This is a subclass of the CPUSolver class which uses memory-aligned data structures and Intel's auto-vectorization. More...

#include "src/VectorizedSolver.h"

Public Member Functions
	VectorizedSolver (TrackGenerator *track_generator=NULL)
	Constructor initializes NULL arrays for source, flux, etc. More...

virtual	~VectorizedSolver ()
	Destructor deletes Track boundary angular flux and and FSR scalar flux and source arrays.

int	getNumVectorWidths ()
	Returns the number of vector lengths required to fit the number of energy groups. More...

void	setFixedSourceByFSR (int fsr_id, int group, FP_PRECISION source)
	Assign a fixed source for a flat source region and energy group. More...

void	setGeometry (Geometry *geometry)
	Sets the Geometry for the Solver and aligns all Material cross-section data for SIMD vector instructions. More...

Public Member Functions inherited from CPUSolver
	CPUSolver (TrackGenerator *track_generator=NULL)
	Constructor initializes array pointers for Tracks and Materials. More...

virtual	~CPUSolver ()
	Destructor deletes array for OpenMP mutual exclusion locks for FSR scalar flux updates, and calls Solver parent class destructor to deletes arrays for fluxes and sources.

int	getNumThreads ()
	Returns the number of shared memory OpenMP threads in use. More...

void	setNumThreads (int num_threads)
	Sets the number of shared memory OpenMP threads to use (>0). More...

void	setFluxes (FP_PRECISION *in_fluxes, int num_fluxes)
	Set the flux array for use in transport sweep source calculations. More...

void	setFixedSourceByFSR (long fsr_id, int group, FP_PRECISION source)
	Assign a fixed source for a flat source region and energy group. More...

void	computeFSRFissionRates (double *fission_rates, long num_FSRs, bool nu=false)
	Computes the volume-averaged, energy-integrated fission or nu-fission rate in each FSR and stores them in an array indexed by FSR ID. More...

void	printInputParamsSummary ()
	A function that prints a summary of the input parameters.

void	tallyScalarFlux (segment curr_segment, int azim_index, int polar_index, FP_PRECISION fsr_flux, float *track_flux)
	Computes the contribution to the FSR scalar flux from a Track segment. More...

void	accumulateScalarFluxContribution (long fsr_id, FP_PRECISION weight, FP_PRECISION *fsr_flux)
	Move the segment(s)' contributions to the scalar flux from the buffer to the global scalar flux array. More...

void	tallyCurrent (segment curr_segment, int azim_index, int polar_index, float track_flux, bool fwd)
	Tallies the current contribution from this segment across the the appropriate CMFD mesh cell surface. More...

void	transferBoundaryFlux (Track track, int azim_index, int polar_index, bool direction, float track_flux)
	Updates the boundary flux for a Track given boundary conditions. More...

void	getFluxes (FP_PRECISION *out_fluxes, int num_fluxes)
	Fills an array with the scalar fluxes. More...

void	initializeFixedSources ()
	Populates array of fixed sources assigned by FSR.

void	printFSRFluxes (std::vector< double > dim1, std::vector< double > dim2, double offset, const char *plane)
	A function that prints the source region fluxes on a 2D mesh grid. More...

void	printFluxesTemp ()
	A function that prints fsr fluxes in xy plane at z=middle. More...

void	printNegativeSources (int iteration, int num_x, int num_y, int num_z)
	A function that prints the number of FSRs with negative sources in the whole geometry subdivided by a 3D lattice. The number of negative sources per energy group is also printed out. More...

Public Member Functions inherited from Solver
	Solver (TrackGenerator *track_generator=NULL)
	Constructor initializes an empty Solver class with array pointers set to NULL. More...

virtual	~Solver ()
	Destructor deletes arrays of boundary angular fluxes, scalar fluxes and sources for each FSR and energy group. More...

void	setGeometry (Geometry *geometry)
	Sets the Geometry for the Solver. More...

Geometry *	getGeometry ()
	Returns a pointer to the Geometry. More...

TrackGenerator *	getTrackGenerator ()
	Returns a pointer to the TrackGenerator. More...

FP_PRECISION	getFSRVolume (long fsr_id)
	Returns the calculated volume for a flat source region. More...

int	getNumPolarAngles ()
	Returns the number of angles used for the polar quadrature. More...

int	getNumIterations ()
	Returns the number of source iterations to converge the source. More...

double	getTotalTime ()
	Returns the total time to converge the source (seconds). More...

double	getKeff ()
	Returns the converged eigenvalue $k_{eff}$ . More...

double	getConvergenceThreshold ()
	Returns the threshold for source/flux convergence. More...

