template<typename _realT>
struct mx::AO::analysis::clAOLinearPredictor< _realT >

Class to manage the calculation of linear predictor coefficients for a closed-loop AO system.

Template Parameters

_realT the real floating point type in which to do all arithmetic.

Definition at line 46 of file clAOLinearPredictor.hpp.

#include <ao/analysis/clAOLinearPredictor.hpp>

Classes
struct	regResult
	Result from one evaluated regularization scale. More...
struct	regularizationReport
	Diagnostic summary of the most recent regularization search. More...

Public Types
enum class	regularizationStatus { notRun , converged , boundaryLimited , invalidControls , iterationLimit , calculationFailure }
	Termination state of the most recent regularization search. More...
typedef _realT	realT
	Floating-point type used for predictor calculations.

Public Member Functions
	clAOLinearPredictor ()=default
	Construct a closed-loop linear-predictor calculator with default search controls.
mx::error_t	calcCoefficients (std::vector< realT > &PSDt, std::vector< realT > &PSDn, realT PSDreg, int Nc, realT condition=0)
	Calculate the LP coefficients for a turbulence PSD and a noise PSD.
template<bool telem>
mx::error_t	_regularizeCoefficients (realT &min_var, realT &min_sc, realT precision, realT max_sc, clGainOpt< realT > &go_lp, std::vector< realT > &PSDt, std::vector< realT > &PSDn, int Nc)
	Worker function for regularizing the PSD for coefficient calculation.
template<bool telem = false>
mx::error_t	regularizeCoefficients (realT &gmax_lp, realT &gopt_lp, realT &var_lp, realT &min_sc, clGainOpt< realT > &go_lp, std::vector< realT > &PSDt, std::vector< realT > &PSDn, int Nc)
	Regularize the PSD and calculate the associated LP coefficients.
template<bool printout = false>
mx::error_t	optimizeNc (realT &gmax_lp, realT &gopt_lp, int &Nc, realT &var_lp, clGainOpt< realT > &go_lp, std::vector< realT > &PSDt, std::vector< realT > &PSDn, int minNc, int maxNc)
	Regularize the PSD and calculate the associated LP coefficients.

Public Attributes
std::vector< realT >	m_PSDtn
	Working memory for the regularized PSD.
std::vector< realT >	m_psd2s
	Working memory for the 2-sided regularized PSD.
std::vector< realT >	m_ac
	Working memory to hold the autocorrelation.
sigproc::autocorrelationFromPSD< realT >	m_acpsd
	Converts the working PSD to an autocorrelation.
sigproc::linearPredictor< realT >	m_lp
	Linear predictor used to calculate coefficients.
realT	m_min_var0 { 0 }
	Initial minimum variance, with zero requesting initialization.
realT	m_min_sc0 { 10 }
	Initial minimum regularization scale in dB.
realT	m_precision0 { 2 }
	Initial regularization scale spacing in dB.
realT	m_max_sc0 { 100 }
	Initial maximum regularization scale in dB.
realT	m_dPrecision { 3 }
	Divisor applied to the spacing during refinement.
realT	m_gmax_lp { 5 }
	The maximum allowable gain for LP.
realT	m_minPrecision { 0.001 }
	Minimum requested regularization spacing in dB.
int	m_maxIts { 100 }
	Maximum number of search refinement iterations.
int	m_extrap { 1 }
	The LP extrapolation length in loop steps. Normally it is 1 step.
std::vector< regResult >	m_regResults
	Per-scale telemetry collected when requested.
regularizationReport	m_regularizationReport
	Diagnostic summary of the latest search.

Member Typedef Documentation

◆ realT

template<typename _realT>

typedef _realT mx::AO::analysis::clAOLinearPredictor< _realT >::realT

Floating-point type used for predictor calculations.

Definition at line 48 of file clAOLinearPredictor.hpp.

Member Enumeration Documentation

◆ regularizationStatus

template<typename _realT>

enum class mx::AO::analysis::clAOLinearPredictor::regularizationStatus

strong

Termination state of the most recent regularization search.

Enumerator
notRun	No regularization search has been attempted.
converged	The requested precision was reached.
boundaryLimited	The optimum remained on the expanded search boundary.
invalidControls	The configured search controls were invalid.
iterationLimit	The search exhausted its iteration limit.
calculationFailure	Coefficient or gain calculation failed.

Definition at line 61 of file clAOLinearPredictor.hpp.

Constructor & Destructor Documentation

◆ clAOLinearPredictor()

template<typename _realT>

mx::AO::analysis::clAOLinearPredictor< _realT >::clAOLinearPredictor ( )

default

Construct a closed-loop linear-predictor calculator with default search controls.

Member Function Documentation

◆ _regularizeCoefficients()

template<typename _realT>

template<bool telem>

mx::error_t mx::AO::analysis::clAOLinearPredictor< _realT >::_regularizeCoefficients	(	realT &	min_var,
		realT &	min_sc,
		realT	precision,
		realT	max_sc,
		clGainOpt< realT > &	go_lp,
		std::vector< realT > &	PSDt,
		std::vector< realT > &	PSDn,
		int	Nc )

inline

Worker function for regularizing the PSD for coefficient calculation.

