functions

Utility functions that can be called from host or device. More...

Functions
template<typename T >
	__attribute__ ((always_inline)) void removeDuplicateVectorElements(vector< T > &data)
	remove duplicate elements from a vector, preserving the order, using sets More...
newData	resize (n-1)
ArrayHandle< T >	h1 (newData, access_location::host, access_mode::overwrite)
	for (int i=0;i< n;++i) = h.data[i]
newData	resize (n-indices.size())
newData	resize (n+extraElements)
ArrayHandle< T >	hnew (newData, access_location::host, access_mode::overwrite)
data	swap (newData)
	if (Nvector > Narray) copydata.resize(Nvector)
ArrayHandle< T >	handle (copydata, access_location::host, access_mode::overwrite)
ArrayHandle< T >	handle (data, access_location::host, access_mode::read)
copydata	resize (n)
HOSTDEVICE scalar	dot (const dVec &p1, const dVec &p2)
	The dot product between d-Dimensional vectors. More...
HOSTDEVICE int	idPow (int i)
	an integer to the dth power... the slow way More...
HOSTDEVICE int	dot (const iVec &p1, const iVec &p2)
	The dot product between d-Dimensional iVecs. More...
HOSTDEVICE scalar	dot (const scalar3 &p1, const scalar3 &p2)
	The dot product between scalar3's. More...
HOSTDEVICE scalar	norm (const scalar3 &p)
	The norm of a scalar3's. More...
HOSTDEVICE scalar	r3Distance (const scalar3 &p1, const scalar3 &p2)
	The dot product between scalar3's. More...
HOSTDEVICE scalar3	cross (const scalar3 &p1, scalar3 &p2)
	The norm of a scalar3's. More...
HOSTDEVICE scalar	norm (const dVec &p)
	The norm of a d-Dimensional vector. More...
HOSTDEVICE scalar	pointSegmentDistance (const scalar3 &p, const scalar3 &s1, const scalar3 &s2, scalar3 &dir)
	point-segment distance in 3D More...
HOSTDEVICE scalar	truncatedPointSegmentDistance (const scalar3 &p, const scalar3 &s1, const scalar3 &s2, scalar3 &dir)
	point-segment distance in 3D More...
HOSTDEVICE int	wrap (int x, int m)
	fit integers into non-negative domains More...
HOSTDEVICE int3	wrap (int3 x, int3 m)
	fit integers into non-negative domains More...
HOSTDEVICE bool	ordered (int a, int b, int c)
HOSTDEVICE int	computeSign (scalar x)
	compute the sign of a scalar, and return zero if x = 0 More...
HOSTDEVICE scalar	computeSignNoCast (scalar x)
	compute the sign of a scalar, and return zero if x = 0...but return a scalar to avoid expensive casts on the GPU More...
__host__ bool	chooseGPU (int USE_GPU, bool verbose=false)
	Get basic stats about the chosen GPU (if it exists) More...
__host__ bool	getAvailableGPUs (vector< string > &devices)
	Get basic stats about the chosen GPU (if it exists) More...
HOSTDEVICE void	qTensorFromDirector (scalar3 n, scalar S0, dVec &q)
	return a qtensor given a director and a value of S0 More...
HOSTDEVICE scalar	TrQ2 (dVec &q)
	Tr(Q^2) = Q_{kl}Q_{lk}. More...
HOSTDEVICE scalar	TrQ3 (dVec &q)
	Tr(Q^3) More...
HOSTDEVICE scalar	TrQ2Squared (dVec &q)
	(Tr(Q^2))^2 More...
HOSTDEVICE dVec	derivativeTrQ2 (dVec &q)
	derivative of Tr(Q^2) w/r/t q[0] .. q[4] More...
HOSTDEVICE dVec	derivativeTrQ3 (dVec &q)
	derivative of Tr(Q^3) w/r/t q[0] .. q[4] More...
HOSTDEVICE dVec	derivativeTrQ2Squared (dVec &q)
	derivative of (Tr(Q^2))^2 w/r/t q[0] .. q[4] More...
HOSTDEVICE dVec	allPhaseComponentForces (dVec &q, scalar &a, scalar &b, scalar &c)
	Phase components combined into one for computational efficiency. More...
HOSTDEVICE dVec	QjkQki (dVec &q)
	Q_{jk}Q_{ki}. More...
HOSTDEVICE scalar	determinantOfQ (dVec &q)
	determinant of a qt matrix More...
HOSTDEVICE scalar	eigFromVecs (vector< scalar > &eVec, scalar ev0, scalar ev1, scalar ev2)
	eVec is Q*e, ev0,ev1,ev2 are components ofe More...
HOSTDEVICE void	eigensystemOfQ (dVec &q, vector< scalar > &eVals, vector< scalar > &eVec1, vector< scalar > &eVec2, vector< scalar > &eVec3)
	get the eigensystem associated with a Q tensor More...
HOSTDEVICE void	eigenvaluesOfQ (dVec &q, scalar &a, scalar &b, scalar &c)
	Get the eigenvalues of a real symmetric traceless 3x3 matrix. More...

