How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
Garside Structure, Normal Forms, and the Center — Examples
1 · Prerequisites
- Binary Operations, Monoids, Groups and Subgroups
- Braided and Symmetric Monoidal Categories
- Conjugacy in Sₙ, Generation, and the Simplicity of Aₙ
- Construction of the Natural Numbers
- Construction of the Real Numbers via Cauchy Sequences
- Construction of the Real Numbers via Dedekind Cuts
- Cosets, Index and Lagrange's Theorem
- Countability and Uncountability
- Finite Counting, Factorials and Binomial Coefficients
- Free Groups and Presentations
- Garside Structure, Normal Forms, and the Center
- Group Actions, Orbits, Stabilisers and Cayley's Theorem
- Group Homomorphisms and the Isomorphism Theorems
- Normal Subgroups and Quotient Groups
- Relations, Functions, and Quotients
- Symmetric Groups, Cycle Decomposition and the Sign Homomorphism
- The ZFC Axioms and the Basic Set Constructions
2 · Summary
These four worked entries make the Garside structure of the companion page concrete, all of them inside , where the half twist is of length and the braid relation also gives . The first example lists the six simple braids and computes their relevant meets and joins in the positive divisibility lattice. There are exactly simple braids, namely , and their images under the permutation homomorphism are the six distinct elements , which both proves the listing and exhibits the bijection of the theory. The two atoms are incomparable on the left with join their braid word, and . For and the left meet is , computed from the divisors of an atom and of a two-letter element, while the right meet is : the two divisibility orders are therefore already different on the four-element subfamily , and the balancedness of — its left and right divisor sets coincide — does not collapse them into one order.
The second example computes a left Garside normal form in full. For the atom complement gives the positive description , and the scaling identity for meets yields , so and the normal form is with , and two proper simple factors , , of permutations and . The pair is left weighted, , and the negative exponent correctly reports that is not a positive braid.
The third example computes the full twist. The braid relation gives , that is, the square of the half twist is the class of a positive word of length ; applying the sliding identity twice shows that each atom commutes with , hence so does every element of and is infinite cyclic by the center theorem for . The half twist itself is not central only because and the atoms are distinct: the passage from to its square in the center is genuine, not an artifact of commuting generators.
The fourth entry is the counterexample. The exponent sum , , is a well-defined homomorphism: every three-term braid relator and every far-commutation relator has equal exponent sum on both sides — and the permutation homomorphism is likewise well defined, with , so in . Yet , and both elements have the same coarse normal-form data: left normal form with exponent and a single proper simple factor. The exponent sum is therefore not a complete braid normal form, and not even the triple determines a braid; what separates the two witnesses is the factor itself, against , i.e. the permutation data that the left Garside normal form retains. Nothing in these four entries uses a choice principle.
3 · Logical flowchart
4 · Definitions, theorems and proofs
None yet.
5 · Examples, counterexamples and false statements
The simple braids and divisibility lattice for b three
Example
Take , so that has the two atoms and the half twist of length . The example verifies:
- the six simple braids of are , with the six distinct endpoint permutations ;
- while ;
- for and one has but ;
- consequently the left and the right divisibility orders on are different, even though by Simple positive braids are indexed by permutations the two orders have the same divisor set on .
Facts & Assumptions
Given: The natural number , the monoid with atoms , half twist of length , and the elements , .
is generated by the atoms, is additive and only for ; the orders are and for some , with length monotone, so a proper divisor of an atom or of or letters has strictly smaller length (Positive braid monoid, Positive artin relations preserve homogeneous length, Left and right divisibility for positive braids).
by the braid relation, and the simple braids — the left divisors of — are in bijection with : they are exactly , , , , , (Simple positive braids are indexed by permutations, The Garside half twist and simple positive braids).
The homomorphism sends to and products to products; , , , , in the composition convention of The symmetric group : the bijections of a set under composition (Reduced adjacent-transposition words have well-defined positive lifts).
