OptimTarget Class Reference

Optimization target specification for quantum control. More...

#include <optimtarget.hpp>

Collaboration diagram for OptimTarget:

Public Member Functions
	OptimTarget ()
	OptimTarget (const Config &config, MasterEq *mastereq, double total_time, Vec rho_t0, bool quietmode_)
	Constructor with full target specification.
	~OptimTarget ()
ObjectiveType	getObjectiveType ()
	Retrieves the objective function type.
int	prepareInitialState (const int iinit, const int ninit, const std::vector< size_t > &nlevels, const std::vector< size_t > &nessential, Vec rho0)
	Prepares the initial condition state.
void	prepareTargetState (const Vec rho)
	Prepares the target state for gate optimization.
void	evalJ (const Vec state, double *J_re_ptr, double *J_im_ptr)
	Evaluates the final-time objective function measure \(J(\rho(T))\).
void	evalJ_diff (const Vec state, Vec statebar, const double J_re_bar, const double J_im_bar)
	Computes derivative of the final-time objective function measure.
double	finalizeJ (const double obj_cost_re, const double obj_cost_im)
	Finalizes the objective function computation.
void	finalizeJ_diff (const double obj_cost_re, const double obj_cost_im, double *obj_cost_re_bar, double *obj_cost_im_bar)
	Derivative of objective function finalization.
double	FrobeniusDistance (const Vec state)
	Computes Frobenius distance between target and current state.
void	FrobeniusDistance_diff (const Vec state, Vec statebar, const double Jbar)
	Derivative of Frobenius distance computation.
void	HilbertSchmidtOverlap (const Vec state, const bool scalebypurity, double *HS_re_ptr, double *Hs_im_ptr)
	Computes Hilbert-Schmidt overlap between state and target.
void	HilbertSchmidtOverlap_diff (Vec statebar, bool scalebypurity, const double HS_re_bar, const double HS_im_bar)
	Derivative of Hilbert-Schmidt overlap computation.

Protected Attributes
PetscInt	dim
	State dimension of full vectorized system: N^2 if Lindblad, N if Schroedinger.
PetscInt	dim_rho
	Dimension of Hilbert space = N.
PetscInt	dim_ess
	Dimension of essential level system = N_e.
int	noscillators
	Number of oscillators in the system.
TargetType	target_type
	Type of optimization target (product state preparation or gate optimization)
ObjectiveType	objective_type
	Type of objective function measure (Frobenius, trace, product-state measure)
Gate *	targetgate
	Pointer to target gate (if gate optimization)
double	purity_rho0
	Purity of initial state Tr(rho(0)^2)
PetscInt	purestateID
	For product state preparation: integer m for preparing the target state \( e_m e_m^{\dagger}\).
Vec	targetstate
	Storage for the target state vector (NULL for product states, \(V\rho V^\dagger\) for gates, density matrix from file)
InitialConditionSettings	initcond
	Initial conditions.
DecoherenceType	decoherence_type
	Type of Lindblad decoherence operators, or NONE for Schroedinger solver.
int	mpisize_petsc
	Size of PETSc communicator.
int	mpirank_petsc
	Rank of PETSc communicator.
PetscInt	localsize_u
	Size of local sub vector u or v in state x=[u,v].
PetscInt	ilow
	First index of the local sub vector u,v.
PetscInt	iupp
	Last index (+1) of the local sub vector u,v.
Vec	aux
	Auxiliary vector for gate optimization objective computation.
bool	quietmode
	Flag for quiet mode operation.

Detailed Description

Optimization target specification for quantum control.

This class manages the target specification for quantum optimal control problems, including gate optimization and product state preparation. It handles target and initial state preparation and the evaluation of the final-time objective function measure.