FP_PRECISION	getMaxOpticalLength ()
	Get the maximum allowable optical length for a track segment. More...

bool	isUsingDoublePrecision ()
	Returns whether the solver is using double floating point precision. More...

bool	isUsingExponentialInterpolation ()
	Returns whether the Solver uses linear interpolation to compute exponentials. More...

bool	is3D ()
	Returns whether the Solver is tackling a 3D problem. More...

void	initializeSolver (solverMode solver_mode)
	Initializes most components of Solver. Mostly needed from the Python side.

void	printFissionRates (std::string fname, int nx, int ny, int nz)
	Prints fission rates to a binary file. More...

void	setKeffFromNeutronBalance ()
	Sets computation method of k-eff from fission, absorption, and leakage rates rather than from fission rates. keff = fission/(absorption + leakage)

void	setResidualByReference (std::string fname)
	Sets residuals to be computed a error relative to a reference. More...

void	dumpFSRFluxes (std::string fname)
	Prints scalar fluxes to a binary file. More...

void	loadInitialFSRFluxes (std::string fname)
	Load the initial scalar flux distribution from a binary file. More...

void	loadFSRFluxes (std::string fname, bool assign_k_eff=false, double tolerance=0.01)
	Load scalar fluxes from a binary file. More...

double	getFlux (long fsr_id, int group)
	Returns the scalar flux for some FSR and energy group. More...

double	getFSRSource (long fsr_id, int group)
	Returns the source for some energy group for a flat source region. More...

void	setTrackGenerator (TrackGenerator *track_generator)
	Sets the Solver's TrackGenerator with characteristic Tracks. More...

void	setConvergenceThreshold (double threshold)
	Sets the threshold for source/flux convergence. More...

virtual void	setFixedSourceByFSR (long fsr_id, int group, double source)
	Assign a fixed source for a flat source region and energy group. More...

void	setFixedSourceByCell (Cell *cell, int group, double source)
	Assign a fixed source for a Cell and energy group. More...

void	setFixedSourceByMaterial (Material *material, int group, double source)
	Assign a fixed source for a Material and energy group. More...

void	setMaxOpticalLength (FP_PRECISION max_optical_length)
	Set the maximum allowable optical length for a track segment. More...

void	setExpPrecision (double precision)
	Set the precision, or maximum allowable approximation error, of the the exponential interpolation table. More...

void	useExponentialInterpolation ()
	Informs the Solver to use linear interpolation to compute the exponential in the transport equation.

void	useExponentialIntrinsic ()
	Informs the Solver to use the exponential intrinsic exp(...) function to compute the exponential in the transport equation.

void	setSolverMode (solverMode solver_mode)
	Choose between direct and adjoint mode. More...

void	setRestartStatus (bool is_restart)
	Informs the Solver that this is a 'restart' calculation and therefore k_eff, track angular and region scalar fluxes should not be reset. More...

void	allowNegativeFluxes (bool negative_fluxes_on)
	Informs the Solver that this calculation may involve negative fluxes for computing higher eigenmodes for example. More...

void	correctXS ()
	Directs OpenMOC to correct unphysical cross-sections. More...

void	stabilizeTransport (double stabilization_factor, stabilizationType stabilization_type=DIAGONAL)
	Directs OpenMOC to use the diagonal stabilizing correction to the source iteration transport sweep. More...

void	setInitialSpectrumCalculation (double threshold)
	Instructs OpenMOC to perform an initial spectrum calculation. More...

void	setCheckXSLogLevel (logLevel log_level)
	Determines which log level to set cross-section warnings. More...

void	setChiSpectrumMaterial (Material *material)
	Sets the chi spectrum for use as an inital flux guess. More...

void	resetMaterials (solverMode mode)
	Returns the Material data to its original state. More...

void	fissionTransportSweep ()
	This method performs one transport sweep using the fission source. More...

void	scatterTransportSweep ()
	This method performs one transport sweep using the scatter source. More...

void	computeFlux (int max_iters=1000, bool only_fixed_source=true)
	Computes the scalar flux distribution by performing a series of transport sweeps. More...

void	computeSource (int max_iters=1000, double k_eff=1.0, residualType res_type=TOTAL_SOURCE)
	Computes the total source distribution by performing a series of transport sweep and source updates. More...

void	computeEigenvalue (int max_iters=1000, residualType res_type=FISSION_SOURCE)
	Computes keff by performing a series of transport sweep and source updates. More...

void	printBGQMemory ()

float *	getBoundaryFlux (long track_id, bool fwd)
	Returns the boundary flux array at the requested indexes. More...

void	setVerboseIterationReport ()
	Sets the solver to print extra information for each iteration.

void	printTimerReport ()
	Prints a report of the timing statistics to the console.

