MultiMesh Refactor #409

mtao · 2023-09-30T05:09:20Z

The current multimesh interface is optimized for tri-tri mappings and not ready for the other mappings we will soon assert. It also has too many public interfaces.

This PR aims to allow for only 2 basic functions:
a) Add a new mesh (for components / apps to use)
b) map a simplex from one mesh to another (which is a one-to-many relationship)

There will also be 2 convenience functions for when users know direct relationships (in debug mode these relationships will be asserted upon):
c) map to parent(parent_mesh, Tuple)
d) map to child(child_mesh, Tuple)

NOTE: I am breaking / giving up on the TriMeshOperations working with multimesh in this PR. Too much further work si required and I want to get this PR in

…ctor

…ing cmakes

multimesh things

…ic map

…h attributes

…ization order

daniel-zint

Looks quite good. There is just quite a lot of commented out code that maybe can be removed.

Otherwise, my main issue is the inconsistency in the usage of Simplex and Tuple. I think that should be unified.

daniel-zint · 2023-10-06T21:41:52Z

src/wmtk/MultiMeshManager.hpp

+    // Simplex maps
+    //===========
+    // generic mapping function that maps a tuple from "this" mesh to the other mesh
+    std::vector<Simplex> map(const Mesh& my_mesh, const Mesh& other_mesh, const Simplex& my_simplex)


I think we should start using Doxygen comments.

daniel-zint · 2023-10-06T21:44:34Z

src/wmtk/MultiMeshManager.hpp


-    static std::vector<Simplex> find_all_simplices_in_child_mesh(const Mesh& parent_mesh, long child_id, const Simplex& simplex_parent);
+    Simplex map_to_parent(const Mesh& my_mesh, const Simplex& my_simplex) const;


You return Simplex in some functions and Tuple in others. I am OK with both but I feel like that should be somehow consistent.

i try to make sure that the tuple variants are called _tuple but this shuld be documented better

daniel-zint · 2023-10-06T21:50:33Z

src/wmtk/MultiMeshManager.hpp

+    };
+
+private:
+    Mesh* m_parent = nullptr;


I would have expected some sort of smart pointer here. I can't tell though if that makes the code any better.

it should be a std::weak_ptr i guess - but doing this would require making every Mesh a shared_ptr (and also disabling public construction of meshes)

daniel-zint · 2023-10-06T21:52:44Z

src/wmtk/MultiMeshManager.hpp

+    // bool is_child_mesh_valid(const Mesh& my_mesh, const Mesh& child_mesh) const;
+
+    //// checks that the map is consistent
+    // bool is_child_map_valid(const Mesh& my_mesh, const ChildData& child) const;


Are these functions still relevant?

no, i will delete - they got moved to unit test (DEBUG_MultiMeshManager)

daniel-zint · 2023-10-06T21:59:30Z

src/wmtk/MultiMeshManager.cpp

+{
+    // this is just to do a little redirection for simpplifying map_to_child (and potentially for a
+    // visitor pattern)
+    return map_to_child_tuples(my_mesh, m_children.at(child_id), my_simplex);


maybe assert existence of child_id in m_children

this is a std::vector and using at in debug mode makes an exception if it's not there

daniel-zint · 2023-10-06T22:24:27Z

src/wmtk/MultiMeshManager.cpp

-    MultiMeshManager::MultiMeshManager(MultiMeshManager&& o) = default;
-    MultiMeshManager& MultiMeshManager::operator=(const MultiMeshManager& o) = default;
-    MultiMeshManager& MultiMeshManager::operator=(MultiMeshManager&& o) = default;
+    auto [source_mesh_base_tuple, target_mesh_base_tuple] =


daniel-zint · 2023-10-06T22:25:47Z

src/wmtk/TriMeshOperationExecutor.cpp

@@ -389,11 +390,14 @@ TriMesh::TriMeshOperationExecutor::prepare_operating_tuples_for_child_meshes() c
            }
        }
    }
+    */


delete old code?

daniel-zint · 2023-10-06T22:29:39Z

src/wmtk/multimesh/utils/find_local_switch_sequence.cpp

+
+namespace {
+
+std::vector<PrimitiveType> find_local_switch_sequence_trimesh(


Not sure if multimesh requires a utils subfolder. To me, everything in the multimesh folder seems like utility functions.