For adjacent atoms, is the least common left multiple and also the least common right multiple, and exists and is unique; more generally every nonempty finite subset of has unique left and right gcds and lcms (Artin atoms have explicit left and right lcms and complements, Positive braids have left and right gcds and lcms).
Verification
The six simple braids and their permutations. By [F2] there are exactly simple braids. Applying [F3] to the six displayed elements gives the images , , , , and in this order; these are the six elements of , so they are pairwise distinct, and since is a function the six braids are pairwise distinct and exhaust the left divisors of . In particular and are proper simple braids, because and by [F2].
The atom pair . The left divisors of are and , since a divisor of has , so either or is an atom equal to ; the same holds for , and the distinct atoms are incomparable, so the common left divisors of and are just : thus . Their least common left multiple is by [F4], which is by [F2], so .
A pair with different left and right meets. For and : left-divides trivially and left-divides , so is a common left divisor; every common left divisor of and divides , hence is or , so : therefore . On the right: the right divisors of are and by [F1], while the right divisors of are — indeed if then ; for one has , so and ; for the element is an atom, and then has length and is an atom as well, so must be a product of two atoms, the four candidates being , which are distinct because every defining relation has length three; only occurs, so the only one-letter right divisor is . Hence the common right divisors of and are , and .
Conclusion. Steps 1.1--1.3 give the six simple braids, , , and the pair with . So already on the four-element subfamily the two orders have different meets, even though on itself the left and right divisor sets coincide by [F2]; the balancedness established in Delta is the lcm of the artin atoms and has the same left and right divisors is therefore a property of and does not identify the two orders. No choice principle is used. ∎
Remarks
- Reading the six permutations. In the convention of The symmetric group : the bijections of a set under composition the product acts with the right factor first, so maps to and to the inverse cycle : the order of the two words is visible in the orientation of the -cycle, and this is what separates the two length-two simple braids.
- Why the right meet is the smaller one here. The right divisors of are its suffixes , while those of are ; the overlap is trivial even though the overlap of the corresponding prefix sets is . This is the smallest instance of the asymmetry between the two orders.
- No choice principle is used; all computations are finite word computations.
A left garside normal form computation in b three
Example
Take , so that has the two generators , the positive monoid has the two atoms , and the half twist is
of length . Let . The example computes the left Garside normal form of of Left garside normal form is unique and checks:
- , hence with ;
- , so in particular ;
- the left normal form of is that is , and the two factors , are proper simple braids with , ;
- the factor pair is left weighted: ;
- consequently , in agreement with .
Facts & Assumptions
Given: The natural number ; the braid group of The braid group by Artin presentation; its positive monoid with atoms and length ; the half twist of length ; and the element .
is generated by the atoms, is additive, forces , and the braid relation gives , so both triangular words for agree: (Positive braid monoid, Positive artin relations preserve homogeneous length, The Garside half twist and simple positive braids).
means for some , and then ; a left divisor of an atom has length or , so the left divisors of are among and those of among (Left and right divisibility for positive braids, Positive artin relations preserve homogeneous length).
is a monoid homomorphism with ; in the composition convention of The symmetric group : the bijections of a set under composition the adjacent transpositions and are distinct, so (Reduced adjacent-transposition words have well-defined positive lifts, The symmetric group : the bijections of a set under composition).
is a submonoid of , and for one has if and only if (The group of fractions of the positive braid monoid is the Artin braid group, Left and right divisibility extend to lattice orders on the braid group).
Scaling identity. For one has , where the meet is the positive left-gcd (Left and right divisibility extend to lattice orders on the braid group).
Every nonempty finite family in has a unique left-gcd and a unique left-lcm; in particular exists for every , and every common left divisor of and left-divides (Positive braids have left and right gcds and lcms).