Template Parameters

telem if true then the results are collected in m_regResults.

On first call (min_var = 0): loop over scale factors from min_sc to max_sc (<=) in steps of precision.

On subsequent calls, when min_var and min_sc are passed back in loop over scale factors from min_sc-precision to max_sc in steps of

Returns: error_t::noerror on success, otherwise the coefficient-calculation error.

Parameters

[in,out]	min_var	the minimum variance found; set to 0 on initial call
[in,out]	min_sc	the scale factor at the minimum variance
[in]	precision	the step size for the scale factor
[in]	max_sc	the maximum scale factor to test
[in]	go_lp	the gain optimization object
[in]	PSDt	the turbulence PSD
[in]	PSDn	the WFS noise PSD
[in]	Nc	the number of coefficients

Definition at line 167 of file clAOLinearPredictor.hpp.

References mx::AO::analysis::clGainOpt< _realT >::a(), mx::AO::analysis::clGainOpt< _realT >::b(), calcCoefficients(), calculationFailure, m_dPrecision, m_gmax_lp, m_lp, m_regResults, m_regularizationReport, mx::AO::analysis::clGainOpt< _realT >::maxStableGain(), mx::noerror, and mx::AO::analysis::clGainOpt< _realT >::optGainOpenLoop().

Referenced by regularizeCoefficients().

◆ calcCoefficients()

template<typename _realT>

mx::error_t mx::AO::analysis::clAOLinearPredictor< _realT >::calcCoefficients	(	std::vector< realT > &	PSDt,
		std::vector< realT > &	PSDn,
		realT	PSDreg,
		int	Nc,
		realT	condition = 0 )

inline

Calculate the LP coefficients for a turbulence PSD and a noise PSD.

This combines the two PSDs, augments to two-sided, and calls the linearPredictor.calcCoefficients method.

A regularization constant can be added to the PSD as well.

Returns: error_t::noerror on success, otherwise error_t::liberr.

Parameters

[in]	PSDt	the turbulence PSD
[in]	PSDn	the WFS noise PSD
[in]	PSDreg	the regularizing constant. Set to 0 to not use.
[in]	Nc	the number of LP coefficients
[in]	condition	the condition number for the SVD. If 0 then levinson recursion is used.

Definition at line 118 of file clAOLinearPredictor.hpp.

References mx::sigproc::augment1SidedPSD(), mx::liberr, m_ac, m_acpsd, m_extrap, m_lp, m_psd2s, m_PSDtn, mx::internal::mxlib_error_report(), and mx::noerror.

Referenced by _regularizeCoefficients(), and regularizeCoefficients().

◆ optimizeNc()

template<typename _realT>

template<bool printout = false>

mx::error_t mx::AO::analysis::clAOLinearPredictor< _realT >::optimizeNc	(	realT &	gmax_lp,
		realT &	gopt_lp,
		int &	Nc,
		realT &	var_lp,
		clGainOpt< realT > &	go_lp,
		std::vector< realT > &	PSDt,
		std::vector< realT > &	PSDn,
		int	minNc,
		int	maxNc )

inline

Regularize the PSD and calculate the associated LP coefficients.

The PSD is regularized by adding a constant to it. This constant is found by minimizing the variance of the residual PSD.

Template Parameters

printout if true then per-scale results are collected in m_regResults.

Returns: error_t::noerror on success, otherwise the regularization error.

Parameters

[out]	gmax_lp	maximum gain for the selected predictor
[out]	gopt_lp	optimum gain for the selected predictor
[out]	Nc	selected number of coefficients
[out]	var_lp	variance at the optimum gain
[in]	go_lp	the gain optimization object
[in]	PSDt	the turbulence PSD
[in]	PSDn	the WFS noise PSD
[in]	minNc	minimum number of coefficients
[in]	maxNc	maximum number of coefficients

Definition at line 400 of file clAOLinearPredictor.hpp.

References mx::noerror, and regularizeCoefficients().

◆ regularizeCoefficients()

template<typename _realT>

template<bool telem = false>

mx::error_t mx::AO::analysis::clAOLinearPredictor< _realT >::regularizeCoefficients	(	realT &	gmax_lp,
		realT &	gopt_lp,
		realT &	var_lp,
		realT &	min_sc,
		clGainOpt< realT > &	go_lp,
		std::vector< realT > &	PSDt,
		std::vector< realT > &	PSDn,
		int	Nc )

inline

Regularize the PSD and calculate the associated LP coefficients.

The PSD is regularized by adding a constant to it. This constant is found by minimizing the variance of the residual PSD.

Template Parameters

telem if true then the results are collected in m_regResults

Returns: error_t::noerror for a converged or boundary-limited search, error_t::invalidconfig for invalid controls, error_t::timeout on iteration exhaustion, or the coefficient-calculation error.

Parameters

[out]	gmax_lp	the maximum gain calculated for the regularized PSD
[out]	gopt_lp	the optimum gain calculated for the regularized PSD
[out]	var_lp	the variance at the optimum gain
[out]	min_sc	the optimum regularization scale factor
[in]	go_lp	the gain optimization object
[in]	PSDt	the turbulence PSD
[in]	PSDn	the WFS noise PSD
[in]	Nc	the number of coefficients