SNES - Low-level Interface

This page documents the low-level, automatically wrapped PETSc SNES (Scalable Nonlinear Equations Solvers) functions available through LibPETSc. These functions provide direct access to the PETSc C API.

For the high-level Julia interface, see SNES.

Overview

The SNES interface includes:

Core SNES operations: Nonlinear solver creation, configuration, and solution (~283 functions in SNES_wrappers.jl)
SNESLineSearch: Line search methods for nonlinear solvers (~50 functions in SNESLineSearch_wrappers.jl)

Usage

All low-level SNES 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

# Create SNES solver
snes = LibPETSc.SNESCreate(petsclib, LibPETSc.PETSC_COMM_SELF)

# Set function and Jacobian (callbacks would need to be defined)
# LibPETSc.SNESSetFunction(petsclib, snes, r, compute_function, ctx)
# LibPETSc.SNESSetJacobian(petsclib, snes, J, J, compute_jacobian, ctx)

# Configure from command line
LibPETSc.SNESSetFromOptions(petsclib, snes)

# Solve
# LibPETSc.SNESSolve(petsclib, snes, C_NULL, x)

# Get convergence info
reason = LibPETSc.SNESGetConvergedReason(petsclib, snes)

iterations = LibPETSc.SNESGetIterationNumber(petsclib, snes)

println("SNES reason: $(reason), iterations: $(iterations)")

# Clean up
LibPETSc.SNESDestroy(petsclib, snes)
PETSc.finalize(petsclib)

Function Reference

PETSc.LibPETSc.SNESAddOptionsChecker — Method

SNESAddOptionsChecker(petsclib::PetscLibType, snescheck::external)

Adds an additional function to check for SNES options.

Not Collective

Input Parameter:

snescheck - function that checks for options

Calling sequence of snescheck:

snes - the SNES object for which it is checking options

Level: developer

See also: SNES, SNESSetFromOptions()

External Links

PETSc Manual: SNES/SNESAddOptionsChecker

source

PETSc.LibPETSc.SNESAppendOptionsPrefix — Method

SNESAppendOptionsPrefix(petsclib::PetscLibType, snes::AbstractSNES, prefix::String)

Appends to the prefix used for searching for all SNES options in the database.

Logically Collective

Input Parameters:

snes - the SNES context
prefix - the prefix to prepend to all option names

Level: advanced

See also: SNESGetOptionsPrefix(), SNESSetOptionsPrefix()

External Links

PETSc Manual: SNES/SNESAppendOptionsPrefix

source

PETSc.LibPETSc.SNESApplyNPC — Method

SNESApplyNPC(petsclib::PetscLibType, snes::AbstractSNES, x::AbstractPetscVec, f::AbstractPetscVec, y::AbstractPetscVec)

Calls SNESSolve() on the preconditioner for the SNES

Collective

Input Parameters:

snes - the SNES context
x - input vector
f - optional; the function evaluation on x

Output Parameter:

y - function vector, as set by SNESSetFunction()

Level: developer

See also: SNES, SNESGetNPC(), SNESSetNPC(), SNESComputeFunction()

External Links

PETSc Manual: SNES/SNESApplyNPC

source

PETSc.LibPETSc.SNESCompositeAddSNES — Method

SNESCompositeAddSNES(petsclib::PetscLibType, snes::AbstractSNES, type::String)

Adds another SNES to the SNESCOMPOSITE

Collective

Input Parameters:

snes - the SNES context of type SNESCOMPOSITE
type - the SNESType of the new solver

Level: developer

See also: SNES, SNESCOMPOSITE, SNESCompositeGetSNES()

External Links

PETSc Manual: SNES/SNESCompositeAddSNES

source

PETSc.LibPETSc.SNESCompositeGetNumber — Method

n::PetscInt = SNESCompositeGetNumber(petsclib::PetscLibType, snes::AbstractSNES)

Get the number of subsolvers in the SNESCOMPOSITE

Logically Collective

Input Parameter:

snes - the SNES context

Output Parameter:

n - the number of subsolvers

Level: developer

See also: SNES, SNESCOMPOSITE, SNESCompositeAddSNES(), SNESCompositeGetSNES()

External Links

PETSc Manual: SNES/SNESCompositeGetNumber

source

PETSc.LibPETSc.SNESCompositeGetSNES — Method

subsnes::SNES = SNESCompositeGetSNES(petsclib::PetscLibType, snes::AbstractSNES, n::PetscInt)

Gets one of the SNES objects in the SNES of SNESType SNESCOMPOSITE

Not Collective

Input Parameters:

snes - the SNES context
n - the number of the composed SNES requested

Output Parameter:

subsnes - the SNES requested

Level: developer

See also: SNES, SNESCOMPOSITE, SNESCompositeAddSNES(), SNESCompositeGetNumber()

External Links

PETSc Manual: SNES/SNESCompositeGetSNES

source

PETSc.LibPETSc.SNESCompositeSetDamping — Method

SNESCompositeSetDamping(petsclib::PetscLibType, snes::AbstractSNES, n::PetscInt, dmp::PetscReal)

Sets the damping of a subsolver when using SNES_COMPOSITE_ADDITIVE with a SNES of SNESType SNESCOMPOSITE

Not Collective

Input Parameters:

snes - the SNES context
n - the number of the sub-SNES object requested
dmp - the damping

Level: intermediate

See also: SNES, SNESCOMPOSITE, SNESCompositeAddSNES(), SNESCompositeGetSNES(), SNES_COMPOSITE_ADDITIVE, SNES_COMPOSITE_MULTIPLICATIVE, SNESCompositeType, SNESCompositeSetType()

External Links

PETSc Manual: SNES/SNESCompositeSetDamping

source

PETSc.LibPETSc.SNESCompositeSetType — Method

SNESCompositeSetType(petsclib::PetscLibType, snes::AbstractSNES, type::SNESCompositeType)

Sets the type of composite preconditioner.

Logically Collective

Input Parameters:

snes - the preconditioner context
type - SNES_COMPOSITE_ADDITIVE (default), SNES_COMPOSITE_MULTIPLICATIVE, or SNES_COMPOSITE_ADDITIVEOPTIMAL

Options Database Key:

-snes_composite_type (multiplicative|additive|additive_optimal) - Sets composite preconditioner type

Level: developer

See also: SNES_COMPOSITE_ADDITIVE, SNES_COMPOSITE_MULTIPLICATIVE, SNESCompositeType, SNESCOMPOSITE, SNES_COMPOSITE_ADDITIVEOPTIMAL, PCCompositeType

External Links

PETSc Manual: SNES/SNESCompositeSetType

source

PETSc.LibPETSc.SNESComputeFunction — Method

SNESComputeFunction(petsclib::PetscLibType, snes::AbstractSNES, x::AbstractPetscVec, f::AbstractPetscVec)

Calls the function that has been set with SNESSetFunction().

Collective

Input Parameters:

snes - the SNES context
x - input vector

Output Parameter:

f - function vector, as set by SNESSetFunction()

Level: developer

See also: SNES, SNESSetFunction(), SNESGetFunction(), SNESComputeMFFunction(), SNESSetFunctionDomainError()

External Links

PETSc Manual: SNES/SNESComputeFunction

source

PETSc.LibPETSc.SNESComputeFunctionDefaultNPC — Method

SNESComputeFunctionDefaultNPC(petsclib::PetscLibType, snes::AbstractSNES, X::AbstractPetscVec, F::AbstractPetscVec)

Compute the residual by applying the attached nonlinear preconditioner when one is present, otherwise defer to SNESComputeFunction()

Collective

Input Parameters:

snes - the SNES context
X - the current iterate

Output Parameter:

F - the residual vector produced by the nonlinear preconditioner (or the standard function evaluation)

Level: developer

See also: SNES, SNESSetNPC(), SNESApplyNPC(), SNESComputeFunction(), SNESGetNPCFunction()

External Links

PETSc Manual: SNES/SNESComputeFunctionDefaultNPC

source

PETSc.LibPETSc.SNESComputeJacobian — Method

SNESComputeJacobian(petsclib::PetscLibType, snes::AbstractSNES, X::AbstractPetscVec, A::AbstractPetscMat, B::AbstractPetscMat)

Computes the Jacobian matrix that has been set with SNESSetJacobian().

Collective

Input Parameters:

snes - the SNES context
X - input vector

Output Parameters:

A - Jacobian matrix
B - optional matrix for building the preconditioner, usually the same as A

Options Database Keys:

-snes_lag_preconditioner lag - how often to rebuild preconditioner
-snes_lag_jacobian lag - how often to rebuild Jacobian
-snes_test_jacobian [threshold] - compare the user provided Jacobian with one compute via finite differences to check for errors.

If a threshold is given, display only those entries whose difference is greater than the threshold.

-snes_test_jacobian_view [viewer] - display the user provided Jacobian, the finite difference Jacobian and the difference between them to help users detect the location of errors in the user provided Jacobian
-snes_compare_explicit - Compare the computed Jacobian to the finite difference Jacobian and output the differences
-snes_compare_explicit_draw - Compare the computed Jacobian to the finite difference Jacobian and draw the result
-snes_compare_explicit_contour - Compare the computed Jacobian to the finite difference Jacobian and draw a contour plot with the result
-snes_compare_operator - Make the comparison options above use the operator instead of the matrix used to construct the preconditioner
-snes_compare_coloring - Compute the finite difference Jacobian using coloring and display norms of difference
-snes_compare_coloring_display - Compute the finite difference Jacobian using coloring and display verbose differences
-snes_compare_coloring_threshold - Display only those matrix entries that differ by more than a given threshold
-snes_compare_coloring_threshold_atol - Absolute tolerance for difference in matrix entries to be displayed by -snes_compare_coloring_threshold
-snes_compare_coloring_threshold_rtol - Relative tolerance for difference in matrix entries to be displayed by -snes_compare_coloring_threshold
-snes_compare_coloring_draw - Compute the finite difference Jacobian using coloring and draw differences
-snes_compare_coloring_draw_contour - Compute the finite difference Jacobian using coloring and show contours of matrices and differences

Level: developer

See also: SNESSetJacobian(), KSPSetOperators(), MatStructure, SNESSetLagPreconditioner(), SNESSetLagJacobian(), SNESSetJacobianDomainError(), SNESCheckJacobianDomainError(), SNESSetCheckJacobianDomainError()

External Links

PETSc Manual: SNES/SNESComputeJacobian

source

PETSc.LibPETSc.SNESComputeJacobianDefault — Method

SNESComputeJacobianDefault(petsclib::PetscLibType, snes::AbstractSNES, x1::AbstractPetscVec, J::AbstractPetscMat, B::AbstractPetscMat, ctx::Ptr{Cvoid})

Computes the Jacobian using finite differences.

Collective

Input Parameters:

snes - the SNES context
x1 - compute Jacobian at this point
ctx - application's function context, as set with SNESSetFunction()

Output Parameters:

J - Jacobian matrix (not altered in this routine)
B - newly computed Jacobian matrix to use with preconditioner (generally the same as J)

Options Database Keys:

-snes_fd - Activates SNESComputeJacobianDefault()
-snes_fd_coloring - Activates a faster computation that uses a graph coloring of the matrix
-snes_test_err etol - Square root of function error tolerance, default square root of machine

epsilon (1.e-8 in double, 3.e-4 in single)

-mat_fd_type (wp|ds) - See MATMFFD_WP and MATMFFD_DS

Level: intermediate

See also: SNES, SNESSetJacobian(), SNESComputeJacobianDefaultColor(), MatCreateSNESMF()

External Links

PETSc Manual: SNES/SNESComputeJacobianDefault

source

PETSc.LibPETSc.SNESComputeJacobianDefaultColor — Method

SNESComputeJacobianDefaultColor(petsclib::PetscLibType, snes::AbstractSNES, x1::AbstractPetscVec, J::AbstractPetscMat, B::AbstractPetscMat, ctx::Ptr{Cvoid})

Computes the Jacobian using finite differences and coloring to exploit matrix sparsity.

