complex_bessel

A C++ library to evaluate Bessel functions of all kinds. More information can be found on the website.

Introduction

C++ library that acts as a wrapper for the Fortran subroutines developed by D.E. Amos. The library provides functionality to compute the Bessel, Hankel and Airy functions of complex argument and real order. Negative orders are implemented via the standard formulae.

We provide a shared object library and header files to be included.

Compilation instructions

Linux

The library uses CMake for compilation. The user should thus install CMake on their machine. On Ubuntu and other Debian-based OSes, this can be done by running

sudo apt-get install cmake

On Arch Linux

sudo pacman -S cmake

We suggest doing an out-of-tree build by first creating a build/ folder, then running cmake, i.e.

cmake -DCMAKE_INSTALL_PREFIX=/path/of/install/dir

The target of the complex_bessel library is exported as complex_bessel::complex_bessel to a package configuration file for this library.

After installation you can find this library in your project CMakeLists.txt with:

find_package(complex_bessel)

or you can just put a copy of complex_bessel source code into your source tree and just add it from the upper level CMakeLists.txt

add_subdirectory(complex_bessel)

To compile and link this library you should have C++14 and Fortran compilers installed, and you should enable CXX Fortran languages in CMakeLists.txt. After that to link with your <target> you can just

target_link_library(<target> complex_bessel::complex_bessel)

Note that CMake will add all additional needed include files to you project compilation automatically.

To run tests you will need to use HDF5, Google Test, Boost 1.6+, and a C compiler (to link with HDF5, so C is present in the list of languages in tests/CMakeLists.txt)

Windows (experimental support)

There is currently experimental support for compiling with Visual Studio Code. The only tested configuration is with the Intel OneAPI Fortran compiler, installed with Visual Studio integration. The project compiles, but nothing else is tested. Feel free to open a PR for better Windows support.

Other similar libraries

The FORTRAN library that is used as the main driver for the computation of Bessel functions is also used in

• scipy.special
• MATLAB

Boost has its own implementation of the Bessel functions, but only supports real values for the argument.

If arbitrary precision is needed, the arb library supports the computation of many special functions, including Bessel functions.

Example