A countable structure is said to be punctual if it is isomorphic to a structure with domain $\omega $ such that all of its functions and relations are uniformly primitive recursive, henceforth p.r. The key challenge in working primitive recursively versus computably is that things must be defined essentially instantly. Such a phenomenon leads to a number of interesting results and techniques in the study of punctual structures different from their analogues in computable structure theory.

A well-known result in computable structure theory is the computable universality of graphs. (See N. Bahzenov and M. Harrison-Trainor, Constructing decidable graphs from decidable structures. $\boldsymbol {Algebra \ and \ Logic}$ , vol. 58 (2019), pp. 369–382.) However, as suggested by the title of the second article under review, graphs are not universal for punctual computability. In fact, in the first article under review, it is shown that no relational structure is p.r. universal.Footnote ¹ This captures the intuition that relational structures are ‘weak’ for punctual computability; pathological properties are generally exhibited by punctual structures with function symbols. This difference is further supported by the fact that a function f and its graph should have the same Turing degree, but f is not p.r. computable from the graph of f in general.

Such an intuition is formally described by the notion of intrinsically punctual functions, functions that are p.r. relative to every presentation $A\cong {\mathcal {A}}$ , for some fixed structure ${\mathcal {A}}$ . Kalimullin, Melnikov, and Montalbán showed that f is intrinsically punctual in ${\mathcal {A}}$ iff f is p.r. bounded, and there is some $\overline {a}\in {\mathcal {A}}^{<\omega }$ and a punctual family of atomic formulas $\{\Phi _{g}\}$ such that $f(x)=y$ iff $\Phi _{g(x,y)}(\overline {a})$ holds in $({\mathcal {A}},\overline {a})$ (plus some technicalities). Intuitively, this means that f is completely determined by some ‘sufficiently large’ initial part $\overline {a}\subseteq {\mathcal {A}}$ . One may think of $\{\Phi _{g}\}$ here as some analogue of the formally $\Sigma _{1}^{0}$ formula (equivalently, a computable family of existential finitary formulas) in the classification of r.i.c.e. relations for computable structures. Although the proofs of both theorems are similar; $\overline {a}\in {\mathcal {A}}^{<\omega }$ fulfills the role of the fixed $\overline {c}$ that generates the formally $\Sigma _{1}^{0}$ formula as presented in C. J. Ash and J. Knight, $\boldsymbol {Computable \ structures \ and \ the \ hyperarithmetical \ hierarchy}$ , Studies in Logic and the Foundations of Mathematics, vol. 144 (2000). We note that the effectiveness condition necessary in the computable setting is not quite present in the punctual analogue above.

Using the classification above, the authors showed that if ${\mathcal {A}}$ is relational, then any f intrinisically punctual in ${\mathcal {A}}$ must also be p.r. Coupled with the observation that any nonrecursive set X is intrinsically punctual in ${\mathcal {A}}=(\omega ,S,0,X)$ , where S is the successor function, $0$ is the unique element not contained in $rng(S)$ , and X is the predicate ‘ $XS^{n}(0)=1$ iff $n\in X$ ’, one can conclude that relational structures cannot be p.r. universal. Observe that if S is replaced with the successor relation, then the characteristic function of X is no longer intrinsically punctual in ${\mathcal {A}}$ .

To complete the discussion on universality, we look at the example of a p.r. universal structure constructed in the second article under review. The authors there showed that the class of structures with a single binary functional symbol is p.r. universal. While the encoding seems standard, extending the domain and using a modified concatenation operation to simulate each relation and function, additional care has to be taken in order to ensure that the ‘recovery’ can be done primitive recursively. Much of this additional complexity boils down to the fact that the inverse of a (bijective) p.r. function is not necessarily p.r., thus requiring a careful analysis of the usage of the minimisation operator in defining the inverse.

Another property of extensive study in computable structure theory is that of categoricity. One approach is to study this ‘on a cone’. The benefit of such an approach is that pathological properties generally disappear (see C. F. D. McCoy, Finite computable dimension does not relativize. $\boldsymbol {Archive \ for \ Mathematical \ Logic}$ , vol. 41 (2002), pp. 309–320 and S. Goncharov, The problem of the number of nonautoequivalent constructivizations. $\boldsymbol {Algebra \ i \ Logika}$ , vol. 19 (1980), pp. 621–639). Along these lines, the authors of the first article under review studied punctual categoricity on a cone. A structure ${\mathcal {A}}$ is punctually categorical on a cone (henceforth p.c.c.) if there is a total function f such that for any $A,B\cong {\mathcal {A}}$ , $f\oplus A\oplus B$ p.r. computes both an isomorphism from A to B and its inverse. We define $\sim _{orb}$ to be an equivalence relation between finite tuples of a structure ${\mathcal {A}}$ such that $\overline {x}\sim _{orb}\overline {y}$ iff there is an isomorphism of ${\mathcal {A}}$ mapping $\overline {x}$ to $\overline {y}$ . The authors then showed that a structure is p.c.c. iff it is $\exists $ -atomic (over a finite set of parameters), uniformly locally finite, and highly local. More specifically, these conditions respectively provide existential formulas $\varphi _{\overline {x}}(\overline {y})$ that hold iff $\overline {x}\sim _{orb}\overline {y}$ ; provide a primitive recursive bound on the sizes of finitely generated substructures; and ensure that each such substructure of sufficiently large size is representative of the global structure, containing representatives from each $\sim _{orb}$ equivalence class.