Collective

Input Parameters:

snes - nonlinear solver object
x1 - location at which to evaluate Jacobian
ctx - MatFDColoring context or NULL

Output Parameters:

J - Jacobian matrix (not altered in this routine)
B - newly computed Jacobian matrix to use with preconditioner (generally the same as J)

Options Database Keys:

-snes_fd_color_use_mat - use a matrix coloring from the explicit matrix nonzero pattern instead of from the DM providing the matrix
-snes_fd_color - Activates SNESComputeJacobianDefaultColor() in SNESSetFromOptions()
-mat_fd_coloring_err err - Sets err (square root of relative error in the function)
-mat_fd_coloring_umin umin - Sets umin, the minimum allowable u-value magnitude
-mat_fd_type - Either wp or ds (see MATMFFD_WP or MATMFFD_DS)
-snes_mf_operator - Use matrix-free application of Jacobian
-snes_mf - Use matrix-free Jacobian with no explicit Jacobian representation

See also: SNES, SNESSetJacobian(), SNESTestJacobian(), SNESComputeJacobianDefault(), SNESSetUseMatrixFree(), MatFDColoringCreate(), MatFDColoringSetFunction()

External Links

PETSc Manual: SNES/SNESComputeJacobianDefaultColor

source

PETSc.LibPETSc.SNESComputeMFFunction — Method

SNESComputeMFFunction(petsclib::PetscLibType, snes::AbstractSNES, x::AbstractPetscVec, y::AbstractPetscVec)

Calls the function that has been set with DMSNESSetMFFunction().

Collective

Input Parameters:

snes - the SNES context
x - input vector

Output Parameter:

y - output vector

Level: developer

See also: SNES, SNESSetFunction(), SNESGetFunction(), SNESComputeFunction(), MatCreateSNESMF(), DMSNESSetMFFunction()

External Links

PETSc Manual: SNES/SNESComputeMFFunction

source

PETSc.LibPETSc.SNESComputeNGS — Method

SNESComputeNGS(petsclib::PetscLibType, snes::AbstractSNES, b::AbstractPetscVec, x::AbstractPetscVec)

Calls the Gauss-Seidel function that has been set with SNESSetNGS().

Collective

Input Parameters:

snes - the SNES context
x - input vector
b - rhs vector

Output Parameter:

x - new solution vector

Level: developer

See also: SNESNGSFn, SNESSetNGS(), SNESComputeFunction(), SNESNGS

External Links

PETSc Manual: SNES/SNESComputeNGS

source

PETSc.LibPETSc.SNESComputeObjective — Method

ob::PetscReal = SNESComputeObjective(petsclib::PetscLibType, snes::AbstractSNES, X::AbstractPetscVec)

Computes the objective function that has been provided by SNESSetObjective()

Collective

Input Parameters:

snes - the SNES context
X - the state vector

Output Parameter:

ob - the objective value

Level: developer

See also: SNESLineSearch, SNES, SNESSetObjective(), SNESGetSolution()

External Links

PETSc Manual: SNES/SNESComputeObjective

source

PETSc.LibPETSc.SNESConverged — Method

SNESConverged(petsclib::PetscLibType, snes::AbstractSNES, it::PetscInt, xnorm::PetscReal, snorm::PetscReal, fnorm::PetscReal)

Run the convergence test and update the SNESConvergedReason.

Collective

Input Parameters:

snes - the SNES context
it - current iteration
xnorm - 2-norm of current iterate
snorm - 2-norm of current step
fnorm - 2-norm of function

Level: developer

See also: SNES, SNESSolve, SNESSetConvergenceTest()

External Links

PETSc Manual: SNES/SNESConverged

source

PETSc.LibPETSc.SNESConvergedCorrectPressure — Method

reason::SNESConvergedReason = SNESConvergedCorrectPressure(petsclib::PetscLibType, snes::AbstractSNES, it::PetscInt, xnorm::PetscReal, gnorm::PetscReal, f::PetscReal, ctx::Ptr{Cvoid})

The regular SNES convergence test that, up on convergence, adds a vector in the nullspace to make the continuum integral of the pressure field equal to zero.

Logically Collective

Input Parameters:

snes - the SNES context
it - the iteration (0 indicates before any Newton steps)
xnorm - 2-norm of current iterate
gnorm - 2-norm of current step
f - 2-norm of function at current iterate
ctx - Optional application context

Output Parameter:

reason - SNES_CONVERGED_ITERATING, SNES_CONVERGED_ITS, or SNES_DIVERGED_FUNCTION_NANORINF

Options Database Key:

-snes_convergence_test correct_pressure - see SNESSetFromOptions()

Level: advanced

See also: SNES, DM, SNESConvergedDefault(), SNESSetConvergenceTest(), DMSetNullSpaceConstructor()

External Links

PETSc Manual: SNES/SNESConvergedCorrectPressure

source

PETSc.LibPETSc.SNESConvergedDefault — Method

reason::SNESConvergedReason = SNESConvergedDefault(petsclib::PetscLibType, snes::AbstractSNES, it::PetscInt, xnorm::PetscReal, snorm::PetscReal, fnorm::PetscReal, ctx::Ptr{Cvoid})

Default convergence test for SNESSolve().

Collective

Input Parameters:

snes - the SNES context
it - the iteration (0 indicates before any Newton steps)
xnorm - 2-norm of current iterate
snorm - 2-norm of current step
fnorm - 2-norm of function at current iterate
ctx - unused context

Output Parameter:

reason - converged reason, see SNESConvergedReason

Options Database Keys:

-snes_convergence_test default - see SNESSetFromOptions()
-snes_stol - convergence tolerance in terms of the norm of the change in the solution between steps
-snes_atol abstol - absolute tolerance of residual norm
-snes_rtol rtol - relative decrease in tolerance norm from the initial 2-norm of the solution
-snes_divergence_tolerance divtol - if the residual goes above divtol*rnorm0, exit with divergence
-snes_max_funcs max_funcs - maximum number of function evaluations, use unlimited for no maximum
-snes_max_fail max_fail - maximum number of line search failures allowed before stopping, default is none
-snes_max_linear_solve_fail - number of linear solver failures before SNESSolve() stops

Level: developer

See also: SNES, SNESSolve(), SNESSetConvergenceTest(), SNESConvergedSkip(), SNESSetTolerances(), SNESSetDivergenceTolerance(), SNESConvergedReason

External Links

PETSc Manual: SNES/SNESConvergedDefault

source

PETSc.LibPETSc.SNESConvergedReasonView — Method

SNESConvergedReasonView(petsclib::PetscLibType, snes::AbstractSNES, viewer::PetscViewer)

Displays the reason a SNES solve converged or diverged to a viewer

Collective

Input Parameters:

snes - iterative context obtained from SNESCreate()
viewer - the viewer to display the reason

Options Database Keys:

-snes_converged_reason - print reason for converged or diverged, also prints number of iterations
-snes_converged_reason ::failed - only print reason and number of iterations when diverged

Level: beginner

See also: SNESConvergedReason, PetscViewer, SNES, SNESCreate(), SNESSetUp(), SNESDestroy(), SNESSetTolerances(), SNESConvergedDefault(), SNESGetConvergedReason(), SNESConvergedReasonViewFromOptions(), PetscViewerPushFormat(), PetscViewerPopFormat()

External Links

PETSc Manual: SNES/SNESConvergedReasonView

source

PETSc.LibPETSc.SNESConvergedReasonViewCancel — Method

SNESConvergedReasonViewCancel(petsclib::PetscLibType, snes::AbstractSNES)

Clears all the reason view functions for a SNES object provided with SNESConvergedReasonViewSet() also removes the default viewer.

Collective

Input Parameter:

snes - the nonlinear iterative solver context obtained from SNESCreate()

Level: intermediate

See also: SNES, SNESCreate(), SNESDestroy(), SNESReset(), SNESConvergedReasonViewSet()

External Links

PETSc Manual: SNES/SNESConvergedReasonViewCancel

source

PETSc.LibPETSc.SNESConvergedReasonViewFromOptions — Method

SNESConvergedReasonViewFromOptions(petsclib::PetscLibType, snes::AbstractSNES)

Processes command line options to determine if/how a SNESConvergedReason is to be viewed at the end of SNESSolve() All the user-provided viewer routines set with SNESConvergedReasonViewSet() will be called, if they exist.

Collective

Input Parameter:

snes - the SNES object

Level: advanced

See also: SNES, SNESConvergedReason, SNESConvergedReasonViewSet(), SNESCreate(), SNESSetUp(), SNESDestroy(), SNESSetTolerances(), SNESConvergedDefault(), SNESGetConvergedReason(), SNESConvergedReasonView()

External Links

PETSc Manual: SNES/SNESConvergedReasonViewFromOptions

source

PETSc.LibPETSc.SNESConvergedReasonViewSet — Method

SNESConvergedReasonViewSet(petsclib::PetscLibType, snes::AbstractSNES, f::external, vctx::Ptr{Cvoid}, reasonviewdestroy::Ptr{Cvoid})

Sets an ADDITIONAL function that is to be used at the end of the nonlinear solver to display the convergence reason of the nonlinear solver.

Logically Collective

Input Parameters:

snes - the SNES context
f - the SNESConvergedReason view function
vctx - [optional] user-defined context for private data for the SNESConvergedReason view function (use NULL if no context is desired)
reasonviewdestroy - [optional] routine that frees the context (may be NULL), see PetscCtxDestroyFn for the calling sequence

Calling sequence of f:

snes - the SNES context
vctx - [optional] context for private data for the function

Options Database Keys:

-snes_converged_reason - sets a default SNESConvergedReasonView()
-snes_converged_reason_view_cancel - cancels all converged reason viewers that have been hardwired into a code by

calls to SNESConvergedReasonViewSet(), but does not cancel those set via the options database.

Level: intermediate

See also: SNES, SNESSolve(), SNESConvergedReason, SNESGetConvergedReason(), SNESConvergedReasonView(), SNESConvergedReasonViewCancel(), PetscCtxDestroyFn

External Links

PETSc Manual: SNES/SNESConvergedReasonViewSet

source

PETSc.LibPETSc.SNESConvergedSkip — Method

reason::SNESConvergedReason = SNESConvergedSkip(petsclib::PetscLibType, snes::AbstractSNES, it::PetscInt, xnorm::PetscReal, snorm::PetscReal, fnorm::PetscReal, ctx::Ptr{Cvoid})

Convergence test for SNES that NEVER returns as converged, UNLESS the maximum number of iteration have been reached.

Logically Collective

Input Parameters:

snes - the SNES context
it - the iteration (0 indicates before any Newton steps)
xnorm - 2-norm of current iterate
snorm - 2-norm of current step
fnorm - 2-norm of function at current iterate
ctx - unused context

Output Parameter:

reason - SNES_CONVERGED_ITERATING, SNES_CONVERGED_ITS, or SNES_DIVERGED_FUNCTION_NANORINF

Options Database Key:

-snes_convergence_test skip - see SNESSetFromOptions()

Level: advanced

See also: SNES, SNESSolve(), SNESConvergedDefault(), SNESSetConvergenceTest(), SNESConvergedReason

External Links

PETSc Manual: SNES/SNESConvergedSkip

source

PETSc.LibPETSc.SNESCreate — Method

outsnes::SNES = SNESCreate(petsclib::PetscLibType, comm::MPI_Comm)

Creates a nonlinear solver context used to manage a set of nonlinear solves

Collective

Input Parameter:

comm - MPI communicator

Output Parameter:

outsnes - the new SNES context

Options Database Keys:

-snes_mf - Activates default matrix-free Jacobian-vector products, and no matrix to construct a preconditioner
-snes_mf_operator - Activates default matrix-free Jacobian-vector products, and a user-provided matrix as set by SNESSetJacobian()
-snes_fd_coloring - uses a relative fast computation of the Jacobian using finite differences and a graph coloring
-snes_fd - Uses (slow!) finite differences to compute Jacobian

Level: beginner

See also: SNES, SNESSolve(), SNESDestroy(), SNESSetLagPreconditioner(), SNESSetLagJacobian()

External Links

PETSc Manual: SNES/SNESCreate

source

PETSc.LibPETSc.SNESDestroy — Method

SNESDestroy(petsclib::PetscLibType, snes::AbstractSNES)

Destroys the nonlinear solver context that was created with SNESCreate().

