Lectures on quantum topology in dimension three. Lectures of the SMF school ``Geometric and quantum topology in dimension 3'', CIRM, Luminy, France, June 2014 (Q2823027)

The four articles of this volume comprise the notes of four lecture series on quantum topology in dimension three: An introduction to the subject and lectures on the AJ conjecture, on Khovanov homology, and on Heegaard Floer homology. The pertinent methods, developed over the past two decades, have brought forth spectacular results in knot theory, such as detection of the unknot, of the knot genus, of knot fibering, and of the Thurston norm. The four parts will be reviewed individually. \newline The first part, written in French by \textit{C. Lescop}, ``Introduction à quelques invariants des variétés de dimension \(3\) et leurs entrelacs'' (Introduction to some invariants of manifolds of dimension \(3\) and their links), supplies an introductory survey of the items announced in the title, leading from the basic notions all the way to the material discussed in the three remaining articles of the book. It provides a thorough and fairly complete inventory of definitions, concepts and results, including a description of the present state of the art and taking into account the historical development. Due to numerous references to the comprehensive literature list, it can serve as an excellent source for the beginner as well as for the expert. Section 1 introduces manifolds of the various categories, mentioning exotic spheres and triangulation related results. Section 2 centers around the fundamental group, classical homology and cobordism. Section 3 deals with problems special to \(4\)-manifold theory, quadratic form and signature, and the difference in classification with respect to the topological and the differentiable cases. The remaining sections are about \(3\)-manifolds, in Section 4 about decompositions (connected sum, Heegaard and Jaco-Shalen-Johannson), thr Poincaré conjecture and the geometrization conjecture. Section 5 defines knots and links, and some classical invariants like genus, unknotting number, Alexander polynomial, knot exterior. Dehn surgery is introduced and various applications of it are explained. Section 6 opens the chapter of post-classical invariants, like new polynomial and finite type invariants, geometrically based on braids, tangles, graphs and diagrams. Further developments concern algebraic and categorical methods leading to quantum invariants and Reshetikhin-Turaev-invariants and, in the framework of the so-called Witten program, to the Kontsevich integral and its extensions such as the LMO-invariant for homology-\(3\)-spheres. Finally, this section touches on the topics of the second and third article in this book, namely the AJ conjecture and Khovanov homology. In the final Section 7 of this introduction, analytic invariants of \(3\)-manifolds are dealt with. After exhibiting the significance of the Casson-invariant, Floer homology and its variations are explained, e.g. Seiberg-Witten-Floer homology, Lagrange-Floer homology, Heegaard-Floer homology. The latter is explicated in the last article of the present book.NEWLINENEWLINE Though this concise introduction is a tour de force through the history of the theory of knots as well as manifolds in dimensions \(3\) and \(4\), it offers a useful summary on what has been achieved and on what is still open. In particular, many highlight results are quoted along the way, e.g. on \(4\)-manifold classification by Donaldson, Freedman and Quinn, on knot classifications by Gordon-Luecke and Waldhausen, and the proof of the Tait conjecture by Kauffman, Murasugi and Thistlethwaite. The choice of details is also meant as a preparation for what follows in the book.NEWLINENEWLINEPart two by \textit{T. T. Q. Lê} (``The colored Jones polynomial and the AJ conjecture''), deals with the AJ conjecture which so far has been proved only for special classes of knots, and it relates the colored Jones polynomial to the \(A\)-polynomial. This can be delineated essentially in the following manner (details below): The coloring is done by attaching a certain complex \(sl_2\)-module to every integer