P. 394 of Theorem List - Metamath Proof Explorer

Theorem List for Metamath Proof Explorer - 39301-39400   *Has distinct variable group(s)

Type	Label	Description
Statement
Theorem	eqvreldisj1 39301*	The elements of the quotient set of an equivalence relation are disjoint (cf. eqvreldisj2 39302, eqvreldisj3 39303). (Contributed by Mario Carneiro, 10-Dec-2016.) (Revised by Peter Mazsa, 3-Dec-2024.)
⊢ ( EqvRel 𝑅 → ∀𝑥 ∈ (𝐴 / 𝑅)∀𝑦 ∈ (𝐴 / 𝑅)(𝑥 = 𝑦 ∨ (𝑥 ∩ 𝑦) = ∅))
Theorem	eqvreldisj2 39302	The elements of the quotient set of an equivalence relation are disjoint (cf. eqvreldisj3 39303). (Contributed by Mario Carneiro, 10-Dec-2016.) (Revised by Peter Mazsa, 19-Sep-2021.)
⊢ ( EqvRel 𝑅 → ElDisj (𝐴 / 𝑅))
Theorem	eqvreldisj3 39303	The elements of the quotient set of an equivalence relation are disjoint (cf. qsdisj2 8739). (Contributed by Mario Carneiro, 10-Dec-2016.) (Revised by Peter Mazsa, 20-Jun-2019.) (Revised by Peter Mazsa, 19-Sep-2021.)
⊢ ( EqvRel 𝑅 → Disj (◡ E ↾ (𝐴 / 𝑅)))
Theorem	eqvreldisj4 39304	Intersection with the converse epsilon relation restricted to the quotient set of an equivalence relation is disjoint. (Contributed by Peter Mazsa, 30-May-2020.) (Revised by Peter Mazsa, 31-Dec-2021.)
⊢ ( EqvRel 𝑅 → Disj (𝑆 ∩ (◡ E ↾ (𝐵 / 𝑅))))
Theorem	eqvreldisj5 39305	Range Cartesian product with converse epsilon relation restricted to the quotient set of an equivalence relation is disjoint. (Contributed by Peter Mazsa, 30-May-2020.) (Revised by Peter Mazsa, 22-Sep-2021.)
⊢ ( EqvRel 𝑅 → Disj (𝑆 ⋉ (◡ E ↾ (𝐵 / 𝑅))))
Theorem	eqvrelqseqdisj2 39306	Implication of eqvreldisj2 39302, lemma for The Main Theorem of Equivalences mainer 39322. (Contributed by Peter Mazsa, 23-Sep-2021.)
⊢ (( EqvRel 𝑅 ∧ (𝐵 / 𝑅) = 𝐴) → ElDisj 𝐴)
Theorem	disjimeldisjdmqs 39307	Disj implies element-disjoint quotient carrier. Supplies the carrier-disjointness half of the Disjs pattern: under Disj 𝑅, the coset family is element-disjoint. (Contributed by Peter Mazsa, 5-Feb-2026.)
⊢ ( Disj 𝑅 → ElDisj (dom 𝑅 / 𝑅))
Theorem	eldisjsim1 39308	An element of the class of disjoint relations is disjoint. (Contributed by Peter Mazsa, 11-Feb-2026.)
⊢ (𝑅 ∈ Disjs → Disj 𝑅)
Theorem	eldisjsim2 39309	An element of the class of disjoint relations is an element of the class of relations. (Contributed by Peter Mazsa, 11-Feb-2026.)
⊢ (𝑅 ∈ Disjs → 𝑅 ∈ Rels )
Theorem	disjsssrels 39310	The class of disjoint relations is a subclass of the class of relations. (Contributed by Peter Mazsa, 11-Feb-2026.)
⊢ Disjs ⊆ Rels
Theorem	eldisjsim3 39311	Disjs implies element-disjoint quotient carrier. Exports the carrier-disjointness property in the ElDisjs packaging. (Contributed by Peter Mazsa, 11-Feb-2026.)
⊢ (𝑅 ∈ Disjs → (dom 𝑅 / 𝑅) ∈ ElDisjs )
Theorem	eldisjsim4 39312	Disjs implies element-disjoint range of QMap. Same as eldisjsim3 39311 but expressed using the block-map range ran QMap 𝑅 (often the more modular expression). (Contributed by Peter Mazsa, 15-Feb-2026.)
⊢ (𝑅 ∈ Disjs → ran QMap 𝑅 ∈ ElDisjs )
Theorem	eldisjsim5 39313	Disjs is closed under QMap. If a relation is "disjoint-structured" (Disjs), then its canonical block map is also "disjoint-structured". This is the second "structure level" in Disjs: it expresses that the property is stable under passing to the canonical block map, a theme that mirrors Pet-grade stability at a different axis. (Contributed by Peter Mazsa, 15-Feb-2026.)
⊢ (𝑅 ∈ Disjs → QMap 𝑅 ∈ Disjs )
Theorem	eldisjs6 39314	Elementhood in the class of disjoints. A relation 𝑅 is in Disjs iff: it is relation-typed, and its quotient-map QMap 𝑅 is itself disjoint, and its quotient-carrier ran QMap 𝑅 = (dom 𝑅 / 𝑅) lies in ElDisjs (element-disjoint carriers). This is the central "stability-by-decomposition" theorem for Disjs: it explains why Disjs is internally well-behaved without adding an external stability clause. It is the exact template that PetParts imitates: for pet 39339, the analogue of "map layer" is the disjointness of the lifted span, the analogue of "carrier layer" is the block-lift fixpoint (BlockLiftFix), and then adds external grade stability (SucMap ShiftStable) which Disjs does not need. (Contributed by Peter Mazsa, 16-Feb-2026.)
⊢ (𝑅 ∈ Disjs ↔ (𝑅 ∈ Rels ∧ (ran QMap 𝑅 ∈ ElDisjs ∧ QMap 𝑅 ∈ Disjs )))
Theorem	eldisjs7 39315*	Elementhood in the class of disjoints. 𝑅 ∈ Disjs iff: 𝑅 ∈ Rels, and every 𝑥 belongs to at most one block 𝑢 in the quotient-carrier (dom 𝑅 / 𝑅) (element-disjointness at the carrier), and every block 𝑢 in the quotient-carrier has a unique representative 𝑡 ∈ dom 𝑅 such that 𝑢 = [𝑡]𝑅. Provides the "fully expanded" quantifier characterization of the same decomposition as eldisjs6 39314, but without explicitly mentioning QMap. This is the "E*/E!"" view that is closest in spirit to suc11reg 9538-style injectivity and to the "unique generator per block" narrative. It is also the right contrast-point to older one-line criteria like dfdisjs4 39170 (the "u R x" style), because it makes the carrier and representation discipline explicit and type-safe. (Contributed by Peter Mazsa, 16-Feb-2026.)
⊢ (𝑅 ∈ Disjs ↔ (𝑅 ∈ Rels ∧ (∀𝑥∃*𝑢 ∈ (dom 𝑅 / 𝑅)𝑥 ∈ 𝑢 ∧ ∀𝑢 ∈ (dom 𝑅 / 𝑅)∃!𝑡 ∈ dom 𝑅 𝑢 = [𝑡]𝑅)))
Theorem	dfdisjs6 39316	Alternate definition of the class of disjoints (via quotient-map stability). Disjs is the class of relations 𝑟 whose quotient-map QMap 𝑟 is again disjoint and whose induced quotient-carrier is element-disjoint. This is the definitional "stability-by-decomposition" packaging of disjointness: it builds Disjs from two internal layers (i) a carrier-layer constraint and (ii) a map-layer closure constraint. This is deliberately different from "u R x" style definitions: it makes the carrier of blocks and the uniqueness-of-representatives discipline first-class and reusable (via QMap) rather than implicit. (Contributed by Peter Mazsa, 16-Feb-2026.)
