Documentation

TauCeti.Algebra.Homology.Contraction.TensorTrick

The tensor trick #

A special contraction of (M, d) onto (N, d') induces a special contraction of the reduced tensor coalgebras Tᶜ(M) = ⨁_{n ≥ 1} M^{⊗ n} onto Tᶜ(N). The endomorphisms on words are the letterwise extensions of d and d', the degree-one graded coderivations ReducedTensorWords.gradedCoderiv G (d ∘ letter) 1 whose only Taylor component is d on single letters; on a word they apply d to one letter at a time, with the Koszul sign of the letters it passes. When d and d' square to zero, these extensions are differentials. The inclusion and projection act letterwise, and the homotopy is

H = ∑_j τ^{⊗ j} ⊗ h ⊗ (i p)^{⊗ (n - j - 1)}

on words of length n, where τ = InternalGrading.koszulTwist G 1 is the Koszul sign of moving the odd map h past a letter. Cross terms of d H + H d cancel because d and h are odd and i p commutes with d and with τ, while the diagonal terms telescope to 1 - (i p)^{⊗ n}. The side conditions of the letters give those of the words.

This is the input of homological transfer along a contraction of an A∞ algebra onto a retract such as its cohomology: the bar differential of the algebra is the letterwise extension of its unary operation plus a perturbation which shortens words, so the perturbation lemma applies to the contraction of bar constructions produced here.

Main definitions #

Main results #

References #

Families of letter maps acting around one slot #

The contraction identity on words of one length #

Operators on reduced tensor words acting length by length #

noncomputable def TauCeti.LinearSpecialContraction.reducedTensorWordsHomotopy {R : Type uR} {M : Type uM} {N : Type uN} [CommRing R] [AddCommGroup M] [Module R M] [AddCommGroup N] [Module R N] {dM : Module.End R M} {dN : Module.End R N} (c : LinearSpecialContraction dM dN) (G : InternalGrading R M) :

The tensor-trick homotopy on reduced tensor words: on words of length n it is ∑_j τ^{⊗ j} ⊗ h ⊗ (i p)^{⊗ (n - j - 1)}, where τ is the Koszul twist of parameter one for the grading G.

Equations
  • One or more equations did not get rendered due to their size.
Instances For
    @[simp]
    theorem TauCeti.LinearSpecialContraction.reducedTensorWordsHomotopy_of {R : Type uR} {M : Type uM} {N : Type uN} [CommRing R] [AddCommGroup M] [Module R M] [AddCommGroup N] [Module R N] {dM : Module.End R M} {dN : Module.End R N} (c : LinearSpecialContraction dM dN) (G : InternalGrading R M) (n : { n : ℕ // 0 < n }) (z : TensorPower R (↑n) M) :
    (c.reducedTensorWordsHomotopy G) ((ReducedTensorWords.of R M n) z) = (ReducedTensorWords.of R M n) ((∑ j ∈ Finset.range ↑n, PiTensorProduct.map fun (i : Fin ↑n) => if ↑i < j then G.koszulTwist 1 else if ↑i = j then c.homotopy else c.incl ∘ₗ c.proj) z)

    The tensor-trick homotopy on an arbitrary word of length n, expanded as the sum of its single-slot actions.

    theorem TauCeti.LinearSpecialContraction.reducedTensorWordsHomotopy_of_tprod {R : Type uR} {M : Type uM} {N : Type uN} [CommRing R] [AddCommGroup M] [Module R M] [AddCommGroup N] [Module R N] {dM : Module.End R M} {dN : Module.End R N} (c : LinearSpecialContraction dM dN) (G : InternalGrading R M) (n : { n : ℕ // 0 < n }) (x : Fin ↑n → M) :
    (c.reducedTensorWordsHomotopy G) ((ReducedTensorWords.of R M n) ((PiTensorProduct.tprod R) x)) = ∑ j ∈ Finset.range ↑n, (ReducedTensorWords.of R M n) ((PiTensorProduct.tprod R) fun (i : Fin ↑n) => if ↑i < j then (G.koszulTwist 1) (x i) else if ↑i = j then c.homotopy (x i) else c.incl (c.proj (x i)))

    The tensor-trick homotopy on a pure tensor word: the sum over positions j of the word with the letters before j Koszul-twisted, h applied at j, and i p applied after j.

    @[simp]

    On a single letter the tensor-trick homotopy is the homotopy of the contraction.

    On a two-letter word the tensor-trick homotopy is h ⊗ i p + τ ⊗ h, where τ is the Koszul twist of parameter one.

    The tensor-trick homotopy lowers the total degree of words by one when the homotopy of the contraction has degree -1 and its inclusion and projection have degree zero.

    The tensor-trick homotopy and deconcatenation #

    The tensor-trick homotopy is a coderivation homotopy. Cutting H z either cuts to the right of the letter carrying h, where the letters already carry i p, or cuts to its left, where the letters passed by h carry the Koszul twist:

    Δ H = (H ⊗ (i p)) Δ + (τ ⊗ H) Δ.

    The tensor-trick homotopy anticommutes with the letterwise Koszul twist: it moves one odd map h past the letters it acts on.

    The tensor trick. A special contraction of (M, dM) onto (N, dN) by graded maps of the expected degrees induces a special contraction of the reduced tensor coalgebras, whose endomorphisms are the letterwise extensions of dM and dN with the Koszul signs of G and H. When dM and dN square to zero, these extensions are differentials. The inclusion and projection act letterwise, and the homotopy is reducedTensorWordsHomotopy.

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For
      @[simp]

      The inclusion of the tensor-trick contraction is the letterwise inclusion.

      @[simp]

      The projection of the tensor-trick contraction is the letterwise projection.

      @[simp]

      The homotopy of the tensor-trick contraction is reducedTensorWordsHomotopy.