VectorBTE Class Reference

Class used to store the "vector" of out-of-equilibrium populations. More...

#include <vector_bte.h>

Inheritance diagram for VectorBTE:

Collaboration diagram for VectorBTE:

Public Member Functions
	VectorBTE (StatisticsSweep &statisticsSweep_, BaseBandStructure &bandStructure_, const int &dimensionality_=3)
	Constructor method, initializes raw buffer data and saves helper variables. More...
	VectorBTE (const VectorBTE &that)
	Copy constructor.
VectorBTE &	operator= (const VectorBTE &that)
	Copy assignment operator.
double &	operator() (const int &iCalc, const int &iDim, const int &iState)
	Get and set operator.
const double &	operator() (const int &iCalc, const int &iDim, const int &iState) const
	Const get and set operator.
Eigen::MatrixXd	dot (const VectorBTE &that)
	Computes the scalar product between two VectorBTE objects. More...
VectorBTE	operator* (VectorBTE &that)
	element wise product between two VectorBTE objects x and y. More...
VectorBTE	operator* (const double &scalar)
	Computes the product of a VectorBTE with a scalar, i.e. More...
VectorBTE	operator* (const Eigen::MatrixXd &vector)
	Computes the product of a VectorBTE with a vector. More...
VectorBTE	operator* (ParallelMatrix< double > &matrix)
	Computes the product of a VectorBTE with a parallel matrix. More...
VectorBTE	operator+ (VectorBTE &that)
	element wise sum between two VectorBTE objects x and y. More...
VectorBTE	operator- (VectorBTE &that)
	element wise difference between two VectorBTE objects x and y. More...
VectorBTE	operator- ()
	Invert the sign of the VectorBTE content i.e. More...
VectorBTE	operator/ (VectorBTE &that)
	element wise division between two VectorBTE objects x and y. More...
VectorBTE	sqrt ()
	Return a new VectorBTE object containing the square root of this one (element-wise x -> sqrt(x) ).
VectorBTE	reciprocal () const
	Return a new VectorBTE object containing the reciprocal of this one (element-wise x -> 1/x).
void	canonical2Population ()
	Convert an out-of-equilibrium population from the canonical form f to the absolute value n, such that n = bose(bose+1)f or n=fermi(1-fermi)f.
void	population2Canonical ()
	Convert an out-of-equilibrium population from the absolute value n to the canonical value n, such that n = bose(bose+1)f or n=fermi(1-fermi)f.
void	outputToJSON (const std::string &outFileName, BaseBandStructure &outerBandStructure)
	Output the contents of a vector BTE object to a JSON file.
int	glob2Loc (const ChemPotIndex &imu, const TempIndex &it, const CartIndex &iDim) const
	glob2Loc and loc2Glob compress/decompress the indices on temperature, chemical potential, and cartesian direction into/from a single index. More...
std::tuple< ChemPotIndex, TempIndex, CartIndex >	loc2Glob (const int &i) const
void	setConst (const double &constant)
	Set the whole content (raw buffer) of BaseVectorBTE to a scalar value. More...
int	getNumStates ()
	return the number of states in this vector BTE object More...

Public Attributes
Eigen::MatrixXd	data
	raw buffer containing the values of the vector The matrix has size (numCalculations, numStates), where numCalculations is the number of pairs of temperature and chemical potentials, and numStates is the number of Bloch states used in the Boltzmann equation.
std::vector< int >	excludeIndices
	List of Bloch states to be excluded from the calculation (i.e. More...
int	dimensionality

Protected Member Functions
VectorBTE	baseOperator (VectorBTE &that, const int &operatorType)
	base class to implement +, -, / and * operations. More...

Protected Attributes
StatisticsSweep &	statisticsSweep
int	numCalculations
int	numStates
int	numChemPots
int	numTemps
BaseBandStructure &	bandStructure
const int	operatorSums = 0
const int	operatorDivs = 1
const int	operatorProd = 2
const int	operatorDiff = 3

Friends
class	ScatteringMatrix

Detailed Description

The vector indices are over the Bloch state. Additionally, there is one of these vector for each pair of (temperatures,chemicalPotentials). Can be used to store both scalar quantities (linewidths) or vectors (like velocities).

Constructor & Destructor Documentation

VectorBTE()

VectorBTE::VectorBTE	(	StatisticsSweep &	statisticsSweep_,
		BaseBandStructure &	bandStructure_,
		const int &	dimensionality_ = 3
	)

Parameters

statisticsSweep	saves the info on how many temperatures/chemical potentials we are evaluating.
bandStructure	saves the underlying band structure for which we compute the out-of-equilibrium populations.
dimensionality	determines the size of the vector on cartesian indices. 1 for scalar quantities like linewidths Gamma(BlochIndices), 3 for vector quantities like phonon populations f(blochIndices,cartesian).

