Installation
Choosing between C and Python
You can use either C or Python when working with REBOUND. Which programming language you want to use depends on your preference and your specific application. In short:
- If you want to set up a planetary system, visualize data with matplotlib, and integrate your simulation with one of the built-in integrators then use the Python version. It's quick and easy to use.
- If you want to run large simulations with millions of particles, develop your own integrator, use the distributed tree code with MPI, OpenMP parallelization, or OpenGL visualization, then use the C version. C gives you the best performance and direct access to all the REBOUND internals.
Note
All the computationally expensive parts of REBOUND are written in C. So even if you use the Python version, your simulation will run very efficiently. If you want to extend REBOUND, for example to include an additional non-gravitational force, you can do that in both C or Python. However, for complicated force routines, a C implementation of your function would most likely be significantly faster.
Installation via pip
Python Wheels
Starting with REBOUND version 3.28, we provide Python Wheels for REBOUND. This makes installing REBOUND easier on a wide variety of systems. For optimal performance, you can compile REBOUND yourself with optimizations flags that specifically target your system.
If you just want to try out REBOUND or don't plan to modify it in any way, then the easiest way to install the python version of REBOUND is pip (the Package Installer for Python). Simply type the following command into a terminal:
pip install rebound
If you have trouble installing a package with pip, consider using a virtual environment. Also, make sure your version of pip is not too old. You can update pip with pip itself:
pip install --upgrade pip
Installation via git
We use the git as a version control system for REBOUND. If you want to use the C version of REBOUND or plan to make any modifications to REBOUND, you can clone the repository to your computer. Make sure you have git installed, then type the following command in a terminal:
git clone https://github.com/hannorein/rebound
This will create a new directory names rebound/
which contains all the source code, examples, and documentation.
To use the python version of REBOUND, go to the rebound/
directory, then install REBOUND with
pip install -e .
Compiling the C version of REBOUND
Examples
If you look at any of the examples in the examples/
sub-directories, you'll see one
problem.c
file and one Makefile
. All the REBOUND code itself is in the
src/
directory. This setup keeps the different projects nicely separated from the shared REBOUND code.
To compile one of the examples, go to the example's directory and type make
.
This triggers the following tasks:
- The Makefile in the example directory sets up various environment variables. These determine settings like compiler optimization flags and which libraries are included (see below).
- Next, the Makefile in the
src/
directory gets called. This compiles the entire REBOUND code into a shared library. - It then creates a symbolic link from the current directory to the
location of the shared library in the
src/
directory. (On Windows the Makefile simply copies the shared library instead of making a symbolic link) - Finally, it compiles your own code, the
problem.c
file and links it to the REBOUND shared library.
You can execute your program with ./rebound
(or rebound.exe
on Windows).
After you edit either the problem.c
file or any file in the src/
directory, you can simply type make
again to recompile your program.
If you change any of the environment variables, clean the build directory first, by executing make clean
.
Your own project
The easiest way to start working on your own problem is to simply copy an example directory that is somewhat similar to what you want to do.
This way, all your project's source and data files will be in one directory, separate from the main REBOUND source files in src/
.
Alternatively, you can also install the shared REBOUND library in a global directory (e.g. /usr/lib/
) and the header file in /usr/include/
.
Doing so will allow you (and any other users on your system) to use REBOUND from any directory.
However, doing so requires root access and some knowledge on how Unix systems work.
By simply replicating and modifying one of the examples, you'll avoid these complications.
Possible issues during compilation
The way we've designed REBOUND should make the compilation process extremely easy. You do not need to install any additional libraries (although you might want to, see below), and you do not need root access. You might nevertheless run into problems. Some of the most common issues are:
- Missing compilers. Make sure you have a C compiler installed. If
you are using a Mac, install the Xcode package which you can
download for free on the App Store. Make sure the command line tools
are installed. if you are on Windows, make sure you install the
compilers that come with Visual studio (
cl.exe
) and have them available in your current command prompt (use the Developer Command Prompt for VS). - Missing glfw3 library. You can compile REBOUND with support for
real-time OpenGL visualizations. This is an optional feature that
requires the glfw3 library. If you are on a Mac, then the easiest
way to install the glfw3 library is with homebrew:
brew tap homebrew/versions && brew install glfw3
. If you are on Linux, you can install it with your package manager, for example withsudo apt-get install libglfw3-dev
. Alternatively, you can disable the OpenGL visualization in the Makefile by settingOPENGL=0
. Then, executemake clean
and try compiling the program again. Note that on some systems theglfw
library is calledglfw3
instead. In that case, change-lglfw
to-lglfw3
in the filesrc/Makefile.defs
. - Compiler optimizations. By default, REBOUND does not use the
compiler flag
-march=native
which tries to optimize the code for the native architecture. If you want to have the most optimized code, add the-march=native
or-mtune=native
flag in the filesrc/Makefile.defs
. If you use the python version, you can add compiler flags tosetup.py
. This might improve performance significantly. - Floating point contractions. Some compilers (e.g. clang) optimize code by
contracting certain floating point operations (e.g. a multiplication
and an addition become one fused multiply-add instruction). This improves performance but might prevent you from
reproducing results exactly. You can turn off fused multiply-add instruction with the
-ffp-contract=off
compiler flag. If you use the python version, you can set theFFP_CONTRACT_OFF
environment variable before installing REBOUND.
Running REBOUND on Windows
There are several ways to run REBOUND on Windows.
Python
You can install the python version of REBOUND using pip:
pip install -e .
Windows Subsystem for Linux (WSL)
You can run the C-version of REBOUND using the Windows Subsystem for Linux (WSL).
You will need make
and a compiler, such as gcc
. These can be installed within WSL with the following command:
sudo apt-install make gcc
git clone https://github.com/hannorein/rebound
cd rebound/examples/simplest
make
./rebound
Native Windows Builds
Note
The native Windows support for REBOUND is relatively new. Several features are currently not supported on native Windows builds: OpenMP, MPI, OpenGL, and AVX512. Please file a bug report on github if you require any of these featured or if you encounter any other problems.
Since version 3.28, you can also run REBOUND natively on Windows. You need to install make and enable the Microsoft Visual Studio compiler. Once you have downloaded the source code of REBOUND, open the Developer Command Prompt for VS or the Windows PowerShell on your system and go to the REBOUND source code. Then, compile and run a simple C-example with the following commands:
cd examples
cd simplest
make
rebound.exe