Installation

Package only

The package can be installed from git repository with pip.

pip install git+https://github.com/engellab/neuralflow

Package and examples

To get the package with the examples, one needs to clone the repository. The examples are provided as Jupyter notebook (ipynb) files, so jupyter package has to be preinstalled. With conda as a package manager, one may opt to create a new environment:

conda create -n neuralflow jupyter pip && conda activate neuralflow

Alternatively, one can work in the base environment (make sure that jupyter package is installed). Clone the repository and go to the repository root directory:

git clone https://github.com/engellab/neuralflow
cd neuralflow

Install the package from a local copy:

pip install .

After that, you should be able to run examples in example folder

CUDA support

Optimization can be performed on CUDA-enabled GPU. For GPU support, cupy package has to be installed on a machine with CUDA-enabled GPU. The package was tested with cupy version 12.2.0. Note that double-precision computations are absolutely necessary for our framework, so optimization benefits from GPU acceleration only if scientific grade GPU (e.g. Tesla V100) is used. A gaming GPU perforamnce is approximately the same as CPU perfromance, since gaming GPUs do not have many double-precision multiprocessors.