The title of this blog post is also the title of a paper on the arXiv today by Ruprecht Machleidt from the University of Idaho: https://arxiv.org/abs/2307.06416.
The phrase ab initio to my mind, and to Ruprecht Machleidt's, is sometimes used quite loosely. He describes is at synonymous with "microscopic" nuclear physics in the following way
The tenet of microscopic nuclear theory is that atomic nuclei can be accurately
described as collections of point-like nucleons interacting via two- and many-body
forces obeying nonrelativistic quantum mechanics—the forces being fixed in free-
space scattering.
The microscopic or ab initio approach to nuclear structure and reactions is then
defined as calculating the properties of nuclei in accordance with the tenet.
To me, a microscopic approach is one which works at the level of individual nucleon wave functions - so e.g. the shell model would be a microscopic approach, while the ab initio approach is the particular microscopic approach that uses free-nucleon forces to build up nuclear structure. In that sense there is such a thing as an ab inito shell model that can be different to the generic shell model.
I also would allow "ab initio" to be used, perhaps even more legitimately, for those theories that consider nucleons to be more than point-like, and to acknowledge the substructure.
One of the points made in the paper is an assessmnet of some so-called ab initio work which is judged, by Machleidt, not to be ab initio by his criteria, and he gives a kind of roadmap of what the future should hold for real ab initio calculations.
I enjoyed his historical perspective of the battle between what-he-calls-microspic and the effective interaction communities, and I don't much mind that people use phrases like ab initio to mean different things, when it can be seen as either a general and somewhat ill-defined class of theories, or as brand marketing.