FP_PRECISION *	getFluxesArray ()
	A function that returns the array of scalar fluxes. More...

void	limitXS ()
	Limits cross-sections so that there are no negative cross-sections. More...

void	setLimitingXSMaterials (std::vector< int > material_ids, int reset_iteration)
	Instructs MOC to limit negative cross-sections for early iterations. More...

void	checkLimitXS (int iteration)
	Checks to see if limited XS should be reset. More...

void	setOTFTransport ()
	Activate On-The-Fly transport, to OTF ray-trace and propagate the track angular fluxes at the same time.

Protected Member Functions
void	initializeExpEvaluator ()
	Allocates memory for the exponential linear interpolation table.

void	initializeFluxArrays ()
	Allocates memory for Track boundary angular flux and leakage and FSR scalar flux arrays. More...

void	initializeSourceArrays ()
	Allocates memory for FSR source arrays. More...

void	normalizeFluxes ()
	Normalizes all FSR scalar fluxes and Track boundary angular fluxes to the total fission source (times $\nu$ ).

void	computeFSRSources (int iteration)
	Computes the total source (fission, scattering, fixed) in each FSR. More...

void	addSourceToScalarFlux ()
	Add the source term contribution in the transport equation to the FSR scalar flux.

void	computeKeff ()
	Compute $k_{eff}$ from the total, fission and scattering reaction rates and leakage. More...

void	tallyScalarFlux (segment curr_segment, int azim_index, FP_PRECISION track_flux, FP_PRECISION *fsr_flux)
	Computes the contribution to the FSR scalar flux from a segment. More...

void	transferBoundaryFlux (int track_id, int azim_index, bool direction, FP_PRECISION *track_flux)
	Updates the boundary flux for a Track given boundary conditions. More...

void	computeExponentials (segment curr_segment, FP_PRECISION exponentials)
	Computes an array of the exponentials in the transport equation, $exp(-\frac{\Sigma_t * l}{sin(\theta)})$ , for each energy group and polar angle for a given Track segment. More...

Protected Member Functions inherited from CPUSolver
virtual void	initializeFSRs ()
	Initializes the FSR volumes and Materials array. More...

void	zeroTrackFluxes ()
	Zero each Track's boundary fluxes for each energy group (and polar angle in 2D) in the "forward" and "reverse" directions.

void	copyBoundaryFluxes ()
	Copies values from the start flux into the boundary flux array for both the "forward" and "reverse" directions.

void	tallyStartingCurrents ()
	Computes the total current impingent on boundary CMFD cells from starting angular fluxes.

virtual void	flattenFSRFluxes (FP_PRECISION value)
	Set the scalar flux for each FSR and energy group to some value. More...

void	flattenFSRFluxesChiSpectrum ()
	Set the scalar flux for each FSR to a chi spectrum.

void	storeFSRFluxes ()
	Stores the FSR scalar fluxes in the old scalar flux array.

void	computeFSRFissionSources ()
	Computes the total fission source in each FSR. More...

void	computeFSRScatterSources ()
	Computes the total scattering source in each FSR. More...

void	transportSweep ()
	This method performs one transport sweep of all azimuthal angles, Tracks, Track segments, polar angles and energy groups. More...

virtual void	computeStabilizingFlux ()
	Computes the stabilizing flux for transport stabilization.

virtual void	stabilizeFlux ()
	Adjusts the scalar flux for transport stabilization.

double	computeResidual (residualType res_type)
	Computes the residual between source/flux iterations. More...

Protected Member Functions inherited from Solver
virtual void	initializeExpEvaluators ()
	Initializes new ExpEvaluator object to compute exponentials.

void	initializeMaterials (solverMode mode)
	Initializes the Material's production matrices. More...

void	countFissionableFSRs ()
	Counts the number of fissionable flat source regions. More...

void	checkXS ()
	All material cross-sections in the geometry are checked for consistency. More...

virtual void	initializeCmfd ()
	Initializes a Cmfd object for acceleration prior to source iteration. More...

void	calculateInitialSpectrum (double threshold)
	Performs a spectrum calculation to update the scalar fluxes. More...

void	clearTimerSplits ()
	Deletes the Timer's timing entries for each timed code section code in the source convergence loop. More...