for every link component. (Dimension \(1\) gives back the by now classical Jones ploynomial.) By the \(q\)-holonomicity of the colored Jones polynomial, the recurrence ideal of a knot \(K\) (the annihilator of the colored Jones polynomial) has a single generator \(\alpha_k(t;L,M)\) which, evaluated at \(t = \pm 1\), is the \(A\)-polynomial up to a factor depending only on \(M\). A prominent feature of the \(A\)-polynomial is that it detects the unknot in \(S^3\). The present article aims at developing the material necessary to describe and understand the AJ conjecture and the so far known cases of its validity to which the author has contributed substantially, while important other results are mentioned along the way. The article begins with a primer on knots, links and braids leading to the colored Jones polynomial via Yang-Baxter (YB) operators. The ingredients are a complex vector space \(V\), representations \(\rho : \mathcal{B}_n \to V^{\otimes n}\) of the braid groups determined by \(\rho(\beta)\) (\(\beta\) being the simplest non-trivial \(2\)-braid), and an invertible operator \(R : V \otimes V \to V \otimes V\) satisfying the YB relations. A trace construction of Turaev yields a link invariant that turns out to be the Jones polynomial, by its characterizing properties. For the coloring, one takes a uniquely determined \(sl_2\)-module \(V_n\) of dimension \(n\), for every \(n\), to obtain the colored Jones polynomial \(J_L (n_1,\cdot\cdot\cdot n_m) = J_L (V_{n_1},\cdot\cdot\cdot V_{n_m}) \in {\mathbb Z} [q^{\pm 1/4}]\). This is an occasion to explain the (meanwhile proved correct) Melvin-Morton conjecture which exhibits the compact limit of a certain series, defined by the colored Jones polynomial, as depending only on the scaled Alexander polynomial of a knot. Considering \(J_K : {\mathbb Z} \to {\mathbb C} [t^{\pm 1}]\) for the colored Jones polynomial of a knot \(K\) and defining operators \(L\) and \(M\) on \(\mathrm{Map} ({\mathbb Z}, V)\) by \((Lf) (n) := f(n+1)\) and \((Mf)(n) := t^{2n} f(n)\), one has the recurrence relation \(LM = t^2 ML\). Then \(q\)-holonomicity of the Jones polynomial expresses the fact that \(J_k : {\mathbb Z}^n \to {\mathbb Z}[t^{\pm 1}]\) has a generator \(\alpha_k(t;L,M)\) of its recurrence (annihilator) ideal. To approach the \(A\)-polynomial of a knot \(K\) the restriction of character varieties \(\chi(\pi_1X) \to \chi(\pi_1 (\partial X))\) is invoked where \(X\) is a nice tubular neighborhood of \(K\). As \(\partial X\) is a torus and \(\chi(\pi_1 (\partial X)) \cong ({\mathbb C}^\ast)^2/\tau\), \(\tau\) the multiplicative inversion of components, the image of the above restriction gives rise to a finite set of complex curves in \({\mathbb C}^2\), defined by a polynomial \(A_K \in {\mathbb Z} [M,L]\), the said \(A\)-polynomial of \(K\). For the AJ conjecture to hold for \(K\), namely that \(\alpha_K | _{t=\pm 1}\) equals \(A_K\) up to a factor depending solely on \(M\), a system of sufficient conditions is given. The proof is only indicated along the lines of sources published by the author and A. T. Tran. Cases known comprise certain pretzel, \(2\)-bridge and double twist knots.\newline The reviewer came across three misprints: p. 58 line 11: replace \(n\) by \(1\) in the denominator; p. 62 line 6: replace \(k\) by \(n\); p. 81 line 15: replace \(\rho\) by \(\vartheta\).NEWLINENEWLINEThe third part, written by \textit{P. Turner}, bears the heading ``A hitchhiker's guide to Khovanov homology''. The latter is an intricate and many-faceted theory. The approach chosen here is to first collect a series of properties expected or considered natural into an existence theorem, draw conclusions on it and later describe realizations. For a link \(L\) given by a diagram, Khovanov homology is a bigraded module \(\mathrm{Kh}^{\ast,\ast}(L)\), functorial with respect to link equivalence, tensor-multiplicative with respect to disjoint union, and equipped with an exact triangle giving rise to a long exact sequence, involving the usual reduction of the number of crossings in a link diagram and thereby permitting inductive calculations. At this stage, one concludes that the Jones polynomial