Bar Recursion Research Articles

Primitive Recursion and the Chain Antichain Principle

Let the chain antichain principle (CAC) be the statement that each partial order on $\mathbb{N}$ possesses an infinite chain or an infinite antichain. Chong, Slaman, and Yang recently proved using forcing over nonstandard models of arithmetic that CAC is $\Pi^1_1$-conservative over $\text{RCA}_0+\Pi^0_1\text{-CP}$ and so in particular that CAC does not imply $\Sigma^0_2$-induction. We provide here a different purely syntactical and constructive proof of the statement that CAC (even together with WKL) does not imply $\Sigma^0_2$-induction. In detail we show using a refinement of Howard's ordinal analysis of bar recursion that $\text{WKL}_0^\omega+\text{CAC}$ is $\Pi^0_2$-conservative over PRA and that one can extract primitive recursive realizers for such statements. Moreover, our proof is finitary in the sense of Hilbert's program. CAC implies that every sequence of $\mathbb{R}$ has a monotone subsequence. This Bolzano-Weierstras}-like principle is commonly used in proofs. Our result makes it possible to extract primitive recursive terms from such proofs. We also discuss the Erdős-Moser principle, which—taken together with CAC—is equivalent to $\text{RT}^2_2$.

Gödel functional interpretation and weak compactness

In recent years, proof theoretic transformations (so-called proof interpretations) that are based on extensions of monotone forms of Gödel’s famous functional (‘Dialectica’) interpretation have been used systematically to extract new content from proofs in abstract nonlinear analysis. This content consists both in effective quantitative bounds as well as in qualitative uniformity results. One of the main ineffective tools in abstract functional analysis is the use of sequential forms of weak compactness. As we recently verified, the sequential form of weak compactness for bounded closed and convex subsets of an abstract (not necessarily separable) Hilbert space can be carried out in suitable formal systems that are covered by existing metatheorems developed in the course of the proof mining program. In particular, it follows that the monotone functional interpretation of this weak compactness principle can be realized by a functional Ω∗ definable from bar recursion (in the sense of Spector) of lowest type. While a case study on the analysis of strong convergence results (due to Browder and Wittmann resp.) that are based on weak compactness indicates that the use of the latter seems to be eliminable, things apparently are different for weak convergence theorems such as the famous Baillon nonlinear ergodic theorem. For this theorem we recently extracted an explicit bound on a metastable (in the sense of T. Tao) version of this theorem that is primitive recursive relative to (a somewhat restricted form of) Ω∗.In the current paper we for the first time give the construction of Ω∗ (in the form needed for the unwinding of Baillon’s theorem). As a corollary to the fine analysis of the use of bar recursion in this construction we obtain that Ω∗ elevates arguments in Tn at most to resulting functionals in Tn+2 (here Tn is the fragment of Gödel’s T with primitive recursion restricted to the type level n). In particular, one can conclude from this that our bound on Baillon’s theorem is at least definable in T4.

The Peirce translation

We develop applications of selection functions to proof theory and computational extraction of witnesses from proofs in classical analysis. The main novelty is a translation of minimal logic plus Peirce law into minimal logic, which we refer to as Peirce translation, as it eliminates uses of Peirce law. When combined with modified realizability, this translation applies to full classical analysis, i.e. Peano arithmetic in the language of finite types extended with countable choice and dependent choice. A fundamental step in the interpretation is the realizability of a strengthening of the double-negation shift via the iterated product of selection functions. In a separate paper, we have shown that such a product of selection functions is equivalent, over system T, to modified bar recursion.

Selection functions, bar recursion and backward induction

Bar recursion arises in constructive mathematics, logic, proof theory and higher-type computability theory. We explain bar recursion in terms of sequential games, and show how it can be naturally understood as a generalisation of the principle of backward induction that arises in game theory. In summary, bar recursion calculates optimal plays and optimal strategies, which, for particular games of interest, amount to equilibria. We consider finite games and continuous countably infinite games, and relate the two. The above development is followed by a conceptual explanation of how the finite version of the main form of bar recursion considered here arises from a strong monad of selections functions that can be defined in any cartesian closed category. Finite bar recursion turns out to be a well-known morphism available in any strong monad, specialised to the selection monad.

