Documentation & User Guide

Welcome to Compack3D

Welcome to the Compack3D user guide. Compack3D is an open-source project developed at Stanford University and supported by the National Science Foundation. Compack3D is relased under the Apache-2.0 License . This package provides optimized implementations of essential algorithms for rapid development of numerical solvers for partial differential equations.

Why do I need it?

Compack3D is a light-weight package, originally designed for computational scientists who use compact finite difference methods (generalized Padé schemes) to accelerate their solvers. It solves the tridiagonal and penta-diagonal systems arising from compact numerical schemes on 3D computational mesh. Compack3D provides efficient solutions for hiearchical parallelization on distributed memory and shared memory and enables affordable large-scale simulations using compact finite difference methods. Compack3D provides both C++ and Fortran APIs, which makes it easy to interface with your own application, even a legacy code.

How to cite our work

If your project benefits from Compack3D, we would appreciate it if you add the following works to your reference list

• Song, H., Matsuno, K. V., West, J. R., Subramaniam, A., Ghate, A. S., & Lele, S. K. (2022). Scalable parallel linear solver for compact banded systems on heterogeneous architectures . Journal of Computational Physics, 468, 111443.

  See BibTex

  @article{song2022scalable,
    title={Scalable parallel linear solver for compact banded systems on heterogeneous architectures},
    author={Song, Hang and Matsuno, Kristen V and West, Jacob R and Subramaniam, Akshay and Ghate, Aditya S and Lele, Sanjiva K},
    journal={Journal of Computational Physics},
    volume={468},
    pages={111443},
    year={2022},
    publisher={Elsevier}
  }
  

• Song, H., Matsuno, K. V., West, J. R., Subramaniam, A, Ghate, A.S., & Lele, S.K. (2024) Scalable parallel linear solver for compact banded systems . In NASA Advanced Modeling and Simulation (AMS) Seminars.

Who to talk to?

This project is led by Prof. Sanjiva K. Lele and the core development team consists of