The chapter is a systematic study of finitely non-standard models of Robinson arithmetic—that is, it is a systematic study of those models of Robinson arithmetic that have a finite non-empty set of non-standard numbers.

In addition to the chapter draft, I have also uploaded to my research section Coq and Python source files corresponding to the sources referenced and included verbatim in the draft.

In the process of completing my thesis, I expect no substantial changes to the uploaded draft’s content. In that sense the draft may be seen as an all-but-finished version of the chapter. A sense in which it may not be seen thus is given by its multitude of typographical mishaps: missing glyphs, inconsistent vertical and horizontal spacings, margin overflows, et cetera. These mishaps will of course be fixed before inclusion in the final version of the thesis. Thus in case anyone wants to refer to the draft: do include the version—that is, do include the date and time printed on the first page of the draft.

I have uploaded slides for each talk to the research page.

At both talks I received useful feedback. Some feedback pointed out some problems with our definitions. Instead of modifying the slides I will address those problems here.

UPDATE 2019-08-29: In the latest summary that I have uploaded to the research page, the following problem has been addressed.
At the FLoV talk, Mattias Granberg Olsson pointed out a problem with our definition of first-order languages of arithmetic and their corresponding standard models. The problem is that a language of arithmetic (as we defined it) can have at most one function symbol (or predicate symbol) for each function (or relation) in the standard model. This makes it very inconvenient (though not impossible) to model proofs showing that two distinct definitions define the same function (or relation). To address this at the JAF 38 talk I simplified our definition of a first-order language of arithmetic and removed the definition of what constitutes the standard model for a first-order language of arithmetic. Since it would still be nice to be able to define the standard model for any language of arithmetic, we will probably have to come up with some other solution.

UPDATE 2020-02-02: I have uploaded to the research page a note reflecting on terminology. This note includes a discussion of the following problem.
At the FLoV talk, Graham Leigh pointed out that the terminology “necessarily non-analytic” might be a bit unfortunate. The reason is that we can construct cases where we have “ψ(x) witnesses that T proves ∀x.φ(x) by necessarily non-analytic induction” and where ψ(x) is a subformula of φ(x). Since analytic proofs are usually taken to be those satisfying a subformula property (in some sense), it might be a bit misleading to use the terminology “non-analytic” when the proof does satisfy a subformula property. We will have to think more about what terminology to use.

I also want to mention a piece of feedback that did not concern any problem. At JAF 38, Matt Kaufmann gave me an example of a non-analytic induction proof. For reasons of efficiency one might want to define the factorial on the natural numbers such that the definition is “tail-recursive”. Showing that this definition is equivalent to the standard one seems to require a non-analytic induction proof. This seems indeed to be the case. I have not verified all the details yet but when I have I will add this example to our summary document (on the research page).

My blog has a feed:

https://anderslundstedt.com/feed.xml

In the future I might add feeds restricted to certain categories of posts.