Mathematical Logic: Proof Theory, Constructive Mathematics

The workshop Mathematical Logic: Proof Theory, Constructive Mathematics was held April 6-12, 2008 and had the following main aims: To promote the interaction of proof theory with core areas of mathematics. This, in particular, refers to uses of proof theoretic techniques (most notably various forms of functional and realizability interpretations) to unwind prima facie ineffective proofs in mathematics. For instance a number of talks presented such applications in the areas of fixed point theory, ergodic theory, topological dynamics and symbolic dynamics resulting in new theorems, surprising to the experts in the respective fields (Avigad, Briseid, Gerhardy, Leuştean, Simpson). Another talk addressed uses of analytic number theory in connection with new unprovability results and phase transitions in provability (Bovykin). To further develop new directions in both the foundations as well as applied forms of constructive mathematics. There were a number of talks focusing on new formal frameworks for constructive mathematics and their proof theoretic properties (Aczel, Iemhoff, J. Moschovakis, Palmgren, Streicher), but also new constructive (and partly computer generated) developments in specific areas such as algebra, analysis, combinatorics and quantum theory (Bauer, Delzell, Lombardi, Paule, Spitters). To explore further the connections between proof theory and computational complexity (e.g. in connection with the study of systems of so-called bounded arithmetic, Beckmann, Pollett, Thapen). Another topic in this area was the study of implicit computational complexity and the design of functional programming languages corresponding to specific complexity classes (Aehlig, Hofmann, Matthes). Yet another approach to the issue of the intrinsic complexity for general classes of algorithmic problems was developed in a talk by Y. Moschovakis. In order to provide the participants with an overview over new developments in the study of functional and realizability interpretations (as mentioned above) two invited lecture series were given. The first one (given by Fernando Ferreira and Paulo Oliva) focussed on novel (so-called monotone and bounded) variants of Gödel's functional interpretation (first published exactly 50 years ago) and its decomposition into an interpretation of linear logic combined with Girard's embedding of intuitionistic logic into linear logic. The second course was delivered by Ulrich Berger and presented a number of proof theoretic approaches for analyzing proofs based on dependent choice (such as Spector's bar recursive solution, the realizability solution by modified bar recursion due to Berger and Oliva and a novel approach based on open induction and its relation to previous work by Berardi, Bezem and Coquand). Moreover, Berger reported on recent experiments (with Schwichtenberg) concerning the automated extraction (based on the MinLog tool) of efficient algorithms from proofs in constructive analysis.

Programs from proofs using classical dependent choice

This article generalises the refined A -translation method for extracting programs from classical proofs [U. Berger,W. Buchholz, H. Schwichtenberg, Refined program extraction from classical proofs, Annals of Pure and Applied Logic 114 (2002) 3–25] to the scenario where additional assumptions such as choice principles are involved. In the case of choice principles, this is done by adding computational content to the ‘translated’ assumptions, an idea which goes back to [S. Berardi, M. Bezem, T. Coquand, On the computational content of the axiom of choice, JSL 63 (2) (1998) 600–622] and has been further elaborated in [U. Berger, P. Oliva, Modified bar recursion and classical dependent choice, in: M. Baaz, S.D. Friedman, J. Kraijcek (Eds.), Logic Colloquium ’01, in: Lecture Notes in Logic, vol. 20, Springer, Berlin, 2005, pp. 89–107]. We further investigate the applicability of this extension by means of a small but non-trivial example, and discuss the formalisation as well as some optimisations necessary in order to extract terminating programs. Both formalisation and term extraction have been implemented in the interactive proof assistant Minlog.

A proof of strong normalisation using domain theory

Ulrich Berger presented a powerful proof of strong normalisation using domains, in particular it simplifies significantly Tait's proof of strong normalisation of Spector's bar recursion. The main contribution of this paper is to show that, using ideas from intersection types and Martin-Lof's domain interpretation of type theory one can in turn simplify further U. Berger's argument. We build a domain model for an untyped programming language where U. Berger has an interpretation only for typed terms or alternatively has an interpretation for untyped terms but need an extra condition to deduce strong normalisation. As a main application, we show that Martin-L\"{o}f dependent type theory extended with a program for Spector double negation shift.