Normal form. Every has a unique expression with , , every a proper simple braid and ; in such an expression is the largest with , the product equals , and is the unique pair with , and . Moreover if and only if , and if then for (Left garside normal form is unique).
Verification
The half twist and the rewriting of . By [F1], has length . In the group , in which sits as a submonoid by [F4], multiplying out gives , so ; substituting this identity into the given element gives , and because it is a product of atoms.
The atoms are coprime. Let be a common left divisor of and . By [F2] write with ; additivity of [F1] gives . If then [F1]; if then , so and . Since also left-divides , write with a possibly different suffix . Length gives , hence and , contradicting [F3]. So the only common left divisor of the two atoms is , and by [F6] their left-gcd is .
The meet with . Apply the scaling identity [F5] to , and ; by [F1] the two products are and , so , the third equality by step 1.2. Hence : if were a common left divisor of and , then by [F6] it would satisfy , and monotonicity of [F2] would force , a contradiction.
The maximal -exponent is . The pair satisfies by step 1.1, by step 1.1 and by step 2.1; by the uniqueness in [F7] of the pair it coincides with , hence and .
The permutation images and left weighting. Since is a homomorphism with [F3], the two factors have images and , which are distinct; in particular , so the two-factor form is not a repetition of one simple braid. Left weighting is the instance of the last assertion of [F7]: , which is literally the computation of step 2.1.
The two greedy factors. Put and , so that by step 3.1. Then by step 2.1, and the unique with is , since [F1]; further because [F1], hence [F6], and the unique with is . So the recursion of [F7] produces with factors . Both are proper simple braids: exhibits and exhibits [F1], while and because and lie in and not in [F1].
The normal form conditions hold. For , step 4.1 gives ; for , step 2.1 gives . So with both proper simple and for ; by the uniqueness in [F7] this is the left normal form of , and [F7] identifies its data as and , in agreement with step 3.1.
Conclusion. The element of has left Garside normal form with and , with proper simple factors and whose images in are and , and with the factor pair left weighted by step 3.2. Moreover , because while characterises positivity by [F7]; this is the qualitative content of the computation, since is visibly written with an inverse letter. No choice principle is used, the only selections being the explicit words displayed above. ∎
Remarks
- Reading off the algorithm. In the notation of Left garside normal form is unique the recursion runs , , , , ; the factor is the maximal simple prefix of because does not divide on the left, and the remainder is already simple.
- Why the -exponent is negative. The computation above gives ; by Left garside normal form is unique (d)(ii), this proves is not positive. The calculation exhibits the witness pair of that theorem's uniqueness clause rather than merely asserting it. Note that has the same length as ; length alone therefore decides neither left divisibility nor the meet, and it is the scaling computation of step 2.1 that shows .
- Conventions. All products are read left to right as words in the generators, and the permutation images are those of the positive monoid map ; the cycle notation is that of The symmetric group : the bijections of a set under composition, so . No choice principle is used: every object in the computation is an explicitly displayed finite word.
The full twist in b three
Example
Take , so that has the generators , the positive monoid has the two atoms , and the half twist is of length . The example verifies:
- , the full twist of The center of b n is generated by the full twist for n greater than two;
- for , hence commutes with every element of , i.e. ;
- together with the centre theorem for this gives , and that centre is infinite cyclic: the full twist is a free generator;
- the half twist itself is not central: while , so and the passage from to its square is genuine.
Facts & Assumptions
Given: The natural number ; the braid group with generators of The braid group by Artin presentation; its positive monoid with atoms , length and half twist of length .
is generated by the atoms, is additive, forces , the braid relation gives , and has length (Positive braid monoid, Positive artin relations preserve homogeneous length, The Garside half twist and simple positive braids).
is a submonoid of and the elements of are multiplied in as in the monoid, so is an invertible element of the group and group cancellation is available (The group of fractions of the positive braid monoid is the Artin braid group).
In the presentation of [F1], every element of is a finite product of the letters (The braid group by Artin presentation).
