SPHinXsys is set to undergo a major transformation, moving from traditional CPU parallelism to a new era of heterogeneous parallelism, where compute-intensive tasks can harness the power of both CPUs and GPUs. This evolution will be driven by SYCL (via Intel's DPC++), enabling us to leverage heterogeneous architectures with standard C++. Importantly, this transformation will be achieved with minimal disruption to the existing codebase, ensuring seamless continuity for current users.
The groundwork for this shift is already laid out. A 2D dambreak test case has been added to the test/test_sycl folder, showcasing the capabilities of our specially designed framework. What's unique about this framework is that it allows for the development and testing of numerical methods even in environments without GPUs or DPC++ installed. If these methods are crafted following our specified guidelines and prove functional, they will seamlessly operate in environments equipped with DPC++ and GPU support.
By embracing this new paradigm, SPHinXsys is positioning itself at the forefront of multi-physics modeling, where performance meets versatility.
Project status
Project communication
The most active development (or default) branch of this repository is master.
This branch gives the updated development of SPHinXsys using heterogeneous (using TBB and SYCL) parallelism.
The most stable branch of the repository is version1.0, which is based on CPU (using TBB) parallelism.
SPHinXsys (pronunciation: s'fink-sis) is an acronym from Smoothed Particle Hydrodynamics for industrial compleX systems. The multi-physics library uses SPH (smoothed particle hydrodynamics) as the underlying numerical method for both particle-based and mesh-based discretization. Due to the unified computational framework, SPHinXsys is able to carry out simulation and optimization at the same time. For more information on the SPHinXsys project, please check the project website: https://www.sphinxsys.org.
Using SPHinXsys library, straightforward and fast multi-physics modeling can be achieved. Here, we present several short examples in flow, solid dynamics, fluid structure interactions (FSI) and dynamic solid contact.
Through the unified computational framework in SPHinXsys, the algorithms for particle methods are full compatible to those in the classical finite volume method (FVM). The following gives an example of the flow around cylinder problem solved by FVM in SPHinXsys.
Note that the code for FVM algorithm is exact the same one for particle interaction in SPHinXsys. The only difference is that SPHinXsys reads a predefined mesh, other than generate particles, before the computation.
The unique target-driven optimization is able to achieve the optimization target and physical solution all-in-once, which is able to accelerate optimization process greatly. The following gives an example of optimizing the conductivity distribution for a thermal domain problem targeting minimum average temperature.
Note that the physical solution of the thermal domain (right) and the optimal distribution of conductivity (left) are obtained at the same time when optimization is finished. Also note that the entire optimization process is very fast and only several times slower than that for a single physical solution with given conductivity distribution.
While SPHinXsys is written in C++, it provides a python interface for users to write python scripts to control the simulation, including carry out regression tests for continuous integration (CI) and other tasks. One example is given below for the dambreak case. Please check the source code of 2D Dambreak case with python interface for the usage.
Main publication on the library:
- C. Zhang, M. Rezavand, Y. Zhu, Y. Yu, D. Wu, W. Zhang, J. Wang, X. Hu,
"SPHinXsys: an open-source multi-physics and multi-resolution library based on smoothed particle hydrodynamics",
Computer Physics Communications, 267, 108066, 2021.
The numerical methods and computational algorithms in SPHinXsys are based on the following publications.
SPHinXsys is cross-platform can be compiled and used in Windows, Linux and McOS systems.
For installation, program manual and tutorials, please check https://www.sphinxsys.org/html/sphinx_index.html. Please check the documentation of the code at https://xiangyu-hu.github.io/SPHinXsys/. For a Docker image, check https://hub.docker.com/r/toshev/sphinxsys.
Thank you for using and supporting our open-source project! We value each feedback.
Your input is crucial to us. We encourage you to report any issues you encounter with the library, including:
- Bug reports
- Poorly written code or algorithm designs
- Benchmark test issues, whether within the library or from literature, especially those highlighting potential deficiencies
- Other issues
We particularly appreciate feedback stemming from practical simulations or projects, as these insights are essential for improving SPHinXsys.
If you don't have a GitHub account yet, please register for one. Fork the SPHinXsys repository to add new features or improve existing ones. Once your changes are ready, commit them and initiate a pull request to have your contributions merged into the main repository.
To ensure efficient and effective development, we prioritize addressing issues raised by active contributors—whether through code, documentation, or other means. We welcome any interaction with SPHinXsys and our team.
You can also join us as a collaborator, enabling you to branch directly within the main repository and review pull requests.
Together, we can build a leading-edge multi-physics library open for all!
If you have any further question, please contact xiangyu.hu@tum.de.