User

Getting started

To have fun with URCA, it is enough the following.

$ python3 -m venv venv_name
$ source venv_name/bin/activate
(venv_name) $ pip install urca

Latest commit

Note that in order to have the last changes or fixes because of a commit which is not a release, you can do the following.

$ python3 -m venv venv_name
$ source venv_name/bin/activate
(venv_name) $ git clone https://github.com/ale-depi/urca
(venv_name) $ cd urca
(venv_name) $ pip install .

From now on, in the virtual environment named venv_name you can execute import urca.

Implementations

The package will be installed with support for both NumPy (CPU) and CuPy (GPU). This dual installation does not cause any compatibility issues on systems without a GPU, since the two implementations are completely isolated. The isolation ensures coherent behavior regardless of the underlying hardware environment, while also enabling efficient utilization of GPU resources when available.

The API documentation is NumPy oriented, but the interfaces, methods, variables, and overall structure remain exactly the same when using the CuPy version. The only differences are found in the implementation details, since not all NumPy interfaces are directly available in CuPy.

Selecting one implementation or the other is simply a matter of the import path: urca.cpu for the CPU-based implementation or urca.gpu for the GPU-based one.