The centre. For one has , and this group is infinite cyclic: and is an isomorphism (The center of b n is generated by the full twist for n greater than two).
is a monoid homomorphism with , and in the composition convention of The symmetric group : the bijections of a set under composition the adjacent transpositions and are distinct, so (Reduced adjacent-transposition words have well-defined positive lifts, The symmetric group : the bijections of a set under composition).
Verification
The half twist and its two words. By [F1], has length and the braid relation gives the second triangular word ; by [F3] the monoid sits inside the group , so may be inverted and cancelled there.
Both generators commute with . Applying the identity of [F2] twice, , and likewise .
The cube of is the square of the half twist. Using associativity and the two words for from step 1.1, .
Centrality of the full twist. By [F4] every is a finite product of the letters . Step 1.2 gives for , and this passes to inverses: from one gets by multiplying the identity on the left and on the right by . Induction on the number of letters therefore gives for every , that is .
The centre is infinite cyclic on the full twist. By [F5] with , , the element is not , and is an isomorphism ; combining this with step 2.1, the centre is infinite cyclic and generated by the full twist , consistently with the local centrality verification of step 2.2.
The half twist alone is not central. By [F2] with one has . If were central then also , so and right cancellation by the invertible element in [F3] gives , contradicting [F6]. Hence , although by step 2.2: the full twist is the square of the half twist and is not the half twist itself.
Conclusion. In the full twist satisfies by step 2.1, it is central by step 2.2, it generates the centre by step 3.1, and the half twist is not central by step 3.2. The four assertions of the Example section are therefore verified. No choice principle is used: the identities are finite word computations and the only inverse taken is the explicit inverse of the displayed word . ∎
Remarks
- Why the name. Geometrically the half twist is the half turn of the strands, whose square is the full turn; the star of the example is the algebraic counterpart: is the full twist of , and The center of b n is generated by the full twist for n greater than two identifies it as the generator of the centre. The words and have exponent sums and respectively, matching their homogeneous lengths and .
- The local and the global verification. Step 2.2 checks centrality directly from the two generator identities, using only that is generated by and ; the appeal to the centre theorem in step 3.1 is then used only to conclude that no other central elements exist. This matches the source's proof of the general theorem, where the same generator identities supply centrality of before the divisibility argument bounds the centre.
- The asymmetry of the half twist. Centrality of and failure of centrality of are both visible in the index reversal : conjugation by permutes the two generators instead of fixing them, whereas its square fixes both. No choice principle is used anywhere.
Exponent sum is not a complete braid normal form
Statement refuted
Refuted claim. The exponent sum , defined on words by , is a complete braid normal form: for all , if then .
The witness is the pair : both braids have exponent sum , and they are distinct. The refutation is robust — the two witnesses agree not only in but also in the -exponent and in the number of simple factors of their left Garside normal forms — so no invariant of the coarse Garside data distinguishes them either. In the left normal form of the theorem of Garside--Elrifai--Morton they differ only in which simple factor occurs, against , that is, in their permutations against .
What is and is not claimed. Nothing here is a claim against the usefulness of the exponent sum: for it is a surjective homomorphism onto (for it is the zero homomorphism on the trivial group), it separates from , and for it also separates from ; being abelian it is insensitive to the braid relations. What is refuted is only completeness. Nor is it claimed that braid groups have no complete normal forms: the left Garside normal form is one, and it is precisely the additional simple-factor data (the first factor versus ) that separates the two witnesses below.
Facts & Assumptions
Given: The braid group , that is, the case of The braid group by Artin presentation; its positive monoid with the two atoms , the length and the half twist of length ; the symmetric group with adjacent transpositions ; and the two elements .
is a presented group in the sense of Group presentation by generators and relations, where consists of the single braid relation (The braid group by Artin presentation). If is a function on the generators whose evaluation sends every relator to the identity, then extends uniquely to a homomorphism of (Von Dyck's theorem: maps of generators that satisfy the relators extend uniquely from a presented group, Group presentation by generators and relations).