Main functionality:

• prepareInitialState prepares and returns the states at time t=0, depending on what the optimization target is
• prepareTargetState prepares and stores the corresponding target state for this initial state
• evalJ for computing the final-time objective function measure

This class contains references to:

• Gate for evaluating the target state for quantum gate optimization

Constructor & Destructor Documentation

OptimTarget() [1/2]

OptimTarget::OptimTarget

(

)

OptimTarget() [2/2]

OptimTarget::OptimTarget	(	const Config &	config,
		MasterEq *	mastereq,
		double	total_time,
		Vec	rho_t0,
		bool	quietmode_
	)

Constructor with full target specification.

Parameters

config	Configuration parameters
mastereq	Pointer to master equation solver
total_time	Total evolution time
rho_t0	Initial state vector
quietmode_	Flag for quiet operation

~OptimTarget()

OptimTarget::~OptimTarget

(

)

Member Function Documentation

evalJ()

void OptimTarget::evalJ	(	const Vec	state,
		double *	J_re_ptr,
		double *	J_im_ptr
	)

Evaluates the final-time objective function measure \(J(\rho(T))\).

The target state must be prepared and stored before calling this function. Returns both real and imaginary parts of the final-time measure. The imaginary part is generally zero except for Schroedinger solver with the trace objective function measure.

Parameters

[in]	state	Current state vector
[out]	J_re_ptr	Pointer to store real part of objective
[out]	J_im_ptr	Pointer to store imaginary part of objective

evalJ_diff()

void OptimTarget::evalJ_diff	(	const Vec	state,
		Vec	statebar,
		const double	J_re_bar,
		const double	J_im_bar
	)

Computes derivative of the final-time objective function measure.

Updates the adjoint state vector for gradient computation.

Parameters

state	Final-time state vector
statebar	Adjoint state vector to update
J_re_bar	Adjoint of real part of objective
J_im_bar	Adjoint of imaginary part of objective

finalizeJ()

double OptimTarget::finalizeJ	(	const double	obj_cost_re,
		const double	obj_cost_im
	)

Finalizes the objective function computation.

Compute the infidelity (1-fidelity).

Parameters

obj_cost_re	Real part of objective cost
obj_cost_im	Imaginary part of objective cost

Returns

double Final objective function value

finalizeJ_diff()

void OptimTarget::finalizeJ_diff	(	const double	obj_cost_re,
		const double	obj_cost_im,
		double *	obj_cost_re_bar,
		double *	obj_cost_im_bar
	)

Derivative of objective function finalization.

Parameters

[in]	obj_cost_re	Real part of objective cost
[in]	obj_cost_im	Imaginary part of objective cost
[out]	obj_cost_re_bar	Pointer to store adjoint of real part
[out]	obj_cost_im_bar	Pointer to store adjoint of imaginary part

FrobeniusDistance()

double OptimTarget::FrobeniusDistance

(

const Vec

state

)

Computes Frobenius distance between target and current state.

Calculates \(F = 1/2 || \rho_{target} - \rho||^2_F\)

Parameters

state

Current state vector

Returns

double Frobenius distance

FrobeniusDistance_diff()

void OptimTarget::FrobeniusDistance_diff	(	const Vec	state,
		Vec	statebar,
		const double	Jbar
	)

Derivative of Frobenius distance computation.

Parameters

state	Current state vector
statebar	Adjoint state vector to update
Jbar	Adjoint seed

getObjectiveType()

ObjectiveType OptimTarget::getObjectiveType ( )

inline

Retrieves the objective function type.

Returns

ObjectiveType Type of objective function

HilbertSchmidtOverlap()

void OptimTarget::HilbertSchmidtOverlap	(	const Vec	state,
		const bool	scalebypurity,
		double *	HS_re_ptr,
		double *	Hs_im_ptr
	)

Computes Hilbert-Schmidt overlap between state and target.

Calculates \( Tr(\rho^\dagger \rho_{target})\), optionally scaled by the purity of the target state.