Collective

Input Parameter:

snes - the SNES context

Level: beginner

See also: SNES, SNESCreate(), SNESSolve()

External Links

PETSc Manual: SNES/SNESDestroy

source

PETSc.LibPETSc.SNESFASCreateCoarseVec — Method

Xcoarse::PetscVec = SNESFASCreateCoarseVec(petsclib::PetscLibType, snes::AbstractSNES)

create a Vec corresponding to a state vector on one level coarser than the current level

Collective

Input Parameter:

snes - SNESFAS object

Output Parameter:

Xcoarse - vector on level one coarser than the current level

Level: developer

See also: SNESFASSetRestriction(), SNESFASRestrict(), SNESFAS

External Links

PETSc Manual: SNESFAS/SNESFASCreateCoarseVec

source

PETSc.LibPETSc.SNESFASCycleGetCorrection — Method

correction::SNES = SNESFASCycleGetCorrection(petsclib::PetscLibType, snes::AbstractSNES)

Gets the coarse correction SNESFAS context for this level

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

correction - the coarse correction solve on this level

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetSmootherUp(), SNESFASCycleGetSmoother()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetCorrection

source

PETSc.LibPETSc.SNESFASCycleGetInjection — Method

mat::PetscMat = SNESFASCycleGetInjection(petsclib::PetscLibType, snes::AbstractSNES)

Gets the injection on a level

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

mat - the restriction operator on this level

Level: advanced

See also: SNES, SNESFAS, SNESFASGetInjection(), SNESFASCycleGetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetInjection

source

PETSc.LibPETSc.SNESFASCycleGetInterpolation — Method

mat::PetscMat = SNESFASCycleGetInterpolation(petsclib::PetscLibType, snes::AbstractSNES)

Gets the interpolation on a level

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

mat - the interpolation operator on this level

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetSmootherUp(), SNESFASCycleGetSmoother()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetInterpolation

source

PETSc.LibPETSc.SNESFASCycleGetRScale — Method

vec::PetscVec = SNESFASCycleGetRScale(petsclib::PetscLibType, snes::AbstractSNES)

Gets the injection scale-factor on a level

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

vec - the restriction operator on this level

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetRestriction(), SNESFASGetRScale()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetRScale

source

PETSc.LibPETSc.SNESFASCycleGetRestriction — Method

mat::PetscMat = SNESFASCycleGetRestriction(petsclib::PetscLibType, snes::AbstractSNES)

Gets the restriction on a level

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

mat - the restriction operator on this level

Level: advanced

See also: SNES, SNESFAS, SNESFASGetRestriction(), SNESFASCycleGetInterpolation()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetRestriction

source

PETSc.LibPETSc.SNESFASCycleGetSmoother — Method

smooth::SNES = SNESFASCycleGetSmoother(petsclib::PetscLibType, snes::AbstractSNES)

Gets the smoother on a particular cycle level.

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

smooth - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetSmootherUp(), SNESFASCycleGetSmootherDown(), SNESFASGetCycleSNES()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetSmoother

source

PETSc.LibPETSc.SNESFASCycleGetSmootherDown — Method

smoothd::SNES = SNESFASCycleGetSmootherDown(petsclib::PetscLibType, snes::AbstractSNES)

Gets the down smoother on a particular cycle level.

Logically Collective

Input Parameter:

snes - SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

smoothd - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetSmootherUp(), SNESFASCycleGetSmoother(), SNESFASGetCycleSNES()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetSmootherDown

source

PETSc.LibPETSc.SNESFASCycleGetSmootherUp — Method

smoothu::SNES = SNESFASCycleGetSmootherUp(petsclib::PetscLibType, snes::AbstractSNES)

Gets the up smoother on a particular cycle level.

Logically Collective

Input Parameter:

snes - the SNESFAS obtained with SNESFASGetCycleSNES()

Output Parameter:

smoothu - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASCycleGetSmoother(), SNESFASCycleGetSmootherDown(), SNESFASGetCycleSNES()

External Links

PETSc Manual: SNESFAS/SNESFASCycleGetSmootherUp

source

PETSc.LibPETSc.SNESFASCycleIsFine — Method

flg::PetscBool = SNESFASCycleIsFine(petsclib::PetscLibType, snes::AbstractSNES)

Determines if a given SNES is the finest level in a SNESFAS

Logically Collective

Input Parameter:

snes - the SNESFAS context obtained with SNESFASGetCycleSNES()

Output Parameter:

flg - indicates if this is the fine level or not

Level: advanced

See also: SNES, SNESFAS, SNESFASSetLevels()

External Links

PETSc Manual: SNESFAS/SNESFASCycleIsFine

source

PETSc.LibPETSc.SNESFASCycleSetCycles — Method

SNESFASCycleSetCycles(petsclib::PetscLibType, snes::AbstractSNES, cycles::PetscInt)

Sets the number of cycles for all levels in a SNESFAS

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
cycles - the number of cycles – 1 for V-cycle, 2 for W-cycle

Level: advanced

See also: SNES, SNESFAS, SNESFASSetCycles()

External Links

PETSc Manual: SNESFAS/SNESFASCycleSetCycles

source

PETSc.LibPETSc.SNESFASFullGetTotal — Method

total::PetscBool = SNESFASFullGetTotal(petsclib::PetscLibType, snes::AbstractSNES)

Use total residual restriction and total interpolation on the initial down and up sweep of full FAS cycles

Logically Collective

Input Parameter:

snes - the SNESFAS nonlinear multigrid context

Output Parameter:

total - whether to use total restriction / interpolatiaon or not (the alternative is defect restriction and correction interpolation)

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothUp(), DMInterpolateSolution(), SNESFullSetTotal()

External Links

PETSc Manual: SNESFAS/SNESFASFullGetTotal

source

PETSc.LibPETSc.SNESFASFullSetDownSweep — Method

SNESFASFullSetDownSweep(petsclib::PetscLibType, snes::AbstractSNES, swp::PetscBool)

Smooth during the initial downsweep for SNESFAS

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
swp - whether to downsweep or not

Options Database Key:

-snes_fas_full_downsweep - Sets whether to smooth on the initial downsweep

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothUp()

External Links

PETSc Manual: SNESFAS/SNESFASFullSetDownSweep

source

PETSc.LibPETSc.SNESFASFullSetTotal — Method

SNESFASFullSetTotal(petsclib::PetscLibType, snes::AbstractSNES, total::PetscBool)

Use total residual restriction and total interpolation on the initial down and up sweep of full SNESFAS cycles

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
total - whether to use total restriction / interpolatiaon or not (the alternative is defect restriction and correction interpolation)

Options Database Key:

-snes_fas_full_total - Use total restriction and interpolation on the initial down and up sweeps for the full SNESFAS cycle

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothUp(), DMInterpolateSolution()

External Links

PETSc Manual: SNESFAS/SNESFASFullSetTotal

source

PETSc.LibPETSc.SNESFASGalerkinFunctionDefault — Method

SNESFASGalerkinFunctionDefault(petsclib::PetscLibType, snes::AbstractSNES, X::AbstractPetscVec, F::AbstractPetscVec, ctx::Ptr{Cvoid})

Computes the Galerkin FAS function

Collective

Input Parameters:

snes - the SNESFAS nonlinear solver context
X - input vector
ctx - the application context

Output Parameter:

F - output vector

Level: developer

See also: SNES, SNESFAS, SNESFASGetGalerkin(), SNESFASSetGalerkin()

External Links

PETSc Manual: SNESFAS/SNESFASGalerkinFunctionDefault

source

PETSc.LibPETSc.SNESFASGetCoarseSolve — Method

coarse::SNES = SNESFASGetCoarseSolve(petsclib::PetscLibType, snes::AbstractSNES)

Gets the coarsest level solver.

Input Parameter:

snes - the SNESFAS nonlinear multigrid context

Output Parameter:

coarse - the coarse-level solver

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASGetCoarseSolve

source

PETSc.LibPETSc.SNESFASGetCycleSNES — Method

lsnes::SNES = SNESFASGetCycleSNES(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the SNES corresponding to a particular level of the SNESFAS hierarchy

Input Parameters:

snes - the SNES nonlinear multigrid context
level - the level to get

Output Parameter:

lsnes - the SNES for the requested level

Level: advanced

See also: SNES, SNESFAS, SNESFASSetLevels(), SNESFASGetLevels()

External Links

PETSc Manual: SNESFAS/SNESFASGetCycleSNES

source

PETSc.LibPETSc.SNESFASGetGalerkin — Method

flg::PetscBool = SNESFASGetGalerkin(petsclib::PetscLibType, snes::AbstractSNES)

Gets if the coarse problems are formed by projection to the fine problem

Not Collective but the result would be the same on all MPI processes

Input Parameter:

snes - the SNESFAS nonlinear solver context

Output Parameter:

flg - PETSC_TRUE if the coarse problem is formed by projection

Level: advanced

See also: SNES, SNESFAS, SNESFASSetLevels(), SNESFASSetGalerkin()

External Links

PETSc Manual: SNESFAS/SNESFASGetGalerkin

source

PETSc.LibPETSc.SNESFASGetInjection — Method

mat::PetscMat = SNESFASGetInjection(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the matrix used to calculate the injection from l-1 to the lth level

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest) to supply [do not supply 0]

Output Parameter:

mat - the injection operator

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASGetRestriction(), SNESFASGetInterpolation(), SNESFASGetRScale()

External Links

PETSc Manual: SNESFAS/SNESFASGetInjection

source

PETSc.LibPETSc.SNESFASGetInterpolation — Method

mat::PetscMat = SNESFASGetInterpolation(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the matrix used to calculate the interpolation from l-1 to the lth level

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest) to supply [do not supply 0]

Output Parameter:

mat - the interpolation operator

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInterpolation(), SNESFASGetInjection(), SNESFASGetRestriction(), SNESFASGetRScale()

External Links

PETSc Manual: SNESFAS/SNESFASGetInterpolation

source

PETSc.LibPETSc.SNESFASGetLevels — Method

levels::PetscInt = SNESFASGetLevels(petsclib::PetscLibType, snes::AbstractSNES)

Gets the number of levels in a SNESFAS, including fine and coarse grids

Input Parameter:

snes - the SNES nonlinear solver context of SNESType SNESFAS

Output Parameter:

levels - the number of levels

Level: advanced

See also: SNES, SNESFAS, SNESFASSetLevels(), PCMGGetLevels()

External Links

PETSc Manual: SNESFAS/SNESFASGetLevels

source

PETSc.LibPETSc.SNESFASGetRestriction — Method

mat::PetscMat = SNESFASGetRestriction(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the matrix used to calculate the restriction from l to the l-1th level

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest) to supply [do not supply 0]

Output Parameter:

mat - the interpolation operator

Level: advanced

See also: SNES, SNESFAS, SNESFASSetRestriction(), SNESFASGetInjection(), SNESFASGetInterpolation(), SNESFASGetRScale()

External Links

PETSc Manual: SNESFAS/SNESFASGetRestriction

source

PETSc.LibPETSc.SNESFASGetSmoother — Method

smooth::SNES = SNESFASGetSmoother(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the default smoother on a level.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest) to supply

Output Parameter:

smooth - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASGetSmoother

source

PETSc.LibPETSc.SNESFASGetSmootherDown — Method

smooth::SNES = SNESFASGetSmootherDown(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the downsmoother on a level.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest) to supply

Output Parameter:

smooth - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASGetSmootherDown

source

PETSc.LibPETSc.SNESFASGetSmootherUp — Method

smooth::SNES = SNESFASGetSmootherUp(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt)

Gets the upsmoother on a level.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
level - the level (0 is coarsest)

Output Parameter:

smooth - the smoother

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASGetSmootherUp

source

PETSc.LibPETSc.SNESFASGetType — Method

fastype::SNESFASType = SNESFASGetType(petsclib::PetscLibType, snes::AbstractSNES)

Gets the update and correction type used for SNESFAS.

Logically Collective

Input Parameter:

snes - SNESFAS context

Output Parameter:

fastype - SNES_FAS_ADDITIVE, SNES_FAS_MULTIPLICATIVE, SNES_FAS_FULL, or SNES_FAS_KASKADE

Level: intermediate

See also: SNES, SNESFAS, PCMGSetType(), SNESFASSetType(), SNES_FAS_ADDITIVE, SNES_FAS_MULTIPLICATIVE, SNES_FAS_FULL, SNES_FAS_KASKADE

External Links

PETSc Manual: SNESFAS/SNESFASGetType

source

PETSc.LibPETSc.SNESFASRestrict — Method

SNESFASRestrict(petsclib::PetscLibType, fine::AbstractSNES, Xfine::AbstractPetscVec, Xcoarse::AbstractPetscVec)

restrict a Vec to the next coarser level

Collective

Input Parameters:

fine - SNES from which to restrict
Xfine - vector to restrict

Output Parameter:

Xcoarse - result of restriction

Level: developer

See also: SNES, SNESFAS, SNESFASSetRestriction(), SNESFASSetInjection(), SNESFASCreateCoarseVec()

External Links

PETSc Manual: SNESFAS/SNESFASRestrict

source

PETSc.LibPETSc.SNESFASSetContinuation — Method

SNESFASSetContinuation(petsclib::PetscLibType, snes::AbstractSNES, continuation::PetscBool)

Sets the SNESFAS cycle to default to using exact Newton solves on the upsweep

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
continuation - whether to use continuation

Options Database Key:

-snes_fas_continuation - sets continuation to true

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothUp()

External Links

PETSc Manual: SNESFAS/SNESFASSetContinuation

source

PETSc.LibPETSc.SNESFASSetCycles — Method

SNESFASSetCycles(petsclib::PetscLibType, snes::AbstractSNES, cycles::PetscInt)

Sets the number of SNESFAS multigrid cycles to use each time a grid is visited. Use SNESFASSetCyclesOnLevel() for more complicated cycling.

