Knot Contact Homology, Chern-Simons Theory, and Topological Strings

Similar documents
Knots and Mirror Symmetry. Mina Aganagic UC Berkeley

Topological Strings, D-Model, and Knot Contact Homology

HIGHER GENUS KNOT CONTACT HOMOLOGY AND RECURSION FOR COLORED HOMFLY-PT POLYNOMIALS

CONORMAL BUNDLES, CONTACT HOMOLOGY, AND KNOT INVARIANTS

Knot Homology from Refined Chern-Simons Theory

Refined Chern-Simons Theory, Topological Strings and Knot Homology

(Quantum) Super-A-polynomial

Knot polynomials, homological invariants & topological strings

Knot invariants, A-polynomials, BPS states

J-holomorphic curves in symplectic geometry

THE QUANTUM CONNECTION

KNOT CONTACT HOMOLOGY, STRING TOPOLOGY, AND THE CORD ALGEBRA

Duality for Legendrian contact homology

Duality Chern-Simons Calabi-Yau and Integrals on Moduli Spaces. Kefeng Liu

Lagrangian Intersection Floer Homology (sketch) Chris Gerig

Super-A-polynomials of Twist Knots

Open Gromov-Witten Invariants from the Augmentation Polynomial

Gauged Linear Sigma Model in the Geometric Phase

Topics in Geometry: Mirror Symmetry

Stringy Topology in Morelia Week 2 Titles and Abstracts

Gauge Theory and Mirror Symmetry

Elements of Topological M-Theory

Ω-deformation and quantization

Volume Conjecture: Refined and Categorified

A Quantitative Look at Lagrangian Cobordisms

Enumerative Invariants in Algebraic Geometry and String Theory

Symplectic topology and algebraic geometry II: Lagrangian submanifolds

Calabi-Yau Geometry and Mirror Symmetry Conference. Cheol-Hyun Cho (Seoul National Univ.) (based on a joint work with Hansol Hong and Siu-Cheong Lau)

Chern-Simons Theory & Topological Strings

LEGENDRIAN CONTACT HOMOLOGY IN THE BOUNDARY OF A SUBCRITICAL WEINSTEIN 4-MANIFOLD

Embedded contact homology and its applications

Chern-Simons gauge theory The Chern-Simons (CS) gauge theory in three dimensions is defined by the action,

Large N Duality, Lagrangian Cycles, and Algebraic Knots

Free fermion and wall-crossing

Knot Floer Homology and the Genera of Torus Knots

KNOT AND BRAID INVARIANTS FROM CONTACT HOMOLOGY I

CHARACTERISTIC CLASSES

Wrapped Fukaya categories

Conjectures on counting associative 3-folds in G 2 -manifolds

Two simple ideas from calculus applied to Riemannian geometry

Dualities and Topological Strings

String Duality and Localization. Kefeng Liu UCLA and Zhejiang University

Topological String Theory

Broken pencils and four-manifold invariants. Tim Perutz (Cambridge)

Homological mirror symmetry via families of Lagrangians

A TOPOLOGICAL INTRODUCTION TO KNOT CONTACT HOMOLOGY

Localization and Conjectures from String Duality

Towards bordered Heegaard Floer homology

Enumerative Geometry: from Classical to Modern

Publication. * are expository articles.

Lagrangian surgery and Rigid analytic family of Floer homologies

NON-ISOTOPIC LEGENDRIAN SUBMANIFOLDS IN R 2n+1

Combinatorial Heegaard Floer Theory

arxiv: v1 [math.gt] 20 Dec 2017

Integrable spin systems and four-dimensional gauge theory

Knots and Physics. Lifang Xia. Dec 12, 2012

Hyperbolic Knots and the Volume Conjecture II: Khov. II: Khovanov Homology

Do Super Cats Make Odd Knots?

DENSITY SPECTRA FOR KNOTS. In celebration of Józef Przytycki s 60th birthday

Higgs Bundles and Character Varieties

2 LEGENDRIAN KNOTS PROBLEM SESSION. (GEORGIA TOPOLOGY CONFERENCE, MAY 31, 2001.) behind the examples is the following: We are Whitehead doubling Legen

arxiv:math/ v1 [math.sg] 15 Jun 2002

Corrigendum to: Geometric Composition in Quilted Floer Theory

A Survey of Graph Cohomology and Perturbative Methods

On Flux Quantization in F-Theory

CHERN-SIMONS THEORY AND LINK INVARIANTS! Dung Nguyen! U of Chicago!

