Normalization, Taylor expansion and rigid approximation of $λ$-terms

Federico Olimpieri
DOI: https://doi.org/10.48550/arXiv.2001.01619
2020-01-06
Abstract:The aim of this work is to characterize three fundamental normalization proprieties in lambda-calculus trough the Taylor expansion of $ \lambda$-terms. The general proof strategy consists in stating the dependence of ordinary reduction strategies on their resource counterparts and in finding a convenient resource term in the Taylor expansion that behaves well under the considered kind of reduction.
Logic in Computer Science
What problem does this paper attempt to address?
### What problem does this paper attempt to solve? This paper aims to characterize the three basic reduction properties in lambda calculus: head normalization, β - normalization, and strong normalization through the Taylor expansion of lambda terms. Specifically, the author introduces a rigid resource calculus and studies these reduction properties through Taylor expansion. #### Main objectives: 1. **Head Normalization**: Research how to characterize whether a lambda term can be head - reduced through Taylor expansion. 2. **β - Normalization**: Explore how to characterize whether a lambda term can be β - reduced through Taylor expansion. 3. **Strong Normalization**: Analyze how to characterize whether a lambda term has the strong normalization property through Taylor expansion. #### Methods: - **Rigid Resource Calculus**: Use lists instead of multisets to represent resources, simplifying definitions, theorems, and proofs. - **Taylor Expansion**: Expand ordinary lambda terms into a series of resource terms, and infer the reduction properties of the original lambda terms through the behavior of these resource terms. #### Innovation points: - Propose a new method to characterize reduction properties through rigid approximation and Taylor expansion. - Through this method, the author can give direct proofs of head reduction and left reduction, demonstrating the finiteness of the operational semantics of the resource calculus. - Extend this method to other fields, such as typability results in type systems, research on execution time, etc. ### Formula summary 1. **Definition of rigid resource terms**: \[ D \ni a, b, c ::= x \mid \lambda x.a \mid \langle c \rangle \vec{d} \mid 0 \] \[ D! \ni \vec{a}, \vec{b}, \vec{d} ::= () \mid (a) :: \vec{d} \] 2. **Rigid Substitution**: \[ e\{\vec{b}/x\} = \begin{cases} \vec{b} & \text{if } e = x \\ y & \text{if } e = y \neq x \\ \lambda y.(s\{\vec{b}/x\}) & \text{if } e = \lambda y.s \\ \langle s\{\vec{b}_0/x\} \rangle \vec{d}\{\vec{b}_1/x\} & \text{if } e = \langle s \rangle \vec{b_0}::\vec{b_1} \\ (a_1\{\vec{b}_1/x\}, \ldots, a_n\{\vec{b}_n/x\}) & \text{if } e = (a_1, \ldots, a_n) \end{cases} \] 3. **Head Reduction**: \[ H(M) = \begin{cases} M & \text{if } M \text{ is in head - reduction form} \\ \lambda x_1 \ldots \lambda x_m.P[Q/x]Q_1 \ldots Q_n & \text{otherwise} \end{cases} \]