Protected Attributes
int	_num_vector_lengths

FP_PRECISION *	_delta_psi

FP_PRECISION *	_thread_taus

FP_PRECISION *	_thread_exponentials

Protected Attributes inherited from CPUSolver
int	_num_threads

omp_lock_t *	_FSR_locks

Protected Attributes inherited from Solver
int	_num_azim

int	_num_groups

long	_num_FSRs

long	_num_fissionable_FSRs

FP_PRECISION *	_FSR_volumes

Material **	_FSR_materials

Material *	_chi_spectrum_material

TrackGenerator *	_track_generator

Geometry *	_geometry

int	_num_materials

Quadrature *	_quad

int	_num_polar

int	_fluxes_per_track

Track **	_tracks

std::vector< int >	_limit_xs_materials

int	_reset_iteration

bool	_limit_xs

std::map< int, Material * >	_original_materials

std::map< int, Material * >	_limit_materials

int ***	_tracks_per_stack

bool	_SOLVE_3D

solverMode	_solver_mode

bool	_is_restart

bool	_user_fluxes

bool	_fixed_sources_on

bool	_correct_xs

bool	_stabilize_transport

bool	_verbose

bool	_calculate_initial_spectrum

double	_initial_spectrum_thresh

bool	_load_initial_FSR_fluxes

bool	_calculate_residuals_by_reference

bool	_negative_fluxes_allowed

std::string	_initial_FSR_fluxes_file

std::string	_reference_file

logLevel	_xs_log_level

segmentationType	_segment_formation

long	_tot_num_tracks

float *	_boundary_flux

float *	_start_flux

float *	_boundary_leakage

FP_PRECISION *	_scalar_flux

FP_PRECISION *	_old_scalar_flux

FP_PRECISION *	_reference_flux

FP_PRECISION *	_stabilizing_flux

FP_PRECISION *	_fixed_sources

std::vector< FP_PRECISION * >	_groupwise_scratch

double *	_regionwise_scratch

std::map< std::pair< int, int >, FP_PRECISION >	_fix_src_FSR_map

std::map< std::pair< Cell *, int >, FP_PRECISION >	_fix_src_cell_map

std::map< std::pair< Material *, int >, FP_PRECISION >	_fix_src_material_map

FP_PRECISION *	_reduced_sources

double	_k_eff

bool	_keff_from_fission_rates

int	_num_iterations

double	_converge_thresh

double	_stabilization_factor

stabilizationType	_stabilization_type

ExpEvaluator ***	_exp_evaluators

int	_num_exp_evaluators_azim

int	_num_exp_evaluators_polar

Timer *	_timer

Cmfd *	_cmfd

std::string	_source_type

bool	_OTF_transport

Detailed Description

This is a subclass of the CPUSolver class which uses memory-aligned data structures and Intel's auto-vectorization.

Note: This class is only compiled if the Intel compiler is used when building OpenMOC. If building OpenMOC with the "--with-icpc" flag, then this class will be available in the "openmoc.intel.single" or "openmoc.intel.double" Python module.

Constructor & Destructor Documentation

◆ VectorizedSolver()

VectorizedSolver::VectorizedSolver ( TrackGenerator * track_generator = NULL )

Constructor initializes NULL arrays for source, flux, etc.

Parameters

track_generator an optional pointer to a TrackGenerator object

Member Function Documentation

◆ computeExponentials()

void VectorizedSolver::computeExponentials	(	segment *	curr_segment,
		FP_PRECISION *	exponentials
	)

protected

Computes an array of the exponentials in the transport equation, $exp(-\frac{\Sigma_t * l}{sin(\theta)})$ , for each energy group and polar angle for a given Track segment.

Parameters

curr_segment	pointer to the Track segment of interest
exponentials	the array to store the exponential values

◆ computeFSRSources()

void VectorizedSolver::computeFSRSources ( int iteration )

protectedvirtual

Computes the total source (fission, scattering, fixed) in each FSR.

This method computes the total source in each FSR based on this iteration's current approximation to the scalar flux.

Reimplemented from CPUSolver.

◆ computeKeff()

void VectorizedSolver::computeKeff ( )

protectedvirtual

Compute $k_{eff}$ from the total, fission and scattering reaction rates and leakage.

This method computes the current approximation to the multiplication factor on this iteration as follows: $k_{eff} = \frac{\displaystyle\sum_{i \in I} \displaystyle\sum_{g \in G} \nu \Sigma^F_g \Phi V_{i}} {\displaystyle\sum_{i \in I} \displaystyle\sum_{g \in G} (\Sigma^T_g \Phi V_{i} - \Sigma^S_g \Phi V_{i} - L_{i,g})}$

Reduce leakage array across tracks, energy groups, polar angles

Reimplemented from CPUSolver.