Towards a modular functor from quantum higher Teichmüller theory

Algebraic structure of the IR limit of massive d=2 N=(2,2) theories. collaboration with Davide Gaiotto & Edward Witten

Compactifying string topology

arxiv: v2 [hep-th] 25 Aug 2017

Knotted Legendrian surfaces with few Reeb chords

Homological Mirror Symmetry and VGIT

A Note on V. I. Arnold s Chord Conjecture. Casim Abbas. 1 Introduction

Web Formalism and the IR limit of massive 2D N=(2,2) QFT. collaboration with Davide Gaiotto & Edward Witten

Constructing compact 8-manifolds with holonomy Spin(7)

LEGENDRIAN SOLID-TORUS LINKS

Four Lectures on Web Formalism and Categorical Wall-Crossing. collaboration with Davide Gaiotto & Edward Witten

Moduli spaces of graphs and homology operations on loop spaces of manifolds

Mirror symmetry. Mark Gross. July 24, University of Cambridge

1 Hochschild Cohomology and A : Jeff Hicks

Refined Chern-Simons Theory. and. Topological String

Geometry of the Calabi-Yau Moduli

HMS Seminar - Talk 1. Netanel Blaier (Brandeis) September 26, 2016

Knot invariants, Hilbert schemes and Macdonald polynomials (joint with A. Neguț, J. Rasmussen)

arxiv:hep-th/ v2 4 Dec 2006

The Goodwillie-Weiss Tower and Knot Theory - Past and Present

CHERN-SIMONS THEORY AND WEYL QUANTIZATION Răzvan Gelca

FROM CLASSICAL THETA FUNCTIONS TO TOPOLOGICAL QUANTUM FIELD THEORY

Pseudoholomorphic Curves and Mirror Symmetry

Factorization Algebras Associated to the (2, 0) Theory IV. Kevin Costello Notes by Qiaochu Yuan

Combinatorial Spanning Tree Models for Knot Homologies

THE EXISTENCE PROBLEM

Instanton calculus for quiver gauge theories

1 Electrons on a lattice, with noisy electric field

KNOT POLYNOMIALS. ttp://knotebook.org (ITEP) September 21, ITEP, Moscow. September 21, / 73

Generalized Global Symmetries

On the Virtual Fundamental Class

Homological mirror symmetry

arxiv: v1 [math.gt] 22 Oct 2017

Knots, BPS states, and algebraic curves

Transcription:

Knot Contact Homology, Chern-Simons Theory, and Topological Strings Tobias Ekholm Uppsala University and Institute Mittag-Leffler, Sweden Symplectic Topology, Oxford, Fri Oct 3, 2014

Plan Reports on joint work with Aganagic, Ng, and Vafa, arxiv:1304.5778 Knot Contact Homology Legendrian DGAs Augmentation varieties Physics Results Relations Chern-Simons theory and topological string The conifold and large N transition GW disk potentials and augmentations Legendrian SFT and recursion relations for the colored HOMFLY

Knot contact homology Let K S 3 be a knot. The Lagrangian conormal of K is L K = {p T K S 3 : p TK = 0} S 1 R 2.

Knot contact homology Let K S 3 be a knot. The Lagrangian conormal of K is L K = {p T K S 3 : p TK = 0} S 1 R 2. The Legendrian conormal is the ideal boundary of L K : Λ K = L K U S 3, where U S 3 is the unit conormal {(q, p) T S 3 : p = 1} S 3 S 2, with the contact form α = pdq, Λ K Sx 1 Sp 1 T 2, where Sx 1 is the longitude and Sp 1 the meridian.

Knot contact homology Knot contact homology is the Legendrian contact homology of the Legendrian conormal of a knot (or link), which we describe next.

Knot contact homology Knot contact homology is the Legendrian contact homology of the Legendrian conormal of a knot (or link), which we describe next. The Reeb vector field R of α: dα(r, ) = 0 and α(r) = 1. For α = p dq, the flow of R is the geodesic flow. A flow line of R beginning and ending on a Legendrian submanifold is a Reeb chord of Λ. For Λ K Reeb chords correspond to binormal geodesic chords on K, and are the critical points of the action functional γ γ α for curves γ with endpoints on Λ K.