Modified bar recursion

This paper studies modified bar recursion, a higher type recursion scheme, which has been used in Berardi et al. (1998) and Berger and Oliva (2005) for a realisability interpretation of classical analysis. A complete clarification of its relation to Spector's and Kohlenbach's bar recursion, the fan functional, Gandy's functional and Kleene's notion of S1–S9 computability is given.

Strong normalisation for applied lambda calculi

We consider the untyped lambda calculus with constructors and recursively defined constants. We construct a domain-theoretic model such that any term not denoting bottom is strongly normalising provided all its `stratified approximations' are. From this we derive a general normalisation theorem for applied typed lambda-calculi: If all constants have a total value, then all typeable terms are strongly normalising. We apply this result to extensions of G\"odel's system T and system F extended by various forms of bar recursion for which strong normalisation was hitherto unknown.

Modified Bar Recursion

We introduce a variant of Spector's bar recursion (called "modified bar recursion'') in finite types to give a realizability interpretation of the classical axiom of countable choice allowing for the extraction of witnesses from proofs of Sigma_1 formulas in classical analysis. As a second application of modified bar recursion we present a bar recursive definition of the fan functional. Moreover, we show that modified bar recursion exists in M (the model of strongly majorizable functionals) and is not S1-S9 computable in C (the model of total functionals). Finally, we show that modified bar recursion defines Spector's bar recursion primitive recursively.

How is it that infinitary methods can be applied to finitary mathematics? Gödel's T: a case study

AbstractInspired by Pohlers' local predicativity approach to Pure Proof Theory and Howard's ordinal analysis of bar recursion of type zero we present a short, technically smooth and constructive strong normalization proof for Gödel's system T of primitive recursive functionals of finite types by constructing an ε0-recursive function []0: T → ω so that a reduces to b implies [a]0 > [b]0. The construction of [ ]0 is based on a careful analysis of the Howard-Schütte treatment of Gödel's T and utilizes the collapsing function ψ: ε0 → ω which has been developed by the author for a local predicativity style proof-theoretic analysis of PA. The construction of [ ]0 is also crucially based on ideas developed in the 1995 paper “A proof of strongly uniform termination for Gödel's T by the method of local predicativity” by the author. The results on complexity bounds for the fragments of T which are obtained in this paper strengthen considerably the results of the 1995 paper.Indeed, for given n let Tn be the subsystem of T in which the recursors have type level less than or equal to n + 2. (By definition, case distinction functionals for every type are also contained in Tn.) As a corollary of the main theorem of this paper we obtain (reobtain?) optimal bounds for the Tn-derivation lengths in terms of ω+2-descent recursive functions. The derivation lengths of type one functionals from Tn (hence also their computational complexities) are classified optimally in terms of <ωn+2 -descent recursive functions.In particular we obtain (reobtain?) that the derivation lengths function of a type one functional a ∈ T0 is primitive recursive, thus any type one functional a in T0 defines a primitive recursive function. Similarly we also obtain (reobtain?) a full classification of T1 in terms of multiple recursion.As proof-theoretic corollaries we reobtain the classification of the IΣn+1-provably recursive functions. Taking advantage from our finitistic and constructive treatment of the terms of Gödel's T we reobtain additionally (without employing continuous cut elimination techniques) that PRA + PRWO(ε0) ⊢ Π20 − Refl(PA) and PRA + PRWO(ωn+2) ⊢ Π20 − Refl(IΣn+1), hence PRA + PRWO(ε0) ⊢ Con(PA) and PRA + PRWO(ωn+2) ⊢ Con(IΣn+1).For programmatic reasons we outline in the introduction a vision of how to apply a certain type of infinitary methods to questions of finitary mathematics and recursion theory. We also indicate some connections between ordinals, term rewriting, recursion theory and computational complexity.

Polymorphic extensions of simple type structures. With an application to a bar recursive minimization

The technical contribution of this paper is threefold. First we show how to encode functionals in a ‘flat’ applicative structure by adding oracles to untyped λ-calculus and mimicking the applicative behaviour of the functionals with an impredicatively defined reduction relation. The main achievement here is a Church-Rosser result for the extended reduction relation. Second, by combining the previous result with the model construction based on partial equivalence relations, we show how to extend a λ-closed simple type structure to a model of the polymorphic λ-calculus. Third, we specialize the previous result to a counter model against a simple minimization. This minimization is realized by a bar recursive functional, which contrasts the results of Spector and Girard which imply that the bar recursive functions are exactly those that are definable in the polymorphic λ-calculus. As a spin-off, we obtain directly the non-conservativity of the extensions of Gödel's T with bar recursion, fan functional, and Luckhardt's minimization functional, respectively. For the latter two extensions these results are new.

