These are the steps to needed to compile the Neutrino Solver using pybind11 in a virtual environment assuming usage on lxplus.

Source a python 3 environment

scl enable rh-python38 bash

Create the virtual environment first by running

python -m venv my_env
source my_env/bin/activate

Install pybind (and numpy for checking output)

pip install pybind11
pip install numpy

Add the virtual environment to the PYTHONPATH

export PYTHONPATH="$WORKINGDIR/my_env/lib/python3.8/site-packages:$PYTHONPATH"

From within the 'compiled' directory, compile the hpp and cpp files to create the pynusolver.so file

bash compile.sh

Then test the resulting file with the following inputs

python
import numpy as np
import pynusolver
jets = np.array([-38.31093216,   2.88590097, -97.31715393, 104.89573669])
lepton = np.array([-119.2567749 ,  -22.68565369, -136.70018005,  182.82168579])
met = np.array([-232.88529968,  -85.29122162])
nu = np.zeros(4)
pynusolver.run_nu_solver(lepton, jets, met, nu)

The resulting nu vector should be

array([ -382.65887115,  -119.84833506,  -266.69686505, 23626.31680084])