- New implementation of boundary conditions
- Flux based schemes
- Fine tuning in MRA
- Multiple adapted grids
- Parallelism
- Release mechanism
Samurai
NSH2
Loïc Gouarin and Pierre Matalon
29 February 2024
This work is licensed under a Creative
Commons Attribution 4.0 International License
Work achieved during 2023
New implementation of boundary conditions
What kind ?
- A constant
- A function in space and time
- A function depending on other fields
New implementation of boundary conditions
New implementation of boundary conditions
New implementation of boundary conditions
New implementation of boundary conditions
- Classical boundary conditions are available (Dirichlet, Neumann).
- A user can easily define its own using a stencil formalism.
- We provide interpolation mechanism for prediction ghosts.
Flux based schemes
Loop on the faces, paying attention to the level jump.
Flux based schemes
Define the stencil.
Flux based schemes
A user can easily define
- a linear, homogeneous operator
- a linear, heterogeneous operator
- a non-linear operator
The flux can be conservative or non-conservative.
These operators can then be used for explicit or implicit time schemes.
Fine tuning in MRA
- Select the fields used for the multiresolution.
- Should work for SOA and AOS storage or a combination of both.
- Adapt the other fields on this new mesh.
Multiple adapted grids
Compression rates
Compression rates
| Level | Num. of cells | p4est | samurai (leaves) | samurai (all) | ratio |
|---|---|---|---|---|---|
| 9 | $66379$ | $2.57$ Mb | $33.68$ Kb | $121$ Kb | $21.24$ |
| 10 | $263767$ | $10.25$ Mb | $66.64$ Kb | $236.8$ Kb | $43.28$ |
| 11 | $1051747$ | $40.96$ Mb | $132.36$ Kb | $467.24$ Kb | $87.66$ |
| 12 | $4200559$ | $163.75$ Mb | $263.6$ Kb | $927$ Kb | $176.64$ |
| 13 | $16789627$ | $654.86$ Mb | $525.9$ Kb | $1.85$ Mb | $353.98$ |
| 14 | $67133575$ | $2.61$ Gb | $1.05$ Mb | $3.68$ Mb | $709.24$ |
Parallelism
- A subdomain has the information of all its neighbors.
- MPI is used.
- The ghosts between subdomains are updated using the algebra of sets.
Release mechanism
- Use of conventional commits
- Increment version automatically
- Create changelog automatically
- Make a release automatically
- Create the conda package on conda-forge
Make the changes available as soon as possible
Roadmap
Roadmap for 2024: numerical methods
- Various finite volume methods
- NSCBC
- Complex time schemes with operator splitting
Roadmap for 2024: code
- Optimization
- Load balancing
- GPU support
- IO optimization and compression
Scientific collaborations
- Lattice Boltzmann methods and multiresolution - Thomas Bellotti (IRMA/Université Strasbourg) and Benjamin Graille (LMO/Université Paris-Saclay)
- Plasma discharges and electric propulsion - Alejandro Alvarez (LPP/Ecole polytechnique) and Louis Reboul (ONERA)
- Direct numerical simulation of lithium-ion batteries based on high-resolution 3D images of porous electrode microstructures - Ali Asad (TotalEnergies) and Laurent François (ONERA)
- Sharp interface method for low Mach two-phase flows - Nicolas Grenier (LISN/Université Paris-Saclay) and Christian Tenaud (EM2C/Université Paris-Saclay)
- Low-Mach reactive flows - Christian Tenaud (EM2C/Université Paris-Saclay)
- Interfacial flow simulation - Giuseppe Orlando (CMAP/Ecole polytechnique) and Marica Pelanti (ENSTA/IP Paris)
- Mathematical modeling and simulation of non-equilibrium plasmas for the prediction of electric propulsion - Zoubaïr Tazakkati (CMAP/Ecole polytechnique)
- Simulation analysis on the Hydrogen risk - Luc Lecointre, Pierre-Alexandre Masset, Etienne Studer (CEA) and Christian Tenaud (EM2C/Université Paris-Saclay)
