solutions.
[in]	n	System dimension.
[in]	m	Number of right-hand sides.

Exceptions

std::runtime_error If a pivot is repeated or has magnitude below 1e-15.

◆ interp1()

static Real orb5x::Util::interp1	(	Real	x,
		const Kokkos::View< Real * > &	xp,
		const Kokkos::View< Real * > &	yp
	)

static

Linearly interpolate a monotone one-dimensional table.

Parameters

[in]	x	Query coordinate in the units of `xp`.
[in]	xp	Increasing tabulation coordinates.
[in]	yp	Values in the same order and with at least `xp.extent(0)` entries.

Returns: Interpolated value; endpoint value outside the table, zero for an empty table, or the left value across a numerically zero-width cell.

◆ log2real()

static Real orb5x::Util::log2real ( bool lin )

static

Convert a logical flag to an exact real indicator.

Parameters

[in] lin Logical input.

Returns: 1.0 for true and 0.0 for false.

◆ mrqcof()

static void orb5x::Util::mrqcof	(	const Kokkos::View< Real * > &	x,
		const Kokkos::View< Real * > &	y,
		const Kokkos::View< Real * > &	sig,
		Kokkos::View< Real * > &	a,
		const Kokkos::View< int * > &	ia,
		Kokkos::View< Real ** > &	alpha,
		Kokkos::View< Real * > &	beta,
		Real &	chisq,
		std::function< void(Real, const HostVectorView &, Real &, HostVectorView &)>	funcs
	)

static

Assemble Levenberg–Marquardt normal-equation coefficients.

Parameters

[in]	x	Independent data coordinates.
[in]	y	Observations.
[in]	sig	One-standard-deviation uncertainties in y units.
[in,out]	a	Model parameters supplied to `funcs`.
[in]	ia	Nonzero entries select parameters to fit.
[out]	alpha	Symmetric curvature matrix for selected parameters.
[out]	beta	Gradient vector for selected parameters.
[out]	chisq	Sum of squared uncertainty-normalized residuals.
[in]	funcs	Callback returning model value and derivatives with respect to every parameter for one x coordinate.

◆ putarr_attime()

template<typename ViewType >

static void orb5x::Util::putarr_attime	(	const ViewType &	arr,
		Real	time,
		const std::string &	name
	)

static

Compatibility placeholder for time-tagged diagnostic array output.

Parameters

[in]	arr	Array that would be written.
[in]	time	Simulation time in ORB5X normalized units.
[in]	name	Diagnostic quantity name.

Note: The current implementation performs no I/O.

◆ zbrent()

static Real orb5x::Util::zbrent	(	std::function< Real(Real)>	func,
		Real	x1,
		Real	x2,
		Real	tol
	)

static

Find a bracketed scalar root with Brent's method.

Parameters

[in]	func	Continuous scalar function.
[in]	x1	First bracket endpoint.
[in]	x2	Second bracket endpoint.
[in]	tol	Absolute convergence tolerance in x units.

Returns: Approximate root, or quiet NaN when the endpoints do not bracket a sign change. After 100 iterations, returns the best current point.

The documentation for this class was generated from the following file:

util.hpp

Public Types

Static Public Member Functions

Detailed Description

Member Typedef Documentation

◆ HostVectorView

Member Function Documentation

◆ A3deriv()

◆ A3interp()

◆ append_stat()

◆ choose_int() [1/2]

◆ choose_int() [2/2]

◆ creatd_stat()

◆ execute_command_line_wrapper()

◆ gaussj()

◆ interp1()

◆ log2real()

◆ mrqcof()

◆ putarr_attime()

◆ zbrent()