⊢ Disjs = {𝑟 ∈ Rels ∣ (ran QMap 𝑟 ∈ ElDisjs ∧ QMap 𝑟 ∈ Disjs )}
Theorem	dfdisjs7 39317*	Alternate definition of the class of disjoints (via carrier disjointness + unique representatives). Ideology-free normal form of dfdisjs6 39316: "blocks cover their elements" (∃*) and "each block has a unique generator" (∃!), expressed entirely at the quotient-carrier level. Same class as dfdisjs6 39316, but presented in fully expanded ∃* / ∃! form over the quotient-carrier (dom 𝑟 / 𝑟). Makes explicit (a) element-disjointness of the quotient-carrier and (b) unique representative existence for each block. These are exactly the two conditions that rule out type-confusions (blocks vs witnesses) and ensure canonical decomposition. This is the form that best supports analogy arguments with df-petparts 39342 and with successor-style uniqueness patterns. (Contributed by Peter Mazsa, 16-Feb-2026.)
⊢ Disjs = {𝑟 ∈ Rels ∣ (∀𝑥∃*𝑢 ∈ (dom 𝑟 / 𝑟)𝑥 ∈ 𝑢 ∧ ∀𝑢 ∈ (dom 𝑟 / 𝑟)∃!𝑡 ∈ dom 𝑟 𝑢 = [𝑡]𝑟)}
Theorem	fences3 39318	Implication of eqvrelqseqdisj2 39306 and n0eldmqseq 39108, see comment of fences 39332. (Contributed by Peter Mazsa, 30-Dec-2024.)
⊢ (( EqvRel 𝑅 ∧ (dom 𝑅 / 𝑅) = 𝐴) → ( ElDisj 𝐴 ∧ ¬ ∅ ∈ 𝐴))
Theorem	eqvrelqseqdisj3 39319	Implication of eqvreldisj3 39303, lemma for the Member Partition Equivalence Theorem mpet3 39324. (Contributed by Peter Mazsa, 27-Oct-2020.) (Revised by Peter Mazsa, 24-Sep-2021.)
⊢ (( EqvRel 𝑅 ∧ (𝐵 / 𝑅) = 𝐴) → Disj (◡ E ↾ 𝐴))
Theorem	eqvrelqseqdisj4 39320	Lemma for petincnvepres2 39336. (Contributed by Peter Mazsa, 31-Dec-2021.)
⊢ (( EqvRel 𝑅 ∧ (𝐵 / 𝑅) = 𝐴) → Disj (𝑆 ∩ (◡ E ↾ 𝐴)))
Theorem	eqvrelqseqdisj5 39321	Lemma for the Partition-Equivalence Theorem pet2 39338. (Contributed by Peter Mazsa, 15-Jul-2020.) (Revised by Peter Mazsa, 22-Sep-2021.)
⊢ (( EqvRel 𝑅 ∧ (𝐵 / 𝑅) = 𝐴) → Disj (𝑆 ⋉ (◡ E ↾ 𝐴)))
Theorem	mainer 39322	The Main Theorem of Equivalences: every equivalence relation implies equivalent comembers. (Contributed by Peter Mazsa, 26-Sep-2021.)
⊢ (𝑅 ErALTV 𝐴 → CoMembEr 𝐴)
Theorem	partimcomember 39323	Partition with general 𝑅 (in addition to the member partition cf. mpet 39327 and mpet2 39328) implies equivalent comembers. (Contributed by Peter Mazsa, 23-Sep-2021.) (Revised by Peter Mazsa, 22-Dec-2024.)
⊢ (𝑅 Part 𝐴 → CoMembEr 𝐴)
Theorem	mpet3 39324	Member Partition-Equivalence Theorem. Together with mpet 39327 mpet2 39328, mostly in its conventional cpet 39326 and cpet2 39325 form, this is what we used to think of as the partition equivalence theorem (but cf. pet2 39338 with general 𝑅). (Contributed by Peter Mazsa, 4-May-2018.) (Revised by Peter Mazsa, 26-Sep-2021.)
⊢ (( ElDisj 𝐴 ∧ ¬ ∅ ∈ 𝐴) ↔ ( CoElEqvRel 𝐴 ∧ (∪ 𝐴 / ∼ 𝐴) = 𝐴))
Theorem	cpet2 39325	The conventional form of the Member Partition-Equivalence Theorem. In the conventional case there is no (general) disjoint and no (general) partition concept: mathematicians have called disjoint or partition what we call element disjoint or member partition, see also cpet 39326. Together with cpet 39326, mpet 39327 mpet2 39328, this is what we used to think of as the partition equivalence theorem (but cf. pet2 39338 with general 𝑅). (Contributed by Peter Mazsa, 30-Dec-2024.)
⊢ (( ElDisj 𝐴 ∧ ¬ ∅ ∈ 𝐴) ↔ ( EqvRel ∼ 𝐴 ∧ (∪ 𝐴 / ∼ 𝐴) = 𝐴))
Theorem	cpet 39326	The conventional form of Member Partition-Equivalence Theorem. In the conventional case there is no (general) disjoint and no (general) partition concept: mathematicians have been calling disjoint or partition what we call element disjoint or member partition, see also cpet2 39325. Cf. mpet 39327, mpet2 39328 and mpet3 39324 for unconventional forms of Member Partition-Equivalence Theorem. Cf. pet 39339 and pet2 39338 for Partition-Equivalence Theorem with general 𝑅. (Contributed by Peter Mazsa, 31-Dec-2024.)
⊢ ( MembPart 𝐴 ↔ ( EqvRel ∼ 𝐴 ∧ (∪ 𝐴 / ∼ 𝐴) = 𝐴))
Theorem	mpet 39327	Member Partition-Equivalence Theorem in almost its shortest possible form, cf. the 0-ary version mpets 39330. Member partition and comember equivalence relation are the same (or: each element of 𝐴 have equivalent comembers if and only if 𝐴 is a member partition). Together with mpet2 39328, mpet3 39324, and with the conventional cpet 39326 and cpet2 39325, this is what we used to think of as the partition equivalence theorem (but cf. pet2 39338 with general 𝑅). (Contributed by Peter Mazsa, 24-Sep-2021.)
⊢ ( MembPart 𝐴 ↔ CoMembEr 𝐴)
Theorem	mpet2 39328	Member Partition-Equivalence Theorem in a shorter form. Together with mpet 39327 mpet3 39324, mostly in its conventional cpet 39326 and cpet2 39325 form, this is what we used to think of as the partition equivalence theorem (but cf. pet2 39338 with general 𝑅). (Contributed by Peter Mazsa, 24-Sep-2021.)
⊢ ((◡ E ↾ 𝐴) Part 𝐴 ↔ ≀ (◡ E ↾ 𝐴) ErALTV 𝐴)
Theorem	mpets2 39329	Member Partition-Equivalence Theorem with binary relations, cf. mpet2 39328. (Contributed by Peter Mazsa, 24-Sep-2021.)
⊢ (𝐴 ∈ 𝑉 → ((◡ E ↾ 𝐴) Parts 𝐴 ↔ ≀ (◡ E ↾ 𝐴) Ers 𝐴))
Theorem	mpets 39330	Member Partition-Equivalence Theorem in its shortest possible form: it shows that member partitions and comember equivalence relations are literally the same. Cf. pet 39339, the Partition-Equivalence Theorem, with general 𝑅. (Contributed by Peter Mazsa, 31-Dec-2024.)
