amrex::LPInfo Struct Reference

Configuration knobs for multilevel linear operators (grid agglomeration, metrics, etc.). More...

#include <AMReX_MLLinOp.H>

Public Member Functions
LPInfo &	setAgglomeration (bool x) noexcept
	Enable or disable grid agglomeration on the coarsest MLMG levels (x = true enables it).
LPInfo &	setConsolidation (bool x) noexcept
	Enable or disable consolidation (MPI rank grouping) on coarse levels (x toggles the feature).
LPInfo &	setSemicoarsening (bool x) noexcept
	Toggle plane-wise semicoarsening instead of full coarsening (x = true selects semicoarsening).
LPInfo &	setAgglomerationGridSize (int x) noexcept
	Override the target grid size x used when agglomerating patches.
LPInfo &	setConsolidationGridSize (int x) noexcept
	Override the consolidation grid cutoff x (cells per MPI task) used to trigger grouping.
LPInfo &	setConsolidationRatio (int x) noexcept
	Set the refinement ratio x between consolidated levels.
LPInfo &	setConsolidationStrategy (int x) noexcept
	Select the heuristic x used when forming consolidated grids.
LPInfo &	setMetricTerm (bool x) noexcept
	Indicate whether metric terms are present so downstream code can skip metric work when absent.
LPInfo &	setMaxCoarseningLevel (int n) noexcept
	Cap how many coarsening steps (standard or semi-) MLMG may perform by setting n.
LPInfo &	setMaxSemicoarseningLevel (int n) noexcept
	Cap the number of semicoarsening steps (when enabled) via n.
LPInfo &	setSemicoarseningDirection (int n) noexcept
	Lock the direction n used for semicoarsening (-1 restores the default heuristic).
LPInfo &	setHiddenDirection (int n) noexcept
	Specify a dimension n that should be treated as “hidden” (e.g., for thin domains).
LPInfo &	setDeterministic (bool x) noexcept
	Enable deterministic reductions even on GPUs (slower but reproducible) by toggling x.
bool	hasHiddenDimension () const noexcept
	True if a hidden dimension was configured via setHiddenDirection().

Static Public Member Functions
static constexpr int	getDefaultAgglomerationGridSize ()
static constexpr int	getDefaultConsolidationGridSize ()

Public Attributes
bool	do_agglomeration = true
bool	do_consolidation = true
bool	do_semicoarsening = false
int	agg_grid_size = -1
int	con_grid_size = -1
int	con_ratio = 2
int	con_strategy = 3
bool	has_metric_term = true
int	max_coarsening_level = 30
int	max_semicoarsening_level = 0
int	semicoarsening_direction = -1
int	hidden_direction = -1
bool	deterministic = false
	Enable deterministic mode for GPU operations.

Detailed Description

Configuration knobs for multilevel linear operators (grid agglomeration, metrics, etc.).

The fluent setters below allow applications to toggle optional capabilities such as agglomeration, semicoarsening, and deterministic reductions.

Member Function Documentation

getDefaultAgglomerationGridSize()

static constexpr int amrex::LPInfo::getDefaultAgglomerationGridSize ( )

inlinestaticconstexpr

getDefaultConsolidationGridSize()

static constexpr int amrex::LPInfo::getDefaultConsolidationGridSize ( )

inlinestaticconstexpr

hasHiddenDimension()

bool amrex::LPInfo::hasHiddenDimension ( ) const

inlinenoexcept

True if a hidden dimension was configured via setHiddenDirection().

setAgglomeration()

LPInfo & amrex::LPInfo::setAgglomeration ( bool  x )

inlinenoexcept

Enable or disable grid agglomeration on the coarsest MLMG levels (x = true enables it).

setAgglomerationGridSize()

LPInfo & amrex::LPInfo::setAgglomerationGridSize ( int  x )

inlinenoexcept

Override the target grid size x used when agglomerating patches.

setConsolidation()

LPInfo & amrex::LPInfo::setConsolidation ( bool  x )

inlinenoexcept

Enable or disable consolidation (MPI rank grouping) on coarse levels (x toggles the feature).

setConsolidationGridSize()

LPInfo & amrex::LPInfo::setConsolidationGridSize ( int  x )

inlinenoexcept

Override the consolidation grid cutoff x (cells per MPI task) used to trigger grouping.

setConsolidationRatio()

LPInfo & amrex::LPInfo::setConsolidationRatio ( int  x )

inlinenoexcept

Set the refinement ratio x between consolidated levels.

setConsolidationStrategy()

LPInfo & amrex::LPInfo::setConsolidationStrategy ( int  x )

inlinenoexcept

Select the heuristic x used when forming consolidated grids.

setDeterministic()

LPInfo & amrex::LPInfo::setDeterministic ( bool  x )

inlinenoexcept

Enable deterministic reductions even on GPUs (slower but reproducible) by toggling x.

setHiddenDirection()

LPInfo & amrex::LPInfo::setHiddenDirection ( int  n )

inlinenoexcept

Specify a dimension n that should be treated as “hidden” (e.g., for thin domains).

setMaxCoarseningLevel()

LPInfo & amrex::LPInfo::setMaxCoarseningLevel ( int  n )

inlinenoexcept

Cap how many coarsening steps (standard or semi-) MLMG may perform by setting n.

setMaxSemicoarseningLevel()

LPInfo & amrex::LPInfo::setMaxSemicoarseningLevel ( int  n )

inlinenoexcept

Cap the number of semicoarsening steps (when enabled) via n.

setMetricTerm()

LPInfo & amrex::LPInfo::setMetricTerm ( bool  x )

inlinenoexcept

Indicate whether metric terms are present so downstream code can skip metric work when absent.

setSemicoarsening()

LPInfo & amrex::LPInfo::setSemicoarsening ( bool  x )

inlinenoexcept

Toggle plane-wise semicoarsening instead of full coarsening (x = true selects semicoarsening).

setSemicoarseningDirection()

LPInfo & amrex::LPInfo::setSemicoarseningDirection ( int  n )

inlinenoexcept

Lock the direction n used for semicoarsening (-1 restores the default heuristic).

Member Data Documentation

agg_grid_size

int amrex::LPInfo::agg_grid_size = -1

con_grid_size

int amrex::LPInfo::con_grid_size = -1

con_ratio

int amrex::LPInfo::con_ratio = 2

con_strategy

int amrex::LPInfo::con_strategy = 3

deterministic

bool amrex::LPInfo::deterministic = false

Enable deterministic mode for GPU operations.

do_agglomeration

bool amrex::LPInfo::do_agglomeration = true

do_consolidation

bool amrex::LPInfo::do_consolidation = true

do_semicoarsening

bool amrex::LPInfo::do_semicoarsening = false

has_metric_term

bool amrex::LPInfo::has_metric_term = true

hidden_direction

int amrex::LPInfo::hidden_direction = -1

max_coarsening_level

int amrex::LPInfo::max_coarsening_level = 30

max_semicoarsening_level

int amrex::LPInfo::max_semicoarsening_level = 0

semicoarsening_direction

int amrex::LPInfo::semicoarsening_direction = -1

---

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

• /home/runner/work/amrex/amrex/Src/LinearSolvers/MLMG/AMReX_MLLinOp.H