◆ getNumVectorWidths()

int VectorizedSolver::getNumVectorWidths ( )

Returns the number of vector lengths required to fit the number of energy groups.

If the number of energy groups is 35 and the vector width is 4, this method will return 9 since 9*4 = 36 is the nearest integer greater than or equal to 35.

Returns: The number of vector widths

◆ initializeFluxArrays()

void VectorizedSolver::initializeFluxArrays ( )

protectedvirtual

Allocates memory for Track boundary angular flux and leakage and FSR scalar flux arrays.

Deletes memory for old flux arrays if they were allocated for a previous simulation.

Reimplemented from CPUSolver.

◆ initializeSourceArrays()

void VectorizedSolver::initializeSourceArrays ( )

protectedvirtual

Allocates memory for FSR source arrays.

Deletes memory for old source arrays if they were allocated for a previous simulation.

Reimplemented from CPUSolver.

◆ setFixedSourceByFSR()

void VectorizedSolver::setFixedSourceByFSR	(	int	fsr_id,
		int	group,
		FP_PRECISION	source
	)

Assign a fixed source for a flat source region and energy group.

Fixed sources should be scaled to reflect the fact that OpenMOC normalizes the scalar flux such that the total energy- and volume-integrated production rate sums to 1.0.

Parameters

fsr_id	the flat source region ID
group	the energy group
source	the volume-averaged source in this group

◆ setGeometry()

void VectorizedSolver::setGeometry ( Geometry * geometry )

Sets the Geometry for the Solver and aligns all Material cross-section data for SIMD vector instructions.

Parameters

geometry a pointer to the Geometry

◆ tallyScalarFlux()

void VectorizedSolver::tallyScalarFlux	(	segment *	curr_segment,
		int	azim_index,
		FP_PRECISION *	track_flux,
		FP_PRECISION *	fsr_flux
	)

protected

Computes the contribution to the FSR scalar flux from a segment.

This method integrates the angular flux for a Track segment across energy groups and polar angles, and tallies it into the FSR scalar flux, and updates the Track's angular flux.

Parameters

curr_segment	a pointer to the Track segment of interest
azim_index	a pointer to the azimuthal angle index for this segment
track_flux	a pointer to the Track's angular flux
fsr_flux	a pointer to the temporary FSR flux buffer

◆ transferBoundaryFlux()

void VectorizedSolver::transferBoundaryFlux	(	int	track_id,
		int	azim_index,
		bool	direction,
		FP_PRECISION *	track_flux
	)

protected

Updates the boundary flux for a Track given boundary conditions.

For reflective boundary conditions, the outgoing boundary flux for the Track is given to the reflecting Track. For vacuum boundary conditions, the outgoing flux tallied as leakage.

Parameters

track_id	the ID number for the Track of interest
azim_index	a pointer to the azimuthal angle index for this segment
direction	the Track direction (forward - true, reverse - false)
track_flux	a pointer to the Track's outgoing angular flux

Member Data Documentation

◆ _delta_psi

FP_PRECISION* VectorizedSolver::_delta_psi

protected

The change in angular flux along a track segment for each energy group

◆ _num_vector_lengths

int VectorizedSolver::_num_vector_lengths

protected

Number of energy groups divided by vector widths (VEC_LENGTH)

◆ _thread_exponentials

FP_PRECISION* VectorizedSolver::_thread_exponentials

protected

An array for the exponential terms in the transport equation for * each thread in each energy group and polar angle

◆ _thread_taus

FP_PRECISION* VectorizedSolver::_thread_taus

protected

An array for the optical length for each thread in each energy group

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

/home/guillaume/Research/OpenMOC/src/VectorizedSolver.h
/home/guillaume/Research/OpenMOC/src/VectorizedSolver.cpp

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ VectorizedSolver()

Member Function Documentation

◆ computeExponentials()

◆ computeFSRSources()

◆ computeKeff()

◆ getNumVectorWidths()

◆ initializeFluxArrays()

◆ initializeSourceArrays()

◆ setFixedSourceByFSR()

◆ setGeometry()

◆ tallyScalarFlux()

◆ transferBoundaryFlux()

Member Data Documentation

◆ _delta_psi

◆ _num_vector_lengths

◆ _thread_exponentials

◆ _thread_taus