⊢ MembParts = CoMembErs
Theorem	mainpart 39331	Partition with general 𝑅 also imply member partition. (Contributed by Peter Mazsa, 23-Sep-2021.) (Revised by Peter Mazsa, 22-Dec-2024.)
⊢ (𝑅 Part 𝐴 → MembPart 𝐴)
Theorem	fences 39332	The Theorem of Fences by Equivalences: all conceivable equivalence relations (besides the comember equivalence relation cf. mpet 39327) generate a partition of the members. (Contributed by Peter Mazsa, 26-Sep-2021.)
⊢ (𝑅 ErALTV 𝐴 → MembPart 𝐴)
Theorem	fences2 39333	The Theorem of Fences by Equivalences: all conceivable equivalence relations (besides the comember equivalence relation cf. mpet3 39324) generate a partition of the members, it alo means that (𝑅 ErALTV 𝐴 → ElDisj 𝐴) and that (𝑅 ErALTV 𝐴 → ¬ ∅ ∈ 𝐴). (Contributed by Peter Mazsa, 15-Oct-2021.)
⊢ (𝑅 ErALTV 𝐴 → ( ElDisj 𝐴 ∧ ¬ ∅ ∈ 𝐴))
Theorem	mainer2 39334	The Main Theorem of Equivalences: every equivalence relation implies equivalent comembers. (Contributed by Peter Mazsa, 15-Oct-2021.)
⊢ (𝑅 ErALTV 𝐴 → ( CoElEqvRel 𝐴 ∧ ¬ ∅ ∈ 𝐴))
Theorem	mainerim 39335	Every equivalence relation implies equivalent coelements. (Contributed by Peter Mazsa, 20-Oct-2021.)
⊢ (𝑅 ErALTV 𝐴 → CoElEqvRel 𝐴)
Theorem	petincnvepres2 39336	A partition-equivalence theorem with intersection and general 𝑅. (Contributed by Peter Mazsa, 31-Dec-2021.)
⊢ (( Disj (𝑅 ∩ (◡ E ↾ 𝐴)) ∧ (dom (𝑅 ∩ (◡ E ↾ 𝐴)) / (𝑅 ∩ (◡ E ↾ 𝐴))) = 𝐴) ↔ ( EqvRel ≀ (𝑅 ∩ (◡ E ↾ 𝐴)) ∧ (dom ≀ (𝑅 ∩ (◡ E ↾ 𝐴)) / ≀ (𝑅 ∩ (◡ E ↾ 𝐴))) = 𝐴))
Theorem	petincnvepres 39337	The shortest form of a partition-equivalence theorem with intersection and general 𝑅. Cf. br1cossincnvepres 38914. Cf. pet 39339. (Contributed by Peter Mazsa, 23-Sep-2021.)
⊢ ((𝑅 ∩ (◡ E ↾ 𝐴)) Part 𝐴 ↔ ≀ (𝑅 ∩ (◡ E ↾ 𝐴)) ErALTV 𝐴)
Theorem	pet2 39338	Partition-Equivalence Theorem, with general 𝑅. This theorem (together with pet 39339 and pets 39340) is the main result of my investigation into set theory, see the comment of pet 39339. (Contributed by Peter Mazsa, 24-May-2021.) (Revised by Peter Mazsa, 23-Sep-2021.)
⊢ (( Disj (𝑅 ⋉ (◡ E ↾ 𝐴)) ∧ (dom (𝑅 ⋉ (◡ E ↾ 𝐴)) / (𝑅 ⋉ (◡ E ↾ 𝐴))) = 𝐴) ↔ ( EqvRel ≀ (𝑅 ⋉ (◡ E ↾ 𝐴)) ∧ (dom ≀ (𝑅 ⋉ (◡ E ↾ 𝐴)) / ≀ (𝑅 ⋉ (◡ E ↾ 𝐴))) = 𝐴))
Theorem	pet 39339	Partition-Equivalence Theorem with general 𝑅 while preserving the restricted converse epsilon relation of mpet2 39328 (as opposed to petincnvepres 39337). A class is a partition by a range Cartesian product with general 𝑅 and the restricted converse element class if and only if the cosets by the range Cartesian product are in an equivalence relation on it. Cf. br1cossxrncnvepres 38916. This theorem (together with pets 39340 and pet2 39338) is the main result of my investigation into set theory. It is no more general than the conventional Member Partition-Equivalence Theorem mpet 39327, mpet2 39328 and mpet3 39324 (because you cannot set 𝑅 in this theorem in such a way that you get mpet2 39328), i.e., it is not the hypothetical General Partition-Equivalence Theorem gpet ⊢ (𝑅 Part 𝐴 ↔ ≀ 𝑅 ErALTV 𝐴), but this one has a general part that mpet2 39328 lacks: 𝑅, which is sufficient for my future application of set theory, for my purpose outside of set theory. (Contributed by Peter Mazsa, 23-Sep-2021.)
⊢ ((𝑅 ⋉ (◡ E ↾ 𝐴)) Part 𝐴 ↔ ≀ (𝑅 ⋉ (◡ E ↾ 𝐴)) ErALTV 𝐴)
Theorem	pets 39340	Partition-Equivalence Theorem with general 𝑅, with binary relations. This theorem (together with pet 39339 and pet2 39338) is the main result of my investigation into set theory, cf. the comment of pet 39339. (Contributed by Peter Mazsa, 23-Sep-2021.)
⊢ ((𝐴 ∈ 𝑉 ∧ 𝑅 ∈ 𝑊) → ((𝑅 ⋉ (◡ E ↾ 𝐴)) Parts 𝐴 ↔ ≀ (𝑅 ⋉ (◡ E ↾ 𝐴)) Ers 𝐴))
Theorem	dmqsblocks 39341*	If the pet 39339 span (𝑅 ⋉ (◡ E ↾ 𝐴)) partitions 𝐴, then every block 𝑢 ∈ 𝐴 is of the form [𝑣] for some 𝑣 that not only lies in the domain but also has at least one internal element 𝑐 and at least one 𝑅-target 𝑏 (cf. also the comments of qseq 39107). It makes explicit that pet 39339 gives active representatives for each block, without ever forcing 𝑣 = 𝑢. (Contributed by Peter Mazsa, 23-Nov-2025.)
⊢ ((dom (𝑅 ⋉ (◡ E ↾ 𝐴)) / (𝑅 ⋉ (◡ E ↾ 𝐴))) = 𝐴 → ∀𝑢 ∈ 𝐴 ∃𝑣 ∈ dom (𝑅 ⋉ (◡ E ↾ 𝐴))∃𝑏∃𝑐(𝑢 = [𝑣](𝑅 ⋉ (◡ E ↾ 𝐴)) ∧ 𝑐 ∈ 𝑣 ∧ 𝑣𝑅𝑏))
21.26.23  Type-safe Partition-Equivalence: PetParts, PetErs, Pet2Parts, Pet2Ers
Definition	df-petparts 39342*	Define the class of partition-side general partition-equivalence spans. ⟨𝑟, 𝑛⟩ ∈ PetParts means: (1) 𝑟 is a set-relation (𝑟 ∈ Rels), and (2) 𝑛 is a membership block-carrier (𝑛 ∈ MembParts), and (3) the block-lift span (𝑟 ⋉ (◡ E ↾ 𝑛)) is a generalized partition on its natural quotient-carrier 𝑛 (i.e. (𝑟 ⋉ (◡ E ↾ 𝑛)) Parts 𝑛). This is the horizontal feasibility base object on the partition side, expressed in the type-safe Parts language. The explicit typing (𝑟 ∈ Rels ∧ 𝑛 ∈ MembParts ) is included at the definition level so later modular refinements can treat typedness as a first-class component (e.g. intersecting a typedness module with disjointness and equilibrium modules) without repeatedly restating it. In particular, it lets decompositions such as dfpetparts2 39346 be written as clean intersections whose first conjunct is exactly the typedness module ( Rels × MembParts ). (Contributed by Peter Mazsa, 19-Feb-2026.) (Revised by Peter Mazsa, 25-Feb-2026.)
