Positive set theory

From Cantor's Attic
Jump to: navigation, search

Positive set theory is the name of a certain group of axiomatic set theories originally created as an example of a (nonstandard) set theories in which the axiom of foundation fails (e.g. there exists $x$ such that $x\in x$). [1] Those theories are based on a weakening of the (inconsistent) comprehension axiom of naive set theory (which asserts that every formula $\phi(x)$ defines a set that contains all $x$ such that $\phi(x)$) by restraining the formulas used to a smaller class of formulas called positive formulas.

While most positive set theories are weaker than $\text{ZFC}$ (and usually mutually interpretable with second-order arithmetic), one of them, $\text{GPK}^+_\infty$ turns out to be very powerful, being mutually interpretable with Morse-Kelley set theory plus an axiom asserting that the class of all ordinals is weakly compact. [2]

Positive formulas

In the first-order language $\{=,\in\}$, we define a BPF formula (bounded positive formula) the following way [2]: For every variable $x$, $y$ and BPF formulas $\varphi$, $\psi$,

  • $x=y$ and $x\in y$ are BPF.
  • $\varphi\land\psi$, $\varphi\lor\psi$, $\exists x\varphi$ and $(\forall x\in y)\varphi$ are BPF.

A formula is then a GPF formula (generalized positive formula) if it is a BPF formula or if it is of the form $\forall x(\theta(x)\Rightarrow\varphi)$ with $\theta(x)$ a GPF formula with exactly one free variable $x$ and no parameter and $\varphi$ is a GPF formula (possibly with parameters). [3]

$\text{GPK}$ positive set theories

The positive set theory $\text{GPK}$ consists of the following axioms:

  • Empty set: $\exists x\forall y(y\not\in x)$.
  • Extensionality: $\forall x\forall y(x=y\Leftrightarrow\forall z(z\in x\Leftrightarrow z\in y))$.
  • GPF comprehension: the universal closure of $\exists x\forall y(y\in x\Leftrightarrow\varphi)$ for every GPF formula $\varphi$ (with parameters) in which $x$ does not occur.

The empty set axiom is necessary, as without it the theory would hold in the trivial model which has only one element satisfying $x=\{x\}$. Note that, while $\text{GPK}$ do proves the existence of a set such that $x\in x$, Olivier Esser proved that it refutes the anti-foundation axiom (AFA). [3]

The theory $\text{GPK}^+$ is obtained by adding the following axiom:

  • Closure: the universal closure of $\exists x(\forall z(\varphi(z)\Rightarrow z\in x)\land\forall y(\forall z(\varphi(z)\Rightarrow z\in y)\Rightarrow x\subset y))$ for every formula $\varphi(z)$ (not necessarily BPF or GPF) with a free variable $z$ (and possibly parameters) such that $x$ does not occur in $\varphi$.

This axiom scheme asserts that for any (possibly proper) class $C=\{x|\varphi(x)\}$ there is a smallest set $X$ containing $C$, i.e. $C\subset X$ and for all sets $Y$ such that $C\subset Y$, one has $X\subset Y$. [4]

Note that replacing GPF comprehension in $\text{GPK}^+$ by BPF comprehension does not make the theory any weaker: BPF comprehension plus Closure implies GPF comprehension.

Both $\text{GPK}$ and $\text{GPK}^+$ are consistent relative to $\text{ZFC}$, in fact mutually interpretable with second-order arithmetic. However a much stronger theory, $\text{GPK}^+_\infty$, is obtained by adding the following axiom:

  • Infinity: the von Neumann ordinal $\omega$ is a set.

By "von Neumann ordinal" we mean the usual definition of ordinals as well-ordered-by-inclusion sets containing all the smaller ordinals. Here $\omega$ is the set of all finite ordinals (the natural numbers). The point of this axiom is not implying the existence of an infinite set; the class $\omega$ exists, so it has a set closure which is certainely infinite. This set closure happens to satisfy the usual axiom of infinity of $\text{ZFC}$ (i.e. it contains 0 and the successor of all its members) but in $\text{GPK}^+$ this is not enough to deduce that $\omega$ itself is a set (an improper class).

Olivier Esser showed that $\text{GPK}^+_\infty$ can not only interpret $\text{ZFC}$ (and prove it consistent), but is in fact mutually interpretable with a much stronger set theory, namely, Morse-Kelley set theory with an axiom asserting that the (proper) class of all ordinals is weakly compact. This theory is powerful enough to prove, for instance, that there exists a proper class of hyper-Mahlo cardinals. [2]

As a topological set theory

To be expanded.

The axiom of choice and $\text{GPK}$ set theories

To be expanded. [5][1]

Other positive set theories and the inconsistency of the axiom of extensionality

To be expanded. [6]


  1. Forti, M and Hinnion, R. The Consistency Problem for Positive Comprehension Principles. J Symbolic Logic 54(4):1401--1418, 1989. bibtex
  2. Esser, Olivier. An Interpretation of the Zermelo-Fraenkel Set Theory and the Kelley-Morse Set Theory in a Positive Theory. Mathematical Logic Quarterly 43:369--377, 1997. www   DOI   bibtex
  3. Esser, Olivier. Inconsistency of GPK+AFA. Mathematical Logic Quarterly 42:104--108, 1996. www   DOI   bibtex
  4. Esser, Olivier. On the Consistency of a Positive Theory. Mathematical Logic Quarterly 45:105--116, 1999. www   DOI   bibtex
  5. Esser, Olivier. Inconsistency of the Axiom of Choice with the Positive Theory $GPK^+_\infty$. Journal of Symbolic Logic 65(4):1911--1916, Dec., 2000. www   DOI   bibtex
  6. Esser, Olivier. On the axiom of extensionality in the positive set theory. Mathematical Logic Quarterly 19:97--100, 2003. www   DOI   bibtex
Main library

    This article is a stub. Please help us to improve Cantor's Attic by adding information.