Growing mesh case #600

fsimonis · 2024-11-27T15:52:10Z

This PR adds a growing mesh case.

Two solvers define the same mesh, which grows at given points in time.

Inspiration for this case is the formation of additional residue layers on the floor of an ocean over time.

@MakisH is this how you envision the separate solver folder to work?

Checklist:

I added a summary of any user-facing changes (compared to the last release) in the changelog-entries/<PRnumber>.md.
I will remember to squash-and-merge, providing a useful summary of the changes of this PR.

MakisH

I know that this is still a draft, but I had a quick look anyway, so I left some comments that I know are probably already on your list.

growing-mesh/README.md

MakisH · 2024-12-03T12:26:29Z

growing-mesh/README.md

+- A who runs first
+- B who runs second


We so far avoid calling participants and data generic names, but try to relate to a specific physical domain and application use case. This is currently only partially described in the naming conventions.

Of course, this is fine for now (and fine for an integration test), but before merging, we should better have more descriptive names.

They both represent the same mesh and don't model any physical behaviour.

Not sure what we would name them.

There is a physical motivation for this scenario. We could borrow names there?

The pyhsical motivation is to couple a hydrogeological and geomechanical model and let them iterate until they agree on effective stress.
Without the iterative scheme, there is no correction and the geomechanical model doesn't have a function.

So, I cannot see obvious names that we can use here.

But then we need to better express the motivation in the description and say that the modeling is (extremely) simplified, as is the case in general in our tutorials.

With the current formulation, it is a great integration test, but it looks a bit odd as a tutorial.

I added the motivation and more details about the solver.

growing-mesh/README.md

growing-mesh/solver/solver.py

MakisH

A few more minor changes. The case starts, I have not run it beyond Compute consistent mapping yet.

I directly edited the run scripts for consistency.

MakisH · 2025-08-29T05:06:23Z

growing-mesh/solver-python/pyproject.toml

@@ -0,0 +1,11 @@
+[project]
+name = "growing"


Could we give a more descriptive name?
Not necessary, but If a user were to install this outside the venv, a growing binary lying around would be quite strange.

One direction could be growing-mesh-solver-python.

MakisH · 2025-08-29T05:07:21Z

growing-mesh/README.md

+
+## Motivation
+
+This case is an extremely simplified version of the scenario presented by Jean-Marc Gratien during Coupled 2023 *Implementing a Comprehensive Hydromechanical Model for Sedimentary Basins by Coupling a 3D Mechanical Code to a Classic Basin Fluid Flow Code with the PreCICE Library*.


Do we need the title case (with PreCICE) here? 😬

MakisH · 2025-08-29T05:10:49Z

growing-mesh/README.md

+
+The problem consists of a unit-square uniformly discretized by 512 x 512 nodes.
+The unit square is partitioned among the ranks of a solver according to an n x m grid, where 512 must be divisible by n and m.
+The mesh starts with 2 nodes in z direction and at a given frequency, 2 nodes are added to the mesh, changing only the load per rank, not the partitioning.


The original was a bit difficult for me to read, maybe this is better:

Suggested change

The mesh starts with 2 nodes in z direction and at a given frequency, 2 nodes are added to the mesh, changing only the load per rank, not the partitioning.

The mesh starts with two nodes in z direction. In regular intervals, two nodes are added to the mesh, changing only the load per rank, but not the partitioning.

MakisH · 2025-08-29T05:15:07Z

growing-mesh/README.md

+The problem consists of a unit-square uniformly discretized by 512 x 512 nodes.
+The unit square is partitioned among the ranks of a solver according to an n x m grid, where 512 must be divisible by n and m.
+The mesh starts with 2 nodes in z direction and at a given frequency, 2 nodes are added to the mesh, changing only the load per rank, not the partitioning.
+


"A and B" are mentioned below, but not introduced. Compromise:

Suggested change

While motivated by a setup with a hydrogeological and a geomechanical model, we name these participants A and B, as this tutorial only aims to demonstrate the any such growing mesh case scenario.

MakisH · 2025-08-29T05:18:21Z

growing-mesh/README.md

+```bash
+cd B
+./run.sh 2
+```


A post-processing section would be nice, hinting to what the user should see in the end. Note that there is no other figure (besides the config visualization) in this tutorial. Someone that does not yet know what "growing mesh" means would need to read and imagine a bit first, before being able to get a first overview.

An image with the mesh in two different times (potentially with partitions shown) would be helpful.

Does this case even produce any results to show how the mesh grows?

MakisH · 2025-08-29T05:23:45Z

growing-mesh/solver-python/solver.py

+            coords = getMeshAtTimeWindow(tw)
+            if rank == 0:
+                print(
+                    f"{participant_name}: Event grows local mesh from {oldCount} to {len(coords)}"


I fixed the formatting of these strings (here and in line 58). Cross-check.