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
cycles - the number of cycles – 1 for V-cycle, 2 for W-cycle

Options Database Key:

-snes_fas_cycles (1|2) - 1 for V-cycle, 2 for W-cycle

Level: advanced

See also: SNES, SNESFAS, SNESFASSetCyclesOnLevel()

External Links

PETSc Manual: SNESFAS/SNESFASSetCycles

source

PETSc.LibPETSc.SNESFASSetGalerkin — Method

SNESFASSetGalerkin(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

Sets coarse problems as formed by projection to the fine problem

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear solver context
flg - PETSC_TRUE to use the projection process

Level: advanced

See also: SNES, SNESFAS, SNESFASSetLevels(), SNESFASGetGalerkin()

External Links

PETSc Manual: SNESFAS/SNESFASSetGalerkin

source

PETSc.LibPETSc.SNESFASSetInjection — Method

SNESFASSetInjection(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt, mat::AbstractPetscMat)

Sets the matrix to be used to inject the solution from level to level-1.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
mat - the injection matrix
level - the level (0 is coarsest) to supply [Do not supply 0]

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInterpolation(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASSetInjection

source

PETSc.LibPETSc.SNESFASSetInterpolation — Method

SNESFASSetInterpolation(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt, mat::AbstractPetscMat)

Sets the Mat to be used to apply the interpolation from l-1 to the lth level

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
mat - the interpolation operator
level - the level (0 is coarsest) to supply [do not supply 0]

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction(), SNESFASSetRScale()

External Links

PETSc Manual: SNESFAS/SNESFASSetInterpolation

source

PETSc.LibPETSc.SNESFASSetLevels — Method

comms::MPI_Comm = SNESFASSetLevels(petsclib::PetscLibType, snes::AbstractSNES, levels::PetscInt)

Sets the number of levels to use with SNESFAS. Must be called before any other SNESFAS routine.

Input Parameters:

snes - the SNES context of SNESType SNESFAS
levels - the number of levels
comms - optional communicators for each level; this is to allow solving the coarser

problems on smaller sets of processors.

Level: intermediate

See also: SNES, SNESFAS, SNESFASGetLevels()

External Links

PETSc Manual: SNESFAS/SNESFASSetLevels

source

PETSc.LibPETSc.SNESFASSetLog — Method

SNESFASSetLog(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

Sets or unsets time logging for various SNESFAS stages on all levels

Logically Collective

Input Parameters:

snes - the SNESFAS context
flg - whether to log or not

Level: advanced

See also: SNES, SNESFAS, SNESFASSetMonitor()

External Links

PETSc Manual: SNESFAS/SNESFASSetLog

source

PETSc.LibPETSc.SNESFASSetMonitor — Method

SNESFASSetMonitor(petsclib::PetscLibType, snes::AbstractSNES, vf::Vector{PetscViewerAndFormat}, flg::PetscBool)

Sets the method-specific cycle monitoring

Logically Collective

Input Parameters:

snes - the SNESFAS context
vf - viewer and format structure (may be NULL if flg is PETSC_FALSE)
flg - monitor or not

Level: advanced

See also: SNES, SNESFAS, SNESMonitorSet(), SNESFASSetCyclesOnLevel()

External Links

PETSc Manual: SNESFAS/SNESFASSetMonitor

source

PETSc.LibPETSc.SNESFASSetNumberSmoothDown — Method

SNESFASSetNumberSmoothDown(petsclib::PetscLibType, snes::AbstractSNES, n::PetscInt)

Sets the number of pre-smoothing steps to use on all levels.

Logically Collective

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
n - the number of smoothing steps to use

Options Database Key:

-snes_fas_smoothdown n - Sets number of pre-smoothing steps

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothUp()

External Links

PETSc Manual: SNESFAS/SNESFASSetNumberSmoothDown

source

PETSc.LibPETSc.SNESFASSetNumberSmoothUp — Method

SNESFASSetNumberSmoothUp(petsclib::PetscLibType, snes::AbstractSNES, n::PetscInt)

Sets the number of post-smoothing steps to use on all levels.

Logically Collective

Input Parameters:

snes - the SNES nonlinear multigrid context
n - the number of smoothing steps to use

Options Database Key:

-snes_fas_smoothup n - Sets number of pre-smoothing steps

Level: advanced

See also: SNES, SNESFAS, SNESFASSetNumberSmoothDown()

External Links

PETSc Manual: SNESFAS/SNESFASSetNumberSmoothUp

source

PETSc.LibPETSc.SNESFASSetRScale — Method

SNESFASSetRScale(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt, rscale::AbstractPetscVec)

Sets the scaling factor of the restriction operator from level l to l-1.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
rscale - the restriction scaling
level - the level (0 is coarsest) to supply [Do not supply 0]

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInjection(), SNESFASSetRestriction()

External Links

PETSc Manual: SNESFAS/SNESFASSetRScale

source

PETSc.LibPETSc.SNESFASSetRestriction — Method

SNESFASSetRestriction(petsclib::PetscLibType, snes::AbstractSNES, level::PetscInt, mat::AbstractPetscMat)

Sets the matrix to be used to restrict the defect from level l to l-1.

Input Parameters:

snes - the SNESFAS nonlinear multigrid context
mat - the restriction matrix
level - the level (0 is coarsest) to supply [Do not supply 0]

Level: advanced

See also: SNES, SNESFAS, SNESFASSetInterpolation(), SNESFASSetInjection()

External Links

PETSc Manual: SNESFAS/SNESFASSetRestriction

source

PETSc.LibPETSc.SNESFASSetType — Method

SNESFASSetType(petsclib::PetscLibType, snes::AbstractSNES, fastype::SNESFASType)

Sets the update and correction type used for SNESFAS.

Logically Collective

Input Parameters:

snes - SNESFAS context
fastype - SNES_FAS_ADDITIVE, SNES_FAS_MULTIPLICATIVE, SNES_FAS_FULL, or SNES_FAS_KASKADE

Level: intermediate

See also: SNES, SNESFAS, PCMGSetType(), SNESFASGetType(), SNES_FAS_ADDITIVE, SNES_FAS_MULTIPLICATIVE, SNES_FAS_FULL, SNES_FAS_KASKADE

External Links

PETSc Manual: SNESFAS/SNESFASSetType

source

PETSc.LibPETSc.SNESFinalizePackage — Method

SNESFinalizePackage(petsclib::PetscLibType)

This function destroys everything in the PETSc interface to the SNES package. It is called from PetscFinalize().

Level: developer

See also: SNESNEWTONTR, NormType

External Links

PETSc Manual: SNES/SNESNewtonTRSetNormType

source

PETSc.LibPETSc.SNESNewtonTRSetPostCheck — Method

SNESNewtonTRSetPostCheck(petsclib::PetscLibType, snes::AbstractSNES, func::external, ctx::Ptr{Cvoid})

Sets a user function that is called after the search step has been determined but before the next function evaluation. Allows the user a chance to change or override the internal decision of the solver

Logically Collective

Input Parameters:

snes - the nonlinear solver object
func - [optional] function evaluation routine, for the calling sequence see SNESNewtonTRPostCheck()
ctx - [optional] user-defined context for private data for the function evaluation routine (may be NULL)

Calling sequence of func:

snes - the nonlinear solver object
X - the current solution value
Y - the tentative update step
W - the tentative new solution value
changed_Y - output, flag indicated Y has been changed by the post-check
changed_W - output, flag indicated W has been changed by the post-check
ctx - the optional application context

Level: intermediate

See also: SNESNEWTONTR, SNESNewtonTRPostCheck(), SNESNewtonTRGetPostCheck(), SNESNewtonTRSetPreCheck(), SNESNewtonTRGetPreCheck()

External Links

PETSc Manual: SNES/SNESNewtonTRSetPostCheck

source

PETSc.LibPETSc.SNESNewtonTRSetPreCheck — Method

SNESNewtonTRSetPreCheck(petsclib::PetscLibType, snes::AbstractSNES, func::external, ctx::Ptr{Cvoid})

Sets a user function that is called before the search step has been determined. Allows the user a chance to change or override the trust region decision.

Logically Collective

Input Parameters:

snes - the nonlinear solver object
func - [optional] function evaluation routine, for the calling sequence see SNESNewtonTRPreCheck()
ctx - [optional] user-defined context for private data for the function evaluation routine (may be NULL)

Calling sequence of func:

snes - the nonlinear solver object
X - the current solution value
Y - the tentative update step
changed - output, flag indicating Y has been changed by the pre-check
ctx - the optional application context

Level: intermediate

See also: SNESNEWTONTR, SNESNewtonTRPreCheck(), SNESNewtonTRGetPreCheck(), SNESNewtonTRSetPostCheck(), SNESNewtonTRGetPostCheck()

External Links

PETSc Manual: SNES/SNESNewtonTRSetPreCheck

source

PETSc.LibPETSc.SNESNewtonTRSetQNType — Method

SNESNewtonTRSetQNType(petsclib::PetscLibType, snes::AbstractSNES, use::SNESNewtonTRQNType)

Specify to use a quasi-Newton model.

Input Parameters:

snes - the nonlinear solver object
use - the type of approximations to be used

Level: intermediate

See also: SNESNEWTONTR, SNESNewtonTRQNType, MATLMVM

External Links

PETSc Manual: SNES/SNESNewtonTRSetQNType

source

PETSc.LibPETSc.SNESNewtonTRSetTolerances — Method

SNESNewtonTRSetTolerances(petsclib::PetscLibType, snes::AbstractSNES, delta_min::PetscReal, delta_max::PetscReal, delta_0::PetscReal)

Sets the trust region parameter tolerances.

Logically Collective

Input Parameters:

snes - the SNES context
delta_min - minimum allowed trust region size
delta_max - maximum allowed trust region size
delta_0 - initial trust region size

Options Database Key:

-snes_tr_deltamin tol - Set minimum size
-snes_tr_deltamax tol - Set maximum size
-snes_tr_delta0 tol - Set initial size

See also: SNES, SNESNEWTONTR, SNESNewtonTRGetTolerances()

External Links

PETSc Manual: SNES/SNESNewtonTRSetTolerances

source

PETSc.LibPETSc.SNESNewtonTRSetUpdateParameters — Method

SNESNewtonTRSetUpdateParameters(petsclib::PetscLibType, snes::AbstractSNES, eta1::PetscReal, eta2::PetscReal, eta3::PetscReal, t1::PetscReal, t2::PetscReal)

Sets the trust region update parameters.

