Benchmarking neighbors list algorithms with PBC

Requirements

• For the rust part, on Ubuntu I first had to install cargo.

sudo apt install cargo

• To get the latest version of the rust-neighborlist do

pip install maturin
git clone https://github.com/mariogeiger/rust-neighborlist.git
cd rust-neighborlist
pip install .

Or for older versions (currently only 0.1.1 is available, which does not support PBC!) just run:

pip install rust-neighborlist

Or all at once with:

python3.9 -m venv venv
source venv/bin/activate
python3 -m pip install --upgrade pip
pip install -r requirements.txt

Findings

• matscipy doesn't support self-interactions, so we leave them out for this comparison.

• In 3D to have:

  • 6(+1) neighbors requires r_cutoff= dx + EPS
  • 18(+1) neighbors requires r_cutoff=sqrt(2) dx + EPS
  • 26(+1) neighbors requires r_cutoff=sqrt(3) dx + EPS -> we selected the 26 neighbors case.