TBH i want to move MultiMeshManager into this folder - and tehre's some more utilities i'm currently working on that might belong here

daniel-zint · 2023-10-06T23:07:52Z

tests/test_multi_mesh.cpp

 }

-TEST_CASE("test_split_multi_mesh","[multimesh][2D]")
+/*
+TEST_CASE("test_split_multi_mesh", "[multimesh][2D]")


Will this test case come back or is it dead code?

working on for next pr

daniel-zint · 2023-10-06T23:10:34Z

tests/tools/DEBUG_MultiMeshManager.cpp

+
+
+namespace wmtk::tests {
+void DEBUG_MultiMeshManager::check_child_mesh_valid(const Mesh& my_mesh, const Mesh& child_mesh)


This should perform some checks in the future, right? In that case I would prefer the "throw a message" thing and a CHECK(false) to make sure that other people are not using it without being aware that it does not do anything.

i dont know - i can't remember who/what the original purpose was for this and just copied it over. tired :(

codecov · 2023-10-07T06:26:41Z

Codecov Report

❗ No coverage uploaded for pull request base (main@68de8a5). Click here to learn what that means.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #409   +/-   ##
=======================================
  Coverage        ?   81.88%           
=======================================
  Files           ?      154           
  Lines           ?     4156           
  Branches        ?        0           
=======================================
  Hits            ?     3403           
  Misses          ?      753           
  Partials        ?        0

Flag	Coverage Δ
wildmeshing	`81.88% <0.00%> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

mtao added 8 commits September 25, 2023 15:01

minor updates

3b16584

Merge remote-tracking branch 'origin/main' into mtao/parent_mesh_refa…

9a4e0a9

…ctor

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

5bba75f

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

1b1ff94

half-baked updates to multimesh manager

b860c02

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

bbcdc7f

moving tuple vec to simplex util out of top level cofaces and separat…

d58d476

…ing cmakes

more updates to multimesh

ef40735

mtao mentioned this pull request Oct 1, 2023

Extract local switching operations from Mesh classes to autogen #411

Closed

mtao added 4 commits October 1, 2023 16:15

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

d3d95dd

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

5be0fef

adding multimesh utility functions to simplify manager implementation

bc151d5

warning suppression

049713a

mtao marked this pull request as draft October 3, 2023 14:34

mtao added 13 commits October 3, 2023 10:53

typo and warning supprssion

40960fc

finishing move of TupleInspector from previous commit

50eb9f4

fixing a namespace ambiguity in top level cofaces and updates to

c8e13a7

multimesh things

fixing reorder warning in simplexcollectoin

46b98af

build of multimesh works, need to unit test the multimesh utils

c218a36

generating interface for multimesh in Mesh

a5c211c

working on rejuvenating multimesh unit tests

fcd749a

merge

6969c8e

adding missed tuple to tuple parent map in multimesh and adding gener…

d77b1da

…ic map

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

b3a38a3

adding an initial implementation of the "identity map" multimesh mapping

fe969ee

moving simplex dim surjection to multimesh but hiding it in a private

b0318d5

updates to map_tuples to fix assert in debug + adding some comments

1546503

mtao self-assigned this Oct 5, 2023

mtao linked an issue Oct 5, 2023 that may be closed by this pull request

MultiMeshRefactor from Tri-Tri specialization #421

Closed

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

e1fb6b0

mtao added 7 commits October 5, 2023 12:36

making multimeshmanager default to -1 for default values

04e925e

adding some convenience functions sto start manually testing multimes…

b2690d0

…h attributes

Merge remote-tracking branch 'base/main' into mtao/parent_mesh_refactor

9b23567

enabled multimeshmanager registration unit and fixed identity initial…

deaccda

…ization order

adding some more unit tests on mulmanager

1415a93

old multimesh validation now in DEBUG_Multimeshmanager

256a3c2

rejuvenated 2d multimesh navigation

195c9fc

mtao requested a review from daniel-zint October 6, 2023 21:32

mtao marked this pull request as ready for review October 6, 2023 21:32

daniel-zint approved these changes Oct 6, 2023

View reviewed changes

comments and constness type changes frm daniel comments

ac0b795

mtao merged commit 19ec160 into wildmeshing:main Oct 7, 2023

mtao deleted the mtao/parent_mesh_refactor branch October 7, 2023 06:26

mtao mentioned this pull request Dec 1, 2023

MultiMeshManager dimensional initialization #392

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

MultiMesh Refactor #409

MultiMesh Refactor #409

mtao commented Sep 30, 2023 •

edited

Loading

daniel-zint left a comment

daniel-zint Oct 6, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

daniel-zint Oct 6, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

daniel-zint Oct 6, 2023

mtao Oct 7, 2023

codecov bot commented Oct 7, 2023


		static std::vector<Simplex> find_all_simplices_in_child_mesh(const Mesh& parent_mesh, long child_id, const Simplex& simplex_parent);
		Simplex map_to_parent(const Mesh& my_mesh, const Simplex& my_simplex) const;


		namespace {

		std::vector<PrimitiveType> find_local_switch_sequence_trimesh(



		namespace wmtk::tests {
		void DEBUG_MultiMeshManager::check_child_mesh_valid(const Mesh& my_mesh, const Mesh& child_mesh)

MultiMesh Refactor #409

MultiMesh Refactor #409

Conversation

mtao commented Sep 30, 2023 • edited Loading

daniel-zint left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented Oct 7, 2023

Codecov Report

mtao commented Sep 30, 2023 •

edited

Loading