Here is the call graph for this function:

Member Function Documentation

baseOperator()

VectorBTE VectorBTE::baseOperator ( VectorBTE &  that, const int &  operatorType  )

protected

It's split separately so that subclasses can create the correct output class, and also because operations are rather similar.

Here is the call graph for this function:

Here is the caller graph for this function:

dot()

Eigen::MatrixXd VectorBTE::dot

(

const VectorBTE &

that

)

The scalar product of x and y, is defined such as z(iCalc) = sum_i x(iCalc,i) y(iCalc,i), where i is an index over Bloch states, and iCalc is an index over temperatures and chemical potentials.

Parameters

that	the second vector used in the scalar product

Here is the call graph for this function:

Here is the caller graph for this function:

getNumStates()

int VectorBTE::getNumStates

(

)

Returns

numStates

glob2Loc()

int VectorBTE::glob2Loc	(	const ChemPotIndex &	imu,
		const TempIndex &	it,
		const CartIndex &	iDim
	)		const

TODO: these indices, and how they are used elsewhere, is rather messy That's because we have to work both with quantities such as line-widths, which are a scalar over the Bloch states, and phonon populations, which are cartesian vectors over the Bloch states. I should probably create two different classes for these.

Here is the caller graph for this function:

operator*() [1/4]

VectorBTE VectorBTE::operator*

(

const double &

scalar

)

all elements of vectorBTE x -> x * scalar.

Parameters

scalar

a double with the constant factor to be used in the element-wise multiplication.

operator*() [2/4]

VectorBTE VectorBTE::operator*

(

const Eigen::MatrixXd &

vector

)

The vector has size equal to the number of calculations (i.e. number of temperatures times the number of chemical potentials) used in the run. Given a calculation index iCalc, the result is an element-wise x(it)*vector(it).

Parameters

vector

a double vector to be used in the product, of size equal to numCalculations.

Here is the call graph for this function:

operator*() [3/4]

VectorBTE VectorBTE::operator*

(

ParallelMatrix< double > &

matrix

)

Only works if the number of temperatures/chemical potentials (numCalculations) is equal to one. At fixed calculation index iCalc, the result is an matrix-vector multiplication x(it,i)*pMatrix(i,j).

Parameters

pMatrix

a parallel distributed double matrix to be used in the product, of size equal to numStates x numStates.

Here is the call graph for this function:

operator*() [4/4]

VectorBTE VectorBTE::operator*

(

VectorBTE &

that

)

If the dimensionality of the two objects is the same, we compute element-wise result = x*y. If y has dimensionality 1, we compute x(every dim)*y(0), and the result has the dimensionality of x.

Parameters

that	the second VectorBTE object y, such that result = *this*y

Here is the call graph for this function:

operator+()

VectorBTE VectorBTE::operator+

(

VectorBTE &

that

)

If the dimensionality of the two objects is the same, we compute element-wise result = x+y. If y has dimensionality 1, we compute x(every dim)+y(0), and the result has the dimensionality of x.

Parameters

that	the second VectorBTE object y, such that result = *this+y

Here is the call graph for this function:

operator-() [1/2]

VectorBTE VectorBTE::operator-

(

)

x -> -x

operator-() [2/2]

VectorBTE VectorBTE::operator-

(

VectorBTE &

that

)

If the dimensionality of the two objects is the same, we compute element-wise result = x-y. If y has dimensionality 1, we compute x(every dim)-y(0), and the result has the dimensionality of x.

Parameters

that	the second VectorBTE object y, such that result = *this-y

Here is the call graph for this function:

operator/()

VectorBTE VectorBTE::operator/

(

VectorBTE &

that

)

If the dimensionality of the two objects is the same, we compute element-wise result = x/y. If y has dimensionality 1, we compute x(every dim)/y(0), and the result has the dimensionality of x.

Parameters

that	the second VectorBTE object y, such that result = *this/y

Here is the call graph for this function:

setConst()

void VectorBTE::setConst

(

const double &

constant

)

Parameters

constant

the value to be used in the set.

Here is the caller graph for this function:

Member Data Documentation

excludeIndices

std::vector<int> VectorBTE::excludeIndices

for which vectorBTE values are 0), for example, the acoustic modes at the gamma point, whose zero frequencies may cause problems.