MedApplication #11419

philbucher · 2023-07-23T11:42:42Z

This PR adds the first version of the interface to the MED-library (which is the mesh/data format used by Salome and Code_Aster). No external source code is added, the MED-library must be installed/compiled separately before this new App can be compiled. I.e. same as we include Trilinos or Metis

This is a full replacement of the KratosSalomePlugin, as then the med-files can be read directly.

Important missing feature is the support for SubModelParts, which I will implement in a follow-up PR. So far it can read and write Main-Modelparts, I tested it with the large examples from my thesis.

Also important to mention is that I work exclusively with Geometries, Elements and Conditions need to be created afterwards, as we also intend to use for other workflows. For this I would use #11370

loumalouomega · 2023-07-23T11:45:00Z

The med lib is this one https://docs.salome-platform.org/latest/dev/MEDCoupling/developer/med-file.html?

loumalouomega · 2023-07-23T11:46:36Z

The med lib is this one https://docs.salome-platform.org/latest/dev/MEDCoupling/developer/med-file.html?

Link is broken :P http://www.salome-platform.org/downloads/current-version

BTW it is LGPL if anyone wonders

loumalouomega · 2023-07-23T11:47:24Z

I will take a look ASAP (I started yesterday my holydays during for one month)

philbucher · 2023-07-23T14:45:52Z

I will take a look ASAP (I started yesterday my holydays during for one month)

no worries, I am not in a hurry

philbucher · 2023-07-23T14:46:40Z

The med lib is this one https://docs.salome-platform.org/latest/dev/MEDCoupling/developer/med-file.html?

yes exactly

applications/MedApplication/med_inc.h

applications/MedApplication/custom_python/add_custom_utilities_to_python.cpp

matekelemen · 2023-07-24T11:10:53Z

applications/MedApplication/custom_utilities/med_testing_utilities.cpp

+            for (auto geom_2 : rModelPart2.Geometries()) {
+                if (have_same_nodes(geom_1, geom_2)) {
+                    CheckGeometriesAreEqual(geom_1, geom_2);
+                    goto here;


I am pretty sure I tried, but break IIRC works only for the inner loop

https://stackoverflow.com/questions/1257744/can-i-use-break-to-exit-multiple-nested-for-loops

Unless I'm missing something, you're just breaking from the inner loop, not the outer one.

Seems so, but I also know that implementing this took me a while (first time I eve used goto)

Since I am a bit out of the loop rn I will leave it as is and revisit it next time I develop

applications/MedApplication/custom_utilities/med_testing_utilities.cpp

matekelemen · 2023-07-24T11:20:58Z

applications/MedApplication/custom_utilities/med_testing_utilities.cpp

+    check_geoms(rModelPart1, rModelPart2);
+    check_geoms(rModelPart2, rModelPart1);


it seems to me that have_same_nodes and CheckGeometriesAreEqual are commutative, so the second check (check_geom(rModelPart2, rModelPart);) is unnecessary.

not sure

ModelPart1 could have less geometries than ModelPart2, which would only be detected in the second call

Checking if the geometries are the same in a modelpart is quite nasty :/

true, but you're already checking whether the two ModelParts have the same number of geometries in line 121. Anyway, performance in a test isn't that important so this is just a minor comment.

hm good point

I will think about this again next time I develop, I remember that I thought long about this

applications/MedApplication/med_application.h

applications/MedApplication/tests/med_files/hexahedral_8N/create_mesh.py

applications/MedApplication/tests/med_files/tetrahedral_4N/create_mesh.py

loumalouomega

Some comments

philbucher · 2023-08-06T13:14:22Z

Thanks guys for the reviews

I will address the comments next week so that we can proceed

Is there anyone interested in testing this?

loumalouomega · 2023-08-06T20:13:43Z

Is there anyone interested in testing this?

Not now, but maybe in the future

philbucher · 2023-09-03T20:57:13Z

ping @matekelemen do you have more comments?

