inciter::Refiner class

Mesh refiner for interfacing the mesh refinement library.

Reference

Public types

enum class RefMode: std::size_t { T0REF = 1, DTREF }: Mode of operation: the way Refiner is used.

Constructors, destructors, conversion operators

Refiner(std::size_t meshid, const CProxy_Transporter& transporter, const CProxy_Sorter& sorter, const tk::CProxy_MeshWriter& meshwriter, const CProxy_Discretization& discretization, const CProxy_RieCG& riecg, const CProxy_LaxCG& laxcg, const CProxy_ZalCG& zalcg, const CProxy_KozCG& kozcg, const tk::RefinerCallback& cbr, const tk::SorterCallback& cbs, const std::vector<std::size_t>& ginpoel, const tk::UnsMesh::CoordMap& coordmap, const std::map<int, std::vector<std::size_t>>& bface, const std::vector<std::size_t>& triinpoel, const std::map<int, std::vector<std::size_t>>& bnode, int nchare) explicit: Constructor.
Refiner(CkMigrateMessage* m) explicit: Migrate constructor.

Public functions

void query(int fromch, const EdgeSet& edges): Incoming query for a list boundary edges for which this chare compiles shared edges.
void recvquery(): Receive receipt of boundary edge lists to quer.
void response(): Respond to boundary edge list queries.
void bnd(int fromch, const std::vector<int>& chares): Receive shared boundary edges for our mesh chunk.
void recvbnd(): Receive receipt of shared boundary edges.
void reorder(): Query Sorter and update local mesh with the reordered one.
void start(): Start new step of initial mesh refinement/derefinement.
void next(): Continue after finishing a refinemen/derefinementt step.
void dtref(const std::map<int, std::vector<std::size_t>>& bface, const std::map<int, std::vector<std::size_t>>& bnode, const std::vector<std::size_t>& triinpoel): Start mesh refinement (during time stepping, t>0)
void refine(): Do a single step of mesh refinemen/derefinementt (only tag edges)
void addRefBndEdges(int fromch, const AMR::EdgeData& ed, const std::unordered_set<size_t>& intermediates): Receive newly added mesh edges and locks on our chare boundary.
void correctref(): Correct refinement to arrive at conforming mesh across chare boundaries.
void comExtra(): Communicate refined edges after a refinement/derefinement step.
void perform(): Perform mesh refinement and decide how to continue.
void sendProxy(): Send Refiner proxy to Discretization objects.
auto refinementFields() const -> std::tuple<std::vector<std::string>, std::vector<std::vector<tk::real>>, std::vector<std::string>, std::vector<std::vector<tk::real>>>: Get refinement field data in mesh cells.