⊢ PetParts = {⟨𝑟, 𝑛⟩ ∣ ((𝑟 ∈ Rels ∧ 𝑛 ∈ MembParts ) ∧ (𝑟 ⋉ (◡ E ↾ 𝑛)) Parts 𝑛)}
Definition	df-peters 39343*	Define the class of equivalence-side general partition-equivalence spans. ⟨𝑟, 𝑛⟩ ∈ PetErs means: (1) 𝑟 is a set-relation (𝑟 ∈ Rels), and (2) 𝑛 is a carrier recognized on the equivalence side of membership (𝑛 ∈ CoMembErs), and (3) the coset relation of the lifted span, ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)), is an equivalence relation on its natural quotient with carrier 𝑛 (i.e. ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)) Ers 𝑛). This packages the equivalence-view of the same lifted construction that underlies PetParts. It is designed to be parallel to PetParts so later proofs can freely choose the partition side (Parts) or the equivalence side (Ers) without rebuilding the bridge each time; the identification is provided by petseq 39350 (using typesafepets 39349 and mpets 39330). The explicit typing (𝑟 ∈ Rels ∧ 𝑛 ∈ CoMembErs ) is included for the same reason as in df-petparts 39342: to make typedness a reusable module. (Contributed by Peter Mazsa, 19-Feb-2026.) (Revised by Peter Mazsa, 25-Feb-2026.)
⊢ PetErs = {⟨𝑟, 𝑛⟩ ∣ ((𝑟 ∈ Rels ∧ 𝑛 ∈ CoMembErs ) ∧ ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)) Ers 𝑛)}
Definition	df-pet2parts 39344	Define the class of grade- and blocklift-stable partition-side general partition-equivalence spans. It consists of those ⟨𝑟, 𝑛⟩ ∈ PetParts such that ⟨𝑟, 𝑛⟩ remains in PetParts after shifting one grade along SucMap (via ShiftStable). Concretely: ⟨𝑟, 𝑛⟩ ∈ PetParts and there exists a predecessor 𝑚 with suc 𝑚 = 𝑛 such that ⟨𝑟, 𝑚⟩ ∈ PetParts (encoded by SucMap ∘ PetParts inside ShiftStable). I.e., it introduces the external (tower/grade) stability axis. This is the "4th level" for pet 39339 (see dfpet2parts2 39347): beyond (i) carrier membership partition, (ii) disjointness, and (iii) semantic equilibrium, we require (iv) stability under a canonical grade shift. PetParts already enforces disjointness and the quotient-carrier equation for the lifted span (hence semantic equilibrium via dfpetparts2 39346). Pet2Parts adds the external grade (tower) stability axis via df-shiftstable 38856 with SucMap. This (iv) is why we need explicit second-level Pet2Parts, while Disjs typically does not: Disjs already packages its own internal two-step consistency (carrier + map) by dfdisjs6 39316 / dfdisjs7 39317, whereas pet 39339 has an additional grade axis that must be imposed separately. (Contributed by Peter Mazsa, 19-Feb-2026.)
⊢ Pet2Parts = ( SucMap ShiftStable PetParts )
Definition	df-pet2ers 39345	Define the class of grade- and blocklift-stable equivalence-side general partition-equivalence spans. The equivalence-side analogue of Pet2Parts: stability of PetErs under one-step grade shift along SucMap. Ensures that the equivalence-side formulation supports the same tower/grade infrastructure as the partition-side formulation. SucMap ShiftStable is the grade axis and does not change the equivalence-vs-partition viewpoint (reinforced by pets2eq 39351). (Contributed by Peter Mazsa, 19-Feb-2026.)
⊢ Pet2Ers = ( SucMap ShiftStable PetErs )
Theorem	dfpetparts2 39346*	Alternate definition of PetParts as typedness + disjoint-span + block-lift equilibrium. This theorem is the key modularization step. It decomposes PetParts into the intersection of three orthogonal modules: (T) typedness: ⟨𝑟, 𝑛⟩ ∈ ( Rels × MembParts ), (D) disjoint-span: (𝑟 ⋉ (◡ E ↾ 𝑛)) ∈ Disjs, (E) semantic equilibrium: ⟨𝑟, 𝑛⟩ ∈ BlockLiftFix, i.e. the carrier 𝑛 is a fixpoint of the induced block-generation operator. Conceptually, (D) provides the disjointness/quotient discipline for the lifted span, while (E) prevents hidden carrier drift (refinement or coarsening of what counts as a block) by enforcing the fixpoint equation. The point of this theorem is that these constraints can be imposed and reused independently by later constructions, while their intersection recovers the intended Parts-based notion. This mirrors the internal packaging of Disjs (see dfdisjs6 39316 / dfdisjs7 39317): for disjoint relations, the "map layer + carrier layer" decomposition is internal via QMap and ElDisjs; for PetParts, the carrier 𝑛 is an external parameter, so the additional carrier stability must be factored explicitly as BlockLiftFix. (Contributed by Peter Mazsa, 20-Feb-2026.) (Revised by Peter Mazsa, 25-Feb-2026.)
⊢ PetParts = ((( Rels × MembParts ) ∩ {⟨𝑟, 𝑛⟩ ∣ (𝑟 ⋉ (◡ E ↾ 𝑛)) ∈ Disjs }) ∩ BlockLiftFix )
Theorem	dfpet2parts2 39347*	Grade stability applied to the decomposed PetParts modules. Pet2Parts is obtained by applying the grade-stability operator SucMap ShiftStable (see df-shiftstable 38856) to the modular intersection from dfpetparts2 39346. This makes the two orthogonal stability axes explicit: (E) semantic stability / equilibrium: BlockLiftFix, (G) grade stability: SucMap ShiftStable, assembled on top of typedness and disjoint-span base modules. This is the principled "extra level" that does not arise for Disjs: disjoint relations already bundle their internal map/carrier consistency via QMap and ElDisjs (see dfdisjs6 39316 / dfdisjs7 39317), while the present construction has an additional external grading axis imposed by the canonical successor map SucMap. (Contributed by Peter Mazsa, 20-Feb-2026.) (Revised by Peter Mazsa, 25-Feb-2026.)
⊢ Pet2Parts = ( SucMap ShiftStable ((( Rels × MembParts ) ∩ {⟨𝑟, 𝑛⟩ ∣ (𝑟 ⋉ (◡ E ↾ 𝑛)) ∈ Disjs }) ∩ BlockLiftFix ))
Theorem	dfpeters2 39348*	Alternate definition of PetErs in fully modular form. This expands the Ers 𝑛 predicate into: (i) a typedness module ( Rels × CoMembErs ), (ii) an equivalence module for the coset relation ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)) ∈ EqvRels, (iii) the corresponding quotient-carrier (domain quotient) equation dom ≀ (...) / ≀ (...) = 𝑛. This is the equivalence-side counterpart of the modular decomposition dfpetparts2 39346 on the partition side. (Contributed by Peter Mazsa, 25-Feb-2026.)
