Abstract:We show how to express intuitionistic Zermelo set theory in deduction modulo (i.e. by replacing its axioms by rewrite rules) in such a way that the corresponding notion of proof enjoys the normalization property. To do so, we first rephrase set theory as a theory of pointed graphs (following a paradigm due to P. Aczel) by interpreting set-theoretic equality as bisimilarity, and show that in this setting, Zermelo's axioms can be decomposed into graph-theoretic primitives that can be turned into rewrite rules. We then show that the theory we obtain in deduction modulo is a conservative extension of (a minor extension of) Zermelo set theory. Finally, we prove the normalization of the intuitionistic fragment of the theory.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to establish a theory for Zermelo set theory (a fundamental set theory) that can be expressed within the framework of "deduction modulo", so that the proofs in this theory possess the normalization property. Specifically, the author hopes to rewrite the intuitionistic Zermelo set theory into a form based on rewrite rules rather than traditional axioms, so that the corresponding proof concept can enjoy the normalization property. To achieve this goal, they adopted the following steps:
1. **Re - formulate the set theory as a pointed graph theory**: By interpreting the equivalence relation in set theory as bisimilarity, and showing that in this setting, Zermelo's axioms can be decomposed into basic elements of graph theory, which can be transformed into rewrite rules.
2. **Prove conservative extension**: Show that the theory obtained in deduction modulo is a conservative extension of the (slightly extended) Zermelo set theory.
3. **Prove the normalization of the intuitionistic fragment**: Finally, prove that the intuitionistic part of the obtained theory has the normalization property.
### Key Formulas and Concepts
- **Bisimilarity**:
\[
a \approx b \equiv \exists r \left( \text{rel}(\text{root}(a), \text{root}(b), r) \land \forall x \forall x' \forall y \left( x' \eta_a x \land \text{rel}(x, y, r) \Rightarrow \exists y' (y' \eta_b y \land \text{rel}(x', y', r)) \right) \land \forall y \forall y' \forall x \left( y' \eta_b y \land \text{rel}(x, y, r) \Rightarrow \exists x' (x' \eta_a x \land \text{rel}(x', y', r)) \right) \right)
\]
- **Membership**:
\[
a \in b \equiv \exists x \left( x \eta_b \text{root}(b) \land a \approx (b/x) \right)
\]
- **Root Relocation**:
\[
\text{root}(a/x) \to x
\]
\[
(a/x)/y \to a/y
\]
\[
x \eta a/z y \to x \eta a y
\]
### Problems Solved
Through the above methods, the authors solved the problems encountered in previous attempts to extend the normalization theorem to set theory. In particular, the counter - example proposed by M. Crabbé shows that directly replacing the axioms of set theory with rewrite rules will destroy the normalization property. The method proposed in this paper successfully bypasses these problems by introducing bisimilarity and rewrite rules, and proves the normalization property in the new theoretical framework.
### Summary
The main contribution of this paper is to provide a new method to deal with the problem of proof normalization in set theory. This method not only avoids existing counter - examples, but also provides potential application prospects for a wider range of logical systems.