Charm++ pack/unpack serializer member functions

std::size_t m_meshid: Mesh ID.
std::size_t m_ncit: Number of parallel-compatibility (mesh ref correction) iterations.
CProxy_Transporter m_host: Host proxy.
CProxy_Sorter m_sorter: Mesh sorter proxy.
tk::CProxy_MeshWriter m_meshwriter: Mesh writer proxy.
CProxy_Discretization m_disc: Discretization base proxy.
CProxy_RieCG m_riecg: Discretization scheme proxy.
CProxy_LaxCG m_laxcg: Discretization scheme proxy.
CProxy_ZalCG m_zalcg: Discretization scheme proxy.
CProxy_KozCG m_kozcg: Discretization scheme proxy.
tk::RefinerCallback m_cbr: Charm++ callbacks associated to compile-time tags for refiner.
tk::SorterCallback m_cbs: Charm++ callbacks associated to compile-time tags for sorter.
std::vector<std::size_t> m_ginpoel: Tetrtahedron element connectivity of our chunk of the mesh (global ids)
tk::UnsMesh::Chunk m_el
std::vector<std::size_t>& m_inpoel: Alias to element connectivity with local node IDs in m_el.
std::vector<std::size_t>& m_gid: Alias to global node IDs of owned elements in m_el.
std::unordered_map<std::size_t, std::size_t>& m_lid: Alias to local node IDs associated to the global ones of owned elements in m_el.
tk::UnsMesh::CoordMap m_coordmap: Coordinates associated to global node IDs of our mesh chunk.
tk::UnsMesh::Coords m_coord: Coordinates of mesh nodes of our chunk of the mesh.
std::map<int, std::vector<std::size_t>> m_bface: List of boundary faces associated to side-set IDs.
std::map<int, std::vector<std::size_t>> m_bnode: List of boundary nodes associated to side-set IDs.
std::vector<std::size_t> m_triinpoel: Boundary face-node connectivity.
int m_nchare: Total number of refiner chares.
RefMode m_mode: True if initial AMR, false if during time stepping.
std::vector<std::string> m_initref: Initial mesh refinement type list (in reverse order)
std::size_t m_ninitref: Number of initial mesh refinement/derefinement steps.
AMR::mesh_adapter_t m_refiner: Mesh refiner (library) object.
std::size_t m_nref: Counter during distribution of newly added nodes to chare-boundary edges.
std::size_t m_nbnd: Counter for number of chares contributing to chare boundary edges.
std::size_t m_extra: Number of chare-boundary newly added nodes that need correction.
std::unordered_set<int> m_ch: Chares we share at least a single edge with.
std::unordered_map<Edge, std::vector<int>, Hash<2>, Eq<2>> m_edgech: Edge->chare map used to build shared boundary edges.
std::unordered_map<int, EdgeSet> m_chedge: Chare->edge map used to build shared boundary edges.
AMR::EdgeData m_localEdgeData: Refinement data associated to edges (edges stored with node-gids)
std::unordered_map<int, std::vector<std::tuple<Edge, int, int, AMR::Edge_Lock_Case>>> m_remoteEdgeData: Refinement data associated to edges shared with other chares (edges stored with node-gids)
std::unordered_map<int, std::vector<Edge>> m_remoteEdges: Edges received from other chares.
std::unordered_set<size_t> m_intermediates: Intermediate nodes.
std::unordered_map<int, std::unordered_set<std::size_t>> m_nodeCommMap: Global mesh node IDs bordering the mesh chunk held by fellow worker chares associated to their chare IDs for the coarse mesh.
TetSet m_oldTets: Tetrahedra before refinement/derefinement step.
std::unordered_map<std::size_t, Edge> m_addedNodes: Newly added mesh nodes (local id) and their parents (local ids)
std::unordered_map<std::size_t, std::size_t> m_addedTets: Newly added mesh cells (local id) and their parent (local id)
std::set<std::size_t> m_removedNodes: Newly removed mesh node local ids.
std::unordered_map<std::size_t, std::size_t> m_amrNodeMap: Node id maps from old mesh to new refined mesh.
std::size_t m_oldntets: Number of tetrahedra in the mesh before refinement/derefinement step.
std::unordered_map<int, FaceSet> m_coarseBndFaces: A unique set of faces associated to side sets of the coarsest mesh.
std::unordered_map<int, std::unordered_set<std::size_t>> m_coarseBndNodes: A unique set of nodes associated to side sets of the coarsest mesh.
std::vector<std::size_t> m_rid: Local -> refiner lib node id map.
std::unordered_map<std::size_t, std::size_t> m_lref: Local -> refiner lib node id map for previous mesh.
CkCallback m_writeCallback: Refiner lib -> local node id map for previous mesh.
void pup(PUP::er& p) override: Pack/Unpack serialize member function.
void operator|(PUP::er& p, Refiner& r): Pack/Unpack serialize operator|.

Enum documentation

enum class inciter::Refiner::RefMode: std::size_t

Mode of operation: the way Refiner is used.

Enumerators
T0REF	Initial (t<0) refinement.
DTREF	During time stepping (t>0)

Function documentation

inciter::Refiner::Refiner(std::size_t meshid, const CProxy_Transporter& transporter, const CProxy_Sorter& sorter, const tk::CProxy_MeshWriter& meshwriter, const CProxy_Discretization& discretization, const CProxy_RieCG& riecg, const CProxy_LaxCG& laxcg, const CProxy_ZalCG& zalcg, const CProxy_KozCG& kozcg, const tk::RefinerCallback& cbr, const tk::SorterCallback& cbs, const std::vector<std::size_t>& ginpoel, const tk::UnsMesh::CoordMap& coordmap, const std::map<int, std::vector<std::size_t>>& bface, const std::vector<std::size_t>& triinpoel, const std::map<int, std::vector<std::size_t>>& bnode, int nchare) explicit

Constructor.