⊢ PetErs = ((( Rels × CoMembErs ) ∩ {⟨𝑟, 𝑛⟩ ∣ ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)) ∈ EqvRels }) ∩ {⟨𝑟, 𝑛⟩ ∣ (dom ≀ (𝑟 ⋉ (◡ E ↾ 𝑛)) / ≀ (𝑟 ⋉ (◡ E ↾ 𝑛))) = 𝑛})
Theorem	typesafepets 39349	Type-safe pets 39340 scheme. On a membership block-carrier 𝐴 ∈ MembParts, the lifted span (𝑅 ⋉ (◡ E ↾ 𝐴)) yields a generalized partition of 𝐴 iff its coset relation yields an equivalence relation on the same carrier 𝐴. This is the type-safe replacement for the earlier broad pets 39340: it explicitly restricts to carriers where 𝐴 is already known to be a block-family (by MembParts). That removes the standard type-safety objection ("are you equating a quotient-carrier of blocks with raw witnesses?") by construction. It is the key bridge used to identify the partition-side and equivalence-side pet classes (petseq 39350), in complete parallel with the membership bridge mpets 39330. This theorem is intentionally not the definition of PetParts; it is the bridge used by petseq 39350 after typedness is enforced by the "Pet*" definitions. (Contributed by Peter Mazsa, 19-Feb-2026.)
⊢ ((𝐴 ∈ MembParts ∧ 𝑅 ∈ 𝑉) → ((𝑅 ⋉ (◡ E ↾ 𝐴)) Parts 𝐴 ↔ ≀ (𝑅 ⋉ (◡ E ↾ 𝐴)) Ers 𝐴))
Theorem	petseq 39350	Generalized partition-equivalence identification. The partition-side scheme PetParts and the equivalence-side scheme PetErs define the same class of spans (pairs ⟨𝑟, 𝑛⟩). This plays the same organizational role for lifted spans that mpets 39330 plays for carriers: mpets 39330 identifies MembParts with CoMembErs at the membership-carrier level, while petseq 39350 identifies the corresponding span-level predicates built from Parts and Ers. Unlike the earlier broad pets 39340, the bridge used here is the type-safe span theorem typesafepets 39349, which restricts to membership block-carriers. Since typedness (𝑟 ∈ Rels and the appropriate carrier condition) is now built directly into PetParts and PetErs, this theorem can be used downstream without repeatedly re-establishing basic typing premises. (Contributed by Peter Mazsa, 19-Feb-2026.)
⊢ PetParts = PetErs
Theorem	pets2eq 39351	Grade-stable generalized partition-equivalence identification. After applying the same grade-stability operator (SucMap ShiftStable) to both sides, the grade-stable pet classes still coincide. Confirms that the grade/tower infrastructure is orthogonal to the partition-vs-equivalence viewpoint: stability is preserved under the PetParts = PetErs identification. This is the level at which we can freely work on whichever side is more convenient (Parts for block discipline, Ers for equivalence reasoning), without changing the stable notion of "pet". (Contributed by Peter Mazsa, 19-Feb-2026.)
⊢ Pet2Parts = Pet2Ers
21.27  Mathbox for Rodolfo Medina
21.27.1  Partitions
Theorem	prtlem60 39352	Lemma for prter3 39381. (Contributed by Rodolfo Medina, 9-Oct-2010.)
⊢ (𝜑 → (𝜓 → (𝜒 → 𝜃)))    &   ⊢ (𝜓 → (𝜃 → 𝜏))    ⇒   ⊢ (𝜑 → (𝜓 → (𝜒 → 𝜏)))
Theorem	bicomdd 39353	Commute two sides of a biconditional in a deduction. (Contributed by Rodolfo Medina, 19-Oct-2010.) (Proof shortened by Andrew Salmon, 29-Jun-2011.)
⊢ (𝜑 → (𝜓 → (𝜒 ↔ 𝜃)))    ⇒   ⊢ (𝜑 → (𝜓 → (𝜃 ↔ 𝜒)))
Theorem	jca2r 39354	Inference conjoining the consequents of two implications. (Contributed by Rodolfo Medina, 17-Oct-2010.)
⊢ (𝜑 → (𝜓 → 𝜒))    &   ⊢ (𝜓 → 𝜃)    ⇒   ⊢ (𝜑 → (𝜓 → (𝜃 ∧ 𝜒)))
Theorem	jca3 39355	Inference conjoining the consequents of two implications. (Contributed by Rodolfo Medina, 14-Oct-2010.)
⊢ (𝜑 → (𝜓 → 𝜒))    &   ⊢ (𝜃 → 𝜏)    ⇒   ⊢ (𝜑 → (𝜓 → (𝜃 → (𝜒 ∧ 𝜏))))
Theorem	prtlem70 39356	Lemma for prter3 39381: a rearrangement of conjuncts. (Contributed by Rodolfo Medina, 20-Oct-2010.)
⊢ ((((𝜓 ∧ 𝜂) ∧ ((𝜑 ∧ 𝜃) ∧ (𝜒 ∧ 𝜏))) ∧ 𝜑) ↔ ((𝜑 ∧ (𝜓 ∧ (𝜒 ∧ (𝜃 ∧ 𝜏)))) ∧ 𝜂))
Theorem	ibdr 39357	Reverse of ibd 270. (Contributed by Rodolfo Medina, 30-Sep-2010.)
⊢ (𝜑 → (𝜒 → (𝜓 ↔ 𝜒)))    ⇒   ⊢ (𝜑 → (𝜒 → 𝜓))
Theorem	prtlem100 39358	Lemma for prter3 39381. (Contributed by Rodolfo Medina, 19-Oct-2010.)
⊢ (∃𝑥 ∈ 𝐴 (𝐵 ∈ 𝑥 ∧ 𝜑) ↔ ∃𝑥 ∈ (𝐴 ∖ {∅})(𝐵 ∈ 𝑥 ∧ 𝜑))
Theorem	prtlem5 39359*	Lemma for prter1 39378, prter2 39380, prter3 39381 and prtex 39379. (Contributed by Rodolfo Medina, 25-Sep-2010.) (Proof shortened by Mario Carneiro, 11-Dec-2016.)
⊢ ([𝑠 / 𝑣][𝑟 / 𝑢]∃𝑥 ∈ 𝐴 (𝑢 ∈ 𝑥 ∧ 𝑣 ∈ 𝑥) ↔ ∃𝑥 ∈ 𝐴 (𝑟 ∈ 𝑥 ∧ 𝑠 ∈ 𝑥))
Theorem	prtlem80 39360	Lemma for prter2 39380. (Contributed by Rodolfo Medina, 17-Oct-2010.)
⊢ (𝐴 ∈ 𝐵 → ¬ 𝐴 ∈ (𝐶 ∖ {𝐴}))
Theorem	brabsb2 39361*	A closed form of brabsb 5480. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ (𝑅 = {⟨𝑥, 𝑦⟩ ∣ 𝜑} → (𝑧𝑅𝑤 ↔ [𝑧 / 𝑥][𝑤 / 𝑦]𝜑))
Theorem	eqbrrdv2 39362*	Other version of eqbrrdiv 5744. (Contributed by Rodolfo Medina, 30-Sep-2010.)
