The functorial braid relation: a supplied statement, and its K₀ shadow #
For an A₂-configuration (A, B) the Seidel--Thomas theorem is the isomorphism of functors
T_A T_B T_A ≅ T_B T_A T_B. It is stated here, never proved and never asserted:
BraidStatement dA dB carries that isomorphism as its single field, and nothing constructs an
inhabitant.
Main definitions and results #
BraidStatement— the supplied isomorphismT_A T_B T_A ≅ T_B T_A T_B.map_braid_of_sphericalPairData— theK₀braid identity for the Euler-form data.BraidStatement.map_braid,BraidStatement.twistK₀_braid— what a supplied isomorphism forces.
What an inhabitant claims #
An inhabitant is a claim about the supplied dA, dB, and Seidel--Thomas supports it only when
those are the genuine twists: SphericalTwistData records no more than a cone-like functor with the
right K₀ shadow, so for padded or otherwise non-genuine data the isomorphism can fail even for a
genuine A₂-configuration. The dg realizations of ObjectTwistData.lean are the genuine ones.
SphericalPairData is the Euler-form shadow of an A₂-configuration, not the configuration
itself: it does not record that Hom^•(A, B) is one-dimensional. The structure here records no
hypothesis at all, so an inhabitant is a claim about a specific pair and must be supplied only for a
genuine A₂-configuration. The isomorphism is of underlying functors; the graded natural
isomorphism of Seidel--Thomas also respects the shift isomorphisms, which is not recorded.
What is proved: only the K₀ shadow, and only in one direction #
The two K₀ facts below are the entire content, and both are far weaker than the functorial
statement, because K₀ sees an object only through its class.
map_braid_of_sphericalPairData: for the Euler-form dataSphericalPairDataofBraid.lean, the induced mapsK₀.map (T_A T_B T_A)andK₀.map (T_B T_A T_B)agree. This istwistK₀_braidread throughSphericalTwistData.map_eq_twistK₀: the lattice identity is a consequence of the geometry'sK₀action, not a parallel story.BraidStatement.map_braid: aBraidStatementimplies the same equality, fromK₀.map_congralone, with no use ofSphericalPairData. It is a consistency check that the supplied isomorphism cannot contradict what is proved: for a pair whoseK₀twists do not braid there is noBraidStatement, so supplying one is a claim a lattice computation can refute.
Nothing projects the other way: the equality of two maps on K₀ gives no isomorphism of
functors.
References #
Seidel--Thomas, arXiv:math/0001043v2. Definition 1.1(b): an
A_m-configuration is a family of m spherical objects with dim Hom^•(E_i, E_j) = 1 for
|i - j| = 1 and 0 for |i - j| ≥ 2, the dimension being total over all degrees. Theorem 1.2
(second part) gives T_{E_i} T_{E_{i+1}} T_{E_i} ≅ T_{E_{i+1}} T_{E_i} T_{E_{i+1}} for such a
family, up to graded natural isomorphism, and Theorem 2.17 is the abstract form. Proposition 2.13
proves the displayed isomorphism for two n-spherical objects with n > 0 and total
dim Hom^•(E_2, E_1) = 1.
The functorial braid relation T_A T_B T_A ≅ T_B T_A T_B, as a supplied statement.
Supplied, not proved; see the module docstring for what an inhabitant claims.
The natural isomorphism, in diagrammatic order:
T_Aapplied first.
Instances For
The K₀ shadow of the braid relation, from the Euler-form data. twistK₀_braid
transported through SphericalTwistData.map_eq_twistK₀. The hypothesis is the product condition
χ(A,B)·χ(B,A) = 1 of SphericalPairData, not χ(A,B) = 1.
The consistency check. A supplied braid isomorphism forces the K₀ identity, without the
Euler-form data.
The same, as the identity τ_A τ_B τ_A = τ_B τ_A τ_B of K₀ twists.