In the adjacent transpositions satisfy , for , and ; in the composition convention of The symmetric group : the bijections of a set under composition the transpositions and are distinct (Reduced adjacent-transposition words have well-defined positive lifts, The symmetric group : the bijections of a set under composition).
is generated by the atoms , is additive with only for , and the half twist of satisfies (Positive braid monoid, Positive artin relations preserve homogeneous length, The Garside half twist and simple positive braids).
means with , and then ; in particular every left divisor of an atom has length or (Left and right divisibility for positive braids, Positive artin relations preserve homogeneous length).
is a submonoid of (The group of fractions of the positive braid monoid is the Artin braid group).
Every has a unique left normal form with , , every a proper simple braid and ; in it is the largest integer with , the product is , and exactly when (Left garside normal form is unique).
Counterexample
The exponent sum is a well-defined homomorphism. Define on the generators of [F1]. The single relator evaluates to . By von Dyck [F1] there is therefore a unique homomorphism with for ; it satisfies and .
The permutation homomorphism exists. Define . By the type-A relation of [F2], sends the three-term relator to ; the further type-A relations and the far commutation are properties of and impose no condition, because the corresponding words are not relators of [F1]. By von Dyck [F1] there is a homomorphism with .
The two atoms are distinct. By step 1.2, and , and these are distinct by [F2]. If held in , then applying the function to the two sides would give , a contradiction; hence .
Their exponent sums agree. By step 1.1, , and since by step 2.1, the assignment is not injective: two distinct braids have equal exponent sum.
Both witnesses are one-factor left normal forms with . Fix . Since [F3], [F6] gives ; also , because a left divisor of the atom has by [F4] while by [F3], so and . For every , , so would imply by transitivity, which was just ruled out. Hence is the largest integer with , that is and . Further for and for by [F3], so , and the greedy first factor of [F6] is , with remainder computed in the group , in which is a submonoid by [F5]; the factor is proper because and by [F3], and it is simple because . So the left normal form of each witness is with simple factor, and the two witnesses also agree in the data and .
Conclusion. The braids and are distinct by step 2.1, while by step 3.1; moreover both have left normal form with one proper simple factor by step 3.2. The exponent sum therefore fails to be a complete normal form, and even the triple fails to determine a braid. The refuted claim is false, already in . No choice principle is used: both homomorphisms are obtained from the explicit generator assignments and , and every computation is a finite word computation. ∎
Remarks
- Why the failure is minimal. The two witnesses are the two distinct generators of the smallest braid group that has more than one of them, ; on one generator there is no such pair to exhibit. The failure is not a defect of the presentation used to define : the homomorphism is well defined by von Dyck's theorem, as step 1.1 checks.
- What the exponent sum does see. It is additive, it takes the value on each inverse letter, it is surjective for because , and for it takes the value on the half twist and on the full twist . Being abelian data, it cannot see the non-abelian structure of the braid group, and the pair is the smallest instance of that insensitivity.
- The extra data that does separate them. The two braids have the same -exponent and the same number of simple factors but different first (and only) simple factors, against , corresponding to the distinct permutations and . This is exactly the refinement supplied by the left Garside normal form of Left garside normal form is unique.
- No choice principle is used; only the explicitly displayed generator assignments and finite word evaluations occur.
Sources
- J. Gonzalez-Meneses, Basic results on braid groups, Section 4, printed pp. 26-29
- J. Birman and T. Brendle, Braids: A Survey, Section 5.1
- J. Gonzalez-Meneses, Basic results on braid groups, Section 4.1, printed pp. 29-30
- J. Gonzalez-Meneses, Basic results on braid groups, Theorem 4.2, printed pp. 30-31
- J. Birman and T. Brendle, Braids: A Survey, Section 5.2