Main page

This is the technical API documentation of the Medusa library. For an overview of the library visit our page or the wiki.

The source code is available through our Gitlab repository.

Our code has all kind of safety assertions built in, and may therefore run slower than expected. These assertions help catch errors during development phase of the PDE solution. After you have verified the correctness of your implementation, the NDEBUG definition can be specified to the compiled (by using e. g. -DNDEBUG) to get better execution times.

Getting started

If you are running a linux system (e.g. Ubuntu), run the usual clone, build, test sequence:

sudo apt-get install git g++ cmake libhdf5-dev
git clone git@gitlab.com:e62Lab/medusa.git --branch master --single-branch
cd medusa
mkdir -p build && cd build
cmake ..
make medusa_run_tests

to set up the library.

For building on other systems and troubleshooting refer to the Installation and building guide

See this page for instructions on using this library in your project.

Concepts

Some concepts used by Medusa library, such as scalar/vector field, approximation engine, RBFs, etc. are described on page Concepts. Users can often supply their own classes instead of the ones included in Medusa for various operations, as long as they conform to the specified interface.

Examples

For first time users, we wrote a tutorial on using Medusa to solve the Poisson equation. If you wish to dive into more field-specific examples, feel free to browse the medusa/examples folder included in the repository. Each example can be run from build/ directory or as a standalone executable. Scripts for plotting are also included. More information on the philosophy of examples and how to run them can be found here. Some examples of solution of basic PDEs using Medusa have additional explanation on the wiki, where you can also find additional use cases.

Finally, some basic examples are located in folder medusa/test/end2end which also serve as unit tests.