Tanaka's Theorem Revisited

Abstract

Tanaka (1997) proved a powerful generalization of Friedman's self-embedding theorem that states that given a countable nonstandard model $(\mathcal{M},\mathcal{A})$ of the subsystem $\mathrm{WKL}_{0}$ of second order arithmetic, and any element $m$ of $\mathcal{M}$, there is a self-embedding $j$ of $(\mathcal{M},\mathcal{A})$ onto a proper initial segment of itself such that $j$ fixes every predecessor of $m$. Here we extend Tanaka's work by establishing the following results for a countable nonstandard model $(\mathcal{M},\mathcal{A})$ of $\mathrm{WKL}_{0} $ and a proper cut $\mathrm{I}$ of $\mathcal{M}$: Theorem A. The following conditions are equivalent: (a) $\mathrm{I}$ is closed under exponentiation. (b) There is a self-embedding $j$ of $(\mathcal{M},\mathcal{A})$ onto a proper initial segment of itself such that $I$ is the longest initial segment of fixed points of $j$. Theorem B. The following conditions are equivalent: (a) $\mathrm{I}$ is a strong cut of $\mathcal{M} $ and $\mathrm{I}\prec _{\Sigma _{1}}\mathcal{M}.$ (b) There is a self-embedding $j$ of $(\mathcal{M},\mathcal{A})$ onto a proper initial segment of itself such that $\mathrm{I} $ is the set of all fixed points of $j$.