Knot contact homology Knot contact homology is the Legendrian contact homology of the Legendrian conormal of a knot (or link), which we describe next. The Reeb vector field R of α: dα(r, ) = 0 and α(r) = 1. For α = p dq, the flow of R is the geodesic flow. A flow line of R beginning and ending on a Legendrian submanifold is a Reeb chord of Λ. For Λ K Reeb chords correspond to binormal geodesic chords on K, and are the critical points of the action functional γ γ α for curves γ with endpoints on Λ K. The grading c of a Reeb chord c is defined by a Maslov index. For binormal geodesics in knot contact homology the grading is the Morse index: min = 0, sad = 1, max = 2.

Knot contact homology Legendrian contact homology is in a sense the Floer homology of the action functional γ α. Because of certain bubbling phenomena this theory becomes more non-linear than in the ordinary case.

Knot contact homology Legendrian contact homology is in a sense the Floer homology of the action functional γ α. Because of certain bubbling phenomena this theory becomes more non-linear than in the ordinary case. The symplectization of U S 3 is U S 3 R with symplectic form d(e t α) where t R. Fix an almost complex structure J such that J( t ) = R, J(ker α) = ker α, and dα(v, Jv) > 0.

Knot contact homology Legendrian contact homology is in a sense the Floer homology of the action functional γ α. Because of certain bubbling phenomena this theory becomes more non-linear than in the ordinary case. The symplectization of U S 3 is U S 3 R with symplectic form d(e t α) where t R. Fix an almost complex structure J such that J( t ) = R, J(ker α) = ker α, and dα(v, Jv) > 0. If c is a Reeb chord of Λ K then c R is a J-holomorphic strip with boundary on the Lagrangian submanifold Λ K R.

Knot contact homology The Legendrian contact homology algbera of Λ K is the free unital (non-commutative) algebra A(Λ K ) = C[H 2 (U S 3, Λ K )] Reeb chords = C[e ±x, e ±p, Q ±1 ] Reeb chords

Knot contact homology The Legendrian contact homology algbera of Λ K is the free unital (non-commutative) algebra A(Λ K ) = C[H 2 (U S 3, Λ K )] Reeb chords = C[e ±x, e ±p, Q ±1 ] Reeb chords A(Λ K ) is a DGA. The differential : A(Λ K ) A(Λ K ) is linear, satisfies Leibniz rule, decreases grading by 1, and is defined on generators through a holomorphic curve count. The DGA (A(Λ K ), ) is invariant under deformations up to homotopy and in particular up to quasi-isomorphism.

Knot contact homology

Knot contact homology

Knot contact homology In general, the knot contact homology can be explicitly computed, via Morse flow trees as an adiabatic limit of holomorphic disks, from a braid presentation of a knot. For a braid on n strands the algebra has n(n 1) generators in degree 0, n(2n 1) in degree 1, and n 2 in degree 2.

Knot contact homology In general, the knot contact homology can be explicitly computed, via Morse flow trees as an adiabatic limit of holomorphic disks, from a braid presentation of a knot. For a braid on n strands the algebra has n(n 1) generators in degree 0, n(2n 1) in degree 1, and n 2 in degree 2. The unknot A(Λ U ) = C[e ±x, e ±p, Q ±1 ] c, e, c = 1, e = 2, e = c c = 0, c = 1 e x e p + Qe x e p

Knot contact homology The right handed trefoil (differential in degree 1): A(Λ T ) = C[e ±x, e ±p, Q ±1 ] a 12, a 21, b 12, b 21, c ij, e ij i,j {1,2}, a ij = 0, b ij = c ij = 1, e ij = 2, b 12 = e x a 12 a 21, b 21 = e x a 21 a 12, c 11 = e p e x e x (2Q e p )a 12 Qa12a 2 21, c 12 = Q e p + e p a 12 + Qa 12 a 21, c 21 = Q e p + e p e x a 21 + Qa 12 a 21, c 22 = e p 1 Qa 21 + e p a 12 a 21,

Augmentation variety Consider A = A(Λ K ) as a family over (C ) 3 of C-algebras, where points in (C ) 3 correspond to values of coefficients (e x, e p, Q).

Augmentation variety Consider A = A(Λ K ) as a family over (C ) 3 of C-algebras, where points in (C ) 3 correspond to values of coefficients (e x, e p, Q). An augmentation of A is a chain map ɛ: A C, ɛ = 1, of DGAs, where we consider C as a DGA with trivial differential concentrated in degree 0.

Augmentation variety Consider A = A(Λ K ) as a family over (C ) 3 of C-algebras, where points in (C ) 3 correspond to values of coefficients (e x, e p, Q). An augmentation of A is a chain map ɛ: A C, ɛ = 1, of DGAs, where we consider C as a DGA with trivial differential concentrated in degree 0. The augmentation variety V K is the algebraic closure of { (e x, e p, Q) (C ) 3 : A has augmentation }.

