Repository navigation
Open
Description
OpenImpala Strategic Development Roadmap (v4.5+)
OpenImpala is a high-performance, block-structured C++17 / AMReX and Python library for computing transport properties and microstructural metrics directly on 3D voxelised images from laptops to multi-GPU / MPI clusters.
Shipped Foundation (v4.0 – v4.6.0)
- Zero-Fortran C++17 AMReX Kernels: Native CPU, MPI, OpenMP, and CUDA/HIP
ParallelForstencils across tortuosity and homogenisation solvers. - Exact Exterior-Face Finite-Volume Numerics &
O(N)MLMG Fast Path (Correct voxel tortuosity boundary conditions and avoid MLMG cut-face aborts #300, perf(tortuosity): GMRES-accelerated MLMG, O(N) CPU flood fill, and auto box sizing #301 —v4.5.0): Half-cell Dirichlet boundary conditions, harmonic-mean face conductances,GMRES(24)-preconditioned matrix-freeMLMGdefault solver,O(N)CPU BFS flood fill, and rank-aware domain decomposition. - Automated CI, V&V & Slim Wheel Packaging (fix(v4.5.0): verified citations, MLMG-ready CLI, canonical V&V, and 2-rank MPI pytest #302, build(v4.5.1): drop Fortran & AMReX EB, add cp313 & sm_89 #303, feat(v4.6.0): full Python API, universal pure-Python wheel, and fast wheel CI #304 —
v4.5.0/v4.6.0):clang-format,clang-tidy, CTest, single- and multi-rank MPIpytest, canonical finite-volume verification, automated PR performance benchmarks, Apptainer.sifbuilds, parallelccachewheel CI, compact Linux CPU (openimpala,4.2 MB) + CUDA 12 (openimpala-cuda,89.7 MB,sm_80/sm_89/compute_75) wheels for Python 3.9–3.13, and a universalpy3-none-anyfallback wheel for macOS, Windows, and ARM64. - Full Python API & Pure-Python Parity (feat(v4.6.0): full Python API, universal pure-Python wheel, and fast wheel CI #304 —
v4.6.0):- Optional 3D concentration/potential field return (
solution_fieldviareturn_field=True) andeffective_diffusivityonoi.tortuosity(). - Multi-phase diffusivity/conductivity maps (
phase_diffusivities) with scale-invariant SI solver tolerances. - Symmetric
$3 \times 3$ effective diffusivity homogenisation tensor (oi.effective_diffusivity()). - Microstructural characterisation suite:
specific_surface_area,connected_components,particle_size_distribution,through_thickness_profile, andrev_study. - Non-cubic / anisotropic voxel spacing (
voxel_size=(dz, dy, dx)and CLI--voxel-size) (Non-cubic voxel field #11).
- Optional 3D concentration/potential field return (
Current Focus: Community Adoption & JOSS (#226)
- External testing, issue reporting, and workflow validation with battery-imaging and porous-media research groups on
v4.6.0. - Updated tutorials, API reference polish, and JOSS software paper submission (Sustainability: Publish OpenImpala in JOSS #226).
Longer-Term Solver & Physics Research Tracks
-
Matrix-free
EffectiveDiffusivityMLMG: Port periodic$3 \times 3$ homogenisation solves ontoGMRES-MLMGfor lower memory footprint and faster GPU/CPU execution, and review strict zero-flux Neumann + nullspace pinning for blocked periodic barriers. - Data & Ecosystem Bridges: Cloud-native OME-Zarr streaming (Modern Data Ecosystem: Support Cloud-Native OME-Zarr I/O #218), built-in thresholding helpers (Usability: Built-in Otsu Thresholding Utility #214), Conda-Forge recipe (Epic: Python Packaging and Distribution Strategy (PyPI & Conda-Forge) #147), and streamlined BPX / PyBaMM parameterisation helpers (Feature Request: Direct Memory-Coupling API for Multi-Scale Frameworks (PyBaMM) #65).
-
Advanced Electrochemistry & Multi-Physics Research (standalone research tracks):
- Frequency-domain impedance / blocking-electrode MacMullin transport (Electrode Tortuosity Factor #12, Feature Request: Implement Transient (Time-Dependent) Solver #35) and grayscale sub-voxel partial-volume conductances (Epic: Direct-from-Grayscale Physics Solvers (Partial Volume Formulation) #213).
- Runtime-configurable boundary conditions and extensible PDE base classes (Refactor: Generalize Boundary Condition Setup for Tortuosity Calculation #16, Refactor: Implement Component-Based Architecture for Multi-Physics Extensibility #15, Modification of Boundary Conditions #5, Heat Equation Computation #4).
Scope Note
External GUI plugins and beamline framework wrappers (e.g., Fiji/ImageJ, napari, Orange Canvas, Trame, Savu) are out of scope for the core OpenImpala repository so core maintenance stays focused on numerical accuracy, solver scalability, and a clean Python/C++ library interface.
Metadata
Metadata
Assignees
Labels
No labels
Activity
[-]# OpenImpala: Strategic Development Roadmap (v2.0)[/-][+]OpenImpala: Strategic Development Roadmap (v2.0)[/+][-]OpenImpala: Strategic Development Roadmap (v2.0)[/-][+]OpenImpala: Strategic Development Roadmap (v3.0)[/+][-]OpenImpala: Strategic Development Roadmap (v3.0)[/-][+]OpenImpala: Strategic Development Roadmap (v4.5+)[/+]