I would like to merge this so that I can continue to add support for sub-meshes :)

matekelemen

Nope, just the minor nitpicking comments from earlier. Thanks the idea and implementation, I'm planning to use it but I have to get familiar with Salome first!

matekelemen

Whoops, I meant to approve :'(

philbucher · 2023-09-04T10:31:46Z

Blocking to prevent accidental merge? 👀😂

I will have another look at you comments before merging, thanks!

matekelemen · 2023-09-19T13:51:49Z

@philbucher alright I'm about try it rn, but I'd need support for submodelparts. Any idea how difficult it is to implement that (or have you already started working on it)?

philbucher · 2023-09-21T07:30:00Z

@philbucher alright I'm about try it rn, but I'd need support for submodelparts. Any idea how difficult it is to implement that (or have you already started working on it)?

great!

Yes I already started, but then I removed it again bcs it grew a lot. (for one PR)

Lets merge this and then I can push my draft in the next days. I am guessing you only need the reading? Writing is a bit more tricky, sadly the med-lib is not very user-friendly, it is pretty confusing when it comes to submodelparts

philbucher · 2023-09-21T07:30:53Z

How does your model look like? Do you only have one level of submodelparts? Or e.g. also SubSubModelParts

matekelemen · 2023-09-21T07:58:13Z

I am guessing you only need the reading?

Yep, reading is fine for now. In fact, I don't think we'll ever want to output MED files. Should we ever manage to finally break away from MDPAs, it'd be best to choose an industry-wide standard format rather than yet another application-specific one.

Do you only have one level of submodelparts?

For now, yes. I have a volume mesh with parts of its surfaces in subpartitions. I'm fine with taking baby steps and initially supporting 1-depth models exclusively, but we should only merge once we support arbitrarily nested models.

Can you plz open a follow-up PR with what you have so far? Maybe I can piggyback off your work and help you with the rest of the impl.

philbucher · 2023-09-21T20:46:17Z

choose an industry-wide standard format

I am honestly not sure this exists 😅. Do you know?

For now, yes. I have a volume mesh with parts of its surfaces in subpartitions. I'm fine with taking baby steps and initially supporting 1-depth models exclusively, but we should only merge once we support arbitrarily nested models.

And do you have entities that are part of more than one SubModelPart?

Can you plz open a follow-up PR with what you have so far? Maybe I can piggyback off your work and help you with the rest of the impl.

Yes, I will do that in a few days. In the meantime, do you (roughly) know how the med-lib stores the mesh? Especially do you know what groups and families are?

philbucher · 2023-09-21T21:12:49Z

merging to avoid blocking the developments

Thanks for the comments everYbody!

matekelemen · 2023-09-22T22:07:39Z

I am honestly not sure this exists 😅. Do you know?

I sure hope there is one, though I have to admit that I have next to no experience with proper modelling and meshing.

I heard about UNV, but it seems to be a bit dated and barely used
how about CAD formats like IGES and STEP? Are they scalable, or were they exclusively meant for CAD?
I'm quite fond of what XDMF does with mapping arbitrary formats, though it's a bit lacking in efficiency.

And do you have entities that are part of more than one SubModelPart?

Polytopes? No. Vertices? Definitely.

In the meantime, do you (roughly) know how the med-lib stores the mesh?

Nope, this is the first time I heard about MED, and I'll have to look everything up.

philbucher · 2023-09-25T11:00:05Z

I am honestly not sure this exists 😅. Do you know?

I sure hope there is one, though I have to admit that I have next to no experience with proper modelling and meshing.

I heard about UNV, but it seems to be a bit dated and barely used

I think so too, afaik it is only used by some older CFD solvers, i.e. not sth we should rely on exclusively

how about CAD formats like IGES and STEP? Are they scalable, or were they exclusively meant for CAD?

I dont think they include any mesh

I'm quite fond of what XDMF does with mapping arbitrary formats, though it's a bit lacking in efficiency.

I think this might be a good option, even though not the most user-friendly. At least it uses hdf under the hood which is nice

And do you have entities that are part of more than one SubModelPart?

Polytopes? No. Vertices? Definitely.

Ok thats fine, enabling this is not too difficult :)

In the meantime, do you (roughly) know how the med-lib stores the mesh?

Nope, this is the first time I heard about MED, and I'll have to look everything up.