Logically Collective

Input Parameters:

snes - the SNES context
eta1 - acceptance tolerance
eta2 - shrinking tolerance
eta3 - enlarging tolerance
t1 - shrink factor
t2 - enlarge factor

Options Database Key:

-snes_tr_eta1 tol - Set eta1
-snes_tr_eta2 tol - Set eta2
-snes_tr_eta3 tol - Set eta3
-snes_tr_t1 tol - Set t1
-snes_tr_t2 tol - Set t2

See also: SNES, SNESNEWTONTR, SNESSetObjective(), SNESNewtonTRGetUpdateParameters()

External Links

PETSc Manual: SNES/SNESNewtonTRSetUpdateParameters

source

PETSc.LibPETSc.SNESObjectiveComputeFunctionDefaultFD — Method

SNESObjectiveComputeFunctionDefaultFD(petsclib::PetscLibType, snes::AbstractSNES, X::AbstractPetscVec, F::AbstractPetscVec, ctx::Ptr{Cvoid})

Computes the gradient of a user provided objective function

Collective

Input Parameters:

snes - the SNES context
X - the state vector
ctx - the (ignored) function context

Output Parameter:

F - the function value

Options Database Keys:

-snes_fd_function_eps - Tolerance for including non-zero entries into the gradient, default is 1.e-6
-snes_fd_function - Computes function from user provided objective function (set with SNESSetObjective()) with finite difference

Level: advanced

See also: SNESSetObjective(), SNESSetFunction(), SNESComputeObjective(), SNESComputeJacobianDefault(), SNESObjectiveFn

External Links

PETSc Manual: SNES/SNESObjectiveComputeFunctionDefaultFD

source

PETSc.LibPETSc.SNESParametersInitialize — Method

SNESParametersInitialize(petsclib::PetscLibType, snes::AbstractSNES)

Sets all the parameters in snes to their default value (when SNESCreate() was called) if they currently contain default values

Collective

Input Parameter:

snes - the SNES object

Level: developer

See also: SNES, SNESSolve(), SNESDestroy(), SNESSetLagPreconditioner(), SNESSetLagJacobian(), PetscObjectParameterSetDefault()

External Links

PETSc Manual: SNES/SNESParametersInitialize

source

PETSc.LibPETSc.SNESPatchSetCellNumbering — Method

SNESPatchSetCellNumbering(petsclib::PetscLibType, snes::AbstractSNES, cellNumbering::PetscSection)

Set the PetscSection that provides a numbering of the cells used to define patches in a SNESPATCH solver

Logically Collective

Input Parameters:

snes - the SNESPATCH solver
cellNumbering - the PetscSection giving the cell numbering; forwarded to the underlying PCPATCH via PCPatchSetCellNumbering()

Level: advanced

See also: SNESPATCH, PCPATCH, PCPatchSetCellNumbering(), PetscSection

External Links

PETSc Manual: SNES/SNESPatchSetCellNumbering

source

PETSc.LibPETSc.SNESPatchSetComputeFunction — Method

SNESPatchSetComputeFunction(petsclib::PetscLibType, snes::AbstractSNES, func::external, ctx::Ptr{Cvoid})

Set the callback used to compute the per-patch nonlinear residual for a SNESPATCH solver

Logically Collective

Input Parameters:

snes - the SNESPATCH solver
func - callback that computes the patch residual; forwarded to the underlying PCPATCH via PCPatchSetComputeFunction()
ctx - optional application context passed to func

Calling sequence of func:

pc - the PC associated with the SNESPATCH solver
point - the point
x - the input solution (not used in linear problems)
f - the patch residual vector
cellIS - an array of the cell numbers
n - the size of dofsArray
dofsArray - the dofmap for the dofs to be solved for
dofsArrayWithAll - the dofmap for all dofs on the patch
ctx - the application context

Level: advanced

See also: SNESPATCH, PCPATCH, PCPatchSetComputeFunction(), SNESPatchSetComputeOperator()

External Links

PETSc Manual: SNES/SNESPatchSetComputeFunction

source

PETSc.LibPETSc.SNESPatchSetComputeOperator — Method

SNESPatchSetComputeOperator(petsclib::PetscLibType, snes::AbstractSNES, func::external, ctx::Ptr{Cvoid})

Set the callback used to assemble the per-patch Jacobian for a SNESPATCH solver

Logically Collective

Input Parameters:

snes - the SNESPATCH solver
func - callback that assembles the patch Jacobian; forwarded to the underlying PCPATCH via PCPatchSetComputeOperator()
ctx - optional application context passed to func

Calling sequence of func:

pc - the PC associated with the SNESPATCH solver
point - the point
x - the input solution (not used in linear problems)
mat - the patch matrix
cellIS - an array of the cell numbers
n - the size of dofsArray
dofsArray - the dofmap for the dofs to be solved for
dofsArrayWithAll - the dofmap for all dofs on the patch
ctx - the application context

Level: advanced

See also: SNESPATCH, PCPATCH, PCPatchSetComputeOperator(), SNESPatchSetComputeFunction()

External Links

PETSc Manual: SNES/SNESPatchSetComputeOperator

source

PETSc.LibPETSc.SNESPatchSetConstructType — Method

SNESPatchSetConstructType(petsclib::PetscLibType, snes::AbstractSNES, ctype::PCPatchConstructType, func::external, ctx::Ptr{Cvoid})

Set the way patches are constructed for a SNESPATCH solver

Logically Collective

Input Parameters:

snes - the SNESPATCH solver
ctype - the PCPatchConstructType selecting the patch construction strategy
func - user callback that builds the patches, used only when ctype is PC_PATCH_USER or PC_PATCH_PYTHON; may be NULL otherwise
ctx - optional application context passed to func

Calling sequence of func:

pc - the PC associated with the SNESPATCH solver
npatch - number of patches
patches - the IS that define each patch
patchIterationSet - how the patches are iterated over
ctx - optional application context

Level: advanced

See also: SNESPATCH, PCPATCH, PCPatchSetConstructType(), PCPatchConstructType

External Links

PETSc Manual: SNES/SNESPatchSetConstructType

source

PETSc.LibPETSc.SNESPatchSetDiscretisationInfo — Method

SNESPatchSetDiscretisationInfo(petsclib::PetscLibType, snes::AbstractSNES, nsubspaces::PetscInt, dms::Vector{<:AbstractPetscDM}, bs::Vector{PetscInt}, nodesPerCell::Vector{PetscInt}, cellNodeMap::PetscInt, subspaceOffsets::Vector{PetscInt}, numGhostBcs::PetscInt, ghostBcNodes::Vector{PetscInt}, numGlobalBcs::PetscInt, globalBcNodes::Vector{PetscInt})

Provide the per-subspace discretisation information required by a SNESPATCH to build patch problems

Logically Collective

Input Parameters:

snes - the SNESPATCH solver
nsubspaces - the number of discretisation subspaces (e.g. fields)
dms - array of length nsubspaces of DMs, one per subspace
bs - array of length nsubspaces giving the block size of each subspace
nodesPerCell - array of length nsubspaces giving the number of nodes per cell for each subspace
cellNodeMap - array of length nsubspaces; entry i is a cell-to-node map for subspace i
subspaceOffsets - array of length nsubspaces + 1 giving the starting global dof offset of each subspace
numGhostBcs - number of ghost (off-process) boundary-condition dofs
ghostBcNodes - array of length numGhostBcs of the ghost boundary-condition dof indices
numGlobalBcs - number of global boundary-condition dofs
globalBcNodes - array of length numGlobalBcs of the global boundary-condition dof indices

Level: advanced

See also: SNESPATCH, PCPATCH, PCPatchSetDiscretisationInfo(), SNESPatchSetComputeOperator(), SNESPatchSetComputeFunction()

External Links

PETSc Manual: SNES/SNESPatchSetDiscretisationInfo

source

PETSc.LibPETSc.SNESPicardComputeFunction — Method

SNESPicardComputeFunction(petsclib::PetscLibType, snes::AbstractSNES, x::AbstractPetscVec, f::AbstractPetscVec, ctx::Ptr{Cvoid})

Compute the residual A(x) x - b(x) using the callbacks registered by SNESSetPicard()

Collective

Input Parameters:

snes - the SNES context
x - the current iterate
ctx - unused application context; the Picard callbacks are retrieved from the attached DMSNES

Output Parameter:

f - the residual vector

Level: developer

See also: SNES, SNESSetPicard(), SNESPicardComputeMFFunction(), SNESPicardComputeJacobian()

External Links

PETSc Manual: SNES/SNESPicardComputeFunction

source

PETSc.LibPETSc.SNESPicardComputeJacobian — Method

SNESPicardComputeJacobian(petsclib::PetscLibType, snes::AbstractSNES, x1::AbstractPetscVec, J::AbstractPetscMat, B::AbstractPetscMat, ctx::Ptr{Cvoid})

Trivial Jacobian assembly callback used by SNESSetPicard(); the Picard operator is filled in by SNESPicardComputeFunction()

Collective

Input Parameters:

snes - the SNES context
x1 - the current iterate (unused)
J - the Jacobian matrix to assemble
B - the preconditioning matrix (unused)
ctx - unused application context

Level: developer

See also: SNES, SNESSetPicard(), SNESPicardComputeFunction(), SNESPicardComputeMFFunction()

External Links

PETSc Manual: SNES/SNESPicardComputeJacobian

source

PETSc.LibPETSc.SNESPicardComputeMFFunction — Method

SNESPicardComputeMFFunction(petsclib::PetscLibType, snes::AbstractSNES, x::AbstractPetscVec, f::AbstractPetscVec, ctx::Ptr{Cvoid})

Matrix-free residual A(x) x - b(x) used by SNESSetPicard() when the operator is applied through -snes_mf_operator

Collective

Input Parameters:

snes - the SNES context
x - the current iterate
ctx - unused application context; the Picard callbacks are retrieved from the attached DMSNES

Output Parameter:

f - the residual vector

Level: developer

See also: SNES, SNESSetPicard(), SNESPicardComputeFunction(), SNESPicardComputeJacobian()

External Links

PETSc Manual: SNES/SNESPicardComputeMFFunction

source

PETSc.LibPETSc.SNESPruneJacobianColor — Method

SNESPruneJacobianColor(petsclib::PetscLibType, snes::AbstractSNES, J::AbstractPetscMat, B::AbstractPetscMat)

Remove nondiagonal zeros in the Jacobian matrix and update the MatMFFD coloring information based on the new nonzero structure

Collective

Input Parameters:

snes - the SNES context
J - Jacobian matrix (not altered in this routine)
B - newly computed Jacobian matrix to use with preconditioner (generally the same as J)

Level: intermediate

See also: SNESComputeJacobianDefaultColor(), MatEliminateZeros(), MatFDColoringCreate(), MatFDColoringSetFunction()

External Links

PETSc Manual: SNES/SNESPruneJacobianColor

source

PETSc.LibPETSc.SNESPythonGetType — Method

pyname::String = SNESPythonGetType(petsclib::PetscLibType, snes::AbstractSNES)

Get the type of a SNES object implemented in Python set with SNESPythonSetType()

Not Collective

Input Parameter:

snes - the nonlinear solver (SNES) context.

Output Parameter:

pyname - full dotted Python name [package].module[.{class|function}]

Level: intermediate

See also: SNES, SNESCreate(), SNESSetType(), SNESPYTHON, PetscPythonInitialize(), SNESPythonSetType()

External Links

PETSc Manual: SNES/SNESPythonGetType

source

PETSc.LibPETSc.SNESPythonSetType — Method

SNESPythonSetType(petsclib::PetscLibType, snes::AbstractSNES, pyname::String)

Initialize a SNES object implemented in Python.

Collective

Input Parameters:

snes - the nonlinear solver (SNES) context.
pyname - full dotted Python name [package].module[.{class|function}]

Options Database Key:

-snes_python_type pyname - python class

Level: intermediate

See also: SNES, SNESCreate(), SNESSetType(), SNESPYTHON, PetscPythonInitialize(), SNESPythonGetType()

External Links

PETSc Manual: SNES/SNESPythonSetType

source

PETSc.LibPETSc.SNESQNSetRestartType — Method

SNESQNSetRestartType(petsclib::PetscLibType, snes::AbstractSNES, rtype::SNESQNRestartType)

Sets the restart type for SNESQN.

Logically Collective

Input Parameters:

snes - the iterative context
rtype - restart type, see SNESQNRestartType

Options Database Keys:

-snes_qn_restart_type (powell|periodic|none) - set the restart type
-snes_qn_m m - sets the number of stored updates and the restart period for periodic

Level: intermediate

See also: SNES, SNESQN, SNESQNRestartType, SNES_QN_RESTART_NONE, SNES_QN_RESTART_POWELL, SNES_QN_RESTART_PERIODIC, SNESQNType, SNESQNScaleType

External Links

PETSc Manual: SNES/SNESQNSetRestartType

source

PETSc.LibPETSc.SNESQNSetScaleType — Method

SNESQNSetScaleType(petsclib::PetscLibType, snes::AbstractSNES, stype::SNESQNScaleType)

Sets the scaling type for the inner inverse Jacobian in SNESQN.

Logically Collective

Input Parameters:

snes - the nonlinear solver context
stype - scale type, see SNESQNScaleType

Options Database Key:

-snes_qn_scale_type (diagonal|none|scalar|jacobian) - Scaling type

Level: intermediate

See also: SNES, SNESQN, SNESLineSearch, SNESQNScaleType, SNESSetJacobian(), SNESQNType, SNESQNRestartType

External Links

PETSc Manual: SNES/SNESQNSetScaleType

source

PETSc.LibPETSc.SNESQNSetType — Method

SNESQNSetType(petsclib::PetscLibType, snes::AbstractSNES, qtype::SNESQNType)

Sets the quasi-Newton variant to be used in SNESQN.

Logically Collective

Input Parameters:

snes - the iterative context
qtype - variant type, see SNESQNType

Options Database Key:

-snes_qn_type (lbfgs|broyden|badbroyden) - quasi-Newton type

Level: intermediate

See also: SNESQN, SNES_QN_LBFGS, SNES_QN_BROYDEN, SNES_QN_BADBROYDEN, SNESQNType, SNESQNScaleType, TAOLMVM, TAOBLMVM

External Links

PETSc Manual: SNES/SNESQNSetType

source

PETSc.LibPETSc.SNESRegister — Method

SNESRegister(petsclib::PetscLibType, sname::String, fnc::external)

Adds a method to the nonlinear solver package.