Equivalence of bar recursors in the theory of functionals of finite type

The main result of this paper is the equivalence of several definition schemas of bar recursion occurring in the literature on functionals of finite type. We present the theory of functionals of finite type, in [T] denoted byqf-WE-HA ω, which is necessary for giving the equivalence proofs. Moreover we prove two results on this theory that cannot be found in the literature, namely the deduction theorem and a derivation of Spector's rule of extensionality from [S]: ifP→T 1=T 2 and Q[X∶≡T1], then P→Q[X∶≡ T2], from the at first sight weaker rule obtained by omitting “P→”.

Ordinal analysis of simple cases of bar recursion

The purpose of the following is to give an ordinal analysis of bar recursion of type 0 for the cases in which the bar recursion operator has type level 3 or 4. We show that the associated ordinals are the ε0th epsilon number and the first ε0-critical number, respectively. Also we analyze a restricted bar recursion operator of level 3 which provides the Gödel functional interpretation of an intuitionistic form of König's lemma. The ordinal in this case is ε0.Let + BR0 be Gödel's theory of primitive recursive functional of finite type extended by functors Φ for bar recursion of type zero. To make bar recursion resemble transfinite recursion more closely, and thereby facilitate the ordinal analysis, we extend + BR0 to a system of terms . By Theorem 1.1a term of + BR0 which is computable by the rules of is already computable by the rules of + BR0, so it is sufficient to give an ordinal analysis of the terms of .A term H of is said to be semi-closed if its free variables have level not exceeding 1. If, in addition, H has level not exceeding 2, then an ordinal measure for H is given by a length for one of its computation trees. We call this a computation size for H. Let F(X) be a term whose free variables have level not exceeding 2, and suppose the variable X of type 2 is the only free variable of level 2.

On bar recursion of types 0 and 1

For general information on bar recursion the reader should consult the papers of Spector [8], where it was introduced, Howard [2] and Tait [11]. In this note we shall prove that the terms of Gödel's theory T(in its extensional version of Spector [8]) are closed under the rule BR0,1 of bar recursion of types 0 and 1. Our method of proof is based on the notion of an infinite term introduced by Tait [9]. The main tools of the proof are (i) the normalization theorem for (notations for) infinite terms and (ii) valuation functionals. Both are elaborated in [6]; for brevity some familiarity with this paper is assumed here. Using (i) and (ii) we reduce BR0,1 to ξ-recursion with ξ < ε0. From this the result follows by work of Tait [10], who gave a reduction of 2ξ-recursion to ξ-recursion at a higher type. At the end of the paper we discuss a perhaps more natural variant of bar recursion introduced by Kreisel in [4].Related results are due to Kreisel (in his appendix to [8]), who obtains results which imply, using the reduction given by Howard [2] of the constant of bar recursion of type τ to the rule of bar recursion of type (0 → τ) → τ, that T is not closed under the rule of bar recursion of a type of level ≥ 2, to Diller [1], who gave a reduction of BR0,1 to ξ-recursion with ξ bounded by the least ω-critical number, and to Howard [3], who gave an ordinal analysis of the constant of bar recursion of type 0. I am grateful to H. Barendregt, W. Howard and G. Kreisel for many useful comments and discussions.

Intensional interpretations of functionals of finite type I

T0 will denote Gödel's theory T[3] of functionals of finite type (f.t.) with intuitionistic quantification over each f.t. added. T1 will denote T0 together with definition by bar recursion of type o, the axiom schema of bar induction, and the schemaof choice. Precise descriptions of these systems are given below in §4. The main results of this paper are interpretations of T0 in intuitionistic arithmetic U0 and of T1 in intuitionistic analysis is U1. U1 is U0 with quantification over functionals of type (0,0) and the axiom schemata AC00 and of bar induction.