Ok no worries

philbucher added 30 commits October 26, 2022 22:10

initial add empty app

a051155

add MedIO

08f0d49

expose to python

f6b0257

Merge remote-tracking branch 'origin/master' into med-application

35988a1

add empty WriteModelpart

af3f067

minor

7e3fb89

prepare C++ testing

0244fcc

missing const in ModelPart

6c289ec

check SubModelParts too

5843bf2

WIP add logo to reame

6b0f2ac

adding const corrected function

76486e6

hide deprecation in BaseClass

f1b0574

using const function

1359f8c

fixes

e23b4ac

Merge remote-tracking branch 'origin/master' into med-application

a42facc

move to utilities

325d4f7

WIP test geometries

b2e1e79

fixes

75d2a52

initial add FindMED

0a74605

use test name as file name

89383be

prepare py tests

70aa9f9

WIP include Med

70db4f8

dummy usage of Med

4f52b32

minor in readme

7d9d992

minor

c8d5b03

draft for partitioning support

003466f

minor in FindMED

3e570f1

handle med file handle types correctly

3146489

missing

4a0b85b

starting to add some version checks

7cf750e

minor

482c36d

matekelemen reviewed Jul 24, 2023

View reviewed changes

loumalouomega reviewed Aug 4, 2023

View reviewed changes

applications/MedApplication/med_application.h Outdated Show resolved Hide resolved

loumalouomega reviewed Aug 4, 2023

View reviewed changes

applications/MedApplication/tests/med_files/hexahedral_8N/create_mesh.py Outdated Show resolved Hide resolved

loumalouomega reviewed Aug 4, 2023

View reviewed changes

applications/MedApplication/tests/med_files/tetrahedral_4N/create_mesh.py Outdated Show resolved Hide resolved

loumalouomega reviewed Aug 4, 2023

View reviewed changes

philbucher added 3 commits August 13, 2023 23:28

only forward declare

f4345c5

Merge remote-tracking branch 'origin/master' into med-application

d1b14e7

review requests

b42fdbd

matekelemen requested changes Sep 4, 2023

View reviewed changes

matekelemen approved these changes Sep 4, 2023

View reviewed changes

philbucher merged commit 635c2c0 into master Sep 21, 2023
15 of 16 checks passed

philbucher deleted the med-application branch September 21, 2023 21:13

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

MedApplication #11419

MedApplication #11419

philbucher commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

philbucher commented Jul 23, 2023

philbucher commented Jul 23, 2023

matekelemen Jul 24, 2023

philbucher Aug 17, 2023

philbucher Sep 3, 2023

matekelemen Sep 4, 2023

philbucher Sep 21, 2023

matekelemen Jul 24, 2023 •

edited

Loading

philbucher Aug 17, 2023

matekelemen Sep 4, 2023

philbucher Sep 21, 2023

loumalouomega left a comment

philbucher commented Aug 6, 2023

loumalouomega commented Aug 6, 2023

philbucher commented Sep 3, 2023

matekelemen left a comment

matekelemen left a comment

philbucher commented Sep 4, 2023

matekelemen commented Sep 19, 2023

philbucher commented Sep 21, 2023 •

edited

Loading

philbucher commented Sep 21, 2023

matekelemen commented Sep 21, 2023 •

edited

Loading

philbucher commented Sep 21, 2023

philbucher commented Sep 21, 2023

matekelemen commented Sep 22, 2023

philbucher commented Sep 25, 2023

		check_geoms(rModelPart1, rModelPart2);
		check_geoms(rModelPart2, rModelPart1);

MedApplication #11419

MedApplication #11419

Conversation

philbucher commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

loumalouomega commented Jul 23, 2023

philbucher commented Jul 23, 2023

philbucher commented Jul 23, 2023

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

matekelemen Jul 24, 2023 • edited Loading

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

loumalouomega left a comment

Choose a reason for hiding this comment

philbucher commented Aug 6, 2023

loumalouomega commented Aug 6, 2023

philbucher commented Sep 3, 2023

matekelemen left a comment

Choose a reason for hiding this comment

matekelemen left a comment

Choose a reason for hiding this comment

philbucher commented Sep 4, 2023

matekelemen commented Sep 19, 2023

philbucher commented Sep 21, 2023 • edited Loading

philbucher commented Sep 21, 2023

matekelemen commented Sep 21, 2023 • edited Loading

philbucher commented Sep 21, 2023

philbucher commented Sep 21, 2023

matekelemen commented Sep 22, 2023

philbucher commented Sep 25, 2023

matekelemen Jul 24, 2023 •

edited

Loading

philbucher commented Sep 21, 2023 •

edited

Loading

matekelemen commented Sep 21, 2023 •

edited

Loading