LBFGS optimizer with optional boxing. More...

#include <Lbfgs.h>

Data Structures
struct	Box
	Type storing boundaries for parameter values. More...

struct	CollectiveRingBuffer
	Class containing hessian information for LBFGS update. More...

struct	OptimizationReturnType
	Type returned from an optimization. More...

struct	StepValues
	Class carrying proposed parameter updates in an optimization. More...

Public Types
using	MatrixType = Eigen::Matrix< FloatType, Eigen::Dynamic, Eigen::Dynamic >

using	VectorType = Eigen::Matrix< FloatType, Eigen::Dynamic, 1 >

Public Member Functions
template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>
OptimizationReturnType	minimize (Eigen::Ref< VectorType > parameters, UpdateFunction &&function, Checker &&check, Observer &&observer=Observer{})
	Find parameters to minimize an objective function. More...

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>
OptimizationReturnType	maximize (Eigen::Ref< VectorType > parameters, UpdateFunction &&function, Checker &&check, Observer &&observer=Observer{})
	Find parameters to maximize an objective function. More...

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>
OptimizationReturnType	minimize (Eigen::Ref< VectorType > parameters, const Box &box, UpdateFunction &&function, Checker &&check, Observer &&observer=Observer{})
	Find parameters to minimize an objective function within bounds. More...

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>
OptimizationReturnType	maximize (Eigen::Ref< VectorType > parameters, const Box &box, UpdateFunction &&function, Checker &&check, Observer &&observer=Observer{})
	Find parameters to maximize an objective function within bounds. More...

Static Public Member Functions
static constexpr bool	isPowerOfTwo (unsigned x)

Data Fields
FloatType	c1 = 0.0001
	1st parameter for the Wolfe conditions.

FloatType	c2 = 0.9
	2nd parameter for the Wolfe conditions. More...

FloatType	stepLength = 1.0
	The initial step length used in the L-BFGS. More...

Private Member Functions
template<typename UpdateFunction , typename Observer , typename... Boxes>
unsigned	adjustStepAlongDirection (UpdateFunction &&function, Observer &&observer, StepValues &step, const VectorType &direction, const Boxes &...boxes)
	Adjusts `stepLength` member and ensures the proposed parameter adjustment lowers the function value. More...

template<typename UpdateFunction , typename Checker , typename Observer , typename... Boxes>
OptimizationReturnType	minimizeBase (Eigen::Ref< VectorType > parameters, UpdateFunction &&function, Checker &&check, Observer &&observer, Boxes &&...boxes)

Detailed Description

template<typename FloatType = double, unsigned ringBufferSize = 16>
class Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >

LBFGS optimizer with optional boxing.

Template Parameters

FloatType	Type to represent floating point numbers
ringBufferSize	Number of gradients to use in Hessian approximation, this is akin to memory

Member Function Documentation

template<typename FloatType = double, unsigned ringBufferSize = 16>

template<typename UpdateFunction , typename Observer , typename... Boxes>

unsigned Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::adjustStepAlongDirection	(	UpdateFunction &&	function,
		Observer &&	observer,
		StepValues &	step,
		const VectorType &	direction,
		const Boxes &...	boxes
	)

inlineprivate

Adjusts stepLength member and ensures the proposed parameter adjustment lowers the function value.

The chosen direction and step length may overshoot or undershoot, and the function value might increase. We try to adapt the step length in order to ensure the function value decreases and step length is more or less optimal.

Parameters

function	Objective function generating value and gradients from parameters
observer	Observer called with parameter updates
step	Suggested step in parameters, values and gradients
direction	Suggested step direction
boxes	Boxes (if present)

Returns: Number of function evaluations made in adjusting the step

template<typename FloatType = double, unsigned ringBufferSize = 16>

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>

OptimizationReturnType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::maximize	(	Eigen::Ref< VectorType >	parameters,
		UpdateFunction &&	function,
		Checker &&	check,
		Observer &&	observer = `Observer {}`
	)

inline

Find parameters to maximize an objective function.