can be obtained from Khovanov homology. The same is true for the reduced version \(\widetilde{\mathrm{Kh}}\vphantom{Kh}^{\ast,\ast}(L)\). While the simplest version of Khovanov homology has coefficients in \({\mathbb Z}_2\), other coefficients like the integers are possible. We give an impression of the laborious construction of Khovanov homology and start with a diagram \(D\) of an oriented link \(L\) and the set \(X_D\) of its crossings. For each subset \(A \subset X_D\), one removes all diagram crossings in \(X_D\) according to a specified rule, to obtain a collection of, say \(\| A \|\), closed curves in the projection plane, called the resolution \(R(A)\) associated to \(A\). Let \(F^j_D(A)\) be the module freely generated by commutative monomials \(x_{\gamma_1} \cdot\cdot\cdot x_{\gamma_m}\) (\(\gamma_k \in R(A)\), relation \({\gamma_k}^2 = 0\)) where \(j = \sharp A + \| A \| - 2m\). This leads to a bigraded cochain complex \((\mathcal{C}^{\ast,\ast}(D),d)\), where \(\mathcal{C}^{i,j}(D) := \bigoplus\limits_{A \subset X_D,\sharp A = i} F^j_D(A)\) and \(d: \mathcal{C}^{i,j}(D) \to \mathcal{C}^{i+1,j}(D)\). Then, independently of the representing diagram \(D\), Khovanov homology can be defined as \(\mathrm{Kh}^{i,j}(L) = H^{i+n_-, \, j+2n_- -n_+}(\mathcal{C}^{\ast,\ast}(D),d)\), where \(D\) has \(n_+\) positive and \(n_-\) negative crossings. The remainder of the article treats variants, extensions and generalizations of Khovanov homology and looks into applications and geometric interpretations. The presentation is concise, but gives numerous references to the pertinent literature. Adopting anticommutativity in the above monomials, written \(x_{\gamma_1}\wedge \cdot\cdot\cdot \wedge x_{\gamma_m}\), creates what is called odd Khovanov homology, also supplying the Jones polynomial. Then a tangle based access to Khovanov homology is introduced. A variant of Khovanov homology arises from replacing the relation \(x_\gamma^2 = 0\) by \(x_\gamma^2 = t+hx_\gamma\), where \(t\) and \(h\) are elements of the ground ring. Specializing to \(h=0\), \(t=1\) over the rationals creates Lee theory, which mirrors linking numbers and to which the Lee-Rasmussen spectral sequence converges. Khovanov-Rozansky \(sl(N)\)-homology \(\mathrm{KR}_N^{\ast,\ast}\) works with coefficients in \(sl(N)\) as a Lie algebra (\(N=2\) is the Khovanov case) and provides the \(sl(N)\)-polynomial which is related to the well-known HOMFLYPT-polynomial. Here, links may be singular, the singularities modelled on the cross \(\times\). Computations have been done for \(2\)-bridge and torus knots, and a HOMFLYPT-version of the theory recovers the HOMFLYPT-polynomial. Each colored Jones polynomial can be obtained from the Jones polynomials of suitable cables. The main application explained concerns concordance theory of knots. The Lee-Rasmussen spectral sequence defines Rasmussen's invariant, which vanishes for smoothly slice knots, whose absolute value is a lower bound for twice the (\(4\)-dimensional) smooth slice genus and which provides a new proof of the Milnor conjecture. Using the \(sl(N)\)-version of Rasmussen's invariant, other lower bounds for the smooth slice genus are found. Furthermore, Khovanov homology detects the unknot. On the geometry side, there are interpretations in symplectic geometry, representation varieties, instanton Floer homology, gauge theory and homotopy theory.\newline A misprint on p. 121 in Theorem 8.12: replace \(n\) by \(N\).NEWLINENEWLINEThe title of the fourth and last part, ``Heegaard Floer homologies'', by \textit{R. Lipshitz}, refers to the notion of Heegaard splitting, i.e., decomposition of \(3\)-manifolds by handlebodies, as well as to Floer homology, a concept created to study symplectic geometry in dimensions \(3\) and \(4\). Heegaard Floer homology was developed by \textit{P. Ozsváth} and \textit{Z. Szabó} [Ann. Math. (2) 159, No. 3, 1027--1158 (2004; Zbl 1073.57009)] and \textit{J. Rasmussen} [Floer homology and knot complements. Harvard University (PhD Thesis) (2003), \url{arXiv:math/0306378}]. The introduction to the subject follows the presentation of sutured Floer homology, \(SFH\), by \textit{A. Juhász} [Algebr. Geom. Topol. 