Not Collective

Input Parameters:

sname - name of a new user-defined solver
function - routine to create method context

Level: advanced

See also: SNESRegisterAll(), SNESRegisterDestroy()

External Links

PETSc Manual: SNES/SNESRegister

source

PETSc.LibPETSc.SNESReset — Method

SNESReset(petsclib::PetscLibType, snes::AbstractSNES)

Resets a SNES context to the state it was in before SNESSetUp() was called and removes any allocated Vec and Mat from its data structures

Collective

Input Parameter:

snes - the nonlinear iterative solver context obtained from SNESCreate()

Level: intermediate

See also: SNES, SNESDestroy(), SNESCreate(), SNESSetUp(), SNESSolve()

External Links

PETSc Manual: SNES/SNESReset

source

PETSc.LibPETSc.SNESResetCounters — Method

SNESResetCounters(petsclib::PetscLibType, snes::AbstractSNES)

Reset counters for linear iterations and function evaluations.

Logically Collective

Input Parameters:

snes - SNES context

Level: developer

See also: SNESGetNumberFunctionEvals(), SNESGetLinearSolveIterations(), SNESGetNPC()

External Links

PETSc Manual: SNES/SNESResetCounters

source

PETSc.LibPETSc.SNESResetFromOptions — Method

SNESResetFromOptions(petsclib::PetscLibType, snes::AbstractSNES)

Sets various SNES and KSP parameters from user options ONLY if the SNESSetFromOptions() was previously called

Collective

Input Parameter:

snes - the SNES context

Level: advanced

See also: SNES, SNESSetFromOptions(), SNESSetOptionsPrefix()

External Links

PETSc Manual: SNES/SNESResetFromOptions

source

PETSc.LibPETSc.SNESSetAlwaysComputesFinalResidual — Method

SNESSetAlwaysComputesFinalResidual(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

tells the SNES to always compute the residual (nonlinear function value) at the final solution

Logically Collective

Input Parameters:

snes - the shell SNES
flg - PETSC_TRUE to always compute the residual

Level: advanced

See also: SNES, SNESFAS, SNESSolve(), SNESGetAlwaysComputesFinalResidual()

External Links

PETSc Manual: SNES/SNESSetAlwaysComputesFinalResidual

source

PETSc.LibPETSc.SNESSetApplicationContext — Method

SNESSetApplicationContext(petsclib::PetscLibType, snes::AbstractSNES, ctx::Ptr{Cvoid})

Sets the optional user-defined context for the nonlinear solvers.

Logically Collective

Input Parameters:

snes - the SNES context
ctx - the application context

Level: intermediate

See also: SNES, SNESSetComputeApplicationContext(), SNESGetApplicationContext()

External Links

PETSc Manual: SNES/SNESSetApplicationContext

source

PETSc.LibPETSc.SNESSetCheckJacobianDomainError — Method

SNESSetCheckJacobianDomainError(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

tells SNESSolve() whether to check if the user called SNESSetJacobianDomainError() to indicate a Jacobian domain error after each Jacobian evaluation.

Logically Collective

Input Parameters:

snes - the SNES context
flg - indicates if or not to check Jacobian domain error after each Jacobian evaluation

Level: advanced

See also: SNES, SNESConvergedReason, SNESCreate(), SNESSetFunction(), SNESFunctionFn, SNESSetFunctionDomainError(), SNESGetCheckJacobianDomainError()

External Links

PETSc Manual: SNES/SNESSetCheckJacobianDomainError

source

PETSc.LibPETSc.SNESSetComputeApplicationContext — Method

SNESSetComputeApplicationContext(petsclib::PetscLibType, snes::AbstractSNES, compute::external, destroy::Ptr{Cvoid})

Sets an optional function to compute a user-defined context for the nonlinear solvers.

Logically Collective; No Fortran Support

Input Parameters:

snes - the SNES context
compute - function to compute the context
destroy - function to destroy the context, see PetscCtxDestroyFn for the calling sequence

Calling sequence of compute:

snes - the SNES context
ctx - context to be computed

Level: intermediate

See also: SNESGetApplicationContext(), SNESSetApplicationContext(), PetscCtxDestroyFn

External Links

PETSc Manual: SNES/SNESSetComputeApplicationContext

source

PETSc.LibPETSc.SNESSetComputeInitialGuess — Method

SNESSetComputeInitialGuess(petsclib::PetscLibType, snes::AbstractSNES, func::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Sets a routine used to compute an initial guess for the nonlinear problem

Logically Collective

Input Parameters:

snes - the SNES context
func - function evaluation routine, see SNESInitialGuessFn for the calling sequence
ctx - [optional] user-defined context for private data for the

function evaluation routine (may be NULL)

Level: intermediate

See also: SNES, SNESSolve(), SNESSetFunction(), SNESGetFunction(), SNESComputeFunction(), SNESSetJacobian(), SNESInitialGuessFn

External Links

PETSc Manual: SNES/SNESSetComputeInitialGuess

source

PETSc.LibPETSc.SNESSetConvergedReason — Method

SNESSetConvergedReason(petsclib::PetscLibType, snes::AbstractSNES, reason::SNESConvergedReason)

Sets the reason the SNES iteration was stopped.

Not Collective

Input Parameters:

snes - the SNES context
reason - negative value indicates diverged, positive value converged, see SNESConvergedReason or the

manual pages for the individual convergence tests for complete lists

Level: developer

See also: SNESGetConvergedReason(), SNESSetConvergenceTest(), SNESConvergedReason

External Links

PETSc Manual: SNES/SNESSetConvergedReason

source

PETSc.LibPETSc.SNESSetConvergenceHistory — Method

SNESSetConvergenceHistory(petsclib::PetscLibType, snes::AbstractSNES, a::Vector{PetscReal}, its::Vector{PetscInt}, na::PetscInt, reset::PetscBool)

Sets the arrays used to hold the convergence history.

Logically Collective

Input Parameters:

snes - iterative context obtained from SNESCreate()
a - array to hold history, this array will contain the function norms computed at each step
its - integer array holds the number of linear iterations for each solve.
na - size of a and its
reset - PETSC_TRUE indicates each new nonlinear solve resets the history counter to zero,

else it continues storing new values for new nonlinear solves after the old ones

Level: intermediate

See also: SNES, SNESSolve(), SNESGetConvergenceHistory()

External Links

PETSc Manual: SNES/SNESSetConvergenceHistory

source

PETSc.LibPETSc.SNESSetConvergenceTest — Method

SNESSetConvergenceTest(petsclib::PetscLibType, snes::AbstractSNES, func::external, ctx::Ptr{Cvoid}, destroy::Ptr{Cvoid})

Sets the function that is to be used to test for convergence of the nonlinear iterative solution.

Logically Collective

Input Parameters:

snes - the SNES context
func - routine to test for convergence
ctx - [optional] context for private data for the convergence routine (may be NULL)
destroy - [optional] destructor for the context (may be NULL; PETSC_NULL_FUNCTION in Fortran)

Calling sequence of func:

snes - the SNES context
it - the current iteration number
xnorm - the norm of the new solution
snorm - the norm of the step
fnorm - the norm of the function value
reason - output, the reason convergence or divergence as declared
ctx - the optional convergence test context

Level: advanced

See also: SNES, SNESConvergedDefault(), SNESConvergedSkip()

External Links

PETSc Manual: SNES/SNESSetConvergenceTest

source

PETSc.LibPETSc.SNESSetCountersReset — Method

SNESSetCountersReset(petsclib::PetscLibType, snes::AbstractSNES, reset::PetscBool)

Sets whether or not the counters for linear iterations and function evaluations are reset every time SNESSolve() is called.

Logically Collective

Input Parameters:

snes - SNES context
reset - whether to reset the counters or not, defaults to PETSC_TRUE

Level: developer

See also: SNESGetNumberFunctionEvals(), SNESGetLinearSolveIterations(), SNESGetNPC()

External Links

PETSc Manual: SNES/SNESSetCountersReset

source

PETSc.LibPETSc.SNESSetDM — Method

SNESSetDM(petsclib::PetscLibType, snes::AbstractSNES, dm::AbstractPetscDM)

Sets the DM that may be used by some SNES nonlinear solvers or their underlying preconditioners

Logically Collective

Input Parameters:

snes - the nonlinear solver context
dm - the DM, cannot be NULL

Level: intermediate

See also: DM, SNES, SNESGetDM(), KSPSetDM(), KSPGetDM()

External Links

PETSc Manual: SNES/SNESSetDM

source

PETSc.LibPETSc.SNESSetDivergenceTolerance — Method

SNESSetDivergenceTolerance(petsclib::PetscLibType, snes::AbstractSNES, divtol::PetscReal)

Sets the divergence tolerance used for the SNES divergence test.

Logically Collective

Input Parameters:

snes - the SNES context
divtol - the divergence tolerance. Use PETSC_UNLIMITED to deactivate the test. If the residual norm F(x^n) \ge divtol * F(x^0) the solver

is stopped due to divergence.

Options Database Key:

-snes_divergence_tolerance divtol - Sets divtol

Level: intermediate

See also: SNES, SNESSolve(), SNESSetTolerances(), SNESGetDivergenceTolerance()

External Links

PETSc Manual: SNES/SNESSetDivergenceTolerance

source

PETSc.LibPETSc.SNESSetErrorIfNotConverged — Method

SNESSetErrorIfNotConverged(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

Causes SNESSolve() to generate an error immediately if the solver has not converged.

Logically Collective

Input Parameters:

snes - iterative context obtained from SNESCreate()
flg - PETSC_TRUE indicates you want the error generated

Options Database Key:

-snes_error_if_not_converged (true|false) - cause an immediate error condition and stop the program if the solver does not converge

Level: intermediate

See also: SNES, SNESGetErrorIfNotConverged(), KSPGetErrorIfNotConverged(), KSPSetErrorIfNotConverged()

External Links

PETSc Manual: SNES/SNESSetErrorIfNotConverged

source

PETSc.LibPETSc.SNESSetForceIteration — Method

SNESSetForceIteration(petsclib::PetscLibType, snes::AbstractSNES, force::PetscBool)

force SNESSolve() to take at least one iteration regardless of the initial residual norm

Logically Collective

Input Parameters:

snes - the SNES context
force - PETSC_TRUE require at least one iteration

Options Database Key:

-snes_force_iteration force - Sets forcing an iteration

Level: intermediate

See also: SNES, TS, SNESSetDivergenceTolerance()

External Links

PETSc Manual: SNES/SNESSetForceIteration

source

PETSc.LibPETSc.SNESSetFromOptions — Method

SNESSetFromOptions(petsclib::PetscLibType, snes::AbstractSNES)

Sets various SNES and KSP parameters from user options.

Collective

Input Parameter:

snes - the SNES context

Options Database Keys:

-snes_type type - newtonls, newtontr, ngmres, ncg, nrichardson, qn, vi, fas, SNESType for complete list
-snes_rtol rtol - relative decrease in tolerance norm from initial
-snes_atol abstol - absolute tolerance of residual norm
-snes_stol stol - convergence tolerance in terms of the norm of the change in the solution between steps
-snes_divergence_tolerance divtol - if the residual goes above divtol*rnorm0, exit with divergence
-snes_max_it max_it - maximum number of iterations
-snes_max_funcs max_funcs - maximum number of function evaluations
-snes_force_iteration force - force SNESSolve() to take at least one iteration
-snes_max_fail max_fail - maximum number of line search failures allowed before stopping, default is none
-snes_max_linear_solve_fail - number of linear solver failures before SNESSolve() stops
-snes_lag_preconditioner lag - how often preconditioner is rebuilt (use -1 to never rebuild)
-snes_lag_preconditioner_persists (true|false) - retains the -sneslagpreconditioner information across multiple SNESSolve()
-snes_lag_jacobian lag - how often Jacobian is rebuilt (use -1 to never rebuild)
-snes_lag_jacobian_persists (true|false) - retains the -sneslagjacobian information across multiple SNESSolve()
-snes_convergence_test (default|skip|correct_pressure) - convergence test in nonlinear solver. default SNESConvergedDefault(). skip SNESConvergedSkip() means continue

iterating until maxit or some other criterion is reached, saving expense of convergence test. correctpressure SNESConvergedCorrectPressure() has special handling of a pressure null space.