In the computable setting, being $\exists $ -atomic over a finite set of parameters is equivalent to being computably categorical on a cone (A. Montalbán, Effectively existentially-atomic structures, in $\boldsymbol {Computability \ and \ Complexity}$ , Lecture Notes in Computer Science, vol. 10010 (2017), pp. 221–237). To see this, given some partial isomorphism $g:\overline {x}\subseteq A\to B$ , and $y\in A$ , search for the appropriate $z\in B$ so that $(B,D)\vDash \varphi _{\overline {x}y}(g(\overline {x})z)$ and then define $g(y)=z$ . Since $\varphi _{\overline {x}y}$ is an existential formula, the process described gives a computable way to perform the back-and-forth argument between A and B. The reader may have already deduced why such an argument fails in the punctual setting: the search for z is unbounded in general. Thus, the additional conditions of uniform local finiteness and high locality are necessary to bound the search for z in the punctual setting.

Expanding upon the theme of categoricity, an approach that is unique to its study in the punctual setting is the notion of punctual degrees. Formally, for a given structure ${\mathcal {A}}$ , the punctual degrees of ${\mathcal {A}}$ , denoted $PR({\mathcal {A}})$ , are obtained by considering all punctual presentations $A\cong {\mathcal {A}}$ and ordering them by defining $A\leq _{pr}B$ if there exists a p.r. isomorphism $f:A\to B$ . Intuitively, such a degree structure measures the different speeds of enumerations of presentations for a fixed structure. For instance, if $PR({\mathcal {A}})$ has a greatest or least element, then there is respectively a ‘fastest’ or ‘slowest’ presentation of ${\mathcal {A}}$ . As noted earlier, since the inverse of a p.r. isomorphism is not necessarily p.r., $PR({\mathcal {A}})$ is generally nontrivial for ‘most’ ${\mathcal {A}}$ .

Using this approach, the order-theoretic properties of the punctual degrees for various structures have been investigated. For instance, it has been shown that finitely generated structures and (computably almost) rigid structures possess dense punctual degrees (N. Bazhenov, I. Kalimullin, A. Melnikov, and K. M. Ng, Online presentations of finitely generated structures. $\boldsymbol {Theoretical \ Comptuer \ Science}$ , vol. 844 (2020), pp. 195–216, and M. Dorzhieva, R. Downey, E. Hammatt, A. Melnikov, and K. M. Ng, Punctually presented structures II: Comparing presentations. $\boldsymbol {Archive \ for \ Mathematical \ Logic}$ , vol. 64 (2025), pp. 159–184). On the other hand, structures with nondense punctual degrees seem to be much rarer; the third article under review constructs the first such example. In fact, it is also the first example of a nondegenerate infinite injury via the tree method in a punctual setting.

Classically, when performing a $\emptyset "$ -priority argument, objects being defined for the $\Pi _{2}^{0}$ -outcome of a requirement may be ‘paused’ for indefinitely long while the $\Sigma _{2}^{0}$ -outcome of the requirement is advancing its strategy. This is never an issue in the computable setting, as the strategy may wait for as long as the corresponding outcome is ‘paused’. However, in the punctual setting, any object once defined must continue to be maintained primitive recursively even when attending to requirements to its right on the tree. In particular, when constructing $A<_{pr}B$ , we may have to enumerate elements into A ‘quickly’ for the sake of $\varphi :M\to A$ for some given $M \leq _{pr}$ -between $A,B$ . This acts in obvious tension with ensuring that $A<_{pr}B$ ; when an element b is enumerated into B, one should expect to keep from enumerating any potential images for b into A as long as possible, but the ‘opponent’ may press us (via $\varphi $ ) into revealing such an image in A by enumerating the preimage of b into M. (Here, ‘press’ is a technical term to indicate that a certain property must be witnessed in p.r. time.)

The key idea in constructing $A<_{pr}B$ with no $M \leq _{pr}$ between them is in analysing the speed of enumeration of M. More formally, given M punctual and p.r. isomorphisms $\psi :A\to M$ and $\theta :M\to B$ , the construction attempts to ‘guess’ a p.r. bound for $\theta ^{-1}$ by testing it against a list of p.r. timestamp functions $\{p_{i}\}_{i\in \omega }$ ; either there exists n such that $\theta ^{-1}(b)[\max _{i<n}p_{i}(b)]\!\downarrow $ for cofinitely many b; or for each n, there exists some b where $\theta ^{-1}(b)[\max _{i\leq n}p_{i}(b)]\!\uparrow $ . In the former case, the construction succeeds in finding some p.r. bound for $\theta ^{-1}$ and thus witnesses that $B\leq _{pr}M$ . By carefully defining $\varphi :M\to A$ , one may succeed in keeping $A<_{pr}B$ . In the latter case, M is (at least intuitively) slow in revealing its elements, ensuring that the construction succeeds in defining $\varphi :M\to A$ surjectively while keeping $A<_{pr}B$ .

The three articles under review provide a good overview of some differences between computable and punctual structures; both in terms of the results and the techniques used in the proofs, suggesting that punctual structures provide a new and interesting perspective. Many aspects of computable structure theory have only begun to be explored in the punctual setting, for instance, notions like degree spectra of relations or categoricity spectra of structures. Some investigations also consider punctual structures up to honest isomorphisms (computable isomorphisms whose graphs are p.r.). Finally, and perhaps most interestingly, almost no general theory has been developed regarding the punctual degrees of various structures. Further investigations could potentially provide insight into connections between algebraic properties and the use of unbounded search in back-and-forth arguments.