MPI-based spline interpolation in distributed 3D volumes.
xspline3d is a Python library for high-efficiency spline
interpolation in 3D volumes. It supports MPI-based memory-distributed
parallel processing and allows large-scale 3D volumes to be decomposed along
the Z-axis (slab decomposition) or both the Y and Z axes (pencil
decomposition). This makes it well-suited for memory-distributed scientific
computing tasks on HPC systems.
- Spline interpolation in segmented 3D volumes
- Support for both slab (Z-axis) and pencil (Y-Z axes) domain decompositions
- MPI-based parallelism for distributed-memory environments
- High efficiency in both interpolator construction and evaluation
- Pure Python interface backed by high-performance libraries (NumPy, SciPy, Numba, etc.)
Install directly from GitHub via pip:
pip install git+https://github.com/TamaLab/xspline3d.gitOr clone the repository and install locally:
git clone https://github.com/TamaLab/xspline3d.git
cd xspline3d
pip install .- Python >= 3.11
- numba >= 0.61
- numpy >= 2.2
- scipy >= 1.15
- threadpoolctl >= 3
- mpi4py
Note on mpi4py: For proper functionality and performance, mpi4py must
be built against and consistent with your system's MPI installation (e.g.,
OpenMPI, MPICH). Please ensure your system's MPI module is loaded and the
MPICC environment variable is set correctly before installing mpi4py.
Refer to the environment.yaml file and its installation
notes for detailed instructions on how to set up mpi4py by building it from
source.
Example scripts demonstrating how to use the interpolators are provided in the examples/ directory.
This project is licensed under the BSD 3-Clause License. See the LICENSE file for details.
If you use xspline3d in your research, please cite the following paper:
Wenyang Zhao, Osamu Miyashita, and Florence Tama. 2026. xspline3d: A Python Library for MPI-Based Spline Interpolation Enforcing Global Continuity in Distributed 3D Volumes. In SCA/HPCAsia 2026 Workshops: Supercomputing Asia and International Conference on High Performance Computing in Asia Pacific Region Workshops (SCA/HPCAsiaWS 2026), January 26–29, 2026, Osaka, Japan. ACM, New York, NY, USA, 7 pages. https://doi.org/10.1145/3784828.3786260
An extended manuscript is also available in the paper/ directory.
- Wenyang Zhao wenyang.zhao@riken.jp
- Osamu Miyashita osamu.miyashita@riken.jp
- Florence Tama florence.tama@riken.jp
Computational Structural Biology Research Team
RIKEN Center for Computational Science