Augmentation variety The augmentation polynomial A K is the polynomial corresponding to the hypersurface V K. It can be computed from the formulas for the differential by elimination theory.

Augmentation variety The augmentation polynomial A K is the polynomial corresponding to the hypersurface V K. It can be computed from the formulas for the differential by elimination theory. The unknot U: A U (e x, e p, Q) = 1 e x e p + Qe x e p.

Augmentation variety The augmentation polynomial A K is the polynomial corresponding to the hypersurface V K. It can be computed from the formulas for the differential by elimination theory. The unknot U: A U (e x, e p, Q) = 1 e x e p + Qe x e p. The trefoil T : A T (e x, e p, Q) = (e 4p e 3p )e 2x + (e 4p Qe 3p + 2Q 2 e 2p 2Qe 2p Q 2 e p + Q 2 )e x + ( Q 3 e p + Q 4 ).

Physics results We apply Witten s argument, equating partition functions in U(N) Chern-Simons gauge theory on a 3-manifold M and in open topological strings in T M with N Lagrangian branes on the 0-section M, to the case of a knot K in M = S 3.

Physics results We apply Witten s argument, equating partition functions in U(N) Chern-Simons gauge theory on a 3-manifold M and in open topological strings in T M with N Lagrangian branes on the 0-section M, to the case of a knot K in M = S 3. Add an additional Lagrangian brane along L K. Integrating the strings in L K S 3 = K out corresponds to inserting det(1 Hol(K)e x ) 1 in the Chern-Simons path integral, where Hol(K) is the holonomy of the U(N)-connection around K and e x the holonomy of the U(1)-connectionon L K.

Physics results Expanding det(1 e x Hol(K)) 1 = k tr Sk Hol(K) e kx then gives Ψ K (e x, e gs, Q) : = Z GW (L K )/Z GW = k tr Sk Hol(K) e kx = k H k (K) e kx, where H k (K) is a HOMFLY-invariant which is polynomial in q = e gs and Q = q N = e gsn. Here the string coupling constant is g s = 2πi N+k, where k is the level in Chern-Simons path integral.

Physics results X = T S 3 is a quadric in C 4, a resolution of a cone. There is another resolution Y, total space of O( 1) O( 1) CP 1.

Physics results Gopakumar-Vafa show: if area(cp 1 ) = t = Ng s, and Q = e t = q N, then Z GW (X ; N, g s ) = Z GW (Y ; g s, Q), relating open curve counts in X to closed curve counts in Y.

Physics results For a knot K S 3, L K can be shifted off the 0-section by a non-exact Lagrangian isotopy. Then L K Y and Gopakumar-Vafa s argument gives Ψ K (e x, e gs, Q) = Z GW (Y ; L K )/Z GW (Y )

Physics results Witten s argument relates constant curves on L K to GL(1)-gauge theory on the solid torus which corresponds to ordinary QM in the periods of the connection. This gives pψ K (x) = g s x Ψ K (x), and the usual asymptotics: ( 1 Ψ K (x) = exp g s ) pdx +....

Physics results Witten s argument relates constant curves on L K to GL(1)-gauge theory on the solid torus which corresponds to ordinary QM in the periods of the connection. This gives pψ K (x) = g s x Ψ K (x), and the usual asymptotics: ( 1 Ψ K (x) = exp g s ) pdx +.... From the GW -perspective Ψ K (x) = exp ( ) 1 W K (x) +..., g s where W K (x) is the disk potential of L K and where... counts curves with 2g 2 + h > 1.

Physics results We conclude that as we vary L K p = W K x.

Physics results We conclude that as we vary L K p = W K x. This equation gives a local parameterization of an algebraic curve that in all examples agree with the augmentation variety (and which is important from a mirror symmetry perspective).

Augmentations and exact Lagrangian fillings Exact Lagrangian fillings L of Λ K in T S 3 induces augmentations by ɛ(a) = M A (a) A. a =0 The map on coefficients are just the induced map on homology.

Augmentations and exact Lagrangian fillings There are two natural exact fillings of Λ K : L K and M K S 3 K. Thus, e p = 1 and e x = 1 belong to V K Q=1 for any K.