Variables
int	index
GPUArray< T >	newData
int	idx = 0
	data = newData
vector< int >	indices
int	n = data.getNumElements()
int	vectorIndex = 0
int	extraElements
GPUArray< T > &	copydata
int	Nvector = data.size()

Detailed Description

Utility functions that can be called from host or device.

Function Documentation

__attribute__()

template<typename T >

__attribute__ ( (always_inline)  ) &

inline

remove duplicate elements from a vector, preserving the order, using sets

copy a GPUarray to a vector

fill the first data.size() elements of a GPU array with elements of the data vector

grow a GPUArray, leaving the current elements the same but with extra capacity at the end of the array

shrink a GPUArray by removing the elements [i1,i2,...in] of a vector and shifting any elements j > i_i into place

shrink a GPUArray by removing the i'th element and shifting any elements j > i into place

References data.

resize() [1/4]

newData resize

(

n-

1

)

Referenced by kernelTuner::kernelTuner().

h1()

ArrayHandle< T > h1	(	newData	,
		access_location::host	,
		access_mode::overwrite
	)

Initial value:

{
    ArrayHandle<T> h(data,access_location::host,access_mode::read)

Referenced by for().

for()

for

(

)

= h.data[i]

References ArrayHandle< T >::data, h1(), idx, and index.

Referenced by Simulation::computePressureTensor().

resize() [2/4]

newData resize

(

n-indices.

size()

)

resize() [3/4]

newData resize

(

n+

extraElements

)

hnew()

ArrayHandle<T> hnew	(	newData	,
		access_location::host	,
		access_mode::overwrite
	)

swap()

data swap

(

newData

)

Referenced by GPUArray< curandState >::swap().

if()

if	(	Nvector	,
		Narray
	)

handle() [1/2]

ArrayHandle<T> handle	(	copydata	,
		access_location::host	,
		access_mode::overwrite
	)

handle() [2/2]

ArrayHandle<T> handle	(	data	,
		access_location::host	,
		access_mode::read
	)

resize() [4/4]

copydata resize

(

n

)

dot() [1/3]

HOSTDEVICE scalar dot	(	const dVec &	p1,
		const dVec &	p2
	)

The dot product between d-Dimensional vectors.

References scalar.

Referenced by energyMinimizerAdam::adamStepCPU(), baseLatticeForce::computeEnergyCPU(), simpleModel::computeInstantaneousTemperature(), simpleModel::computeKineticEnergy(), NISymmetricEigensolver3x3::Dot(), energyMinimizerFIRE::fireStepCPU(), gpu_compute_neighbor_list_kernel(), gpu_compute_neighbor_list_TPC_kernel(), gpu_compute_neighbor_list_TPP_kernel(), gpu_dot_dVec_vectors_kernel(), gpu_dVec_dot_products_kernel(), gpu_scalar_times_dVec_squared_kernel(), gpu_set_random_spins_kernel(), gpu_vec_dot_product_kernel(), energyMinimizerGradientDescent::gradientDescentCPU(), energyMinimizerNesterovAG::nesterovStepCPU(), norm(), IndexDD::operator()(), pointSegmentDistance(), poissonDiskSampling::sample_annulus_point(), cubicLattice::setSpinsRandomly(), simpleModel::setVelocitiesMaxwellBoltzmann(), and truncatedPointSegmentDistance().

idPow()

HOSTDEVICE int idPow

(

int

i

)

an integer to the dth power... the slow way

Referenced by hyperrectangularCellList::getCellNeighbors().

dot() [2/3]

HOSTDEVICE int dot	(	const iVec &	p1,
		const iVec &	p2
	)

The dot product between d-Dimensional iVecs.