6, 1429--1457 (2006; Zbl 1129.57039)]. Based on that, Heegaard Floer homology \(\widehat{HF}(Y)\) of a closed \(3\)-manifold \(Y\) and Heegaard Floer knot homology \(\widehat{HFK}(K)\) of a knot \(K\) are defined. \(\widehat{HFK}(K)\) comes with a double grading, \(\widehat{HFK}_i(K,j)\), where \(i\) is a homological grading (Maslow grading) and \(j\) is called Alexander grading. A striking result is that \(\widehat{HFK}\) detects the genus \(g(K)\) of a classical knot (hence it detects the unknot) by \(g(K) = \max \{j \mid \widehat{HFK}_\ast(K,j) \neq 0\}\) (Theorem 4.4). A proof is sketched according to \textit{A. Juhász} [Geom. Topol. 12, No. 1, 299--350 (2008; Zbl 1167.57005)]. It is also mentioned that \(\widehat{HFK}(K)\) detects the Alexander-Conway polynomial of \(K\): \(\Delta_K(t) = \sum\limits_j t^j \sum\limits_i (-1)^i \mathrm{rank} (\widehat{HFK}_i(K,j))\). Moreover, \(\widehat{HFK}\) detects fibered knots: \(S^3 \setminus K\) is fibered over \(S^1\) if and only if \(\sum\limits_i \dim \widehat{HFK}_i(K,g(K)) = 1\) (Theorem 4.6). Finally, for a closed \(3\)-manifold \(Y\), \(\widehat{HF}(Y)\) decomposes over the possible \(\mathrm{spin}^c\)-structures \(\mathfrak{s}\) on \(Y\), and with that it detects the Thurston norm \(x: H_2Y \to {\mathbb Z}\) by \(x(h) = \max\{\langle c_1(\mathfrak{s}),h \rangle \mid \widehat{HF}(Y,\mathfrak{s}) \neq 0\}\) (Theorem 4.5). Here, \(x(h) := \min \{x(F) \mid F \mathrm{ an embedded \,\! surface \,\! representing } h \}\) where \(x(F) := \sum_i |\chi(F_i)|\), \(F_i\) the components of \(F\) with \(\chi(F_i) < 0\). (The Thurston norm is important for foliations of \(3\)-manifolds.) \(\widehat{HF}\) and \(\widehat{HFK}\) are defined as sutured Floer homology \(SFH(\mathcal{H})\) where \(\mathcal{H}\) is a sutured Heegaard diagram of the spaces \(Y \setminus {\mathbb D}^3\) and \(Y \setminus \mathrm{nbd } \!(K)\), resp., with additional data. D. Gabai's sutured \(3\)-manifolds and their Heegaard diagrams are explained in detail in the text. The data of \(\mathcal{H} = (\Sigma, \alpha_1,\ldots,\alpha_n,\beta_1,\ldots ,\beta_n)\) allow a model for the \(3\)-manifold in question. \(\Sigma\) is a compact oriented surface with non-empty boundary and the \(\alpha_i\) and \(\beta_j\) are (certain) closed curves in its interior, in general position, and both the \(\alpha_i\) and the \(\beta_j\) are supposed to be mutually disjoint sets. The \(3\)-manifold is obtained by thickening \(\Sigma\) and capping the \(\alpha_i\) to one side and the \(\beta_j\) to the other side by \(2\)-disks and subsequently thickening these, too. The appropriate chain complex \(SFH(\mathcal{H})\) is constructed as follows. \(\mathrm{Sym}^n \! (\Sigma)\) is the symmetric product, i.e., the orbit space of the natural action of the symmetric group on the \(n\)-fold product of \(\Sigma\). Let \(T_\alpha = \alpha _1 \times \cdot\cdot\cdot \times \alpha_n\), \(T_\beta = \beta_1 \times \cdot\cdot\cdot \times \beta_n\) be the images in \(\mathrm{Sym}^n \! (\Sigma)\). The chain groups are freely generated over \({\mathbb Z}_2\) by the points of \(T_\alpha \cap T_\beta\). For the differential \(\partial\), symplectic geometry comes in. \(\mathrm{Sym}^n \! (\Sigma)\) has a natural almost complex structure. For generators \(\mathbf{x,y} \in T_\alpha \cap T_\beta\), let \(\mathcal{M}(\mathbf{x,y})\) consist of (certain) holomorphic disks \({\mathbb D}^2 \to \mathrm{Sym}^n \! (\Sigma)\), with a natural action of \({\mathbb R}\), namely \(1\)-parameter conformal deformations of \({\mathbb D}^2\) fixing \(\pm i\). Then one defines \(\partial (\mathbf{x}) = \sum\limits_\mathbf{y} (\sharp (\mathcal{M}(\mathbf{x,y}) / {\mathbb R})) \mathbf{y}\), where one disregards infinite summands. (This is called ``counting holomorphic disks''.) Gabai's sutured manifold theory is employed in showing that \(\partial\) is a proper differential and that \(SFH(\mathcal{H})\) is well-defined, the Lagrangian Floer homology of \(T_\alpha\), \(T_\beta\) in \(\mathrm{Sym}^n \! (\Sigma)\). A few examples are displayed. A final section discusses spectral sequence connections between reduced Khovanov homology and Heegaard Floer homology.