Positive Set Theory (GPK⁺∞)

Origin. Helen Skala and Isaac Malitz proposed positive comprehension in the 1970s; Forti and Hinnion, "The consistency problem for positive comprehension principles" (JSL 54, 1989), settled the basic questions and connected the theory to hyperuniverses (Forti–Honsell). Olivier Esser gave the theory its modern form and its metatheory in a sequence of papers (1996–2004), including the interpretation of ZF and Kelley–Morse in GPK⁺∞ (1997) and the inconsistency of choice with it (2000).

Models. A universal set obtained by restricting comprehension to positive formulas — those built without negation. Russell's paradox needs ¬(x ∈ x), which is not positive, so {x : x = x} is a set and {x : x ∉ x} is not even a candidate. This is the second route to a universal set, and it is not the New Foundations route: NF restricts by stratification, GPK by polarity, and the two theories are incomparable in method, model, and strength. The natural models are hyperuniverses — compact topological spaces in which every closed subset is a point.

Formalism.

Language: ∈, =.

Positive formulas. BPF (bounded positive formulas) is the smallest class containing the atomic formulas x ∈ y and x = y and closed under ∧, ∨, ∀x ∈ y, ∃x ∈ y, ∃x, ∀x. No negation, no →, no unbounded ¬. GPF (generalized positive formulas): BPF, plus formulas ∀x(θ(x) → φ) where θ is a GPF with exactly one free variable and no parameters, and φ is GPF with parameters.

Axioms of GPK:

• Extensionality: ∀x∀y (x = y ↔ ∀z (z ∈ x ↔ z ∈ y))
• Empty set: ∃x ∀y (y ∉ x)
• GPF comprehension: the universal closure of ∃x ∀y (y ∈ x ↔ φ), for every GPF formula φ (with parameters) in which x does not occur.

GPK⁺∞: GPK plus a form of the axiom of infinity and a closure scheme (Esser 1997).

Immediate consequences: V = {x : x = x} is a set, since x = x is positive. {x : x ∉ x} is not formable — ∉ is not positive. The paradox is blocked at the syntax, not by a size restriction. Every class defined by a positive condition has a set intersection-closure; the "positive closure" operator is the topological closure in the hyperuniverse model.

Strength (Esser): 1997: GPK⁺∞ interprets ZF, and interprets Kelley–Morse class theory. 1999: GPK⁺∞ + AC_WF (choice restricted to well-founded sets) and KM + global choice + "On is ramifiable" are mutually interpretable. Since "On is ramifiable" implies a proper class of inaccessibles, GPK⁺∞ is a strong theory — it proves the consistency of ZFC. Equivalently stated: GPK⁺∞ is mutually interpretable with Morse–Kelley plus "the class of all ordinals is weakly compact"; this proves a proper class of hyper-Mahlo cardinals exists. 2000: AC is inconsistent with GPK⁺∞. Choice fails, not by omission but by refutation. 1996: GPK + AFA is inconsistent — anti-foundation and positive comprehension do not combine.

Weakenings: Fackler's topological set theory (2012), a common weakening of ZF, GPK, and hyperuniverse theory, with the consistency strength of ZF; the universal set axiom is what lifts it to GPK's strength.

Symbols.

Metatheory. GPK⁺∞ answers the same question as New Foundations — how to have a universal set — with a different mechanism, and the comparison is the reason both belong in the division. NF restricts comprehension by stratification, has an unsettled consistency status until Holmes's proof, and refutes choice as a theorem (Specker). GPK restricts by polarity, has natural topological models, is strong rather than weak — it interprets KM plus weak compactness of On and hence proves Con(ZFC) — and also refutes choice, by Esser's theorem, for entirely different reasons. That two such different restrictions both kill choice is a fact about universal sets, not a coincidence of either method. The hyperuniverse semantics is what makes GPK tractable: positive comprehension is topological closure, and the models are compact spaces where every closed set is a point.

Applies to. Set theory with a universal set, as the non-stratified alternative to NF. Hyperuniverses and topological set theory. Non-well-founded phenomena, where positive conditions naturally define self-membered sets. The consistency-strength hierarchy above ZFC, since GPK⁺∞ sits at weak compactness of On. Paraconsistent and positive-logic set theories, where the polarity restriction is the shared idea.

Limitations. Choice is refutable, so ordinary mathematics does not transfer. The comprehension class is defined syntactically and is delicate — GPF's admission of ∀x(θ(x) → φ) is a specific patch whose motivation is technical rather than principled. Consistency requires large cardinals: GPK⁺∞ proves Con(ZFC), so it cannot be justified from ZFC. Anti-foundation is inconsistent with it. Much of the theory's development is Esser's alone, and the literature is thin relative to NF's.

© 2026 Lingenic LLC