References iVec::x.

dot() [3/3]

HOSTDEVICE scalar dot	(	const scalar3 &	p1,
		const scalar3 &	p2
	)

The dot product between scalar3's.

norm() [1/2]

HOSTDEVICE scalar norm

(

const scalar3 &

p

)

The norm of a scalar3's.

References dot().

Referenced by neighborList::computeCPU(), multirankSimulation::createCylindricalObject(), qTensorLatticeModel::createSimpleSpherialColloid(), multirankSimulation::createSphericalCavity(), multirankSimulation::createSphericalColloid(), multirankSimulation::createSpherocylinder(), gpu_update_spins_kernel(), cubicLattice::moveParticles(), NISymmetricEigensolver3x3::operator()(), poissonDiskSampling::poissonDiskSampling(), r3Distance(), and multirankSimulation::setDipolarField().

r3Distance()

HOSTDEVICE scalar r3Distance	(	const scalar3 &	p1,
		const scalar3 &	p2
	)

The dot product between scalar3's.

References norm(), and scalar3.

Referenced by pointSegmentDistance(), and truncatedPointSegmentDistance().

cross()

HOSTDEVICE scalar3 cross	(	const scalar3 &	p1,
		scalar3 &	p2
	)

The norm of a scalar3's.

References scalar3.

Referenced by NISymmetricEigensolver3x3::Cross().

norm() [2/2]

HOSTDEVICE scalar norm

(

const dVec &

p

)

The norm of a d-Dimensional vector.

References dot().

pointSegmentDistance()

HOSTDEVICE scalar pointSegmentDistance	(	const scalar3 &	p,
		const scalar3 &	s1,
		const scalar3 &	s2,
		scalar3 &	dir
	)

point-segment distance in 3D

References dot(), r3Distance(), scalar, and scalar3.

Referenced by multirankSimulation::createSpherocylinder().

truncatedPointSegmentDistance()

HOSTDEVICE scalar truncatedPointSegmentDistance	(	const scalar3 &	p,
		const scalar3 &	s1,
		const scalar3 &	s2,
		scalar3 &	dir
	)

point-segment distance in 3D

References dot(), r3Distance(), scalar, and scalar3.

Referenced by multirankSimulation::createCylindricalObject().

wrap() [1/2]

HOSTDEVICE int wrap	(	int	x,
		int	m
	)

fit integers into non-negative domains

Referenced by qTensorLatticeModel::createBoundaryFromFile(), multirankSimulation::createMultirankBoundaryObject(), qTensorLatticeModel::createSimpleSpherialColloid(), cubicLattice::getNeighbors(), gpu_get_six_neighbors(), gpu_lattice_spin_force_nn_kernel(), and wrap().

wrap() [2/2]

HOSTDEVICE int3 wrap	(	int3	x,
		int3	m
	)

fit integers into non-negative domains

References wrap().

ordered()

HOSTDEVICE bool ordered	(	int	a,
		int	b,
		int	c
	)

Referenced by multirankQTensorLatticeModel::positionToIndex().

computeSign()

HOSTDEVICE int computeSign

(

scalar

x

)

compute the sign of a scalar, and return zero if x = 0

computeSignNoCast()

HOSTDEVICE scalar computeSignNoCast

(

scalar

x

)

compute the sign of a scalar, and return zero if x = 0...but return a scalar to avoid expensive casts on the GPU

chooseGPU()

__host__ bool chooseGPU ( int  USE_GPU, bool  verbose = false  )

inline

Get basic stats about the chosen GPU (if it exists)

getAvailableGPUs()

__host__ bool getAvailableGPUs ( vector< string > &  devices )

inline

Get basic stats about the chosen GPU (if it exists)

qTensorFromDirector()

HOSTDEVICE void qTensorFromDirector	(	scalar3	n,
		scalar	S0,
		dVec &	q
	)

return a qtensor given a director and a value of S0

Referenced by multirankSimulation::createCylindricalObject(), qTensorLatticeModel::createSimpleSpherialColloid(), multirankSimulation::createSphericalCavity(), multirankSimulation::createSphericalColloid(), multirankSimulation::createSpherocylinder(), gpu_set_random_nematic_qTensors_kernel(), multirankSimulation::setDipolarField(), and qTensorLatticeModel::setNematicQTensorRandomly().

