Trixi.jl is a numerical simulation framework for hyperbolic conservation laws written in Julia. A key objective for the framework is to be useful to both scientists and students. Therefore, next to having an extensible design with a fast implementation, Trixi is focused on being easy to use for new or inexperienced users, including the installation and postprocessing procedures. Its features include:

  • Hierarchical quadtree/octree grid with adaptive mesh refinement
  • Native support for 1D, 2D, and 3D simulations
  • High-order accuracy in space in time
  • Nodal discontinuous Galerkin spectral element methods
    • Kinetic energy-preserving and entropy-stable split forms
    • Entropy-stable shock capturing
    • Positivity-preserving limiting
  • Compatible with the SciML ecosystem for ordinary differential equations
  • Square/cubic domains with periodic and weakly-enforced boundary conditions
  • Multiple governing equations:
    • Compressible Euler equations
    • Magnetohydrodynamics equations
    • Hyperbolic diffusion equations for elliptic problems
    • Lattice-Boltzmann equations (D2Q9 and D3Q27 schemes)
    • Scalar advection
  • Multi-physics simulations
  • Shared-memory parallelization via multithreading
  • Visualization and postprocessing of the results
    • In-situ and a posteriori visualization with Plots.jl
    • Postprocessing with ParaView/VisIt via Trixi2Vtk


If you have not yet installed Julia, please follow the instructions for your operating system. Trixi works with Julia v1.5.

For users

Trixi and its related tools are registered Julia packages. Hence, you can install Trixi, the visualization tool Trixi2Vtk, OrdinaryDiffEq.jl, and Plots.jl by executing the following commands in the Julia REPL:

julia> import Pkg

julia> Pkg.add("Trixi"); Pkg.add("Trixi2Vtk"); Pkg.add("OrdinaryDiffEq"); Pkg.add("Plots")

You can copy and paste all commands to the REPL including the leading julia> prompts - they will automatically be stripped away by Julia. The package OrdinaryDiffEq.jl provides time integration schemes used by Trixi, while Plots.jl can be used to directly visualize Trixi's results from the REPL.

Note on package versions: If some of the examples for how to use Trixi do not work, verify that you are using a recent Trixi release by comparing the installed Trixi version from

julia> import Pkg; Pkg.update("Trixi"); Pkg.status("Trixi")

to the latest release. If the installed version does not match the current release, please check the Troubleshooting section.

For developers

If you plan on editing Trixi itself, you have two options: installing it as a dev package or cloning it to a local folder.

  1. Install Trixi as a dev package: You can install Trixi as a dev package by running

    julia> import Pkg

    This will download and install Trixi to a designated package development directory (usually .julia/dev, but you can determine the path on your system with Pkg.devdir()). As opposed to using the Pkg.add method as described above, packages installed via may have their files edited locally.

  2. Install Trixi in a local folder: Alternatively, you can download Trixi locally and use it from within the cloned directory:

    git clone
    cd Trixi.jl
    julia --project=@. -e 'import Pkg; Pkg.instantiate()' # Install Trixi's dependencies

    The last line can also be used to update the dependencies if they have changed since you first installed Trixi.

    If you installed Trixi this way, you always have to start Julia with the --project flag set to your local Trixi clone, e.g.,

    julia --project=@.

Either way, since the postprocessing tool Trixi2Vtk typically does not need to be modified, it is recommended to install it as a normal package by executing

julia> import Pkg

julia> Pkg.add("Trixi2Vtk")

in the Julia REPL. Likewise, you can install OrdinaryDiffEq.jl and Plots.jl as ordinary packages with the following REPL commands:

julia> import Pkg

julia> Pkg.add("OrdinaryDiffEq"); Pkg.add("Plots")

Example: Installing Trixi as a package

Please note that the playback speed is set to 3x, thus the entire installation procedure lasts around 45 seconds in real time (depending on the performance of your computer and on how many dependencies had already been installed before).


