linalg.utilities#

Functions#

`array_to_psydac`(x, V)	Convert a NumPy array to a Vector of the space V.
`petsc_to_psydac`(x, Xh[, out])	Convert a PETSc.Vec object to a StencilVector or BlockVector.

Details#

array_to_psydac(x, V)[source]#

Convert a NumPy array to a Vector of the space V. This function is designed to be the inverse of the method .toarray() of the class Vector.

Parameters:

xnumpy.ndarray: Array to be converted. It only contains the true data, the ghost regions must not be included.
Vpsydac.linalg.stencil.StencilVectorSpace or psydac.linalg.block.BlockVectorSpace: Space of the final PSYDAC Vector.

Returns:

upsydac.linalg.stencil.StencilVector or psydac.linalg.block.BlockVector: Element of space V, the coefficients of which (excluding ghost regions) are the entries of x. The ghost regions of u are up to date.

Notes

This function works in parallel but it is very costly and should be avoided if performance is a priority.

petsc_to_psydac(x, Xh, out=None)[source]#

Convert a PETSc.Vec object to a StencilVector or BlockVector. It assumes that PETSc was installed with the configuration for complex numbers. It uses the index conversion functions in psydac.linalg.topetsc.

Parameters:

xPETSc.Vec: PETSc vector
Xhpsydac.linalg.stencil.StencilVectorSpace | psydac.linalg.block.BlockVectorSpace: Space of the coefficients of the PSYDAC vector.
outpsydac.linalg.stencil.StencilVector | psydac.linalg.block.BlockVector, optional: The PSYDAC vector where to store the result.

Returns:

upsydac.linalg.stencil.StencilVector | psydac.linalg.block.BlockVector: PSYDAC vector. In the case of a BlockVector, the blocks must be of type StencilVector. The general case is not yet implemented.