⊢ (((Rel 𝐴 ∧ Rel 𝐵) ∧ 𝜑) → (𝑥𝐴𝑦 ↔ 𝑥𝐵𝑦))    ⇒   ⊢ (((Rel 𝐴 ∧ Rel 𝐵) ∧ 𝜑) → 𝐴 = 𝐵)
Theorem	prtlem9 39363*	Lemma for prter3 39381. (Contributed by Rodolfo Medina, 25-Sep-2010.)
⊢ (𝐴 ∈ 𝐵 → ∃𝑥 ∈ 𝐵 [𝑥] ∼ = [𝐴] ∼ )
Theorem	prtlem10 39364*	Lemma for prter3 39381. (Contributed by Rodolfo Medina, 14-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ( ∼ Er 𝐴 → (𝑧 ∈ 𝐴 → (𝑧 ∼ 𝑤 ↔ ∃𝑣 ∈ 𝐴 (𝑧 ∈ [𝑣] ∼ ∧ 𝑤 ∈ [𝑣] ∼ ))))
Theorem	prtlem11 39365	Lemma for prter2 39380. (Contributed by Rodolfo Medina, 12-Oct-2010.)
⊢ (𝐵 ∈ 𝐷 → (𝐶 ∈ 𝐴 → (𝐵 = [𝐶] ∼ → 𝐵 ∈ (𝐴 / ∼ ))))
Theorem	prtlem12 39366*	Lemma for prtex 39379 and prter3 39381. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ ( ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)} → Rel ∼ )
Theorem	prtlem13 39367*	Lemma for prter1 39378, prter2 39380, prter3 39381 and prtex 39379. (Contributed by Rodolfo Medina, 13-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (𝑧 ∼ 𝑤 ↔ ∃𝑣 ∈ 𝐴 (𝑧 ∈ 𝑣 ∧ 𝑤 ∈ 𝑣))
Theorem	prtlem16 39368*	Lemma for prtex 39379, prter2 39380 and prter3 39381. (Contributed by Rodolfo Medina, 14-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ dom ∼ = ∪ 𝐴
Theorem	prtlem400 39369*	Lemma for prter2 39380 and also a property of partitions . (Contributed by Rodolfo Medina, 15-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ ¬ ∅ ∈ (∪ 𝐴 / ∼ )
Syntax	wprt 39370	Extend the definition of a wff to include the partition predicate.
wff Prt 𝐴
Definition	df-prt 39371*	Define the partition predicate. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ (Prt 𝐴 ↔ ∀𝑥 ∈ 𝐴 ∀𝑦 ∈ 𝐴 (𝑥 = 𝑦 ∨ (𝑥 ∩ 𝑦) = ∅))
Theorem	erprt 39372	The quotient set of an equivalence relation is a partition. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ ( ∼ Er 𝑋 → Prt (𝐴 / ∼ ))
Theorem	prtlem14 39373*	Lemma for prter1 39378, prter2 39380 and prtex 39379. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ (Prt 𝐴 → ((𝑥 ∈ 𝐴 ∧ 𝑦 ∈ 𝐴) → ((𝑤 ∈ 𝑥 ∧ 𝑤 ∈ 𝑦) → 𝑥 = 𝑦)))
Theorem	prtlem15 39374*	Lemma for prter1 39378 and prtex 39379. (Contributed by Rodolfo Medina, 13-Oct-2010.)
⊢ (Prt 𝐴 → (∃𝑥 ∈ 𝐴 ∃𝑦 ∈ 𝐴 ((𝑢 ∈ 𝑥 ∧ 𝑤 ∈ 𝑥) ∧ (𝑤 ∈ 𝑦 ∧ 𝑣 ∈ 𝑦)) → ∃𝑧 ∈ 𝐴 (𝑢 ∈ 𝑧 ∧ 𝑣 ∈ 𝑧)))
Theorem	prtlem17 39375*	Lemma for prter2 39380. (Contributed by Rodolfo Medina, 15-Oct-2010.)
⊢ (Prt 𝐴 → ((𝑥 ∈ 𝐴 ∧ 𝑧 ∈ 𝑥) → (∃𝑦 ∈ 𝐴 (𝑧 ∈ 𝑦 ∧ 𝑤 ∈ 𝑦) → 𝑤 ∈ 𝑥)))
Theorem	prtlem18 39376*	Lemma for prter2 39380. (Contributed by Rodolfo Medina, 15-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (Prt 𝐴 → ((𝑣 ∈ 𝐴 ∧ 𝑧 ∈ 𝑣) → (𝑤 ∈ 𝑣 ↔ 𝑧 ∼ 𝑤)))
Theorem	prtlem19 39377*	Lemma for prter2 39380. (Contributed by Rodolfo Medina, 15-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (Prt 𝐴 → ((𝑣 ∈ 𝐴 ∧ 𝑧 ∈ 𝑣) → 𝑣 = [𝑧] ∼ ))
Theorem	prter1 39378*	Every partition generates an equivalence relation. (Contributed by Rodolfo Medina, 13-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (Prt 𝐴 → ∼ Er ∪ 𝐴)
Theorem	prtex 39379*	The equivalence relation generated by a partition is a set if and only if the partition itself is a set. (Contributed by Rodolfo Medina, 15-Oct-2010.) (Revised by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (Prt 𝐴 → ( ∼ ∈ V ↔ 𝐴 ∈ V))
Theorem	prter2 39380*	The quotient set of the equivalence relation generated by a partition equals the partition itself. (Contributed by Rodolfo Medina, 17-Oct-2010.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ (Prt 𝐴 → (∪ 𝐴 / ∼ ) = (𝐴 ∖ {∅}))
Theorem	prter3 39381*	For every partition there exists a unique equivalence relation whose quotient set equals the partition. (Contributed by Rodolfo Medina, 19-Oct-2010.) (Proof shortened by Mario Carneiro, 12-Aug-2015.)
⊢ ∼ = {⟨𝑥, 𝑦⟩ ∣ ∃𝑢 ∈ 𝐴 (𝑥 ∈ 𝑢 ∧ 𝑦 ∈ 𝑢)}    ⇒   ⊢ ((𝑆 Er ∪ 𝐴 ∧ (∪ 𝐴 / 𝑆) = (𝐴 ∖ {∅})) → ∼ = 𝑆)
21.28  Mathbox for Norm Megill We are sad to report the passing of Metamath creator and long-time contributor Norm Megill (1950 - 2021). Norm of course was the author of the Metamath proof language, the specification, all of the early tools (and some of the later ones), and the foundational work in logic and set theory for set.mm. His tools, now at https://github.com/metamath/metamath-exe, include a proof verifier, a proof assistant, a proof minimizer, style checking and reformatting, and tools for searching and displaying proofs. One of his key insights was that formal proofs can exist not only to be verified by computers, but also to be read by humans. Both the specification of the proof format (which stores full proofs, as opposed to the proof templates used by most proof assistants) and the generated web display of Metamath proofs, one of its distinctive features, contribute to this double objective. Metamath innovated both by using a very simple substitution rule (and then using that to build more complicated notions like free and bound variables) and also by taking the axiom schemas found in many theories and taking them to the next level - by making all axioms, theorems and proofs operate in terms of schemas. Not content to create Metamath for his own amusement, he also published it for the world and encouraged the development of a community of people who contributed to it and created their own tools. He was an active participant in the Metamath mailing list and other forums until days before his passing. It is often our custom to supply a quote from someone memorialized in a mathbox entry. And it is difficult to select a quote for someone who has written so much about Metamath over the years. But here is one quote from the Metamath web page which illustrates not just his clear thinking about what Metamath can and cannot do but also his desire to encourage students at all levels: Q: Will Metamath help me learn abstract mathematics? A: Yes, but probably not by itself. In order to follow a proof in an advanced math textbook, you may need to know prerequisites that could take years to learn. Some people find this frustrating. In contrast, Metamath uses a single, simple substitution rule that allows you to follow any proof mechanically. You can actually jump in anywhere and be convinced that the symbol string you see in a proof step is a consequence of the symbol strings in the earlier steps that it references, even if you don't understand what the symbols mean. But this is quite different from understanding the meaning of the math that results. Metamath alone probably will not give you an intuitive feel for abstract math, in the same way it can be hard to grasp a large computer program just by reading its source code, even though you may understand each individual instruction. However, the Bibliographic Cross-Reference lets you compare informal proofs in math textbooks and see all the implicit missing details "left to the reader."
