Parameter Search¶

ssqp::param_search provides online dynamics parameter identification and SSQP hyperparameter search. See Parameter Search in Concepts for the conceptual explanation. The reference below is generated directly from include/param_search/search.hpp's own doc comments.

`dynamics_params::Search`¶

The abstract base for dynamics parameter searchers (namespace ssqp::param_search::dynamics_params):

Class ssqp::param_search::dynamics_params::Search¶

ClassList > ssqp > param_search > dynamics_params > Search

Inherited by the following classes: ssqp::param_search::dynamics_params::GaussNewton, ssqp::param_search::dynamics_params::GradientDescent

Public Functions¶

Type	Name
	Search (const std::shared_ptr< model::stagewise::StageDynamics > & dynamics, const double armijo_beta=0.5, const double armijo_c1=1e-4)
bool	addData (const VecX & x, const VecX & u, const VecX & x_next)
void	clearData ()

Public Functions Documentation¶

function Search¶

ssqp::param_search::dynamics_params::Search::Search (
    const std::shared_ptr< model::stagewise::StageDynamics > & dynamics,
    const double armijo_beta=0.5,
    const double armijo_c1=1e-4
)

function addData¶

bool ssqp::param_search::dynamics_params::Search::addData (
    const VecX & x,
    const VecX & u,
    const VecX & x_next
)

function clearData¶

void ssqp::param_search::dynamics_params::Search::clearData ()

The documentation for this class was generated from the following file /home/runner/work/ssqp/ssqp/include/param_search/search.hpp

`GradientDescent`¶

Class ssqp::param_search::dynamics_params::GradientDescent¶

ClassList > ssqp > param_search > dynamics_params > GradientDescent

Inherits the following classes: ssqp::param_search::dynamics_params::Search

Public Functions¶

Type	Name
	GradientDescent (const std::shared_ptr< model::stagewise::StageDynamics > & dynamics, const uint min_data_size, const double armijo_beta=0.5, const double armijo_c1=1e-4)
void	setMinimumDataSize (const uint data_size)

Public Functions Documentation¶

function GradientDescent¶

ssqp::param_search::dynamics_params::GradientDescent::GradientDescent (
    const std::shared_ptr< model::stagewise::StageDynamics > & dynamics,
    const uint min_data_size,
    const double armijo_beta=0.5,
    const double armijo_c1=1e-4
)

function setMinimumDataSize¶

void ssqp::param_search::dynamics_params::GradientDescent::setMinimumDataSize (
    const uint data_size
)

The documentation for this class was generated from the following file /home/runner/work/ssqp/ssqp/include/param_search/search.hpp

`GaussNewton`¶

Class ssqp::param_search::dynamics_params::GaussNewton¶

ClassList > ssqp > param_search > dynamics_params > GaussNewton

Inherits the following classes: ssqp::param_search::dynamics_params::Search

Public Functions¶

Type	Name
	GaussNewton (const std::shared_ptr< model::stagewise::StageDynamics > & dynamics, const uint window_size=100, const double armijo_beta=0.5, const double armijo_c1=1e-4)
void	setWindowSize (const uint window_size, bool prune_last=false)

Public Functions Documentation¶

function GaussNewton¶

ssqp::param_search::dynamics_params::GaussNewton::GaussNewton (
    const std::shared_ptr< model::stagewise::StageDynamics > & dynamics,
    const uint window_size=100,
    const double armijo_beta=0.5,
    const double armijo_c1=1e-4
)

function setWindowSize¶

void ssqp::param_search::dynamics_params::GaussNewton::setWindowSize (
    const uint window_size,
    bool prune_last=false
)

Sets the window size to the given value Parameters:

window_size: The size to resize the window to
prune_last: Whether to remove the most recent data points, in case the window currently holds more data than the provided size; if false, an algorithm is used to instead prune the most similar data points has no effect if the provided size is larger than the data currently held

The documentation for this class was generated from the following file /home/runner/work/ssqp/ssqp/include/param_search/search.hpp

`HyperParamSearch`¶

Declared directly in ssqp::param_search (not under dynamics_params):

Class ssqp::param_search::HyperParamSearch¶

ClassList > ssqp > param_search > HyperParamSearch

Public Functions¶

Type	Name
	HyperParamSearch () = default
	HyperParamSearch (const std::vector< solvers::ssqp::Params > & hyperparams, const std::vector< std::shared_ptr< model::Model > > & problems)
void	addHyperparams (const solvers::ssqp::Params & sqp_params)
void	addProblem (const std::shared_ptr< model::Model > & model)
const std::vector< std::shared_ptr< model::Model > > &	problems () const
const solvers::ssqp::Params	search () const

Public Functions Documentation¶

function HyperParamSearch [1/2]¶

ssqp::param_search::HyperParamSearch::HyperParamSearch () = default

function HyperParamSearch [2/2]¶

ssqp::param_search::HyperParamSearch::HyperParamSearch (
    const std::vector< solvers::ssqp::Params > & hyperparams,
    const std::vector< std::shared_ptr< model::Model > > & problems
)

function addHyperparams¶

void ssqp::param_search::HyperParamSearch::addHyperparams (
    const solvers::ssqp::Params & sqp_params
)

function addProblem¶

void ssqp::param_search::HyperParamSearch::addProblem (
    const std::shared_ptr< model::Model > & model
)

function problems¶

const std::vector< std::shared_ptr< model::Model > > & ssqp::param_search::HyperParamSearch::problems () const

function search¶

const solvers::ssqp::Params ssqp::param_search::HyperParamSearch::search () const

The documentation for this class was generated from the following file /home/runner/work/ssqp/ssqp/include/param_search/search.hpp