Parameters

state	Current state vector
scalebypurity	Flag to scale by purity of target state
HS_re_ptr	Pointer to store real part of overlap
Hs_im_ptr	Pointer to store imaginary part of overlap

HilbertSchmidtOverlap_diff()

void OptimTarget::HilbertSchmidtOverlap_diff	(	Vec	statebar,
		bool	scalebypurity,
		const double	HS_re_bar,
		const double	HS_im_bar
	)

Derivative of Hilbert-Schmidt overlap computation.

Parameters

statebar	Adjoint state vector to update
scalebypurity	Flag to scale by purity of target state
HS_re_bar	Adjoint of real part of overlap
HS_im_bar	Adjoint of imaginary part of overlap

prepareInitialState()

int OptimTarget::prepareInitialState	(	const int	iinit,
		const int	ninit,
		const std::vector< size_t > &	nlevels,
		const std::vector< size_t > &	nessential,
		Vec	rho0
	)

Prepares the initial condition state.

Parameters

iinit	Index in processor range [rank * ninit_local .. (rank+1) * ninit_local - 1]
ninit	Total number of initial conditions
nlevels	Number of levels per oscillator
nessential	Number of essential levels per oscillator
rho0	Vector to store the initial condition

Returns

int Identifier for this initial condition (element number in matrix vectorization)

prepareTargetState()

void OptimTarget::prepareTargetState

(

const Vec

rho

)

Prepares the target state for gate optimization.

For gate optimization, computes the rotated target state \(V \rho V^{\dagger}\) for a given initial state \(\rho\) and stores it locally as a class member. Also stores the purity of \(rho\) needed for scaling the Hilbert-Schmidt overlap in the trace objective function.

Parameters

rho	Initial state vector

Member Data Documentation

aux

Vec OptimTarget::aux

protected

Auxiliary vector for gate optimization objective computation.

decoherence_type

DecoherenceType OptimTarget::decoherence_type

protected

Type of Lindblad decoherence operators, or NONE for Schroedinger solver.

dim

PetscInt OptimTarget::dim

protected

State dimension of full vectorized system: N^2 if Lindblad, N if Schroedinger.

dim_ess

PetscInt OptimTarget::dim_ess

protected

Dimension of essential level system = N_e.

dim_rho

PetscInt OptimTarget::dim_rho

protected

Dimension of Hilbert space = N.

ilow

PetscInt OptimTarget::ilow

protected

First index of the local sub vector u,v.

initcond

InitialConditionSettings OptimTarget::initcond

protected

Initial conditions.

iupp

PetscInt OptimTarget::iupp

protected

Last index (+1) of the local sub vector u,v.

localsize_u

PetscInt OptimTarget::localsize_u

protected

Size of local sub vector u or v in state x=[u,v].

mpirank_petsc

int OptimTarget::mpirank_petsc

protected

Rank of PETSc communicator.

mpisize_petsc

int OptimTarget::mpisize_petsc

protected

Size of PETSc communicator.

noscillators

int OptimTarget::noscillators

protected

Number of oscillators in the system.

objective_type

ObjectiveType OptimTarget::objective_type

protected

Type of objective function measure (Frobenius, trace, product-state measure)

purestateID

PetscInt OptimTarget::purestateID

protected

For product state preparation: integer m for preparing the target state \( e_m e_m^{\dagger}\).

purity_rho0

double OptimTarget::purity_rho0

protected

Purity of initial state Tr(rho(0)^2)

quietmode

bool OptimTarget::quietmode

protected

Flag for quiet mode operation.

target_type

TargetType OptimTarget::target_type

protected

Type of optimization target (product state preparation or gate optimization)

targetgate

Gate* OptimTarget::targetgate

protected

Pointer to target gate (if gate optimization)

targetstate

Vec OptimTarget::targetstate

protected

Storage for the target state vector (NULL for product states, \(V\rho V^\dagger\) for gates, density matrix from file)

---

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

• /home/runner/work/quandary/quandary/include/optimtarget.hpp
• /home/runner/work/quandary/quandary/src/optimtarget.cpp