Make sure TimerType has the required interface. More...

Classes
class	Tsqr
	Parallel Tall Skinny QR (TSQR) factorization. More...
class	ApplyType
	NoTranspose, Transpose, or ConjugateTranspose. More...
class	CacheBlocker
class	CacheBlockingStrategy
class	Combine
	TSQR's six computational kernels. More...
class	CombineDefault
	Default implementation of TSQR::Combine. More...
class	CombineFortran
	Interface to Fortran 2003 back end of TSQR::Combine. More...
class	CombineFortran< Scalar, false >
class	CombineFortran< Scalar, true >
class	CombineNative
	Interface to C++ back end of TSQR::Combine. More...
class	CombineNative< Ordinal, Scalar, false >
class	CombineNative< Ordinal, Scalar, true >
class	DistTsqr
	Internode part of TSQR. More...
class	DistTsqrHelper
	Implementation of the internode part of TSQR. More...
class	DistTsqrRB
	Reduce-and-Broadcast (RB) version of DistTsqr. More...
class	Matrix
	A column-oriented dense matrix. More...
class	MatView
class	ConstMatView
class	MessengerBase
class	RMessenger
	Send, receive, and broadcast square R factors. More...
class	ScalarTraits
	Map from Scalar type to its arithmetic properties. More...
class	StatTimeMonitor
	Like Teuchos::TimeMonitor, but collects running stats. More...
class	TimeStats
	Collect running statistics. More...
class	TrivialTimer
	Satisfies TimerType concept trivially. More...
class	TwoLevelDistTsqr
	Interprocess part of TSQR. More...
class	ScalarPrinter
	Print a Scalar value to the given output stream. More...
Functions
template<class Ordinal , class Scalar >
std::pair< typename ScalarTraits< Scalar > ::magnitude_type, typename ScalarTraits< Scalar > ::magnitude_type >	local_verify (const Ordinal nrows, const Ordinal ncols, const Scalar const A, const Ordinal lda, const Scalar const Q, const Ordinal ldq, const Scalar *const R, const Ordinal ldr)
template<class MatrixViewType , class ConstMatrixViewType >
void	scatterStack (const ConstMatrixViewType &R_stack, MatrixViewType &R_local, const Teuchos::RCP< MessengerBase< typename MatrixViewType::scalar_type > > &messenger)
	Distribute a stack of R factors.

Detailed Description

Make sure TimerType has the required interface.

verifyTimerConcept Our TSQR benchmarks are templated on TimerType, in order to avoid depending on a particular timer implementation. TimerType should support the following three methods (modeled after Trilinos' Teuchos::Time class):

Construction using a const std::string& or something convertible to that (to name the timer). Semantically, the constructor should not start the timer.
start(bool reset=false) (returns nothing, starts the timer)
stop() (returns a double-precision floating-point value, which is the number of seconds elapsed since calling start())

TimerType need not be able to handle recursive calls; the model is that start() and stop() wrap some timing loop, and the loop does not reference the timer at all. We include this concept check in all of our TSQR benchmark routines via

 verifyTimerConcept< TimerType >();

If TimerType does not satisfy this interface, that line of code will fail to compile. The compiler should give an informative error message about a missing method.

Function Documentation

template<class Ordinal , class Scalar >

std::pair< typename ScalarTraits< Scalar >::magnitude_type, typename ScalarTraits< Scalar >::magnitude_type > TSQR::local_verify	(	const Ordinal	nrows,
		const Ordinal	ncols,
		const Scalar *const	A,
		const Ordinal	lda,
		const Scalar *const	Q,
		const Ordinal	ldq,
		const Scalar *const	R,
		const Ordinal	ldr
	)

Test accuracy of the computed QR factorization of the matrix A

Parameters:

nrows	[in] Number of rows in the A and Q matrices; nrows >= ncols >= 1
ncols	[in] Number of columns in the A, Q, and R matrices; nrows >= ncols >= 1
A	[in] Column-oriented nrows by ncols matrix with leading dimension lda
lda	[in] Leading dimension of the matrix A; lda >= nrows
Q	[in] Column-oriented nrows by ncols matrix with leading dimension ldq; computed Q factor of A
ldq	[in] Leading dimension of the matrix Q; ldq >= nrows
R	[in] Column-oriented upper triangular ncols by ncols matrix with leading dimension ldr; computed R factor of A
ldr	[in] Leading dimension of the matrix R; ldr >= ncols

Returns:: $\| A - Q R \|_F / \|A\|_F$ and $\| I - Q^* Q \|_F$ -- the relative residual of the QR factorization and a measure of the orthogonality of the resulting Q factor, both in the Frobenius (square root of (sum of squares of the matrix entries) norm.

Note:: The reason for the elaborate "magnitude_type" construction is because this function returns norms, and norms always have real-valued type. Scalar may be complex. We could simply set the imaginary part to zero, but it seems more sensible to enforce the norm's value property in the type system. Besides, one could imagine more elaborate Scalars (like rational functions, which do form a field) that have different plausible definitions of magnitude -- this is not just a problem for complex numbers (that are isomorphic to pairs of real numbers).

Definition at line 236 of file Tsqr_LocalVerify.hpp.

template<class MatrixViewType , class ConstMatrixViewType >

void TSQR::scatterStack	(	const ConstMatrixViewType &	R_stack,
		MatrixViewType &	R_local,
		const Teuchos::RCP< MessengerBase< typename MatrixViewType::scalar_type > > &	messenger
	)

Distribute a stack of R factors.

Parameters:

R_stack	[in] nprocs*ncols by ncols stack of square upper triangular matrices. The whole stack is stored in column-major order.
R_local	[out] ncols by ncols upper triangular matrix, stored in column-major order (in unpacked form).
messenger	[in/out] Object that handles communication

Definition at line 177 of file Tsqr_RMessenger.hpp.

Classes

Functions

Detailed Description

Function Documentation