tboudreaux/stroid
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(topology): vacuum coordinate and multi block

Browse Source 
Two major changes in this version. First stroid now embeds a vacuum coordinate as part of its StroidMesh file (this is a packed set of mfem meshes and GridFunction). This is a logical coordinate from 0 at the stellar surface to 1 at the mesh surface / compactified infinity which can be used by consumers to much more stablly infer position in the vacuum region. Second, there is a new topology backend, multi_block, which has been made the default. See the readme for more information but the basic jist is that multi_block addes 6 transition blocks onto the edge of the core domain. This allows for a much more well conditioned transition from the internal cartesien region to the external spherical region. The mesh conditioning improves by roughly a factor of 1000 for the same refinement level when compared to the legacy topology. The legacy topology is maintained as a option if core_mapping is set to spherified in the config.
This commit is contained in:
Emily Boudreaux
2026-09-09 08:11:08 -04:00
parent db727ebd7b
commit a5905e5fed
27 changed files with 1693 additions and 183 deletions
Download Patch File Download Diff File Expand all files Collapse all files

73
readme.md
View File

@@ -92,6 +92,8 @@ inf_bdr_id = 2
core_id = 1
envelope_id = 2
vacuum_id = 3
core_mapping = "multi_block"
[main.optimization_methods]
tmop = false
@@ -99,29 +101,53 @@ smoothstep = true
```
<!-- Table of what these parameters do -->
| Parameter | Description | Default |
|---------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|---------|
| refinement_levels | Number of uniform refinement levels to apply to the mesh after generation | 4 |
| order | The polynomial order of the finite elements in the mesh | 3 |
| include_external_domain | Whether to include an external domain extending to r_infinity | true |
| r_core | The radius of the core region of the star | 1.5 |
| r_star | The radius of the star | 5.0 |
| flattening | The flattening factor of the star (0 for spherical, >0 for oblate) | 0 |
| r_infinity | The outer radius of the external domain (if included) | 6.0 |
| r_instability | The radius at which no transformations are applied to the initial topology (to avoid singularities) | 1e-14 |
| core_steepness | The steepness of the transition between the core and envelope regions of the star | 1.0 |
| surface_bdr_id | The boundary ID to assign to the surface of the star | 1 |
| inf_bdr_id | The boundary ID to assign to the outer boundary of the external domain (if included) | 2 |
| core_id | The material ID to assign to the core region of the star | 1 |
| envelope_id | The material ID to assign to the envelope region of the star | 2 |
| vacuum_id | The material ID to assign to the vacuum region of the star (if included) | 3 |
| optimization_methods.tmop | The tmop flag enables or disables the use of TMOP ideal shape unit size metric optimization during mesh generation. This can help improve the quality of the generated mesh, but will dramatically increase the time required for mesh generation. | false |
| optimization_methods.smoothstep | The smoothstep flag enables or disables the use of a smoothstep function to transition between the core and envelope regions of the star. This can help improve the quality of the generated mesh | true |
| Parameter | Description | Default |
|---------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|---------------|
| refinement_levels | Number of uniform refinement levels to apply to the mesh after generation | 4 |
| order | The polynomial order of the finite elements in the mesh | 3 |
| include_external_domain | Whether to include an external domain extending to r_infinity | true |
| r_core | The radius of the core region of the star | 1.5 |
| r_star | The radius of the star | 5.0 |
| flattening | The flattening factor of the star (0 for spherical, >0 for oblate) | 0 |
| r_infinity | The outer radius of the external domain (if included) | 6.0 |
| r_instability | The radius at which no transformations are applied to the initial topology (to avoid singularities) | 1e-14 |
| core_steepness | The steepness of the transition between the core and envelope regions of the star | 1.0 |
| surface_bdr_id | The boundary ID to assign to the surface of the star | 1 |
| inf_bdr_id | The boundary ID to assign to the outer boundary of the external domain (if included) | 2 |
| core_id | The material ID to assign to the core region of the star | 1 |
| envelope_id | The material ID to assign to the envelope region of the star | 2 |
| vacuum_id | The material ID to assign to the vacuum region of the star (if included) | 3 |
| optimization_methods.tmop | The tmop flag enables or disables the use of TMOP ideal shape unit size metric optimization during mesh generation. This can help improve the quality of the generated mesh, but will dramatically increase the time required for mesh generation. | false |
| optimization_methods.smoothstep | The smoothstep flag enables or disables the use of a smoothstep function to transition between the core and envelope regions of the star. This can help improve the quality of the generated mesh | true |
| core_mapping | The core mapping strategy to use for the mesh generation. Options are "spherified" (legacy) or "multi_block" (conditioned). The multi_block strategy is strongly preferred for its improved condition number. | "multi_block" |
If no configuration file is provided, stroid will use the default parameters listed above. Further, configuration files
need only include parameters that differ from the defaults, any parameters not specified will use the default values.
### Conditioned core mapping
There are two core mapping strategies, spherified and multi_block. Generally multi_block should be strongly preferred. The
`core_mapping = "multi_block"` strategy avoids the radial rank loss at the eight corners of the spherified core
block. It uses a Cartesian center plus six transition blocks inside the core. The inner cube has circumscribed radius
`r_core / 2`; its six faces connect linearly to the existing spherical `r_core` interface. If enabled, spheroidal flattening is
applied afterwards.
```python
cfg = stroid.config.MeshConfig(core_mapping="multi_block", refinement_levels=2)
cfg.optimization_methods = stroid.config.OptimizationMethods(tmop=False)
mesh = stroid.GenerateMesh(cfg)
```
The optional, non-installed `geometry_quality_experiment` target may be used to measure the actual high-order geometry
at quadrature points, vertices, edges, and near-corner probes. You may build and run it explicitly:
```bash
meson compile -C build geometry_quality_experiment
build/tools/geometry_quality_experiment --orders 4 --refinements 2 \
--contraction-probe --probe-order 3 --output core_comparison.csv
```
### C++ Interface
Stroid can be used as a library in C++ projects. After installation, include the stroid header and link against the stroid library.
@@ -153,7 +179,14 @@ int main() {
## Example Meshes
An example mesh with the default configuration parameters is shown below (coloration indicates attribute IDs of different regions):
![Example Mesh](assets/imgs/ExampleMesh.png)
![Example Mesh](assets/imgs/ExampleMesh_multi-block.png)
The legacy spherified core mapping strategy is shown below as well
![Example Spheried Mesh](assets/imgs/ExampleMesh_spherified.png)
Note that both of these meshes are shown with 3 levels of refinement and polynomial order 3. Blue shows the stellar
domain while purple shows the vacuum domain.
## Funding
Stroid is developed as part of the 4D-STAR project.
@@ -161,4 +194,4 @@ Stroid is developed as part of the 4D-STAR project.
4D-STAR is funded by European Research Council (ERC) under the Horizon Europe programme (Synergy Grant agreement No.
101071505: 4D-STAR)
Work for this project is funded by the European Union. Views and opinions expressed are however those of the author(s)
only and do not necessarily reflect those of the European Union or the European Research Council.
only and do not necessarily reflect those of the European Union or the European Research Council.