-snes_monitor [ascii][:filename][:viewer format] - prints residual norm at each iteration. if no filename given prints to stdout
-snes_monitor_solution [ascii binary draw][:filename][:viewer format] - plots solution at each iteration
-snes_monitor_residual [ascii binary draw][:filename][:viewer format] - plots residual (not its norm) at each iteration
-snes_monitor_solution_update [ascii binary draw][:filename][:viewer format] - plots update to solution at each iteration
-snes_monitor draw::draw_lg - plots residual norm at each iteration
-snes_monitor_lg_range - plots function range at each iteration
-snes_monitor_pause_final - Pauses all monitor drawing after the solver ends
-snes_fd - use finite differences to compute Jacobian; very slow, only for testing
-snes_fd_color - use finite differences with coloring to compute Jacobian
-snes_mf_ksp_monitor - if using matrix-free multiply then print h at each KSP iteration
-snes_converged_reason - print the reason for convergence/divergence after each solve
-npc_snes_type type - the SNES type to use as a nonlinear preconditioner
-snes_test_jacobian [threshold] - compare the user provided Jacobian with one computed via finite differences to check for errors.

If a threshold is given, display only those entries whose difference is greater than the threshold.

-snes_test_jacobian_view - display the user provided Jacobian, the finite difference Jacobian and the difference between them

to help users detect the location of errors in the user provided Jacobian.

Options Database Keys for Eisenstat-Walker method:

-snes_ksp_ew - use Eisenstat-Walker method for determining linear system convergence
-snes_ksp_ew_version ver - version of Eisenstat-Walker method
-snes_ksp_ew_rtol0 rtol0 - Sets rtol0
-snes_ksp_ew_rtolmax rtolmax - Sets rtolmax
-snes_ksp_ew_gamma gamma - Sets gamma
-snes_ksp_ew_alpha alpha - Sets alpha
-snes_ksp_ew_alpha2 alpha2 - Sets alpha2
-snes_ksp_ew_threshold threshold - Sets threshold

Level: beginner

See also: SNESType, SNESSetOptionsPrefix(), SNESResetFromOptions(), SNES, SNESCreate(), MatCreateSNESMF(), MatFDColoring

External Links

PETSc Manual: SNES/SNESSetFromOptions

source

PETSc.LibPETSc.SNESSetFunction — Method

SNESSetFunction(petsclib::PetscLibType, snes::AbstractSNES, r::AbstractPetscVec, f::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Sets the function evaluation routine and function vector for use by the SNES routines in solving systems of nonlinear equations.

Logically Collective

Input Parameters:

snes - the SNES context
r - vector to store function values, may be NULL
f - function evaluation routine; for calling sequence see SNESFunctionFn
ctx - [optional] user-defined context for private data for the

function evaluation routine (may be NULL)

Level: beginner

See also: SNES, SNESGetFunction(), SNESComputeFunction(), SNESSetJacobian(), SNESSetPicard(), SNESFunctionFn

External Links

PETSc Manual: SNES/SNESSetFunction

source

PETSc.LibPETSc.SNESSetFunctionDomainError — Method

SNESSetFunctionDomainError(petsclib::PetscLibType, snes::AbstractSNES)

tells SNES that the input vector, a proposed new solution, to your function you provided to SNESSetFunction() is not in the function's domain. For example, a step with negative pressure.

Not Collective

Input Parameter:

snes - the SNES context

Level: advanced

See also: SNESCreate(), SNESSetFunction(), SNESFunctionFn, SNESSetJacobianDomainError(), SNESVISetVariableBounds(), SNESVISetComputeVariableBounds(), SNESLineSearchSetPreCheck(), SNESLineSearchSetPostCheck(), SNESConvergedReason, SNESGetConvergedReason(), SNES_DIVERGED_FUNCTION_DOMAIN, SNESSetObjectiveDomainError(), SNES_DIVERGED_OBJECTIVE_DOMAIN

External Links

PETSc Manual: SNES/SNESSetFunctionDomainError

source

PETSc.LibPETSc.SNESSetFunctionNorm — Method

SNESSetFunctionNorm(petsclib::PetscLibType, snes::AbstractSNES, norm::PetscReal)

Sets the last computed residual norm.

Logically Collective

Input Parameters:

snes - the SNES context
norm - the value of the norm

Level: developer

See also: SNES, SNESGetNormSchedule(), SNESComputeFunction(), VecNorm(), SNESSetFunction(), SNESSetInitialFunction(), SNESNormSchedule

External Links

PETSc Manual: SNES/SNESSetFunctionNorm

source

PETSc.LibPETSc.SNESSetFunctionType — Method

SNESSetFunctionType(petsclib::PetscLibType, snes::AbstractSNES, type::SNESFunctionType)

Sets the SNESFunctionType of the SNES method.

Logically Collective

Input Parameters:

snes - the SNES context
type - the function type

Level: developer

Values of the function type:

SNES_FUNCTION_DEFAULT - the default for the given SNESType
SNES_FUNCTION_UNPRECONDITIONED - an unpreconditioned function evaluation (this is the function provided with SNESSetFunction()
SNES_FUNCTION_PRECONDITIONED - a transformation of the function provided with SNESSetFunction()

See also: SNES, SNESFunctionType, SNESGetNormSchedule(), SNESComputeFunction(), VecNorm(), SNESSetFunction(), SNESSetInitialFunction(), SNESNormSchedule

External Links

PETSc Manual: SNES/SNESSetFunctionType

source

PETSc.LibPETSc.SNESSetGridSequence — Method

SNESSetGridSequence(petsclib::PetscLibType, snes::AbstractSNES, steps::PetscInt)

sets the number of steps of grid sequencing that SNES will do

Logically Collective

Input Parameters:

snes - the SNES context
steps - the number of refinements to do, defaults to 0

Options Database Key:

-snes_grid_sequence steps - Use grid sequencing to generate initial guess

Level: intermediate

See also: SNES, SNESGetLagPreconditioner(), SNESSetLagJacobian(), SNESGetLagJacobian(), SNESGetGridSequence(), SNESSetDM(), SNESSolve()

External Links

PETSc Manual: SNES/SNESSetGridSequence

source

PETSc.LibPETSc.SNESSetInitialFunction — Method

SNESSetInitialFunction(petsclib::PetscLibType, snes::AbstractSNES, f::AbstractPetscVec)

Set an already computed function evaluation at the initial guess to be reused by SNESSolve().

Logically Collective

Input Parameters:

snes - the SNES context
f - vector to store function value

Level: developer

See also: SNES, SNESFAS, SNESSetFunction(), SNESComputeFunction(), SNESSetInitialFunctionNorm()

External Links

PETSc Manual: SNES/SNESSetInitialFunction

source

PETSc.LibPETSc.SNESSetIterationNumber — Method

SNESSetIterationNumber(petsclib::PetscLibType, snes::AbstractSNES, iter::PetscInt)

Sets the current iteration number.

Not Collective

Input Parameters:

snes - SNES context
iter - iteration number

Level: developer

See also: SNESGetLinearSolveIterations()

External Links

PETSc Manual: SNES/SNESSetIterationNumber

source

PETSc.LibPETSc.SNESSetJacobian — Method

SNESSetJacobian(petsclib::PetscLibType, snes::AbstractSNES, Amat::AbstractPetscMat, Pmat::AbstractPetscMat, J::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Sets the function to compute Jacobian as well as the location to store the matrix.

Logically Collective

Input Parameters:

snes - the SNES context
Amat - the matrix that defines the (approximate) Jacobian
Pmat - the matrix to be used in constructing the preconditioner, usually the same as Amat.
J - Jacobian evaluation routine (if NULL then SNES retains any previously set value), see SNESJacobianFn for details
ctx - [optional] user-defined context for private data for the

Jacobian evaluation routine (may be NULL) (if NULL then SNES retains any previously set value)

Level: beginner

See also: SNES, KSPSetOperators(), SNESSetFunction(), MatMFFDComputeJacobian(), SNESComputeJacobianDefaultColor(), MatStructure, SNESSetPicard(), SNESJacobianFn, SNESFunctionFn

External Links

PETSc Manual: SNES/SNESSetJacobian

source

PETSc.LibPETSc.SNESSetJacobianDomainError — Method

SNESSetJacobianDomainError(petsclib::PetscLibType, snes::AbstractSNES)

tells SNES that the function you provided to SNESSetJacobian() at the proposed step. For example there is a negative element transformation.

Logically Collective

Input Parameter:

snes - the SNES context

Level: advanced

See also: SNESCreate(), SNESSetFunction(), SNESFunctionFn, SNESSetFunctionDomainError(), SNESVISetVariableBounds(), SNESVISetComputeVariableBounds(), SNESLineSearchSetPreCheck(), SNESLineSearchSetPostCheck(), SNESConvergedReason, SNESGetConvergedReason()

External Links

PETSc Manual: SNES/SNESSetJacobianDomainError

source

PETSc.LibPETSc.SNESSetKSP — Method

SNESSetKSP(petsclib::PetscLibType, snes::AbstractSNES, ksp::AbstractKSP)

Sets a KSP context for the SNES object to use

Not Collective, but the SNES and KSP objects must live on the same MPI_Comm

Input Parameters:

snes - the SNES context
ksp - the KSP context

Level: developer

See also: SNES, KSP, KSPGetPC(), SNESCreate(), KSPCreate()

External Links

PETSc Manual: SNES/SNESSetKSP

source

PETSc.LibPETSc.SNESSetLagJacobian — Method

SNESSetLagJacobian(petsclib::PetscLibType, snes::AbstractSNES, lag::PetscInt)

Set when the Jacobian is rebuilt in the nonlinear solve. See SNESSetLagPreconditioner() for determining how often the preconditioner is rebuilt.

Logically Collective

Input Parameters:

snes - the SNES context
lag - -1 indicates NEVER rebuild, 1 means rebuild every time the Jacobian is computed within a single nonlinear solve, 2 means every second time

the Jacobian is built etc. -2 means rebuild at next chance but then never again

Options Database Keys:

-snes_lag_jacobian_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_jacobian (-2|1|2|...) - sets the lag
-snes_lag_preconditioner_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_preconditioner (-2|1|2|...) - sets the lag.

Level: intermediate

See also: SNES, SNESGetLagPreconditioner(), SNESSetLagPreconditioner(), SNESGetLagJacobianPersists(), SNESSetLagPreconditionerPersists()

External Links

PETSc Manual: SNES/SNESSetLagJacobian

source

PETSc.LibPETSc.SNESSetLagJacobianPersists — Method

SNESSetLagJacobianPersists(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

Set whether or not the Jacobian lagging persists through multiple nonlinear solves

Logically collective

Input Parameters:

snes - the SNES context
flg - jacobian lagging persists if true

Options Database Keys:

-snes_lag_jacobian_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_jacobian (-2|1|2|...) - sets the lag
-snes_lag_preconditioner_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_preconditioner (-2|1|2|...) - sets the lag

Level: advanced

See also: SNES, SNESSetLagPreconditionerPersists(), SNESSetLagJacobian(), SNESGetLagJacobian(), SNESGetNPC()

External Links

PETSc Manual: SNES/SNESSetLagJacobianPersists

source

PETSc.LibPETSc.SNESSetLagPreconditioner — Method

SNESSetLagPreconditioner(petsclib::PetscLibType, snes::AbstractSNES, lag::PetscInt)

Sets when the preconditioner is rebuilt in the nonlinear solve SNESSolve().

