We develop a systematic algorithmic framework that unites global and local classification problems for functional separable spaces and apply it to attack classification problems concerning the Banach space C[0,1] of real-valued continuous functions on the unit interval. We prove that the classification problem for continuous (binary) regular functions among almost everywhere linear, pointwise linear-time Lipshitz functions is Σ 0 2 -complete. We show that a function f ∶ [0, 1] → R is (binary) transducer if and only if it is continuous regular; interestingly, this peculiar and nontrivial fact was overlooked by experts in automata theory. As one of many consequences, our Σ 0 2 -completeness result covers the class of transducer functions as well. Finally, we show that the Banach space C[0, 1] of real-valued continuous functions admits an arithmetical classification among separable Banach spaces. Our proofs combine methods of abstract computability theory, automata theory, and functional analysis.
Contents1. Introduction 1 2. Background on regular functions 5 2.1. Regular functions 6 2.2. Transducer functions 10 2.3. Linear time computability 11 3. Proof of Theorem 1.1 12 3.1. The technical propositions 14 4. Proof of Theorem 1.2 20 5. Proof of Theorem 1.3 21 References 25Note that computability implies continuity. An adaptation of polynomial-time computability to continuous functions has previously been studied; the foundations can be found in [Ko91]. However, it is not necessarily clear that polynomial-time computability is the right notion of efficiency when it comes to infinite objects; see [BDKM19, Rou19] for a discussion. It is thus not surprising that other models of efficient computability for members of C[0, 1] have been tested as well; we will be focused on two such notions.