Augmentations and exact Lagrangian fillings There are two natural exact fillings of Λ K : L K and M K S 3 K. Thus, e p = 1 and e x = 1 belong to V K Q=1 for any K. For the unknot A U (e x, e p, Q = 1) = (1 e x )(1 e p ).

Augmentations and exact Lagrangian fillings There are two natural exact fillings of Λ K : L K and M K S 3 K. Thus, e p = 1 and e x = 1 belong to V K Q=1 for any K. For the unknot A U (e x, e p, Q = 1) = (1 e x )(1 e p ). For the trefoil A T (e x, e p, Q = 1) = (1 e x )(1 e p )(e 3p 1).

Augmentations and non-exact Lagrangian fillings The Lagrangian conormal can be shifted off of the 0-section. It becomes non-exact but a similar version of compactness still holds: closed disks lie in a compact, and disks at infinity are as for Λ K R.

Augmentations and non-exact Lagrangian fillings The previous definition of a chain map does not work because of new boundary phenomena. Compare the family of real curves in C 2, xy = ɛ, ɛ 0.

Augmentations and non-exact Lagrangian fillings We resolve this problem by using bounding chains: fix a chain σ D for each rigid disk D that connects its boundary in L K to a multiple of a standard homology generator at infinity.

Augmentations and non-exact Lagrangian fillings We introduce quantum corrected holomorphic disks with punctures: these are ordinary holomorphic disks with all possible insertions of σ along the boundary. (Compare Fukaya s unforgettable marked points.) In the moduli space M A (a; σ) of quantum corrected disks, boundary bubbling become interior points.

Augmentations and non-exact Lagrangian fillings Analyzing the boundary then shows that ɛ(a) = M A (a; σ)e A a =1 is a chain map provided p = W K x.

Augmentations and non-exact Lagrangian fillings We find that p = W K x parameterizes a branch of the augmentation variety.

Augmentations and non-exact Lagrangian fillings We find that p = W K x parameterizes a branch of the augmentation variety. Here W K is the Gromov-Witten disk potential which is a sum over weighted trees, where there is a holomorphic disk at each vertex of the tree and where each edge is weighted by the linking number of the boundaries of the disks at the vertices of its endpoints, as measured by the bounding chains.

Many component links For an m-component link, the GW-disk potential of the conormal filling sees only the individual knot components. We thus need other Lagrangian fillings of the conormal tori. Such fillings indeed exist in special cases. In the general case there are candidates. The GW-disk potential W K of such a filling gives a Lagrangian (locally x j = W p j ) augmentation variety V K in (C ) m (C ) m.

Physical quantization of V K To quantize V K, consider the topological A-model on T T 2m with a co-isotropic A-brane B on the whole space and a Lagrangian brane L on V K. The algebra of open string states from B to B gives the Weyl algebra. Strings with one end on B and the other on L has a unique ground state which gives a fermion ψ K (x) on V K. The theory of this fermion is the D-model on V K. Conjecture: the expectation value of this fermion living on V K is the (HOMFLY) function, ψ K (x) = Ψ K (x) = Ψ K (x 1,..., x m ). The action of B B strings on B V K strings gives a D-module with characteristic variety V K, this is in line with the intersection behavior of various components of V K.

Physical quantization of V K To quantize V K, consider the topological A-model on T T 2m with a co-isotropic A-brane B on the whole space and a Lagrangian brane L on V K. The algebra of open string states from B to B gives the Weyl algebra. Strings with one end on B and the other on L has a unique ground state which gives a fermion ψ K (x) on V K. The theory of this fermion is the D-model on V K. Conjecture: the expectation value of this fermion living on V K is the (HOMFLY) function, ψ K (x) = Ψ K (x) = Ψ K (x 1,..., x m ). The action of B B strings on B V K strings gives a D-module with characteristic variety V K, this is in line with the intersection behavior of various components of V K. This is in line with Garoufalidis result that the colored HOMFLY polynomial is q-holonomic.

Mathematical quantization of V K Legendrian SFT It is possible to define a version of Legendrian SFT in the current set up which we conjecture gives the D-module defined by the colored HOMFLY polynomial.

Mathematical quantization of V K Legendrian SFT It is possible to define a version of Legendrian SFT in the current set up which we conjecture gives the D-module defined by the colored HOMFLY polynomial. Let the degree 1 Reeb chords be denoted b 1,... b m and the degree 0 Reeb chords a 1,..., a n

Mathematical quantization of V K Legendrian SFT It is possible to define a version of Legendrian SFT in the current set up which we conjecture gives the D-module defined by the colored HOMFLY polynomial. Let the degree 1 Reeb chords be denoted b 1,... b m and the degree 0 Reeb chords a 1,..., a n Additional data: a Morse function f on L K which gives obstruction chains. A 4-chain C K for L K with C K = 2[L K ] which looks like ±J f near the boundary.