In the Julia REPL, first load the package Trixi

julia> using Trixi

Then start a simulation by executing

julia> trixi_include(default_example())

To visualize the results, load the package Plots

julia> using Plots

and generate a heatmap plot of the results with

julia> plot(sol) # No trailing semicolon, otherwise no plot is shown

This will open a new window with a 2D visualization of the final solution: <p align="center"> <img width="300px" src=""> </p>

The method trixi_include(...) expects a single string argument with the path to a Trixi elixir, i.e., a text file containing Julia code necessary to set up and run a simulation. To quickly see Trixi in action, default_example() returns the path to an example elixir with a short, two-dimensional problem setup. A list of all example elixirs packaged with Trixi can be obtained by running get_examples(). Alternatively, you can also browse the examples/ subdirectory. If you want to modify one of the elixirs to set up your own simulation, download it to your machine, edit the configuration, and pass the file path to trixi_include(...).

Example: Running a simulation with Trixi

Note on performance: Julia uses just-in-time compilation to transform its source code to native, optimized machine code at the time of execution and caches the compiled methods for further use. That means that the first execution of a Julia method is typically slow, with subsequent runs being much faster. For instance, in the example above the first execution of trixi_include takes about 20 seconds, while subsequent runs require less than 60 milliseconds.

Performing a convergence analysis

To automatically determine the experimental order of convergence (EOC) for a given setup, execute

julia> convergence_test(default_example(), 4)

This will run a convergence test with the elixir default_example(), using four iterations with different initial refinement levels. The initial iteration will use the elixir unchanged, while for each subsequent iteration the initial_refinement_level parameter is incremented by one. Finally, the measured $l^2$ and $l^\infty$ errors and the determined EOCs will be displayed like this:

error     EOC
9.14e-06  -
5.69e-07  4.01
3.55e-08  4.00
2.22e-09  4.00

mean      4.00
error     EOC
6.44e-05  -
4.11e-06  3.97
2.58e-07  3.99
1.62e-08  4.00

mean      3.99

An example with multiple variables looks like this:

julia> convergence_test(joinpath(examples_dir(), "2d", "elixir_euler_source_terms.jl"), 3)
rho                 rho_v1              rho_v2              rho_e
error     EOC       error     EOC       error     EOC       error     EOC
8.52e-07  -         1.24e-06  -         1.24e-06  -         4.28e-06  -
6.49e-08  3.71      8.38e-08  3.88      8.38e-08  3.88      2.96e-07  3.85
4.33e-09  3.91      5.39e-09  3.96      5.39e-09  3.96      1.93e-08  3.94

mean      3.81      mean      3.92      mean      3.92      mean      3.90
rho                 rho_v1              rho_v2              rho_e
error     EOC       error     EOC       error     EOC       error     EOC
8.36e-06  -         1.03e-05  -         1.03e-05  -         4.50e-05  -
5.58e-07  3.90      6.58e-07  3.97      6.58e-07  3.97      2.92e-06  3.94
3.77e-08  3.89      4.42e-08  3.90      4.42e-08  3.90      1.91e-07  3.93

mean      3.90      mean      3.93      mean      3.93      mean      3.94


If you use Trixi in your own research or write a paper using results obtained with the help of Trixi, please cite the following paper:

  title={A purely hyperbolic discontinuous {G}alerkin approach for
         self-gravitating gas dynamics},
  author={Schlottke-Lakemper, Michael and Winters, Andrew R and
          Ranocha, Hendrik and Gassner, Gregor J},

In addition, you can also refer to Trixi directly as

  title={{T}rixi.jl: A tree-based numerical simulation framework
         for hyperbolic {PDE}s written in {J}ulia},
  author={Schlottke-Lakemper, Michael and Gassner, Gregor J and
          Ranocha, Hendrik and Winters, Andrew R},


Trixi was initiated by Michael Schlottke-Lakemper and Gregor Gassner (both University of Cologne, Germany). Together with Hendrik Ranocha (KAUST, Saudi Arabia) and Andrew Winters (Linköping University, Sweden), they are the principal developers of Trixi. The full list of contributors can be found under Authors.

License and contributing

Trixi is licensed under the MIT license (see License). Since Trixi is an open-source project, we are very happy to accept contributions from the community. Please refer to Contributing for more details.