More info
COMPAS is developed by the Block Research Group of ETH Zurich, with the support of the National Centre for Competence in Research (NCCR) in Digital Fabrication. It's main features are:
- a pure Python base with flexible datastructures, algorithms, and methods geared towards applications in architecture, engineering, and fabrication;
- geometry processing independent of CAD tools;
- interoperability with C/C++ code and libraries such as ShapeOp, libigl and Eigen;
- methods and solvers for numerical computation built around NumPy and SciPy;
- high peformance computing through GPU acceleration and JIT compilation;
- plotters and viewers for two-dimensional and basic three-dimensional visualization; and
- interfaces to common CAD software and ecosystems.
Multi-disciplinary research
Architecture is a highly multi-disciplinary field, combining research from computer science, robotics, mathematics, automation, and several other scientific areas. The target audience of the COMPAS framework is therefore very diverse.
To deal with the different academic backgrounds, programming skills, computational experience, and best/accepted practices of its users and their respective fields, COMPAS is implemented primarily in Python and designed to be entirely independent of the functionality of CAD software. As a result, it can be used on different platforms and in combination with external software and libraries, and at the same time take advantage of the various scientific and non-scientific libraries available in the Python ecosystem itself. Furthermore, and perhaps more importantly, it ensures that research based on COMPAS is not tied to a specific CAD-based ecosystem, as this can hinder effective collaboration between different users.
Public, Private, Shared
COMPAS is divided into a main library and a pool of user-contributed research packages. The main library is entirely open source and subject to the MIT license. The user-contributed packages are released with access levels defined by the respective authors. They can be private, public, or shared with a specific group of users.
The reason for this setup is related to the objectives of COMPAS to facilitate the transfer of knowledge generated through research in the fields of architecture, engineering, digital fabrication, and related fields, and to facilitate collaboration.
By building specialised research packages on the main library, these packages are compatible with each other, allowing expertise and know-how to be shared with, and reviewed and extended by a large community of researchers and practicioners.
The possibility of defining packages as private, public, or shared, and the possibility to change this state at any given time, allows researchers to publish their work in a way that corresponds to the state of their research, to the sensitivities of collaborations, or to the requirements of their funding.
Main library
The main library consists of a core package (compas) and several additional packages for integration of the core functionality in CAD software (compas_rhino, compas_blender, compas_maya). The core package defines all real functionality. The CAD packages simply provide a unified framework for processing, visualising, and interacting with datastructures and geometrical objects, and for building user interfaces in different CAD software.
The documentation of the main library provides many more resources for further exploration. The API reference contains detailed information about the functionality of the core package and the CAD packages.
Additional packages
The additional packages supplement the main library with functionality related to specialised topics. These packages can be related to ongoing or completed research, or simply provide additional functionality, as a service to other users. For an overview of available packages, see https://compas-dev.github.io/packages/.
Citing
If you use the main library of COMPAS in a project, please refer to the GitHub repository.
@misc{compas-dev,
title = {{compas}: A framework for computational research in architecture and structures.},
author = {Tom Van Mele and Andrew Liew and Tomas Mendéz Echenagucia and Matthias Rippmann and others},
note = {http://compas-dev.github.io/compas/},
year = {2017},
}