nLab exhaustible set

Redirected from "omniscient set".

Contents

Definition

In constructive mathematics, a set AA is exhaustible or omniscient if it satisfies a version of the limited principle of omniscience for the set AA rather than the natural numbers \mathbb{N}: For all functions ff from AA to the booleans {0,1}\{0, 1\}, either there exists an element xAx \in A such that f(x)=1f(x) = 1, or for all elements xAx \in A, f(x)=0f(x) = 0.

One can equivalently use decidable subsets instead of functions into the booleans. A set AA is exhaustible or omniscient if the existential quantification of any decidable proposition on AA is again decidable. That is,

(xA,P(x)¬P(x))(xA,P(x))¬(xA,P(x)), (\forall x \in A, P(x) \vee \neg{P(x)}) \Rightarrow (\exists x \in A, P(x)) \vee \neg(\exists x \in A, P(x)) ,

or equivalently

(xA,P(x)¬P(x))(xA,P(x))(xA,¬P(x)). (\forall x \in A, P(x) \vee \neg{P(x)}) \Rightarrow (\exists x \in A, P(x)) \vee (\forall x \in A, \neg{P(x)}) .

If you take the set AA to be the subsingleton corresponding to a given truth value and apply this principle to the constantly true proposition, then the subsingleton is decidable follows, since xA,P(x)\forall x \in A, P(x) holds, then either the subsingleton is inhabited, in which xA,P(x)\exists x \in A, P(x) holds, or the subsingleton is empty, in which xA,¬P(x)\forall x \in A, \neg P(x) holds. Thus, that every subsingleton is exhaustible implies EMEM; conversely, EMEM implies that every set is exhaustible (over any domain). Note that propositions of the form x,P(x)\exists x, P(x) when PP is decidable are the open truth values in the Rosolini dominance.

We can also state the principle set-theoretically: a set AA is exhaustible or omniscient if for all subsets P𝒫(A)P \in \mathcal{P}(A), if PP is a decidable subset of AA (PP¯=AP \cup \bar{P} = A), then either PP is inhabited or P¯=A\bar{P} = A, where P¯\bar{P} is the complement of PP.

Examples

Properties

This section appears to be original research, as there doesn’t seem to be any existing literature on these properties. Note that the material in this section originally appeared on the article limited principle of omniscience before being moved here, so please check that article for the history of this material.

Theorem

A set AA is exhaustible if and only if the function set 𝟚 A\mathbb{2}^A has decidable tight apartness, where the tight apartness relation on the function set 𝟚 A\mathbb{2}^A be defined as

f#gxA.f(x)g(x)f \# g \coloneqq \exists x \in A.f(x) \neq g(x)

Proof

Since the boolean domain is a boolean ring, the function set 𝟚 A\mathbb{2}^A is also a boolean ring via pointwise addition, subtraction and multiplication of boolean-valued functions. Thus, for functions ff and gg from AA to 𝟚\mathbb{2}, the tight apartness relation on the function set 𝟚 A\mathbb{2}^A, defined by

f#gxA.f(x)g(x)f \# g \coloneqq \exists x \in A.f(x) \neq g(x)

is logically equivalent to xA.f(x)g(x)=1\exists x \in A.f(x) - g(x) = 1, since f(x)g(x)f(x) \neq g(x) is logically equivalent to f(x)=g(x)+1f(x) = g(x) + 1.

Suppose that AA is exhaustible holds. Then for functions ff and gg from AA to 𝟚\mathbb{2}, define the function hh by h(x)=f(x)g(x)h(x) = f(x) - g(x) for all xx \in \mathbb{N}. Then since AA being exhaustible says that xA.h(x)=1\exists x \in A.h(x) = 1 is decidable, and xA.h(x)=1\exists x \in A.h(x) = 1 is logically equivalent to f#gf \# g, then f#gf \# g is decidable.

Conversely, assume that f#gf \# g is decidable. Then take gg to be the constant function at the boolean 00, and f#λxA.0f \# \lambda x \in A.0 is logically equivalently to xA.f(x)0=1\exists x \in A.f(x) - 0 = 1 and thus xA.f(x)=1\exists x \in A.f(x) = 1. Since f#gf \# g is decidable, then xA.f(x)=1\exists x \in A.f(x) = 1 is decidable, which is precisely that AA is exhaustible.

The next two statements relate a set AA being exhaustible with axioms that the tight apartness relation on certain other function sets are a decidable tight apartness.

Theorem

More generally, given a finite set SS of cardinality n>1n \gt 1, a set AA is exhaustible if and only if that the function set S AS^A has decidable tight apartness, where the tight apartness relation on the function set S AS^A be defined as

f#gxA.f(x)g(x)f \# g \coloneqq \exists x \in A.f(x) \neq g(x)

Theorem

A set AA is exhaustible if and only if the function set A\mathbb{N}^A has decidable tight apartness, where the tight apartness relation on the function set \mathbb{N}^\mathbb{N} be defined as

f#gxA.f(x)g(x)f \# g \coloneqq \exists x \in A.f(x) \neq g(x)

Exhaustible subsingletons and excluded middle

There is a constant function x1x \mapsto 1 from every subsingleton AA to the boolean domain 𝟚\mathbb{2} taking every element xAx \in A to the boolean 1𝟚1 \in \mathbb{2}.

