Multi-dimenstional Kernel Density Estimation (KDE) including adaptive bandwidths and C and CUDA implementations for specific cases.
Sebastian Schoenen (schoenen@physik.rwth-aachen.de) and Martin Leuermann for the IceCube collaboration.
Download the software into directory
. There should be a subdirectory named "kde" within the directory.To install in a location independent of your system Python files, install via the following command:
$ pip install
[cuda] --userwhere [cuda] is optional, ensuring support for GPU.
To install with references to the source code where it is downloaded (so that changes in the sourcecode are reflected immediately):
$ pip install -e
[cuda] --user