Freedom251 › Physics Simulation Software
Kinetic.jl
What Kinetic.jl is
Kinetic.jl is a free computational fluid dynamics toolbox written in Julia. It combines mechanical and neural network models in a differentiable programming framework and supports simultaneous simulations in one, two, or three dimensions on CPUs and GPUs. The toolbox models particle transport and advection-diffusion-type equations for particles including molecules, photons, plasmas, and neutrons. Documented equation models include Boltzmann, Fokker-Planck-Landau, direct simulation Monte Carlo, advection-diffusion, Burgers, Euler, Navier-Stokes, magnetohydrodynamics, and Maxwell equations. Kinetic.jl is intended for kinetic theory education and scientific research, with API levels designed to cooperate with Julia packages. Its modular structure separates portable components to reduce LLVM just-in-time compiler latency. KitML adds scientific machine learning interfaces aligned with Flux.jl and DiffEqFlux.jl, while universal Boltzmann functions are documented for DifferentialEquations.jl solvers. A Python wrapper, optional Fortran package, and Julia distributed-computing guidance are also listed.
Who it suits
Kinetic.jl suits researchers, educators, and Julia users working with fluid dynamics, kinetic theory, particle transport, or scientific machine learning. It also fits projects needing CPU/GPU simulations or the listed Python and Fortran options.
What it does well
- Free Julia toolbox
- Simulates one to three dimensions
- Runs on CPUs and GPUs
- Supports many documented equation models
- Includes KitML and Python interface
What to know first
- No lesson library
- No authoring or assessment tools
- No 3D visuals
Verdict
Kinetic.jl provides a broad Julia-based environment for fluid, kinetic, and particle simulations, with neural-model integration and multiple interfaces. Its scope is technical and research-oriented rather than a general teaching package.
Facts from Kinetic.jl's own pages
- Applications
- It supports particle transport and advection-diffusion-type equations involving particles such as molecules, photons, plasmas, and neutrons. · xiaotianbai.com, 8 Oct 2026
- Assessment tools
- No · github.com, 24 Sept 2026
- Authoring tools
- No · github.com, 24 Sept 2026
- Compute
- The documentation says simultaneous one to three dimensional simulations can run on CPUs and GPUs. · xiaotianbai.com, 8 Oct 2026
- Delivery mode
- desktop · github.com, 24 Sept 2026
- Design
- The project describes its modular design as supporting educational use in kinetic theory and scientific research through APIs at different levels. · xiaotianbai.com, 8 Oct 2026
- Distributed computing
- The distributed computing documentation uses Julia's Distributed module and describes its @distributed macro as providing MPI-type parallelization. · xiaotianbai.com, 7 Oct 2026
- Education and research
- The interface is intended for educational use in kinetic theory and scientific research, with multiple API levels designed to cooperate with Julia ecosystem packages. · xiaotianbai.com, 7 Oct 2026
- Fortran backend
- KitFort is an optional Fortran backend that can be manually installed when execution efficiency is the priority; it is not a default Kinetic submodule. · xiaotianbai.com, 8 Oct 2026
- Free plan
- Yes · github.com, 24 Sept 2026
- Julia ecosystem
- The documentation describes Kinetic as a 100% Julia stack designed to integrate with existing Julia packages and avoid a two-language workflow. · xiaotianbai.com, 8 Oct 2026
- Lesson library
- No · github.com, 24 Sept 2026
- Machine learning
- KitML is a scientific machine learning toolbox intended to fuse mechanical and neural models, including a neural network enhanced Boltzmann equation. · xiaotianbai.com, 8 Oct 2026
- ML integrations
- KitML documentation says its training interfaces are consistent with the Flux.jl and DiffEqFlux.jl ecosystems and that its universal Boltzmann equation works with DifferentialEquations.jl solvers. · xiaotianbai.com, 8 Oct 2026
- ODE integration
- The universal Boltzmann equation functions are documented as working with modern ODE solvers in DifferentialEquations.jl. · xiaotianbai.com, 7 Oct 2026
- Offline access
- Yes · github.com, 24 Sept 2026
- Parallel computing
- The distributed computing guide uses Julia's Distributed module and describes MPI-type parallelization across workers. · xiaotianbai.com, 8 Oct 2026
- Product
- Kinetic.jl is a computational fluid dynamics toolbox written in Julia that combines mechanical and neural network models in a differentiable programming framework. · xiaotianbai.com, 8 Oct 2026
- Purpose
- Kinetic is a computational fluid dynamics toolbox written in Julia that combines mechanical and neural network models in a unified framework. · xiaotianbai.com, 7 Oct 2026
- Python
- The project lists kineticpy as a Python interface built on pyjulia. · xiaotianbai.com, 8 Oct 2026
- Python interface
- The kineticpy wrapper lets users locate Kinetic methods from Python and includes installation and usage instructions. · xiaotianbai.com, 7 Oct 2026
- Scientific machine learning
- KitML is a scientific machine learning toolbox for combining mechanical and neural models, and its documented training interfaces are consistent with Flux.jl and DiffEqFlux.jl. · xiaotianbai.com, 7 Oct 2026
- Scope
- The framework models particle transport and can solve advection-diffusion-type equations for particles including molecules, photons, plasmas, and neutrons. · xiaotianbai.com, 7 Oct 2026
- Simulation dimensions
- It supports simultaneous numerical simulations in 1–3 dimensions on CPUs and GPUs. · xiaotianbai.com, 7 Oct 2026
Where Kinetic.jl stands on Freedom251
- Physics Simulation Software#35 of 89