TrQ2()

HOSTDEVICE scalar TrQ2

(

dVec &

q

)

Tr(Q^2) = Q_{kl}Q_{lk}.

Referenced by computeBoundaryEnergy(), qTensorLatticeModel::computeDefectMeasures(), landauDeGennesLC::computeEnergyCPU(), gpu_degenerateEigenvalue_kernel(), gpu_energyPerSite_kernel(), and TrQ2Squared().

TrQ3()

HOSTDEVICE scalar TrQ3

(

dVec &

q

)

Tr(Q^3)

Referenced by landauDeGennesLC::computeEnergyCPU(), and gpu_energyPerSite_kernel().

TrQ2Squared()

HOSTDEVICE scalar TrQ2Squared

(

dVec &

q

)

(Tr(Q^2))^2

References scalar, and TrQ2().

Referenced by landauDeGennesLC::computeEnergyCPU(), and gpu_energyPerSite_kernel().

derivativeTrQ2()

HOSTDEVICE dVec derivativeTrQ2

(

dVec &

q

)

derivative of Tr(Q^2) w/r/t q[0] .. q[4]

derivative of Tr(Q^2) w/r/t q[0] .. q[5]

Referenced by landauDeGennesLC::computeAllDistortionTermsBoundaryCPU(), landauDeGennesLC::computeAllDistortionTermsBulkCPU(), landauDeGennesLC::computeL1BoundaryCPU(), landauDeGennesLC::computeL1BulkCPU(), and gpu_phase_force().

derivativeTrQ3()

HOSTDEVICE dVec derivativeTrQ3

(

dVec &

q

)

derivative of Tr(Q^3) w/r/t q[0] .. q[4]

derivative of Tr(Q^3) w/r/t q[0] .. q[5]

Referenced by landauDeGennesLC::computeAllDistortionTermsBoundaryCPU(), landauDeGennesLC::computeAllDistortionTermsBulkCPU(), landauDeGennesLC::computeL1BoundaryCPU(), landauDeGennesLC::computeL1BulkCPU(), and gpu_phase_force().

derivativeTrQ2Squared()

HOSTDEVICE dVec derivativeTrQ2Squared

(

dVec &

q

)

derivative of (Tr(Q^2))^2 w/r/t q[0] .. q[4]

References scalar.

Referenced by landauDeGennesLC::computeAllDistortionTermsBoundaryCPU(), landauDeGennesLC::computeAllDistortionTermsBulkCPU(), landauDeGennesLC::computeL1BoundaryCPU(), landauDeGennesLC::computeL1BulkCPU(), and gpu_phase_force().

allPhaseComponentForces()

HOSTDEVICE dVec allPhaseComponentForces	(	dVec &	q,
		scalar &	a,
		scalar &	b,
		scalar &	c
	)

Phase components combined into one for computational efficiency.

References scalar.

QjkQki()

HOSTDEVICE dVec QjkQki

(

dVec &

q

)

Q_{jk}Q_{ki}.

determinantOfQ()

HOSTDEVICE scalar determinantOfQ

(

dVec &

q

)

determinant of a qt matrix

Referenced by qTensorLatticeModel::computeDefectMeasures(), eigenvaluesOfQ(), gpu_computeDeterminant_kernel(), and gpu_degenerateEigenvalue_kernel().

eigFromVecs()

HOSTDEVICE scalar eigFromVecs	(	vector< scalar > &	eVec,
		scalar	ev0,
		scalar	ev1,
		scalar	ev2
	)

eVec is Q*e, ev0,ev1,ev2 are components ofe

References scalar.

eigensystemOfQ()

HOSTDEVICE void eigensystemOfQ	(	dVec &	q,
		vector< scalar > &	eVals,
		vector< scalar > &	eVec1,
		vector< scalar > &	eVec2,
		vector< scalar > &	eVec3
	)

get the eigensystem associated with a Q tensor

eigenvaluesOfQ()

HOSTDEVICE void eigenvaluesOfQ	(	dVec &	q,
		scalar &	a,
		scalar &	b,
		scalar &	c
	)

Get the eigenvalues of a real symmetric traceless 3x3 matrix.

