CCL header file for learning state dependent constraint. More...

#include <ccl_math.h>
#include <gsl/gsl_linalg.h>
#include <gsl/gsl_blas.h>
#include <gsl/gsl_math.h>
#include <gsl/gsl_sf.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>

Data Structures
struct	LEARN_A_MODEL
	This structure describes a "LEARN_A_MODEL" (a learn alpha model). This structure constains the dimentionality of the defined problems and the model parameters. It should always initialised with ccl_learn_alpha_model_alloc, and destroyed with ccl_learn_alpha_model_free. More...

struct	OPTION
	This structure defines the OPTION for the optimizer. More...

struct	SOLVE_LM_WS
	This structure defines the workspace variables for solving the non-linear LM optimization. More...

Macros
#define	NUM_CONSTRAINT 3

Functions
int	ccl_learn_alpha_model_alloc (LEARN_A_MODEL *model)
	Allocates the memory for the learn_alpha model. More...

int	ccl_learn_alpha_model_free (LEARN_A_MODEL *model)
	Free the memory for the learn_alpha model. More...

int	ccl_solve_lm_ws_alloc (const LEARN_A_MODEL model, SOLVE_LM_WS lm_ws)
	Allocates the workspace memory for the LM solver. More...

int	ccl_solve_lm_ws_free (SOLVE_LM_WS *lm_ws)
	Free the memory for the LM solver. More...

void	ccl_learn_alpha (const double Un, const double X, const int dim_b, const int dim_r, const int dim_n, const int dim_x, const int dim_u, LEARN_A_MODEL optimal)
	Main computation routine for learning state dependent constraint A. More...

void	search_learn_alpha (const double BX, const double RnUn, LEARN_A_MODEL *model)
	Main computation routine for learning a model of A. More...

void	obj_AUn (const LEARN_A_MODEL model, const double W, const double BX, const double RnUn, double *fun_out)
	Computation routine for learning constraint by increasing the dimensionality of k. More...

void	ccl_get_unit_vector_from_matrix (const double theta, int dim_n, int dim_t, double alpha)
	Computation routine for generating unit vector for row vector alpha by given theta anlges. More...

void	ccl_solve_lm (const LEARN_A_MODEL model, const double RnUn, const double BX, const OPTION option, SOLVE_LM_WS lm_ws_param, double *W)
	Computation routine for optimizing non-linear objective functions using LM approach. More...

void	findjac (const LEARN_A_MODEL model, const int dim_x, const double BX, const double RnUn, const double y, const double x, double epsx, double J)
	Computation routine for numerical approximation of the objevtive jacobian matrix. More...

void	ccl_get_rotation_matrix (const double theta, const double currentRn, const LEARN_A_MODEL model, const int alpha_id, double Rn)
	Computation routine for rotation matrix after finding the k^th constraint vector. More...

void	ccl_make_given_matrix (const double theta, int i, int j, int dim, double *G)
	Computation routine for generating roation matrix of a plane rotation of degree theta in an arbitrary plane and dimension R. More...

void	predict_proj_alpha (double x, LEARN_A_MODEL model, double centres, double variance, double Iu, double *A)
	Computation routine for prediction of the A matrix. More...

Detailed Description

CCL header file for learning state dependent constraint.

Macro Definition Documentation

#define NUM_CONSTRAINT 3

NUM_CONSTRAINT is the default full rank of the task constraint

Function Documentation

void ccl_get_rotation_matrix	(	const double *	theta,
		const double *	currentRn,
		const LEARN_A_MODEL *	model,
		const int	alpha_id,
		double *	Rn
	)

Computation routine for rotation matrix after finding the k^th constraint vector.

Parameters

[in]	theta	Theta from kth constrains, must point to an arrary of 1 dim_t doubles*
[in]	currentRn	Rotation matrix from the last iteration, must point to an array of dim_u dim_u of doubles*
[in]	model	Must be pointer to a valid LEARN_A_MODEL structure
[in]	alpha_id	Current searching dimension of alpha
[out]	Rn	Returned rotation matrix, must point to an array of dim_u dum_u doubles*

void ccl_get_unit_vector_from_matrix	(	const double *	theta,
		int	dim_n,
		int	dim_t,
		double *	alpha
	)

Computation routine for generating unit vector for row vector alpha by given theta anlges.

Parameters

[in]	theta	Learned constraint parameters, must point to an array of dim_n dim_t doubles*
[in]	dim_n	Number of data samples
[in]	dim_t	Number of constraint parameters
[out]		A uit vector of constraint basis, must point to an array of dim_n dim_u doubles*

void ccl_learn_alpha	(	const double *	Un,
		const double *	X,
		const int	dim_b,
		const int	dim_r,
		const int	dim_n,
		const int	dim_x,
		const int	dim_u,
		LEARN_A_MODEL	optimal
	)

Main computation routine for learning state dependent constraint A.

Parameters

[in]	Un	Observed actions, Must point to an array of dim_u dim_n of doubles*
[in]	X	Observed state variables, Must point to an array of dim_x dim_n of doubles*
[in]	dim_b	Number of basis functions
[in]	dim_r	Dimensionality of the task space
[in]	dim_n	NUmber of data points
[in]	dim_x	Dimensionality of the state variables
[in]	dim_u	Dimensionality of the action space
[out]	optimal,optimal	model paramters

int ccl_learn_alpha_model_alloc ( LEARN_A_MODEL * model )

Allocates the memory for the learn_alpha model.

