ALIEN: CUDA-Powered Artificial Life Simulation with Evolving Digital Organisms

What ALIEN Simulates and Why

ALIEN (Artificial Life Environment) is a simulation tool for studying how complexity and life-like behavior can emerge from simple physical rules. The README identifies three distinct motivations for using it.

The first is scientific curiosity: with self-replication and mutation enabled, the simulation produces evolutionary dynamics autonomously. The second is play: the physics engine is fast enough to interact with hundreds of thousands of machines in real time using the mouse cursor, which the README describes as "like playing god in your own universe with your own rules." The third is academic research: the project is oriented toward questions about the conditions for pre-biotic evolution, how ecosystems adapt to environmental change, and how to find conditions that support open-ended evolution.

A fourth use the README mentions is generative art. The evolutionary dynamics produce forms and behaviors that no human designer specified, which makes it interesting for artists who work with emergent processes. The project won the ALIFE 2024 Virtual Creatures Competition, as noted in the README's demo video link.

The Particle Physics Engine and How Organisms Are Represented

The physics layer models soft and rigid body mechanics, fluids, heat dissipation, adhesion and damage using a network of particles. Each simulated body is a network of particles connected by bonds, not a rigid geometric shape. This means organisms deform, collide and break in ways that follow from the particle interactions rather than from scripted animations.

The simulation code runs entirely in CUDA, and the README states it is "optimized for large-scale real-time simulations with millions of particles." Rendering and post-processing use OpenGL via CUDA-OpenGL interoperability, which allows the GPU to render frames directly from simulation data without transferring particle positions back to the CPU on each frame.

On top of the physics layer, higher-level cellular functions are assigned to individual particles in the network. These include sensors for detecting proximity and other environmental conditions, muscles for applying force, weapons for attacking, constructors for building offspring, and neural networks for orchestrating these functions. The README describes these as a kind of programmable cell biology: "The bodies can be thought of as agents or digital organisms operating in a common environment."

Genomes store the blueprint for how a constructor builds an organism cell by cell. When reproduction is enabled and mutations are active, offspring inherit genomes with random variations, and the evolutionary loop runs without further human input. Color is used as a configurable marker to distinguish cell types within an organism, allowing users to map different functions to different visual identities. The simulation also supports spatially varying parameters, which means the environment itself can have regions with different physics properties, adding another dimension of selection pressure to an evolutionary run.

Running ALIEN on Windows and Building the Docker Image

The most direct way to run ALIEN on Windows is to download and unpack the nightly build from alien-project.org/files/alien-develop.zip. The README notes that the archive includes two executables: `alien.exe` for NVIDIA GPUs and `alien-amd.exe` for AMD GPUs (RDNA2, RDNA3 and RDNA4). The README warns to start the executable from the unpacked folder, not from a shortcut in another directory, because the program must find its resource folder at runtime.

For long simulation runs without local hardware, the Dockerfile in the repository is used to build a cloud-ready image. The comment in the Dockerfile shows the build command used in CI:

docker build -f Dockerfile -t <user>/alien:nightly docker/payload

The build context is the `docker/payload` directory, which holds the pre-compiled binary staged by the CI pipeline alongside the `resources/` directory. The Dockerfile base image is `vastai/base-image:cuda-13.0.3-auto`, and CUDA 13.0 requires an NVIDIA driver of version 580 or newer. On a cloud GPU rental service such as vast.ai, the driver version is a filter criterion when selecting an instance.

The headless build (`cli`) runs without a display, making it suitable for unattended server runs. The Dockerfile installs `libgl1`, `libglu1-mesa`, `libx11-6`, `libxext6`, `libxrandr2`, `libxinerama1`, `libxcursor1` and `libxi6` as runtime dependencies, because the simulation engine links OpenGL and X11 even in headless mode through its CUDA-OpenGL interop code.

GPU Requirements and Platform Support

The simulation runs entirely on GPU. For NVIDIA hardware, compute capability 7.5 or higher is required, which means GeForce RTX 20 series or newer. The README links to the CUDA GPU compatibility list on Wikipedia for a full enumeration. For AMD hardware, RDNA2 or newer is required.

