Welcome to HIFIR Project

CI	Documentation

Introduction

Welcome to the HIFIR package! HIFIR stands for Hybrid Incomplete Factorization with Iterative Refinement, which is a multilevel preconditioner for ill-conditioned and (nearly) singular systems. HIFIR has (near) linear time complexity in its factorization and solve and is robust due to its multilevel structure, scalability-oriented dropping, and careful control of the condition numbers of the triangular factors.

Detailed documentation of HIFIR can be found at https://hifirworks.github.io/hifir.

Installation

To download the latest version of the code, use the command

git clone -b release -–depth 1 https://github.com/hifirworks/hifir.git hifir

Use git pull to download any new changes that have been added since git clone or last git pull. Alternatively, use git checkout v[GLOBAL].[MAJOR].[MINOR] to download a specific version.

Another way is to download the current archive at https://github.com/hifirworks/hifir/releases.

The C++ interface of HIFIR is header-only. To install the code, you can simply do the following on a UNIX system

sudo cp -r /path/to/hifir/src/* /usr/local/include

or copy /path/to/hifir/src/* to any user-level directory, for example,

cp -r /path/to/hifir/src/* $HOME/.local/include

and then add -I$HOME/.local/include to the command line of your C++ compiler.

For Windows users, the simplest and cleanest way is to use Windows Subsystem for Linux (WSL) or Cygwin.

Optionally, one can also use the C interface by building the runtime library libhifir; see the README therein for more details.

Third-Party Dependency: LAPACK

The only third-party dependency of HIFIR is LAPACK, which HIFIR uses for its dense-level factorization and solve. We recommend using OpenBLAS, ATLAS, or Intel oneAPI Math Kernel Library (formerly Intel MKL). Assuming you have system-administrator permussions, you can install OpenBLAS using the apt and yum command on Debian and RedHat-based Linux systems, such as

sudo apt install libopenblas-dev

and

sudo yum install libopenblas-devel

If you prefer to use ATLAS, the corresponding libraries are libatlas-dev and libatlas-devel on Debian and RedHat-based systems, respectively. Note that if you have multiple versions of BLAS and LAPACK installed, you may need to use the update-alternatives --config or alternatives --config command to config your system for the proper versions of BLAS and LAPACK; see https://wiki.debian.org/DebianScience/LinearAlgebraLibraries and https://www.redhat.com/sysadmin/alternatives-command for more detail.

For Intel MKL, please refer to https://software.intel.com/content/www/us/en/develop/articles/installing-intel-free-libs-and-python-apt-repo.html and https://software.intel.com/content/www/us/en/develop/articles/installing-intel-free-libs-and-python-yum-repo.html for installation on Debian and RedHat-based systems, respectively.

For other platforms, please refer to the official documentation of OpenBLAS, ATLAS, Intel MKL, or the vendor-provided LAPACK for your platform, respectively.

Copyright and Licenses

The HIFIR software suite (including hifir4m and hifir4py) are developed by the NumGeom Research Group at Stony Brook University.

The HIFIR software package is released under a dual-license model. For academic users, individual users, or open-source software developers, you can use HIFIR under the GNU Affero General Public License version 3 (AGPLv3+, see LICENSE) free of charge for research and evaluation purpose. For commerical users, separate commerical licenses are available through the Stony Brook University. For inquiries regarding commercial licenses, please contact Prof. Xiangmin Jiao at xiangmin.jiao@stonybrook.edu.

How to Cite HIFIR

If you use HIFIR in your research for nonsingular systems, please cite the HILUCSI paper:

@Article{chen2021hilucsi,
  author  = {Chen, Qiao and Ghai, Aditi and Jiao, Xiangmin},
  title   = {{HILUCSI}: Simple, robust, and fast multilevel {ILU} for
             large-scale saddle-point problems from {PDE}s},
  journal = {Numer. Linear Algebra Appl.},
  year    = {2021},
  number  = {6},
  pages   = {e2400},
  volume  = {28},
  doi     = {10.1002/nla.2400}
}

If you plan to use HIFIR in solving singular and ill-conditioned systems, please cite the following papers.

@Article{jiao2022approximate,
  author  = {Xiangmin Jiao and Qiao Chen},
  journal = {SIAM J. Matrix Anal. Appl.},
  title   = {Approximate generalized inverses with iterative refinement for
             $\epsilon$-accurate preconditioning of singular systems},
  year    = {2022},
  number  = {1},
  pages   = {40--67},
  volume  = {43},
  doi     = {10.1137/20M1364126}
}

@Article{chen2022hifir,
  author  = {Chen, Qiao and Jiao, Xiangmin},
  title   = {{HIFIR}: Hybrid incomplete factorization with iterative refinement
             for preconditioning ill-conditioned and singular systems},
  journal = {ACM Trans. Math. Softw.},
  year    = {2022},
  doi     = {10.1145/3536165}
}

Contacts

• Qiao Chen, qiao.chen@stonybrook.edu, benechiao@gmail.com
• Xiangmin Jiao, xiangmin.jiao@stonybrook.edu, xmjiao@gmail.com