Logically Collective

Input Parameters:

snes - the SNES context
lag - 1 means rebuild every time the Jacobian is computed within a single nonlinear solve, 2 means every second time

the Jacobian is built etc. -2 indicates rebuild preconditioner at next chance but then never rebuild after that

Options Database Keys:

-snes_lag_jacobian_persists (true|false) - sets the persistence through multiple SNESSolve()
-snes_lag_jacobian (-2|1|2|...) - sets the lag
-snes_lag_preconditioner_persists (true|false) - sets the persistence through multiple SNESSolve()
-snes_lag_preconditioner (-2|1|2|...) - sets the lag

Level: intermediate

See also: SNESGetLagPreconditioner(), SNESSetLagJacobian(), SNESGetLagJacobian(), SNESSetLagPreconditionerPersists(), SNESSetLagJacobianPersists(), SNES, SNESSolve()

External Links

PETSc Manual: SNES/SNESSetLagPreconditioner

source

PETSc.LibPETSc.SNESSetLagPreconditionerPersists — Method

SNESSetLagPreconditionerPersists(petsclib::PetscLibType, snes::AbstractSNES, flg::PetscBool)

Set whether or not the preconditioner lagging persists through multiple nonlinear solves

Logically Collective

Input Parameters:

snes - the SNES context
flg - preconditioner lagging persists if true

Options Database Keys:

-snes_lag_jacobian_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_jacobian (-2|1|2|...) - sets the lag
-snes_lag_preconditioner_persists (true|false) - sets the persistence through multiple SNES solves
-snes_lag_preconditioner (-2|1|2|...) - sets the lag

Level: developer

See also: SNES, SNESSetLagJacobianPersists(), SNESSetLagJacobian(), SNESGetLagJacobian(), SNESGetNPC(), SNESSetLagPreconditioner()

External Links

PETSc Manual: SNES/SNESSetLagPreconditionerPersists

source

PETSc.LibPETSc.SNESSetLineSearch — Method

SNESSetLineSearch(petsclib::PetscLibType, snes::AbstractSNES, linesearch::SNESLineSearch)

Sets the SNESLineSearch to be used for a given SNES

Collective

Input Parameters:

snes - iterative context obtained from SNESCreate()
linesearch - the linesearch object

Level: developer

See also: SNES, SNESLineSearch, SNESGetLineSearch()

External Links

PETSc Manual: SNES/SNESSetLineSearch

source

PETSc.LibPETSc.SNESSetMaxLinearSolveFailures — Method

SNESSetMaxLinearSolveFailures(petsclib::PetscLibType, snes::AbstractSNES, maxFails::PetscInt)

the number of failed linear solve attempts allowed before SNES returns with a diverged reason of SNES_DIVERGED_LINEAR_SOLVE

Logically Collective

Input Parameters:

snes - SNES context
maxFails - maximum allowed linear solve failures, use PETSC_UNLIMITED to have no limit on the number of failures

Options Database Key:

-snes_max_linear_solve_fail num - The number of failures before the solve is terminated

Level: intermediate

See also: SNESSetErrorIfNotConverged(), SNESGetLinearSolveFailures(), SNESGetMaxLinearSolveFailures(), SNESGetLinearSolveIterations()

External Links

PETSc Manual: SNES/SNESSetMaxLinearSolveFailures

source

PETSc.LibPETSc.SNESSetMaxNonlinearStepFailures — Method

SNESSetMaxNonlinearStepFailures(petsclib::PetscLibType, snes::AbstractSNES, maxFails::PetscInt)

Sets the maximum number of unsuccessful steps attempted by the nonlinear solver before it gives up and returns unconverged or generates an error

Not Collective

Input Parameters:

snes - SNES context
maxFails - maximum of unsuccessful steps allowed, use PETSC_UNLIMITED to have no limit on the number of failures

Options Database Key:

-snes_max_fail n - maximum number of unsuccessful steps allowed

Level: intermediate

See also: SNESSetErrorIfNotConverged(), SNESGetMaxLinearSolveFailures(), SNESGetLinearSolveIterations(), SNESSetMaxLinearSolveFailures(), SNESGetLinearSolveFailures(), SNESGetMaxNonlinearStepFailures(), SNESGetNonlinearStepFailures(), SNESCheckLineSearchFailure()

External Links

PETSc Manual: SNES/SNESSetMaxNonlinearStepFailures

source

PETSc.LibPETSc.SNESSetNGS — Method

SNESSetNGS(petsclib::PetscLibType, snes::AbstractSNES, f::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Sets the user nonlinear Gauss-Seidel routine for use with composed nonlinear solvers.

Input Parameters:

snes - the SNES context, usually of the SNESType SNESNGS
f - function evaluation routine to apply Gauss-Seidel, see SNESNGSFn for calling sequence
ctx - [optional] user-defined context for private data for the smoother evaluation routine (may be NULL)

Level: intermediate

See also: SNESNGS, SNESGetNGS(), SNESNCG, SNESGetFunction(), SNESComputeNGS(), SNESNGSFn

External Links

PETSc Manual: SNES/SNESSetNGS

source

PETSc.LibPETSc.SNESSetNPC — Method

SNESSetNPC(petsclib::PetscLibType, snes::AbstractSNES, npc::AbstractSNES)

Sets the nonlinear preconditioner to be used.

Collective

Input Parameters:

snes - iterative context obtained from SNESCreate()
npc - the SNES nonlinear preconditioner object

Options Database Key:

-npc_snes_type type - set the type of the SNES to use as the nonlinear preconditioner

Level: developer

See also: SNES, SNESNGS, SNESFAS, SNESGetNPC(), SNESHasNPC()

External Links

PETSc Manual: SNES/SNESSetNPC

source

PETSc.LibPETSc.SNESSetNPCSide — Method

SNESSetNPCSide(petsclib::PetscLibType, snes::AbstractSNES, side::PCSide)

Sets the nonlinear preconditioning side used by the nonlinear preconditioner inside SNES.

Logically Collective

Input Parameter:

snes - iterative context obtained from SNESCreate()

Output Parameter:

side - the preconditioning side, where side is one of

$PC_LEFT - left preconditioning PC_RIGHT - right preconditioning (default for most nonlinear solvers)$

Options Database Key:

-snes_npc_side (right|left) - nonlinear preconditioner side

Level: intermediate

See also: SNES, SNESGetNPC(), SNESNRICHARDSON, SNESNCG, SNESType, SNESGetNPCSide(), KSPSetPCSide(), PC_LEFT, PC_RIGHT, PCSide

External Links

PETSc Manual: SNES/SNESSetNPCSide

source

PETSc.LibPETSc.SNESSetNormSchedule — Method

SNESSetNormSchedule(petsclib::PetscLibType, snes::AbstractSNES, normschedule::SNESNormSchedule)

Sets the SNESNormSchedule used in convergence and monitoring of the SNES method, when norms are computed in the solving process

Logically Collective

Input Parameters:

snes - the SNES context
normschedule - the frequency of norm computation

Options Database Key:

-snes_norm_schedule (none|always|initialonly|finalonly|initialfinalonly) - set the schedule

Level: advanced

See also: SNESNormSchedule, SNESGetNormSchedule(), SNESComputeFunction(), VecNorm(), SNESSetFunction(), SNESSetInitialFunction()

External Links

PETSc Manual: SNES/SNESSetNormSchedule

source

PETSc.LibPETSc.SNESSetObjective — Method

SNESSetObjective(petsclib::PetscLibType, snes::AbstractSNES, obj::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Sets the objective function minimized by some of the SNES linesearch methods, used instead of the 2-norm of the residual in the line search

Logically Collective

Input Parameters:

snes - the SNES context
obj - objective evaluation routine; see SNESObjectiveFn for the calling sequence
ctx - [optional] user-defined context for private data for the objective evaluation routine (may be NULL)

Level: intermediate

See also: SNES, SNESLineSearch(), SNESGetObjective(), SNESComputeObjective(), SNESSetFunction(), SNESSetJacobian(), SNESObjectiveFn, SNESSetObjectiveDomainError()

External Links

PETSc Manual: SNES/SNESSetObjective

source

PETSc.LibPETSc.SNESSetObjectiveDomainError — Method

SNESSetObjectiveDomainError(petsclib::PetscLibType, snes::AbstractSNES)

tells SNES that the input vector, a proposed new solution, to your function you provided to SNESSetObjective() is not in the function's domain. For example, a step with negative pressure.

Not Collective

Input Parameter:

snes - the SNES context

Level: advanced

See also: SNESCreate(), SNESSetFunction(), SNESFunctionFn, SNESSetJacobianDomainError(), SNESVISetVariableBounds(), SNESVISetComputeVariableBounds(), SNESLineSearchSetPreCheck(), SNESLineSearchSetPostCheck(), SNESConvergedReason, SNESGetConvergedReason(), SNES_DIVERGED_OBJECTIVE_DOMAIN, SNESSetFunctionDomainError(), SNES_DIVERGED_FUNCTION_DOMAIN

External Links

PETSc Manual: SNES/SNESSetObjectiveDomainError

source

PETSc.LibPETSc.SNESSetOptionsPrefix — Method

SNESSetOptionsPrefix(petsclib::PetscLibType, snes::AbstractSNES, prefix::String)

Sets the prefix used for searching for all SNES options in the database.

Logically Collective

Input Parameters:

snes - the SNES context
prefix - the prefix to prepend to all option names

Level: advanced

See also: SNES, SNESSetFromOptions(), SNESAppendOptionsPrefix()

External Links

PETSc Manual: SNES/SNESSetOptionsPrefix

source

PETSc.LibPETSc.SNESSetPicard — Method

SNESSetPicard(petsclib::PetscLibType, snes::AbstractSNES, r::AbstractPetscVec, bp::Ptr{Cvoid}, Amat::AbstractPetscMat, Pmat::AbstractPetscMat, J::Ptr{Cvoid}, ctx::Ptr{Cvoid})

Use SNES to solve the system A(x) x = bp(x) + b via a Picard type iteration (Picard linearization)

Logically Collective

Input Parameters:

snes - the SNES context
r - vector to store function values, may be NULL
bp - function evaluation routine, may be NULL, for the calling sequence see SNESFunctionFn
Amat - matrix with which A(x) x - bp(x) - b is to be computed
Pmat - matrix from which preconditioner is computed (usually the same as Amat)
J - function to compute matrix values, for the calling sequence see SNESJacobianFn
ctx - [optional] user-defined context for private data for the function evaluation routine (may be NULL)

Level: intermediate

See also: SNES, SNESGetFunction(), SNESSetFunction(), SNESComputeFunction(), SNESSetJacobian(), SNESGetPicard(), SNESLineSearchPreCheckPicard(), SNESFunctionFn, SNESJacobianFn

External Links

PETSc Manual: SNES/SNESSetPicard

source

PETSc.LibPETSc.SNESSetSolution — Method

SNESSetSolution(petsclib::PetscLibType, snes::AbstractSNES, u::AbstractPetscVec)

Sets the solution vector for use by the SNES routines.

Logically Collective

Input Parameters:

snes - the SNES context obtained from SNESCreate()
u - the solution vector

Level: beginner

See also: SNES, SNESSolve(), SNESGetSolution(), Vec

External Links

PETSc Manual: SNES/SNESSetSolution

source

PETSc.LibPETSc.SNESSetTolerances — Method

SNESSetTolerances(petsclib::PetscLibType, snes::AbstractSNES, abstol::PetscReal, rtol::PetscReal, stol::PetscReal, maxit::PetscInt, maxf::PetscInt)

Sets various parameters used in SNES convergence tests.

Logically Collective

Input Parameters:

snes - the SNES context
abstol - the absolute convergence tolerance, F(x^n) \le abstol
rtol - the relative convergence tolerance, F(x^n) \le reltol * F(x^0)
stol - convergence tolerance in terms of the norm of the change in the solution between steps, || delta x || < stol*|| x ||
maxit - the maximum number of iterations allowed in the solver, default 50.
maxf - the maximum number of function evaluations allowed in the solver (use PETSC_UNLIMITED indicates no limit), default 10,000

Options Database Keys:

-snes_atol abstol - Sets abstol
-snes_rtol rtol - Sets rtol
-snes_stol stol - Sets stol
-snes_max_it maxit - Sets maxit
-snes_max_funcs maxf - Sets maxf (use unlimited to have no maximum)

Level: intermediate

See also: SNESSolve(), SNES, SNESSetDivergenceTolerance(), SNESSetForceIteration()

External Links

PETSc Manual: SNES/SNESSetTolerances

source

PETSc.LibPETSc.SNESSetType — Method

SNESSetType(petsclib::PetscLibType, snes::AbstractSNES, type::String)

Sets the algorithm/method to be used to solve the nonlinear system with the given SNES

Collective

Input Parameters:

snes - the SNES context
type - a known method

Options Database Key:

-snes_type type - Sets the method; see SNESType

Level: intermediate

See also: SNES, SNESSolve(), SNESType, SNESCreate(), SNESDestroy(), SNESGetType(), SNESSetFromOptions()

External Links

PETSc Manual: SNES/SNESSetType

source

PETSc.LibPETSc.SNESSetUp — Method

SNESSetUp(petsclib::PetscLibType, snes::AbstractSNES)

Sets up the internal data structures for the later use of a nonlinear solver SNESSolve().

Collective

Input Parameter:

snes - the SNES context

Level: advanced

See also: SNES, SNESCreate(), SNESSolve(), SNESDestroy(), SNESSetFromOptions()

External Links

PETSc Manual: SNES/SNESSetUp

source

PETSc.LibPETSc.SNESSetUpMatrices — Method

SNESSetUpMatrices(petsclib::PetscLibType, snes::AbstractSNES)

ensures that matrices are available for SNES Newton-like methods, this is called by SNESSetUp_XXX()

Collective

Input Parameter:

snes - SNES object to configure

Level: developer

Hand-written helpers

PETSc.LibPETSc.SNESGetJacobianMat — Method

SNESGetJacobianMat(petsclib, snes) -> PetscMat

Return the assembled system (A) matrix from snes, passing NULL for the preconditioner matrix, Jacobian function, and context. Use this when only the system matrix is needed (e.g. to attach a null space via MatSetNullSpace).

source