TSQR::Trilinos::Randomizer< S, LO, GO, MV, Gen > Class Template Reference

Generates random test problems for TSQR. More...

#include <TsqrRandomizer.hpp>

Public Member Functions
virtual void	randomMultiVector (multivector_type &A, const magnitude_type singularValues[])
	Fill A with a (pseudo)random (distributed) matrix. More...

Protected Member Functions
void	init (const multivector_type &mv, const normalgen_ptr &pGen)

Detailed Description

template<class S, class LO, class GO, class MV, class Gen>
class TSQR::Trilinos::Randomizer< S, LO, GO, MV, Gen >

Generates random test problems for TSQR.

Randomizer knows how to fill in an instance of the given MultiVector class MV with a (pseudo)random test problem, using a generator of type Gen.

• S type of the matrix entries
• LO local ordinal type
• GO global ordinal type
• MV MultiVector type
• Gen normal(0,1) pseudorandom generator type

Definition at line 71 of file TsqrRandomizer.hpp.

Member Function Documentation

template<class S , class LO , class GO , class MV , class Gen >

virtual void TSQR::Trilinos::Randomizer< S, LO, GO, MV, Gen >::randomMultiVector ( multivector_type &  A, const magnitude_type  singularValues[]  )

inlinevirtual

Fill A with a (pseudo)random (distributed) matrix.

Fill the MultiVector A with a (pseudo)random (distributed) matrix with the given singular values. Given the same singular values and the same pseudorandom number sequence (produced by the Gen object), this function will always produce the same matrix, no matter the number of processors. It achieves this at the cost of scalability; only Proc 0 invokes the pseudorandom number generator.

Definition at line 101 of file TsqrRandomizer.hpp.

template<class S , class LO , class GO , class MV , class Gen >

void TSQR::Trilinos::Randomizer< S, LO, GO, MV, Gen >::init ( const multivector_type &  mv, const normalgen_ptr &  pGen  )

inlineprotected

Like the constructor, except you're not supposed to call the constructor of a pure virtual class.

Parameters

mv	[in] Only used to extract the underlying communication object (e.g., Epetra_Comm or Teuchos::Comm<int>).
pGen	[in/out] Pointer to generator of pseudorandom normal(0,1) sequence.

Definition at line 127 of file TsqrRandomizer.hpp.

---

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

• TsqrRandomizer.hpp