| Mathbox for BJ |
< Previous
Next >
Nearby theorems |
||
| Mirrors > Home > MPE Home > Th. List > Mathboxes > bj-axnul | Structured version Visualization version GIF version | ||
| Description: Over the base theory ax-1 6--
ax-5 1942, the axiom of separation implies
the weak emptyset axiom.
By "weak emptyset axiom", we mean the axiom asserting existence of an empty set (which can be called "the" empty set when the axiom of extensionality ax-ext 2732 is posited) provided existence of a set (the True truth constant existentially quantified over a fresh variable, extru 2007). This is the conclusion of bj-axnul 38124. The weak emptyset axiom implies ⊢ (∃𝑥⊤ → ∃𝑦⊤) without DV conditions hence also the same statement as the weak emptyset axiom without DV conditions on 𝑥, but only on 𝑦, 𝑧. By "axiom of separation", we mean the universal closure of ax-sep 5248, simulated here by its instance with ⊥ substituted for 𝜑 (and with the variable used to assert existence in the weak emptyset axiom substituted for the containing set) as the hypothesis of bj-axnul 38124. In particular, the axiom of existence extru 2007 and the axiom of separation together imply the emptyset axiom (and conversely, the emptyset axiom implies the axiom of existence). Note: this theorem does not require a disjointness condition on 𝑦, 𝑧, although both axioms should be stated with all variables disjoint. This proof only uses an instance of the axiom of separation with a bounded formula, so is valid in a constructive setting (see the CZF section in the "Intuitionistic Logic Explorer" iset.mm). (Contributed by BJ, 8-Mar-2026.) (Proof modification is discouraged.) |
| Ref | Expression |
|---|---|
| bj-axnul.axsep | ⊢ ∀𝑥∃𝑦∀𝑧(𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) |
| Ref | Expression |
|---|---|
| bj-axnul | ⊢ (∃𝑥⊤ → ∃𝑦∀𝑧 ∈ 𝑦 ⊥) |
| Step | Hyp | Ref | Expression |
|---|---|---|---|
| 1 | bj-bisimpr 37547 | . . . . . 6 ⊢ ((𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) → (𝑧 ∈ 𝑦 → ⊥)) | |
| 2 | 1 | alimi 1843 | . . . . 5 ⊢ (∀𝑧(𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) → ∀𝑧(𝑧 ∈ 𝑦 → ⊥)) |
| 3 | 2 | ralrid 3084 | . . . 4 ⊢ (∀𝑧(𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) → ∀𝑧 ∈ 𝑦 ⊥) |
| 4 | 3 | eximi 1867 | . . 3 ⊢ (∃𝑦∀𝑧(𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) → ∃𝑦∀𝑧 ∈ 𝑦 ⊥) |
| 5 | bj-axnul.axsep | . . 3 ⊢ ∀𝑥∃𝑦∀𝑧(𝑧 ∈ 𝑦 ↔ (𝑧 ∈ 𝑥 ∧ ⊥)) | |
| 6 | 4, 5 | bj-alimii 37611 | . 2 ⊢ ∀𝑥∃𝑦∀𝑧 ∈ 𝑦 ⊥ |
| 7 | bj-spvw 37658 | . 2 ⊢ (∃𝑥⊤ → (∃𝑦∀𝑧 ∈ 𝑦 ⊥ ↔ ∀𝑥∃𝑦∀𝑧 ∈ 𝑦 ⊥)) | |
| 8 | 6, 7 | mpbiri 261 | 1 ⊢ (∃𝑥⊤ → ∃𝑦∀𝑧 ∈ 𝑦 ⊥) |
| Colors of variables: wff setvar class |
| This proof depends on syntax axioms: → wi 4 ↔ wb 209 ∧ wa 401 ∀wal 1567 ⊤wtru 1570 ⊥wfal 1581 ∃wex 1811 ∀wral 3076 |
| This proof depends on axioms: ax-mp 5 ax-1 6 ax-2 7 ax-3 8 ax-gen 1827 ax-4 1841 ax-5 1942 |
| This proof depends on definitions: df-bi 210 df-an 402 df-ex 1812 df-ral 3077 |
| This theorem is used by: (None) |
| Copyright terms: Public domain | W3C validator |