Template Parameters

UpdateFunction	A callable object taking arguments (const VectorType&, double&, Ref<VectorType>). Any return values are ignored.
Checker	An object implementing one methods: shouldContinue: (const unsigned iterations, const StepValues& step) -> bool
Observer	An observer callable object taking parameters. Return values are ignored.

Parameters

parameters	The initial parameters to optimize. Resulting optimization parameters are written to this argument.
function	The objective function to optimize.
check	The Checker instance
observer	The Observer instance

Returns: The number of calls to function made during optimization

template<typename FloatType = double, unsigned ringBufferSize = 16>

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>

OptimizationReturnType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::maximize	(	Eigen::Ref< VectorType >	parameters,
		const Box &	box,
		UpdateFunction &&	function,
		Checker &&	check,
		Observer &&	observer = `Observer {}`
	)

inline

Find parameters to maximize an objective function within bounds.

Template Parameters

UpdateFunction	A callable object taking arguments (const VectorType&, double&, Ref<VectorType>). Any return values are ignored.
Checker	An object implementing one methods: shouldContinue: (const unsigned iterations, const StepValues& step) -> bool
Observer	An observer callable object taking parameters. Return values are ignored.

Parameters

parameters	The initial parameters to optimize. Resulting optimization parameters are written to this argument.
box	Bounds on the parameters
function	The objective function to optimize.
check	The Checker instance
observer	The Observer instance

Returns: The number of calls to function made during optimization

template<typename FloatType = double, unsigned ringBufferSize = 16>

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>

OptimizationReturnType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::minimize	(	Eigen::Ref< VectorType >	parameters,
		UpdateFunction &&	function,
		Checker &&	check,
		Observer &&	observer = `Observer {}`
	)

inline

Find parameters to minimize an objective function.

Template Parameters

UpdateFunction	A callable object taking arguments (const VectorType&, double&, Ref<VectorType>). Any return values are ignored.
Checker	An object implementing one methods: shouldContinue: (const unsigned iterations, const StepValues& step) -> bool
Observer	An observer callable object taking parameters. Return values are ignored.

Parameters

parameters	The initial parameters to optimize. Resulting optimization parameters are written to this argument.
function	The objective function to optimize.
check	The Checker instance
observer	The Observer instance

Returns: A OptimizationReturnType instance

template<typename FloatType = double, unsigned ringBufferSize = 16>

template<typename UpdateFunction , typename Checker , typename Observer = Detail::DoNothingObserver>

OptimizationReturnType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::minimize	(	Eigen::Ref< VectorType >	parameters,
		const Box &	box,
		UpdateFunction &&	function,
		Checker &&	check,
		Observer &&	observer = `Observer {}`
	)

inline

Find parameters to minimize an objective function within bounds.

Template Parameters

UpdateFunction	A callable object taking arguments (const VectorType&, double&, Ref<VectorType>). Any return values are ignored.
Checker	An object implementing one methods: shouldContinue: (const unsigned iterations, const StepValues& step) -> bool
Observer	An observer callable object taking parameters. Return values are ignored.

Parameters

parameters	The initial parameters to optimize. Resulting optimization parameters are written to this argument.
box	Bounds on the parameters
function	The objective function to optimize.
check	The Checker instance
observer	The Observer instance

Note: Gradients passed to check do not contain bounded components

Returns: The number of calls to function made during optimization

Field Documentation

template<typename FloatType = double, unsigned ringBufferSize = 16>

FloatType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::c2 = 0.9

2nd parameter for the Wolfe conditions.

A value as low as 0.1 can be used.

template<typename FloatType = double, unsigned ringBufferSize = 16>

FloatType Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >::stepLength = 1.0

The initial step length used in the L-BFGS.

Note: the first step is a gradient descent with 0.1 times the steplength.

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

Molassembler/Temple/Optimization/Lbfgs.h

Data Structures

Public Types

Public Member Functions

Static Public Member Functions

Data Fields

Private Member Functions

Detailed Description

template<typename FloatType = double, unsigned ringBufferSize = 16> class Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >

Member Function Documentation

Field Documentation

template<typename FloatType = double, unsigned ringBufferSize = 16>
class Scine::Molassembler::Temple::Lbfgs< FloatType, ringBufferSize >