Software
Over the past decades, ab initio many-body methods have known a tremendous development, moving from the light sector of the nuclear chart up to the Tin isotopic line. With progress being made in nuclear Hamiltonians as well as numerical techniques, high-precision calculations using high-order formalisms are now within reach.
But with the complexity coming from high-order methods comes an increased risk of mistakes when derivations are made by hand, as well as an increased time spent in the design phase of the formalism. As such, trying to automatise the process as much as possible will help to both speed up the formal development and make it safer.
This is what is being accomplished with the Automatic Diagram Generator ADG, of which I am the main developer and maintainer. ADG is developed in Python and relies on external libraries Numpy, Scipy and NetworkX. It uses methods of graph theory to generate the diagrams for a given formalism and extract the associated expression, thus providing the user directly with the expressions to be implemented in their numerical code. The code is open-source, available on GitHub and distributed via the Python Package Index. Code quality is ensured via Continuous Integration, and the code documentation is available online.
Up to now, ADG has known three major releases associated to three articles, extending the set of supported formalisms to:
- Bogoliubov Many-Body Perturbation Theory (BMBPT), as well as support for Hartree-Fock MBPT limited to two-body interaction
- Projected BMBPT
- Bogoliubov In-Medium Similarity Renormalization Group (BIMSRG)
The goal is for ADG to be extended to other formalisms in the years to come, including support for Gorkov Self-Consistent Green’s Function theory, as well as support for three-body forces in MBPT.