Legendrian SFT

Legendrian SFT We define the GW-potential e F as the generating function of oriented graphs with holomorphic curves at vertices, and intersections with chains at the edges weighted by ± 1 2 : F = C χ,m,i gs χ e mx a i1... a ir

Legendrian SFT Let H(b j ) denote the count of rigid holomorphic curves in the symplectization with a positive puncture at b j : H(b j ) = C χ,m,n,i,j g χ s e mx e np a i1... a ir gs l... a j1 Then if p = g s x, e F H(b j )e F counts ends of a 1-dimensional moduli space and in particular: H(b j )e F = 0. a jl. We conjecture that the above system give a generator ÂK with ÂK e F 0 = 0 for the D-module ideal corresponding to the colored HOMFLY of K (generators Âj K if K is a link).

Legendrian SFT and HOMFLY For the unknot there are no (formal) higher genus curves and the operator equation is  U (e x, e p, Q) = (1 e x e p Qe x e p )Ψ U = 0.

Legendrian SFT and HOMFLY For the unknot there are no (formal) higher genus curves and the operator equation is  U (e x, e p, Q) = (1 e x e p Qe x e p )Ψ U = 0. For the Hopf link L Reeb chord generators are as for the trefoil. The relevant parts for the operator H is as follows: H(c 11 ) = (1 e x 1 e p 1 + Qe x 1 e p 1 ) + g 2 s a12 a21 + O(a), H(c 22 ) = (1 e x 2 e p 2 + Qe x 2 e p 2 ) + Qe x 2 e p 2 g 2 s a12 a21 + O(a), H(c 12 ) = (e p 2 e p 1 Qe x 2 e p 2 )g s a12 + gs 1 (e gs 1)(1 e x 2 )a 21 + O(a 2 ), H(c 21 ) = (Qe x 1 e p 1 e gs e p 1 e p 2 )g s a21 + gs 1 ( (e g s (e gs 1) e 2gs (e gs 1)e x 1 )e p 1 e p 2 +(e gs 1)Qe x 1 e p 1 gs 2 ) a12 a21 a12 + O(a 2 ).

Legendrian SFT and HOMFLY After the change of variables, e x 1 = e g s e x 1, e p 1 = e g s e p 1 ; e x 2 = Q 1 e x 2, e p 2 = e g s Q 1 e p 2 ; Q = e gs Q, g s = g s, we find the D-module ideal generators  1 L = (ex 1 e x 2 ) + (e p 1 e p 2 ) Q(e x 1 e p 1 e x 2 e p 2 )  2 L = (1 e gs e x 1 e p 1 + Qe x 1 e p 1 )(e x 1 e p 2 )  3 L = (1 e gs e x 2 e p 2 + Qe x 2 e p 2 )(e x 2 e p 1 ), in agreement with HOMFLY.

Legendrian SFT and HOMFLY Similarly, for the trefoil we get the D-module ideal generator  T = (Q 2 Qe p )e gs (Q e gs e 2p )Q 2 + ( (Q 2 Qe p )e 5gs (Q e gs e 2p )e 2p) e x + ( (qe 2p e p )e 3gs (Q e 3gs e 2p )Q 2) e x + ( e gs (Q e gs e 2p )(Q e 2gs e 2p )(Q e 3gs e 2p ) ) e x + ( (e gs e 2p e p )e 7gs (Q e 3gs e 2p )e 2p) e 2x. which after a change of variables is in agreement with HOMFLY.

Legendrian SFT and HOMFLY Similarly, for the trefoil we get the D-module ideal generator  T = (Q 2 Qe p )e gs (Q e gs e 2p )Q 2 + ( (Q 2 Qe p )e 5gs (Q e gs e 2p )e 2p) e x + ( (qe 2p e p )e 3gs (Q e 3gs e 2p )Q 2) e x + ( e gs (Q e gs e 2p )(Q e 2gs e 2p )(Q e 3gs e 2p ) ) e x + ( (e gs e 2p e p )e 7gs (Q e 3gs e 2p )e 2p) e 2x. which after a change of variables is in agreement with HOMFLY. Note that the classical limit g s 0 of this is (Q e 2p )A T (e x, e p, Q).