Theorem

Given a subsingleton AA, suppose that AA is exhaustible: either there exists xAx \in A such that (x1)(x)=1(x \mapsto 1)(x) = 1, or for all xAx \in A, (x1)(x)=0(x \mapsto 1)(x) = 0. Then AA is a decidable subsingleton.

Proof

We prove by case analysis.

Suppose that there exists x:Ax:A such that (x1)(x)=1(x \mapsto 1)(x) = 1. Then AA is a inhabited subsingleton and thus a decidable subsingleton.

Then suppose that for all x:Ax:A, (x1)(x)=0(x \mapsto 1)(x) = 0. Then AA is empty and thus a decidable subsingleton.

This exhausts all options for decidable subsingletons, and exhausts all possible conditions in the hypothesis.

Theorem

Suppose that for sets AA and BB with decidable tight apartness relations, the tight apartness relation on the function set B AB^A, defined by f#gx:A.f(x)g(x)f \# g \coloneqq \exists x:A.f(x) \neq g(x), is decidable. Then excluded middle holds.

Proof

Every subsingleton AA has a decidable tight apartness relation where x#yx \# y is always false. The boolean domain 𝟚\mathbb{2} also has a decidable tight apartness relation where x#yx \# y is given by the denial inequality xyx \neq y. We proved earlier in the article that a set AA being exhaustible is equivalent to the tight apartness relation on the function set 𝟚 A\mathbb{2}^A being decidable, and every subsingleton AA being exhaustible implies that every subsingleton AA is a decidable subsingleton, which is precisely the condition of excluded middle.

There is also a constant function x0x \mapsto 0 from every subsingleton AA to the boolean domain 𝟚\mathbb{2} taking every element xAx \in A to the boolean 0𝟚0 \in \mathbb{2}.

Theorem

Given a subsingleton AA, suppose that either there exists xAx \in A such that (x1)(x)(x0)(x)(x \mapsto 1)(x) \neq (x \mapsto 0)(x), or for all xAx \in A, (x1)(x)=(x0)(x)(x \mapsto 1)(x) = (x \mapsto 0)(x). Then AA is a decidable subsingleton.

Proof

We prove by case analysis.

Suppose that there exists xAx \in A such that (x1)(x)(x0)(x)(x \mapsto 1)(x) \neq (x \mapsto 0)(x). Then AA is a inhabited subsingleton and thus a decidable subsingleton.

Then suppose that for all xAx \in A, (x1)(x)=(x0)(x)(x \mapsto 1)(x) = (x \mapsto 0)(x). Then AA is empty and thus a decidable subsingleton, since (x1)=(x0)(x \mapsto 1) = (x \mapsto 0) by function extensionality.

This exhausts all options for decidable subsingletons, and exhausts all possible conditions in the hypothesis.

In dependent type theory

In the context of dependent type theory, a type AA being exhaustible or omniscient is expressed as

f:A𝟚(x:A.f(x)=1)( x:Af(x)=0)\prod_{f:A \to \mathbb{2}} (\exists x:A.f(x) = 1) \vee \left(\prod_{x:A} f(x) = 0\right)

Note that x:Af(x)=0\prod_{x:A} f(x) = 0 is always a proposition by weak function extensionality.

However, there are in general two ways of interpreting predicate logic, the traditional interpretation using propositional truncations, and the BHK interpretation which do not use propositional truncations. This results in four different interpretations of an exhaustible or omniscient type, depending on whether the existential quantifier is truncated or untruncated, and whether the disjunction is truncated or untruncated.

  • When the disjunction is truncated and the existential quantifier is truncated, this results in the usual concept of an exhaustible or omniscient type AA

  • When the disjunction is untruncated and the existential quantifier is truncated, by using the sum type, this results in the concept of a \exists-compact type AA (Escardó 18)

f:A𝟚(x:A.f(x)=1)+( x:Af(x)=0)\prod_{f:A \to \mathbb{2}} \left(\exists x:A.f(x) = 1\right) + \left(\prod_{x:A} f(x) = 0\right)
f:A𝟚(( x:Af(x)=1)+ x:Af(x)=0)\prod_{f:A \to \mathbb{2}} \left(\left(\sum_{x:A} f(x) = 1\right) + \prod_{x:A} f(x) = 0\right)
  • The case where the disjunction is truncated and the existential quantifier is untruncated remains undefined in the literature.
f:𝟚( x:Af(x)=1)( x:Af(x)=0)\prod_{f:\mathbb{N} \to \mathbb{2}} \left(\sum_{x:A} f(x) = 1\right) \vee \left(\prod_{x:A} f(x) = 0\right)

Note that Escardó 11 and Escardó 18 uses ¬ x:Af(x)=1\neg \sum_{x:A} f(x) = 1 instead of x:Af(x)=0\prod_{x:A} f(x) = 0. But one can show that ¬ x:Af(x)=1\neg \sum_{x:A} f(x) = 1 is logically equivalent to x:A¬f(x)=1\prod_{x:A} \neg f(x) = 1 in Martin-Lof type theory, see here for example, and ¬f(x)=1\neg f(x) = 1 is f(x)=0f(x) = 0. In addition, the two sources define a decidable type using a sum type A+¬AA + \neg A rather than disjunction A¬AA \vee \neg A.

References

Last revised on June 19, 2026 at 07:26:58. See the history of this page for a list of all contributions to it.