Mat - Low-level Interface

This page documents the low-level, automatically wrapped PETSc Mat (matrix) functions available through LibPETSc. These functions provide direct access to the PETSc C API.

For the high-level Julia interface, see Mat.

Overview

The Mat interface includes:

Core Mat operations: Creation, manipulation, and mathematical operations on matrices (~632 functions in Mat_wrappers.jl)
Mat utilities: Additional matrix operations and conversions (~124 functions in Mataddons_wrappers.jl)

Usage

All low-level Mat functions require a petsclib parameter as the first argument:

using PETSc

# Get the library instance
petsclib = PETSc.petsclibs[1]
PETSc.initialize(petsclib)

# Get PETSc types
PetscInt = petsclib.PetscInt
PetscScalar = petsclib.PetscScalar

# Create a sparse matrix
mat = LibPETSc.MatCreate(petsclib, LibPETSc.PETSC_COMM_SELF)
LibPETSc.MatSetSizes(petsclib, mat, PetscInt(10), PetscInt(10), PetscInt(10), PetscInt(10))
LibPETSc.MatSetType(petsclib, mat, "seqaij")
LibPETSc.MatSetUp(petsclib, mat)

# Set values (0-based indexing!)
row = PetscInt[0]
cols = PetscInt[0, 1]
vals = PetscScalar[2.0, -1.0]
LibPETSc.MatSetValues(petsclib, mat, PetscInt(1), row, PetscInt(2), cols, vals, LibPETSc.INSERT_VALUES)

# Assemble
LibPETSc.MatAssemblyBegin(petsclib, mat, LibPETSc.MAT_FINAL_ASSEMBLY)
LibPETSc.MatAssemblyEnd(petsclib, mat, LibPETSc.MAT_FINAL_ASSEMBLY)

# Clean up
LibPETSc.MatDestroy(petsclib, mat)
PETSc.finalize(petsclib)

Function Reference

PETSc.LibPETSc.MatADot — Method

val::PetscScalar = MatADot(petsclib::PetscLibType, mat::AbstractPetscMat, x::AbstractPetscVec, y::AbstractPetscVec)

Computes the inner product with respect to a matrix, i.e., (x, y)_A = y^H A x where A is symmetric (Hermitian when using complex) positive definite.

Collective

Input Parameters:

mat - matrix used to define the inner product
x - first vector
y - second vector

Output Parameter:

val - the dot product with respect to A

Level: intermediate

See also: Mat, MatANorm(), VecDot(), VecNorm(), MatMult(), MatMultAdd(), MatMultTransposeAdd()

External Links

PETSc Manual: Mat/MatADot

source

PETSc.LibPETSc.MatAIJGetLocalMat — Method

A_loc::PetscMat = MatAIJGetLocalMat(petsclib::PetscLibType, A::AbstractPetscMat)

Creates a MATSEQAIJ from a MATAIJ matrix.

Not Collective

Input Parameter:

A - the matrix

Output Parameter:

A_loc - the local sequential matrix generated

Level: developer

See also: Mat, MatMPIAIJGetLocalMat()

External Links

PETSc Manual: Mat/MatAIJGetLocalMat

source

PETSc.LibPETSc.MatANorm — Method

val::PetscReal = MatANorm(petsclib::PetscLibType, mat::AbstractPetscMat, x::AbstractPetscVec)

Computes the norm with respect to a matrix, i.e., (x, x)_A^{1/2} = (x^H A x)^{1/2} where A is symmetric (Hermitian when using complex) positive definite.

Collective

Input Parameters:

mat - matrix used to define norm
x - the vector to compute the norm of

Output Parameter:

val - the norm with respect to A

Level: intermediate

See also: Mat, MatADot(), VecDot(), VecNorm(), MatMult(), MatMultAdd(), MatMultTransposeAdd()

External Links

PETSc Manual: Mat/MatANorm

source

PETSc.LibPETSc.MatAXPY — Method

MatAXPY(petsclib::PetscLibType, Y::AbstractPetscMat, a::PetscScalar, X::AbstractPetscMat, str::MatStructure)

Computes Y = a*X + Y.

Logically Collective

Input Parameters:

a - the scalar multiplier
X - the first matrix
Y - the second matrix
str - either SAME_NONZERO_PATTERN, DIFFERENT_NONZERO_PATTERN, UNKNOWN_NONZERO_PATTERN, or SUBSET_NONZERO_PATTERN (nonzeros of X is a subset of Y's)

Level: intermediate