Interestingly, while the solver messages show up in the correct order when running with two ranks, they show up separately at the end when running serially:

--[precice] time-window 11 (max: 12), t 10, Dt 1, max-dt 1 ---[precice] Time window completed ---[precice] Mapping "Data-B" for t=11 from "B-Mesh" to "A-Mesh" ---[precice] time-window 12 (max: 12), t 11, Dt 1, max-dt 1 ---[precice] Reinitializing Participant ---[precice] Close distributed communication channels ---[precice] Setting up primary communication to coupling partner/s ---[precice] Primary ranks are connected ---[precice] Setting up preliminary secondary communication to coupling partner/s ---[precice] Prepare partition for mesh A-Mesh ---[precice] Gather mesh A-Mesh ---[precice] Send global mesh A-Mesh ---[precice] Receive global mesh B-Mesh ---[precice] Setting up secondary communication to coupling partner/s ---[precice] Secondary ranks are connected ---[precice] Time window completed ---[precice] Computing "nearest-neighbor" mapping from mesh "B-Mesh" to mesh "A-Mesh" in "read" direction. ---[precice] Mapping distance min:0 max:0 avg: 0 var: 0 cnt: 2621440 ---[precice] Mapping "Data-B" for t=12 from "B-Mesh" to "A-Mesh" ---[precice] Reached end at: final time-window: 12, final time: 12 ---[precice] Implicitly finalizing in destructor ---[precice] Close communication channels Rank 0/1 has partition (0, 0)/(1, 1) Each of 1 partitions has node size 512x512 = 262144 for a total of 262144 nodes on the base A: data: 0.0 * 524288 A: data: 1.0 * 524288 A: data: 2.0 * 524288 A: Event grows local mesh from 524288 to 1048576 and global mesh from 524288 to 1048576 A: data: 3.0 * 1048576 A: data: 4.0 * 1048576 A: data: 5.0 * 1048576 A: Event grows local mesh from 1048576 to 1572864 and global mesh from 1048576 to 1572864 A: data: 6.0 * 1572864 A: data: 7.0 * 1572864 A: data: 8.0 * 1572864 A: Event grows local mesh from 1572864 to 2097152 and global mesh from 1572864 to 2097152 A: data: 9.0 * 2097152 A: data: 10.0 * 2097152 A: data: 11.0 * 2097152 A: Event grows local mesh from 2097152 to 2621440 and global mesh from 2097152 to 2621440

growing-mesh/precice-config.xml

fsimonis self-assigned this Nov 27, 2024

fsimonis requested a review from MakisH November 27, 2024 15:53

This was referenced Nov 27, 2024

Missing requirements.txt and venv in some Python-based examples #547

Open

Create mini packages for python-based solvers #568

Open

MakisH reviewed Dec 3, 2024

View reviewed changes

fsimonis force-pushed the growing-mesh-case branch from e322653 to 4ea4cd4 Compare March 14, 2025 09:43

fsimonis force-pushed the growing-mesh-case branch from 4ea4cd4 to dcbe41c Compare August 6, 2025 12:04

fsimonis added 3 commits August 6, 2025 15:12

Added growing mesh case

7713f2d

Format with black

998d17b

Fix formatting

c7dd907

fsimonis force-pushed the growing-mesh-case branch from 48f2c5c to c7dd907 Compare August 6, 2025 13:12

Add motivation and update docs

9f7667b

fsimonis marked this pull request as ready for review August 14, 2025 12:15

MakisH added 4 commits August 29, 2025 06:53

Merge branch 'develop' into growing-mesh-case

1d899a3

Add logging in run scripts and switch shell to bash

3f76611

Add pip freeze step in run scripts

c18fa51

Fix Python string formatting

cff528b

MakisH reviewed Aug 29, 2025

View reviewed changes

growing-mesh/precice-config.xml Outdated Show resolved Hide resolved

MakisH added 3 commits August 29, 2025 09:34

Add cleaning scripts

00cd678

Fix strings format

c0f2629

Disable debug logging

4b58c77


		## Motivation

		This case is an extremely simplified version of the scenario presented by Jean-Marc Gratien during Coupled 2023 Implementing a Comprehensive Hydromechanical Model for Sedimentary Basins by Coupling a 3D Mechanical Code to a Classic Basin Fluid Flow Code with the PreCICE Library.

	The mesh starts with 2 nodes in z direction and at a given frequency, 2 nodes are added to the mesh, changing only the load per rank, not the partitioning.
	The mesh starts with two nodes in z direction. In regular intervals, two nodes are added to the mesh, changing only the load per rank, but not the partitioning.



	While motivated by a setup with a hydrogeological and a geomechanical model, we name these participants A and B, as this tutorial only aims to demonstrate the any such growing mesh case scenario.

Uh oh!

Growing mesh case #600

Are you sure you want to change the base?

Growing mesh case #600

Uh oh!

Conversation

fsimonis commented Nov 27, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MakisH left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fsimonis Aug 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

MakisH left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

fsimonis commented Nov 27, 2024 •

edited

Loading

fsimonis Aug 6, 2025 •

edited

Loading

MakisH left a comment •

edited

Loading