Typedefs
typedef struct MboNumOp_t *	MboNumOp
	Data type for numerical operators. More...

Functions
MBO_EXPORT MBO_STATUS	mboNumOpCompile (MboTensorOp op, MboNumOp *numOp)
	Build a numerical operator from a tensor operator. More...

MBO_EXPORT void	mboNumOpDestroy (MboNumOp *op)
	Deallocate all resources associated with a MboNumOp. More...

MBO_EXPORT MboProdSpace	mboNumOpGetSpace (MboNumOp op)
	Returns a (non-owning) pointer to the space of an operator. More...

MBO_EXPORT MBO_STATUS	mboNumOpMatVec (struct MboAmplitude alpha, MboNumOp a, struct MboAmplitude x, struct MboAmplitude beta, struct MboAmplitude y)

MBO_EXPORT double	mboNumOpFlops (MboNumOp a)
	Returns an estimate for the number of floating point operations in a MatVec. More...

MBO_EXPORT void	mboNumOpDenseMatrix (MboNumOp a, struct MboAmplitude *mat)
	Dense matrix representation of operator. More...

MBO_EXPORT void	mboNumOpRowOffsets (MboNumOp op, MboGlobInd rmin, MboGlobInd rmax, MboGlobInd *i)
	Compute row offsets of sparse matrix. More...

MBO_EXPORT MBO_STATUS	mboNumOpSparseMatrix (MboNumOp op, MboGlobInd rmin, MboGlobInd rmax, MboGlobInd i, MboGlobInd j, struct MboAmplitude *a)
	Create a sparse matrix representation for a numerical operator. More...

MBO_EXPORT void	mboNumOpDiagonal (MboNumOp op, MboGlobInd rmin, MboGlobInd rmax, struct MboAmplitude *diag)
	Get diagonal entries from a numerical operator. More...

MBO_EXPORT void	mboNumOpDeleteDiagonal (MboNumOp op)
	Erase the diagonal from an operator. More...

Detailed Description

Typedef Documentation

typedef struct MboNumOp_t* MboNumOp

Data type for numerical operators.

MboNumOp objects can be created from MboTensorOps using mboNumOpCompile

See also: mboNumOpCompile, MboTensorOp

Function Documentation

MBO_EXPORT MBO_STATUS mboNumOpCompile	(	MboTensorOp	op,
		MboNumOp *	numOp
	)

Build a numerical operator from a tensor operator.

Parameters

op	Operator to compile.
numOp	On exit this pointer contains the compiled operator. The generated MboNumOp has to be destroyed with mboNumOpDestroy.

See also: MboNumOp, MboTensorOp, mboNumOpDestroy

MBO_EXPORT void mboNumOpDeleteDiagonal ( MboNumOp op )

Erase the diagonal from an operator.

Parameters

op	Operator from which to delete the diagonal.

See also: MboNumOp, mboTensorOpDiagonal

MBO_EXPORT void mboNumOpDenseMatrix	(	MboNumOp	a,
		struct MboAmplitude *	mat
	)

Dense matrix representation of operator.

This function fills the array mat with the dense matrix representation of the tensor operator. The matrix is written in row major order

Parameters

a	The tensor operator.
mat	User allocated array for the dense matrix. Has to be at least of length dim * dim, where dim is the dimension of the Hilbert space of the operator.

MBO_EXPORT void mboNumOpDestroy ( MboNumOp * op )

Deallocate all resources associated with a MboNumOp.

See also: mboNumOpCompile, MboNumOp

MBO_EXPORT void mboNumOpDiagonal	(	MboNumOp	op,
		MboGlobInd	rmin,
		MboGlobInd	rmax,
		struct MboAmplitude *	diag
	)

Get diagonal entries from a numerical operator.

Parameters

op	The operator.
rmin	First row for which to get the diagonal.
rmax	Last row for which to get the diagonal.
diag	The diagonal. Has to point to memory for at least rmax - rmin entries.

See also: MboNumOp, mboNumOpDeleteDiagonal

MBO_EXPORT double mboNumOpFlops ( MboNumOp a )

Returns an estimate for the number of floating point operations in a MatVec.

A complex multiply-add is counted as 8 floating point operations.

Parameters

a	The numerical operator for which the flops are to be computed.

MBO_EXPORT MboProdSpace mboNumOpGetSpace ( MboNumOp op )

Returns a (non-owning) pointer to the space of an operator.

Parameters

op	The operator.

Returns: The product space.

See also: MboProdSpace, MboNumOp

MBO_EXPORT MBO_STATUS mboNumOpMatVec	(	struct MboAmplitude	alpha,
		MboNumOp	a,
		struct MboAmplitude *	x,
		struct MboAmplitude	beta,
		struct MboAmplitude *	y
	)

y <- alpha * a * x + beta * y

MBO_EXPORT void mboNumOpRowOffsets	(	MboNumOp	op,
		MboGlobInd	rmin,
		MboGlobInd	rmax,
		MboGlobInd *	i
	)

Compute row offsets of sparse matrix.

This method computes the row offsets of the sparse matrix representing the tensor operator. This corresponds to the i array in aij or Yale sparse matrix format. We have i[0] = 0, i[1] = nnz[rmin], i[2] = nnz[rmin] + nnz[rmin + 1], etc. In particular, i[rmax - rmin] contains the total number of non-zeros in the row range [rmin, rmax). Note that this method, like mboNumOpSparseMatrix, counts duplicate entries separately. The i array is suitable for input to mboNumOpSparseMatrix.

Parameters

op	The tensor.
rmin	Start of row range for which to compute the offsets.
rmax	End of row range for which to compute the offsets.
i	On exit this array contains the row offsets. The offset of the first row in the range is set to 0. Thus, in order to obtain the global offsets one has to add the global offset of the first row to all offsets in i.

See also: mboNumOpSparseMatrix

MBO_EXPORT MBO_STATUS mboNumOpSparseMatrix	(	MboNumOp	op,
		MboGlobInd	rmin,
		MboGlobInd	rmax,
		MboGlobInd *	i,
		MboGlobInd *	j,
		struct MboAmplitude *	a
	)

Create a sparse matrix representation for a numerical operator.

After a call to mboNumOpRowOffsets and mboNumOpSparseMatrix the arrays i, j, and a contain the compressed sparse row representation of the matrix for the rows [rmin, rmax).

Parameters

op	Create sparse matrix for this operator.
rmin	Start of row range for which to compute the sparse matrix.
rmax	End of row range.
i	Offsets array computed by mboNumOpRowOffsets.
j	Column indices of nonzero entries.
a	Non-zero entries.

Returns: Error code.

See also: mboNumOpDenseMatrix, mboNumOpRowOffsets.

Typedefs

Functions

Detailed Description

Typedef Documentation

Function Documentation