Parameters

[in] model Must point to a valid LEARN_A_MODEL structure

Returns

1 in case of succes
0 in case of failure (e.g. memory could not be allocated)

int ccl_learn_alpha_model_free ( LEARN_A_MODEL * model )

Free the memory for the learn_alpha model.

Parameters

[in] model Must point to a valid LEARN_A_MODEL structure

Returns

1 in case of succes
0 in case of failure (e.g. memory could not be freed)

void ccl_make_given_matrix	(	const double	theta,
		int	i,
		int	j,
		int	dim,
		double *	G
	)

Computation routine for generating roation matrix of a plane rotation of degree theta in an arbitrary plane and dimension R.

Parameters

[in]	theta	Degree of rotation, must point to a scalar doubles
[in]	i	Row index of matrix
[in]	j	Col index of matrix
[in]	dim	Dimensionality of the rotation matrix
[out]	G	Returned rotation matrix, must point to an array of dim_u dum_u doubles*

void ccl_solve_lm	(	const LEARN_A_MODEL *	model,
		const double *	RnUn,
		const double *	BX,
		const OPTION	option,
		SOLVE_LM_WS *	lm_ws_param,
		double *	W
	)

Computation routine for optimizing non-linear objective functions using LM approach.

Parameters

[in]	model	Must be pointer to a valid LEARN_A_MODEL structure.
[in]	RnUn	Pre-calculated dot product of Rn and Un, must point to an array of dim_u dim_n of doubles*
[in]	BX	Transformed state variables, must point to an array of dim_b dim_n of doubles*
[in]	option	Options for optimizer
[in]	lm_ws_param	Must point to a valid structure of SOLVE_LM_WS for LM optimizer workspace variables
[out]	W	Returned model parameter, must point to an array of dim_u-dum_k doubles

int ccl_solve_lm_ws_alloc	(	const LEARN_A_MODEL *	model,
		SOLVE_LM_WS *	lm_ws
	)

Allocates the workspace memory for the LM solver.

Parameters

[in]	model	Must point to a valid LEARN_A_MODEL structure
[in]	lm_ws	Must point to a valid lm workspace structure

Returns

1 in case of succes
0 in case of failure (e.g. memory could not be allocated).

int ccl_solve_lm_ws_free ( SOLVE_LM_WS * lm_ws )

Free the memory for the LM solver.

Parameters

[in] lm_ws Must point to a valid lm workspace structure

Returns

1 in case of succes
0 in case of failure (e.g. memory could not be freed).

void findjac	(	const LEARN_A_MODEL *	model,
		const int	dim_x,
		const double *	BX,
		const double *	RnUn,
		const double *	y,
		const double *	x,
		double	epsx,
		double *	J
	)

Computation routine for numerical approximation of the objevtive jacobian matrix.

Parameters

[in]	model	Must be pointer to a valid LEARN_A_MODEL structure
[in]	dim_x	Dimensionality of the model parameters
[in]	BX	Transformed state variables, must point to an array of dim_b dim_n of doubles*
[in]	RnUn	Pre-calculated dot product of Rn and Un, must point to an array of dim_u dim_n of doubles*
[in]	y	Residual calculated at x, must point to a scalar of double.
[in]	x	Current model parameter x, must point to an array of dim_x doubles, note here dim_x is not the dimension of the state variable
[in]	epsx	Tolerance of the model parameters
[out]	J	Returned jocabian matrix, must point to an array of dim_n dim_x doubles*

void obj_AUn	(	const LEARN_A_MODEL *	model,
		const double *	W,
		const double *	BX,
		const double *	RnUn,
		double *	fun_out
	)

Computation routine for learning constraint by increasing the dimensionality of k.

Parameters

[in]	model	Must be pointer to a valid LEARN_A_MODEL structure.
[in]	W	Current model parameter, must point to an array of dim_u-dum_k doubles.
[in]	RnUn	Pre-calculated dot product of Rn and Un, must point to an array of dim_u dim_n of doubles*
[out]		Residual value returned from the objective function, must point to a scalar of double

void predict_proj_alpha	(	double *	x,
		LEARN_A_MODEL *	model,
		double *	centres,
		double	variance,
		double *	Iu,
		double *	A
	)

Computation routine for prediction of the A matrix.

Parameters

[in]	x	Input state variable, must point to an array of dim_x doubles
[in]	model	Must be pointer to a valid LEARN_A_MODEL structure
[in]	centres	Rbf centers, must point to an array of 1 dim_b doubles.*
[in]	variance	Variance of the rbf
[in]	Iu	Identity matrix, must point to an array of dim_u dim_u doubles*
[out]	A	Constraint A matrix, must point to an array of dim_k dum_u doubles*

void search_learn_alpha	(	const double *	BX,
		const double *	RnUn,
		LEARN_A_MODEL *	model
	)

Main computation routine for learning a model of A.

Parameters

[in]	BX	Transformed state variables, must point to an array of dim_b dim_n of doubles*
[in]	RnUn	Pre-calculated dot product of Rn and Un, must point to an array of dim_u dim_n of doubles*
[in]	model	model paramters, must point to a valid LEARN_A_MODEL structure
[out]	model	model paramters, must point to a valid LEARN_A_MODEL structure

Data Structures

Macros

Functions

Detailed Description

Macro Definition Documentation

Function Documentation