A class for Adam optimization according to https://towardsdatascience.com/how-to-implement-an-adam-optimizer-from-scratch-76e7b217f1cc. More...

#include <Adam.h>

Public Member Functions
	Adam ()
	Nullary constructor of the class. More...

	Adam (double beta1_in, double beta2_in, double epsilon_in, double eta_in)
	Contructor of the class. More...

double	get_decreasing_test ()
	????????????? More...

void	initialize_moment_and_variance (int parameter_num)
	????????????? More...

void	reset ()
	????????????? More...

int	update (Matrix_real &parameters, Matrix_real &grad, const double &f0)
	Call to set the number of gate blocks to be optimized in one shot. More...

virtual	~Adam ()
	Destructor of the class. More...

Public Attributes
double	eta

Protected Attributes
double	beta1
	parameter beta1 of the Adam algorithm More...

double	beta1_t
	beta1^t More...

double	beta2
	parameter beta2 of the Adam algorithm More...

double	beta2_t
	beta2^t More...

int	decreasing_idx = 0
	current index in the decreasing_vec array More...

double	decreasing_test
	decreasing_test More...

matrix_base< int >	decreasing_vec
	vector containing 1 if cost function decreased from previous value, and -1 if it increased More...

double	epsilon

int	f0_idx
	current index in the f0_vec array More...

double	f0_mean
	Mean of the latest cost function values to test local minimum. More...

double	f0_prev
	previous value of the cost function More...

Matrix_real	f0_vec
	vector stroing the lates values of cost function values to test local minimum More...

int64_t	iter_t
	iteration index More...

Matrix_real	mom
	momentum parameter of the Adam algorithm More...

int	num_threads
	Store the number of OpenMP threads. (During the calculations OpenMP multithreading is turned off.) More...

Matrix_real	var
	variance parameter of the Adam algorithm More...

Detailed Description

A class for Adam optimization according to https://towardsdatascience.com/how-to-implement-an-adam-optimizer-from-scratch-76e7b217f1cc.

Definition at line 36 of file Adam.h.

Constructor & Destructor Documentation

◆ Adam() [1/2]

Adam::Adam ( )

Nullary constructor of the class.

Returns: An instance of the class

Definition at line 31 of file common/Adam.cpp.

Here is the call graph for this function:

◆ Adam() [2/2]

Adam::Adam	(	double	beta1_in,
		double	beta2_in,
		double	epsilon_in,
		double	eta_in
	)

Contructor of the class.

Constructor of the class.

Parameters

Umtx_in	The unitary matrix to be decomposed
qbit_num_in	The number of qubits spanning the unitary to be decomposed.
initial_guess_in	Type to guess the initial values for the optimization. Possible values: ZEROS=0, RANDOM=1, CLOSE_TO_ZERO=2

Returns: An instance of the class

Constructor of the class.

Parameters

???????????????????????????????
???????????????????????????????
???????????????????????????????

Returns: An instance of the class

Definition at line 58 of file common/Adam.cpp.

Here is the call graph for this function:

◆ ~Adam()

Adam::~Adam ( )

virtual

Destructor of the class.

Definition at line 82 of file common/Adam.cpp.

Member Function Documentation

◆ get_decreasing_test()

double Adam::get_decreasing_test ( )

?????????????

Definition at line 273 of file common/Adam.cpp.

◆ initialize_moment_and_variance()

void Adam::initialize_moment_and_variance ( int parameter_num )

?????????????

Definition at line 123 of file common/Adam.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ reset()

void Adam::reset ( )

?????????????

Definition at line 90 of file common/Adam.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ update()

int Adam::update	(	Matrix_real &	parameters,
		Matrix_real &	grad,
		const double &	f0
	)

Call to set the number of gate blocks to be optimized in one shot.

Parameters

optimization_block_in	The number of gate blocks to be optimized in one shot
optimization_block_in	The number of gate blocks to be optimized in one shot

Returns: 0 optimizer is in decreasing stage, 1 converged to local minumum

Definition at line 138 of file common/Adam.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

Member Data Documentation

◆ beta1

double Adam::beta1

protected

parameter beta1 of the Adam algorithm

Definition at line 50 of file Adam.h.

◆ beta1_t

double Adam::beta1_t

protected

beta1^t

Definition at line 64 of file Adam.h.

◆ beta2

double Adam::beta2

protected

parameter beta2 of the Adam algorithm

Definition at line 52 of file Adam.h.

◆ beta2_t

double Adam::beta2_t

protected

beta2^t

Definition at line 66 of file Adam.h.

◆ decreasing_idx

int Adam::decreasing_idx = 0

protected

current index in the decreasing_vec array

Definition at line 78 of file Adam.h.

◆ decreasing_test

double Adam::decreasing_test

protected

decreasing_test

Definition at line 80 of file Adam.h.

◆ decreasing_vec

matrix_base<int> Adam::decreasing_vec

protected

vector containing 1 if cost function decreased from previous value, and -1 if it increased

Definition at line 76 of file Adam.h.

◆ epsilon

double Adam::epsilon

protected

Definition at line 54 of file Adam.h.

◆ eta

double Adam::eta

Definition at line 42 of file Adam.h.

◆ f0_idx

int Adam::f0_idx

protected

current index in the f0_vec array

Definition at line 73 of file Adam.h.

◆ f0_mean

double Adam::f0_mean

protected

Mean of the latest cost function values to test local minimum.

Definition at line 71 of file Adam.h.

◆ f0_prev

double Adam::f0_prev

protected

previous value of the cost function

Definition at line 82 of file Adam.h.

◆ f0_vec

Matrix_real Adam::f0_vec

protected

vector stroing the lates values of cost function values to test local minimum

Definition at line 69 of file Adam.h.

◆ iter_t

int64_t Adam::iter_t

protected

iteration index

Definition at line 61 of file Adam.h.

◆ mom

Matrix_real Adam::mom

protected

momentum parameter of the Adam algorithm

Definition at line 57 of file Adam.h.

◆ num_threads

int Adam::num_threads

protected

Store the number of OpenMP threads. (During the calculations OpenMP multithreading is turned off.)

Definition at line 48 of file Adam.h.

◆ var

Matrix_real Adam::var

protected

variance parameter of the Adam algorithm

Definition at line 59 of file Adam.h.

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

Public Member Functions

Public Attributes

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ Adam() [1/2]

◆ Adam() [2/2]

◆ ~Adam()

Member Function Documentation

◆ get_decreasing_test()

◆ initialize_moment_and_variance()

◆ reset()

◆ update()

Member Data Documentation

◆ beta1

◆ beta1_t

◆ beta2

◆ beta2_t

◆ decreasing_idx

◆ decreasing_test

◆ decreasing_vec

◆ epsilon

◆ eta

◆ f0_idx

◆ f0_mean

◆ f0_prev

◆ f0_vec

◆ iter_t

◆ mom

◆ num_threads

◆ var