Creating Networks

pynucastro can output the righthand side functions for ODE integration in Python, pure C++, or in a form for the AMReX-Astrophysics Microphysics package in C++.

To create the python or C++ code for a network, you start by building the library representing the rates that you want and then you construct the appropriate network class:

PythonNetwork for a pure python network.
SimpleCxxNetwork for a basic, pure C++ network.
AmrexAstroCxxNetwork for an AMReX-Astro C++ network.

Usually the steps are something like:

Create a working directory for your network (this does not have to be in the pynucastro/ directory tree).
Write a short python script to generate your network, e.g. mynet.py. As an example, for a network with just He4, C12, and O16, we would do:
```
import pynucastro as pyna

rl = pyna.ReacLibLibrary()
lib = rl.linking_nuclei(["he4", "c12", "o16"])

net = pyna.PythonNetwork(libraries=[lib])
net.write_network()
```
and we can replace PythonNetwork with one of the other network types to get the output in a different format.

Run your python script

python mynet.py

Python network

A PythonNetwork will output a python module with all the data needed for constructing the righthand side and Jacobian of the ODE system. To integrate a Python network using the SciPy integration routines, customize examples/CNO/burn.py to initialize and run your network using the right hand side module you generated above.

Simple C++ network

A SimpleCxxNetwork is a very basic C++ network. Currently, the following features are not supported:

tabular rates
screening
partition functions
NSE
temperature evolution
plasma neutrino losses

A simple driver, main.cpp and a GNUmakefile will be generated that will simply evaluate the righthand side and Jacobian, and the energy release for a single thermodynamic state. The generated code is intended to be used to interface with a hydrodynamics code. That code will be responsible for providing the equation of state and adding any desired temperature / energy evolution to the network.

Note

A C++17 compiler is required

The test driver can be built by simply doing:

make

AMReX-Astro Microphysics network

An AmrexAstroCxxNetwork is intended to be used with an AMReX-Astro code like Castro or MAESTROeX. All pynucastro features are supported and these networks can run on GPUs in the AMReX Astro framework.

The examples/triple-alpha/triple-alpha-cxx.py example builds the right hand side, Jacobian, and helper C++ modules to copy into the networks/ subdirectory of the AMReX-Astro Microphysics repository.

No additional customization is required after running the steps for all networks above.