Parameters
meshid in	Mesh ID
transporter in	Transporter (host) proxy
sorter in	Mesh reordering (sorter) proxy
meshwriter in	Mesh writer proxy
discretization in	Discretization base proxy
riecg in	Discretization scheme proxy
laxcg in	Discretization scheme proxy
zalcg in	Discretization scheme proxy
kozcg in	Discretization scheme proxy
cbr in	Charm++ callbacks for Refiner
cbs in	Charm++ callbacks for Sorter
ginpoel in	Mesh connectivity (this chare) using global node IDs
coordmap in	Mesh node coordinates (this chare) for global node IDs
bface in	File-internal elem ids of side sets
triinpoel in	Triangle face connectivity with global IDs
bnode in	Node lists of side sets
nchare in	Total number of refiner chares (chare array elements)

void inciter::Refiner::query(int fromch, const EdgeSet& edges)

Incoming query for a list boundary edges for which this chare compiles shared edges.

Parameters
fromch in	Sender chare ID
edges in	Chare-boundary edge list from another chare

void inciter::Refiner::bnd(int fromch, const std::vector<int>& chares)

Receive shared boundary edges for our mesh chunk.

Parameters
fromch in	Sender chare ID
chares in	Chare ids we share edges with

void inciter::Refiner::dtref(const std::map<int, std::vector<std::size_t>>& bface, const std::map<int, std::vector<std::size_t>>& bnode, const std::vector<std::size_t>& triinpoel)

Start mesh refinement (during time stepping, t>0)

Parameters
bface in	Boundary-faces mapped to side set ids
bnode in	Boundary-node lists mapped to side set ids
triinpoel in	Boundary-face connectivity

void inciter::Refiner::refine()

Do a single step of mesh refinemen/derefinementt (only tag edges)

During initial (t<0, t0ref) mesh refinement, this is a single step in a potentially multiple-entry list of initial adaptive mesh refinement steps. Distribution of the chare-boundary edges must have preceded this step, so that boundary edges (shared by multiple chares) can agree on a refinement that yields a conforming mesh across chare boundaries.

During-timestepping (t>0, dtref) mesh refinement this is called once, as we only do a single step during time stepping.

void inciter::Refiner::addRefBndEdges(int fromch, const AMR::EdgeData& ed, const std::unordered_set<size_t>& intermediates)

Receive newly added mesh edges and locks on our chare boundary.

Parameters
fromch in	Chare call coming from
ed in	Edges on chare boundary
intermediates in	Intermediate nodes

Receive edges on our chare boundary from other chares Other than update remoteEdge data, this function also updates locking information for such edges whos nodes are marked as intermediate by neighboring chares.

void inciter::Refiner::correctref()

Correct refinement to arrive at conforming mesh across chare boundaries.

This function is called repeatedly until there is not a a single edge that needs correction for the whole distributed problem to arrive at a conforming mesh across chare boundaries during a mesh refinement step.

void inciter::Refiner::perform()

Perform mesh refinement and decide how to continue.

First the mesh refiner object is called to perform a single step of mesh refinement. Then, if this function is called during a step (potentially multiple levels of) initial AMR, it evaluates whether to do another one. If it is called during time stepping, this concludes the single mesh refinement step and the new mesh is sent to the PDE worker (Discretization).

void inciter::Refiner::sendProxy()

Send Refiner proxy to Discretization objects.

This should be called when bound Discretization chare array elements have already been created.

std::tuple<std::vector<std::string>, std::vector<std::vector<tk::real>>, std::vector<std::string>, std::vector<std::vector<tk::real>>> inciter::Refiner::refinementFields() const

Get refinement field data in mesh cells.

Returns	Names and fields of mesh refinement data in mesh cells and nodes

void inciter::Refiner::pup(PUP::er& p) override

Pack/Unpack serialize member function.

Parameters
p in/out	Charm++'s PUP::er serializer object reference

void inciter::Refiner::operator|(PUP::er& p, Refiner& r)

Pack/Unpack serialize operator|.

Parameters
p in/out	Charm++'s PUP::er serializer object reference
r in/out	Refiner object reference

Variable documentation

tk::UnsMesh::Chunk inciter::Refiner::m_el

Elements of the mesh chunk we operate on

The first vector is the element connectivity (local IDs), the second vector is the global node IDs of owned elements, while the third one is a map of global->local node IDs.

std::unordered_map<std::size_t, std::size_t> inciter::Refiner::m_lref

Local -> refiner lib node id map for previous mesh.

Refiner lib -> local node id map

CkCallback inciter::Refiner::m_writeCallback

Refiner lib -> local node id map for previous mesh.

Child -> parent tet map for previous mesh Function to continue with after writing field output