21.28.1  Obsolete schemes ax-c4,c5,c7,c10,c11,c11n,c15,c9,c14,c16 These older axiom schemes are obsolete and should not be used outside of this section. They are proved above as theorems axc4 , sp 2195, axc7 2326, axc10 2393, axc11 2438, axc11n 2434, axc15 2430, axc9 2390, axc14 2471, and axc16 2273.
Axiom	ax-c5 39382	Axiom of Specialization. A universally quantified wff implies the wff without the universal quantifier (i.e., an instance, or special case, of the generalized wff). In other words, if something is true for all 𝑥, then it is true for any specific 𝑥 (that would typically occur as a free variable in the wff substituted for 𝜑). (A free variable is one that does not occur in the scope of a quantifier: 𝑥 and 𝑦 are both free in 𝑥 = 𝑦, but only 𝑥 is free in ∀𝑦𝑥 = 𝑦.) Axiom scheme C5' in [Megill] p. 448 (p. 16 of the preprint). Also appears as Axiom B5 of [Tarski] p. 67 (under his system S2, defined in the last paragraph on p. 77). Note that the converse of this axiom does not hold in general, but a weaker inference form of the converse holds and is expressed as rule ax-gen 1802. Conditional forms of the converse are given by ax-13 2380, ax-c14 39390, ax-c16 39391, and ax-5 1917. Unlike the more general textbook Axiom of Specialization, we cannot choose a variable different from 𝑥 for the special case. In our axiomatization, that requires the assistance of equality axioms, and we deal with it later after we introduce the definition of proper substitution (see stdpc4 2079). An interesting alternate axiomatization uses axc5c711 39417 and ax-c4 39383 in place of ax-c5 39382, ax-4 1816, ax-10 2152, and ax-11 2168. This axiom is obsolete and should no longer be used. It is proved above as Theorem sp 2195. (Contributed by NM, 3-Jan-1993.) Use sp 2195 instead. (New usage is discouraged.)
⊢ (∀𝑥𝜑 → 𝜑)
Axiom	ax-c4 39383	Axiom of Quantified Implication. This axiom moves a universal quantifier from outside to inside an implication, quantifying 𝜓. Notice that 𝑥 must not be a free variable in the antecedent of the quantified implication, and we express this by binding 𝜑 to "protect" the axiom from a 𝜑 containing a free 𝑥. Axiom scheme C4' in [Megill] p. 448 (p. 16 of the preprint). It is a special case of Lemma 5 of [Monk2] p. 108 and Axiom 5 of [Mendelson] p. 69. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc4 2330. (Contributed by NM, 3-Jan-1993.) (New usage is discouraged.)
⊢ (∀𝑥(∀𝑥𝜑 → 𝜓) → (∀𝑥𝜑 → ∀𝑥𝜓))
Axiom	ax-c7 39384	Axiom of Quantified Negation. This axiom is used to manipulate negated quantifiers. Equivalent to axiom scheme C7' in [Megill] p. 448 (p. 16 of the preprint). An alternate axiomatization could use axc5c711 39417 in place of ax-c5 39382, ax-c7 39384, and ax-11 2168. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc7 2326. (Contributed by NM, 10-Jan-1993.) (New usage is discouraged.)
⊢ (¬ ∀𝑥 ¬ ∀𝑥𝜑 → 𝜑)
Axiom	ax-c10 39385	A variant of ax6 2392. Axiom scheme C10' in [Megill] p. 448 (p. 16 of the preprint). This axiom is obsolete and should no longer be used. It is proved above as Theorem axc10 2393. (Contributed by NM, 10-Jan-1993.) (New usage is discouraged.)
⊢ (∀𝑥(𝑥 = 𝑦 → ∀𝑥𝜑) → 𝜑)
Axiom	ax-c11 39386	Axiom ax-c11 39386 was the original version of ax-c11n 39387 ("n" for "new"), before it was discovered (in May 2008) that the shorter ax-c11n 39387 could replace it. It appears as Axiom scheme C11' in [Megill] p. 448 (p. 16 of the preprint). This axiom is obsolete and should no longer be used. It is proved above as Theorem axc11 2438. (Contributed by NM, 10-May-1993.) (New usage is discouraged.)
⊢ (∀𝑥 𝑥 = 𝑦 → (∀𝑥𝜑 → ∀𝑦𝜑))
Axiom	ax-c11n 39387	Axiom of Quantifier Substitution. One of the equality and substitution axioms of predicate calculus with equality. Appears as Lemma L12 in [Megill] p. 445 (p. 12 of the preprint). The original version of this axiom was ax-c11 39386 and was replaced with this shorter ax-c11n 39387 ("n" for "new") in May 2008. The old axiom is proved from this one as Theorem axc11 2438. Conversely, this axiom is proved from ax-c11 39386 as Theorem axc11nfromc11 39425. This axiom was proved redundant in July 2015. See Theorem axc11n 2434. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc11n 2434. (Contributed by NM, 16-May-2008.) (New usage is discouraged.)
⊢ (∀𝑥 𝑥 = 𝑦 → ∀𝑦 𝑦 = 𝑥)
Axiom	ax-c15 39388	Axiom ax-c15 39388 was the original version of ax-12 2189, before it was discovered (in Jan. 2007) that the shorter ax-12 2189 could replace it. It appears as Axiom scheme C15' in [Megill] p. 448 (p. 16 of the preprint). It is based on Lemma 16 of [Tarski] p. 70 and Axiom C8 of [Monk2] p. 105, from which it can be proved by cases. To understand this theorem more easily, think of "¬ ∀𝑥𝑥 = 𝑦 →..." as informally meaning "if 𝑥 and 𝑦 are distinct variables then..." The antecedent becomes false if the same variable is substituted for 𝑥 and 𝑦, ensuring the theorem is sound whenever this is the case. In some later theorems, we call an antecedent of the form ¬ ∀𝑥𝑥 = 𝑦 a "distinctor". Interestingly, if the wff expression substituted for 𝜑 contains no wff variables, the resulting statement can be proved without invoking this axiom. This means that even though this axiom is metalogically independent from the others, it is not logically independent. Specifically, we can prove any wff-variable-free instance of Axiom ax-c15 39388 (from which the ax-12 2189 instance follows by Theorem ax12 2431.) The proof is by induction on formula length, using ax12eq 39440 and ax12el 39441 for the basis steps and ax12indn 39442, ax12indi 39443, and ax12inda 39447 for the induction steps. (This paragraph is true provided we use ax-c11 39386 in place of ax-c11n 39387.) This axiom is obsolete and should no longer be used. It is proved above as Theorem axc15 2430, which should be used instead. (Contributed by NM, 14-May-1993.) (New usage is discouraged.)