References determinantOfQ(), PI, and scalar.

Referenced by qTensorLatticeModel::computeDefectMeasures(), qTensorLatticeModel::getAverageEigenvalues(), and gpu_largestEigenvalue_kernel().

Variable Documentation

index

int index

Initial value:

{
    int n = data.getNumElements()

Referenced by for(), getBufferInt3(), and multirankQTensorLatticeModel::getBufferInt3FromIndex().

newData

GPUArray< T > newData

idx

int idx = 0

Referenced by landauDeGennesLC::computeFirstDerivatives(), cubicLattice::cubicLattice(), for(), getBufferInt3(), multirankQTensorLatticeModel::getBufferInt3FromIndex(), gpu_adam_step_kernel(), gpu_compute_cell_list_kernel(), gpu_compute_neighbor_list_kernel(), gpu_computeDeterminant_kernel(), gpu_copy_boundary_object_kernel(), gpu_copy_gpuarray_kernel(), gpu_copyReceivingBuffer_kernel(), gpu_degenerateEigenvalue_kernel(), gpu_displacement_vv_kernel(), gpu_dot_dVec_vectors_kernel(), gpu_dVec_plusEqual_dVec_kernel(), gpu_dVec_times_scalar_kernel(), gpu_energyPerSite_kernel(), gpu_largestEigenvalue_kernel(), gpu_lattice_spin_force_nn_kernel(), gpu_move_boundary_object_kernel(), gpu_move_particles_kernel(), gpu_nesterovAG_step_kernel(), gpu_prepareSendingBuffer_kernel(), gpu_qTensor_computeBoundaryForcesGPU_kernel(), gpu_qTensor_computeObjectForceFromStresses_kernel(), gpu_qTensor_multiConstantForce_kernel(), gpu_qTensor_oneConstantForce_kernel(), gpu_qTensor_uniformFieldForcekernel(), gpu_scalar_times_dVec_squared_kernel(), gpu_set_array_kernel(), gpu_set_random_nematic_qTensors_kernel(), gpu_set_random_spins_kernel(), gpu_update_qTensor_simple_kernel(), gpu_update_spins_kernel(), gpu_update_spins_simple_kernel(), gpu_update_velocity_FIRE_kernel(), gpu_update_velocity_kernel(), gpu_vec_dot_product_kernel(), gpu_vec_dot_product_unrolled_kernel(), multirankQTensorLatticeModel::indexToPosition(), initialize_RNG_array_kernel(), and multirankSimulation::saveState().

data

data = newData

Referenced by __attribute__(), and multirankSimulation::sumUpdaterData().

indices

vector<int> indices

Initial value:

{
    std::sort(indices.begin(),indices.end())

Referenced by neighborList::computeCPU(), neighborList::computeGPU(), gpu_compute_neighbor_list(), gpu_compute_neighbor_list_kernel(), gpu_compute_neighbor_list_TPC_kernel(), and gpu_compute_neighbor_list_TPP_kernel().

n

int n = data.getNumElements()

Referenced by landauDeGennesLC::computeStressTensors(), qTensorLatticeModel::getAverageEigenvalues(), getNumBlocksAndThreads(), gpu_dot_dVec_vectors_kernel(), gpu_dVec_plusEqual_dVec_kernel(), gpu_dVec_times_scalar_kernel(), gpu_scalar_times_dVec_squared_kernel(), gpu_set_random_nematic_qTensors_kernel(), gpuReduction(), simpleModel::initializeSimpleModel(), reduce6(), qTensorLatticeModel::setNematicQTensorRandomly(), force::setNThreads(), updater::setNThreads(), Simulation::setNThreads(), simpleModel::setNThreads(), and simpleModel::simpleModel().

vectorIndex

int vectorIndex = 0

extraElements

int extraElements

Initial value:

{
    int n = data.getNumElements()

copydata

vector< T > & copydata

Initial value:

{
    int Narray = copydata.getNumElements()

Nvector

int Nvector = data.size()