Building from source on Linux requires a separate step for AMD support via HIP or SCALE, as the nightly build is for Windows only. Cloud instances only support NVIDIA. There is no documented CPU fallback mode; ALIEN requires GPU hardware to run at the simulation scales it targets.

The build system uses CMake with presets defined in CMakePresets.json. The vcpkg.json file lists C++ dependencies. The repository includes a build-windows-ninja.bat script for Windows builds. On Linux, build instructions are in the repository for developers who need the source build.

Editing Tools, Networking and the Genetic Editor

ALIEN includes tools for designing and modifying organisms within the simulation. The README lists a graph editor for manipulating every particle and connection, freehand and geometric drawing tools, a genetic editor for designing organisms, and mass-operations including scaling functions. These tools operate on a live simulation, so changes take effect immediately and the user can observe the physical consequences in real time.

The genetic editor gives direct access to the genome that controls how a constructor builds offspring, which is the primary tool for seeding intentional designs into an otherwise evolutionary simulation. Users can define an organism's body plan in the genome and introduce it into a simulation to observe how evolution acts on it over time.

A built-in simulation browser with upload and download support allows sharing simulation files. Users can rate simulations by giving stars. This creates a library of community-contributed simulations accessible from within the application. The YouTube channel linked in the README provides videos showing evolved ecosystems, which also function as informal documentation of what the simulator can produce.

ALIEN Versus Simpler Artificial Life Environments

Conway's Game of Life is the best-known reference point for emergence from simple rules, but it operates on a binary grid with no physics, no continuous space, and no concept of organism identity. ALIEN operates in a continuous 2D space with particle physics, which makes organism morphology a physical constraint rather than a pattern on a grid.

Lenia, a continuous cellular automaton, is a closer comparison in that it uses continuous space and real-valued states. Lenia produces organism-like patterns through its update rules, but it does not include explicit energy, combat, reproduction with genomes, or the neural network machinery ALIEN provides. ALIEN's particle-based physical substrate and explicit genome system make it closer to a true artificial life environment than a pure continuous automaton.

For research purposes, ALIEN's combination of physics simulation, genetic system and neural network controllers in one tool distinguishes it from most available ALife environments, which tend to emphasize one of these dimensions at the expense of the others. The ALIFE 2024 Virtual Creatures Competition win mentioned in the README provides one data point of external scientific recognition, though the project's primary goal is exploratory simulation rather than competition performance.

One practical limitation worth noting is documentation currency. The README states explicitly that the older gitbook documentation at alien-project.gitbook.io covers the previous major version and is "no longer up to date." The in-program help windows and tooltips are the current reference for the features introduced in recent releases. Users approaching ALIEN for serious research should plan to explore the interface interactively rather than relying on external documentation.

Licence, Citation and Documentation State

ALIEN is released under the BSD-3-Clause licence, which permits academic and commercial use, modification and redistribution with attribution. The paper for academic citation is Heinemann (2008) in Informatik-Spektrum, referenced in the README with full BibTeX. The README notes that this paper describes the early system and that the current simulator differs substantially, so the repository itself is an equally valid citation target.

The last push was on 2026-09-27. Releases are named by feature rather than by strict version number, with v4.12.3 being the most recent tagged release from December 2024. The in-program help windows and tooltips provide the most current user guidance. The older gitbook documentation at alien-project.gitbook.io covers the previous major version and is described in the README as no longer fully up to date.

Editorial conclusion

ALIEN is the right tool for researchers studying pre-biotic evolution, emergent complexity or open-ended evolution, and for generative artists who want a physics-driven creative environment. It requires an NVIDIA GPU with compute capability 7.5 or higher (GeForce RTX 20 series or newer) or an AMD RDNA2+ GPU. The Dockerfile and cloud instructions make it feasible to run on rented GPU instances for long simulation runs. The project is at an active development stage, with the last push on 2026-09-27, though the documentation notes that the gitbook documentation for the previous major version is no longer fully up to date. The license is BSD-3-Clause, which permits academic and commercial use with attribution.

Frequently asked questions

Official sources

1. chrxh/alien on GitHub
2. License: BSD-3-Clause
3. Project website
4. README
5. Releases