⊢ (¬ ∀𝑥 𝑥 = 𝑦 → (𝑥 = 𝑦 → (𝜑 → ∀𝑥(𝑥 = 𝑦 → 𝜑))))
Axiom	ax-c9 39389	Axiom of Quantifier Introduction. One of the equality and substitution axioms of predicate calculus with equality. Informally, it says that whenever 𝑧 is distinct from 𝑥 and 𝑦, and 𝑥 = 𝑦 is true, then 𝑥 = 𝑦 quantified with 𝑧 is also true. In other words, 𝑧 is irrelevant to the truth of 𝑥 = 𝑦. Axiom scheme C9' in [Megill] p. 448 (p. 16 of the preprint). It apparently does not otherwise appear in the literature but is easily proved from textbook predicate calculus by cases. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc9 2390. (Contributed by NM, 10-Jan-1993.) (New usage is discouraged.)
⊢ (¬ ∀𝑧 𝑧 = 𝑥 → (¬ ∀𝑧 𝑧 = 𝑦 → (𝑥 = 𝑦 → ∀𝑧 𝑥 = 𝑦)))
Axiom	ax-c14 39390	Axiom of Quantifier Introduction. One of the equality and substitution axioms for a non-logical predicate in our predicate calculus with equality. Axiom scheme C14' in [Megill] p. 448 (p. 16 of the preprint). It is redundant if we include ax-5 1917; see Theorem axc14 2471. Alternately, ax-5 1917 becomes unnecessary in principle with this axiom, but we lose the more powerful metalogic afforded by ax-5 1917. We retain ax-c14 39390 here to provide completeness for systems with the simpler metalogic that results from omitting ax-5 1917, which might be easier to study for some theoretical purposes. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc14 2471. (Contributed by NM, 24-Jun-1993.) (New usage is discouraged.)
⊢ (¬ ∀𝑧 𝑧 = 𝑥 → (¬ ∀𝑧 𝑧 = 𝑦 → (𝑥 ∈ 𝑦 → ∀𝑧 𝑥 ∈ 𝑦)))
Axiom	ax-c16 39391*	Axiom of Distinct Variables. The only axiom of predicate calculus requiring that variables be distinct (if we consider ax-5 1917 to be a metatheorem and not an axiom). Axiom scheme C16' in [Megill] p. 448 (p. 16 of the preprint). It apparently does not otherwise appear in the literature but is easily proved from textbook predicate calculus by cases. It is a somewhat bizarre axiom since the antecedent is always false in set theory (see dtru 5383), but nonetheless it is technically necessary as you can see from its uses. This axiom is redundant if we include ax-5 1917; see Theorem axc16 2273. Alternately, ax-5 1917 becomes logically redundant in the presence of this axiom, but without ax-5 1917 we lose the more powerful metalogic that results from being able to express the concept of a setvar variable not occurring in a wff (as opposed to just two setvar variables being distinct). We retain ax-c16 39391 here to provide logical completeness for systems with the simpler metalogic that results from omitting ax-5 1917, which might be easier to study for some theoretical purposes. This axiom is obsolete and should no longer be used. It is proved above as Theorem axc16 2273. (Contributed by NM, 10-Jan-1993.) (New usage is discouraged.)
⊢ (∀𝑥 𝑥 = 𝑦 → (𝜑 → ∀𝑥𝜑))
21.28.2  Rederive new axioms ax-4, ax-10, ax-6, ax-12, ax-13 from old Theorems ax12fromc15 39404 and ax13fromc9 39405 require some intermediate theorems that are included in this section.
Theorem	axc5 39392	This theorem repeats sp 2195 under the name axc5 39392, so that the Metamath program "MM> VERIFY MARKUP" command will check that it matches axiom scheme ax-c5 39382. (Contributed by NM, 18-Aug-2017.) (Proof modification is discouraged.) Use sp 2195 instead. (New usage is discouraged.)
⊢ (∀𝑥𝜑 → 𝜑)
Theorem	ax4fromc4 39393	Rederivation of Axiom ax-4 1816 from ax-c4 39383, ax-c5 39382, ax-gen 1802 and minimal implicational calculus { ax-mp 5, ax-1 6, ax-2 7 }. See axc4 2330 for the derivation of ax-c4 39383 from ax-4 1816. (Contributed by NM, 23-May-2008.) (Proof modification is discouraged.) Use ax-4 1816 instead. (New usage is discouraged.)
⊢ (∀𝑥(𝜑 → 𝜓) → (∀𝑥𝜑 → ∀𝑥𝜓))
Theorem	ax10fromc7 39394	Rederivation of Axiom ax-10 2152 from ax-c7 39384, ax-c4 39383, ax-c5 39382, ax-gen 1802 and propositional calculus. See axc7 2326 for the derivation of ax-c7 39384 from ax-10 2152. (Contributed by NM, 23-May-2008.) (Proof modification is discouraged.) Use ax-10 2152 instead. (New usage is discouraged.)
⊢ (¬ ∀𝑥𝜑 → ∀𝑥 ¬ ∀𝑥𝜑)
Theorem	ax6fromc10 39395	Rederivation of Axiom ax-6 1974 from ax-c7 39384, ax-c10 39385, ax-gen 1802 and propositional calculus. See axc10 2393 for the derivation of ax-c10 39385 from ax-6 1974. Lemma L18 in [Megill] p. 446 (p. 14 of the preprint). (Contributed by NM, 14-May-1993.) (Proof modification is discouraged.) Use ax-6 1974 instead. (New usage is discouraged.)
⊢ ¬ ∀𝑥 ¬ 𝑥 = 𝑦
Theorem	hba1-o 39396	The setvar 𝑥 is not free in ∀𝑥𝜑. Example in Appendix in [Megill] p. 450 (p. 19 of the preprint). Also Lemma 22 of [Monk2] p. 114. (Contributed by NM, 24-Jan-1993.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ (∀𝑥𝜑 → ∀𝑥∀𝑥𝜑)
Theorem	axc4i-o 39397	Inference version of ax-c4 39383. (Contributed by NM, 3-Jan-1993.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ (∀𝑥𝜑 → 𝜓)    ⇒   ⊢ (∀𝑥𝜑 → ∀𝑥𝜓)
Theorem	equid1 39398	Proof of equid 2019 from our older axioms. This is often an axiom of equality in textbook systems, but we don't need it as an axiom since it can be proved from our other axioms (although the proof, as you can see below, is not as obvious as you might think). This proof uses only axioms without distinct variable conditions and requires no dummy variables. A simpler proof, similar to Tarski's, is possible if we make use of ax-5 1917; see the proof of equid 2019. See equid1ALT 39424 for an alternate proof. (Contributed by NM, 10-Jan-1993.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ 𝑥 = 𝑥
Theorem	equcomi1 39399	Proof of equcomi 2024 from equid1 39398, avoiding use of ax-5 1917 (the only use of ax-5 1917 is via ax7 2023, so using ax-7 2015 instead would remove dependency on ax-5 1917). (Contributed by BJ, 8-Jul-2021.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ (𝑥 = 𝑦 → 𝑦 = 𝑥)
Theorem	aecom-o 39400	Commutation law for identical variable specifiers. The antecedent and consequent are true when 𝑥 and 𝑦 are substituted with the same variable. Lemma L12 in [Megill] p. 445 (p. 12 of the preprint). Version of aecom 2435 using ax-c11 39386. Unlike axc11nfromc11 39425, this version does not require ax-5 1917 (see comment of equcomi1 39399). (Contributed by NM, 10-May-1993.) (Proof modification is discouraged.) (New usage is discouraged.)
⊢ (∀𝑥 𝑥 = 𝑦 → ∀𝑦 𝑦 = 𝑥)