Higher Teichmüller Theory: Frontiers¶

Job: survey the research frontiers of higher Teichmüller theory: positivity and Fock–Goncharov cluster coordinates, the Anosov / Morse program into general Lie groups, relatively Anosov representations, the parametrization of \((G,X)\)-structures on closed manifolds, dynamics on character varieties (entropy, pressure metric), and the link to the Langlands program and the Hitchin system.

Core¶

Positivity and Fock–Goncharov cluster coordinates¶

For \(G=\mathrm{PSL}_n(\mathbb{R})\) the Hitchin component \(\mathrm{Hit}_n(\Sigma)\) admits, beyond the Hitchin-section parametrization, a positive structure: Fock–Goncharov construct coordinates that identify \(\mathrm{Hit}_n(\Sigma)\) with a component of a positive moduli space of framed local systems, with the transition maps between coordinate charts given by cluster mutations (of a cluster algebra of type \(\mathrm{PSL}_n\)). Concretely, for a triangulation of \(\Sigma\) and a Hitchin representation with its Frenet boundary map \(\xi:\partial\pi_1\Sigma\to\mathbb{RP}^{n-1}\), the cross-ratio invariants of triples / quadruples of boundary points give positive (totally positive matrix) coordinates — the "golden ratio" / Fock–Goncharov coordinates.

Object (Fock–Goncharov)	Structure
\(\mathcal{A}\)-moduli \(\mathcal{A}_{G,\Sigma}\)	framed local systems; cluster \(\mathcal{A}\)-coordinates; mutation-invariant Poisson
\(\mathcal{X}\)-moduli \(\mathcal{X}_{G,\Sigma}\)	framed local systems up to frame gauge; cluster \(\mathcal{X}\)-coordinates
Positive locus \(\mathcal{X}_{G,\Sigma}^+(\mathbb{R})\)	connected component cut out by positivity; for \(\mathrm{PSL}_n\) isomorphic to \(\mathrm{Hit}_n(\Sigma)\times\mathrm{flag}\)-frame
Cluster algebra	type \(\mathrm{PSL}_n\) (or the Langlands dual \(\mathrm{PGL}_n\)); seed = triangulation; mutations = flips

Theorem (Fock–Goncharov). For split real \(G\) (with dual \({}^LG\)), the positive component \(\mathcal{X}_{G,\Sigma}^+(\mathbb{R})\) of the higher Teichmüller \(\mathcal{X}\)-moduli is a cluster variety, and for \(G=\mathrm{PSL}_n(\mathbb{R})\) it identifies with the Hitchin component (with a choice of boundary frame). The positivity condition is stronger than being Hitchin: for \(\mathrm{SL}_n\), a representation is Fock–Goncharov positive iff it is in the Hitchin component and the boundary curve is positive in the Lusztig sense.

The cluster structure gives a canonical quantization (Bridgeland cluster variety \(\mathcal{X}\)-quantization) and connects higher Teichmüller theory to the theory of cluster algebras, quantum groups, and the dualities expected from the Langlands program.

The Anosov / Morse program into general Lie groups¶

The Guichard–Wienhard program (and collaborators: Labourie, Minski, Kapovich–Leeb–Porti) extends the Anosov framework to:

Uniformly Morse representations (Kapovich–Leeb–Porti; Schrader–Wyatt): a representation \(\rho:\Gamma\to G\) is uniformly Morse if \(\rho\)-images of geodesics in \(\Gamma\) are Morse quasi-geodesics in \(G/K\) with uniform Morse constants; this is parabolic-independent and equivalent to being \(P\)-Anosov for some \(P\) in many cases.
Anosov into general \(G\): characterize, for each semisimple \(G\), which \(P\) give meaningful Anosov conditions and which components of \(\mathrm{Rep}(\Gamma,G)\) are Anosov; the classification is open for exceptional \(G\) and for \(G\) of higher real rank.
Stability under deformations: the structural stability theorem (Guichard–Wienhard) holds in this generality, but the geometry of the Anosov locus (its boundary, its compactification) is an active area.

Relatively Anosov representations¶

For \(\Gamma\) relatively hyperbolic (with peripheral subgroups \(\{H_1,\dots,H_m\}\)), one defines relatively \(P\)-Anosov representations \(\rho:\Gamma\to G\) by requiring the Anosov boundary map on the Bowditch boundary \(\partial\Gamma\) (which agrees with \(\partial_{\mathrm{Gromov}}\Gamma\) outside the parabolic fixed points), with no Anosov condition imposed on the peripheral directions. This relaxes the word-hyperbolic hypothesis and covers:

Free products \(\Gamma=H_1*\cdots*H_m\) (relatively hyperbolic);
Geometrically finite Kleinian groups (relatively hyperbolic, the \(H_i\) are the cusp stabilizers);
Mapping class groups of surfaces (relatively hyperbolic relative to Dehn twists — conjecturally).

Relatively Anosov representations are discrete, structurally stable, and form an open subset of \(\mathrm{Rep}(\Gamma,G)\); they are the natural setting for representations of groups that are not themselves hyperbolic but are hyperbolic relative to a well-understood peripheral structure.

\((G,X)\)-structures and the parametrization theorem¶

A \((G,X)\)-structure on a closed manifold \(M\) is an atlas with charts in \(X\) and transition maps in \(G\); the holonomy gives a representation \(\rho:\pi_1 M\to G\). Guichard–Wienhard show that for appropriate \(G\) and \(X\) (e.g. \(G=\mathrm{PSL}_n(\mathbb{R})\), \(X=\mathbb{RP}^{n-1}\) for convex projective structures; \(G=\mathrm{SO}_0(n,1)\), \(X=\mathbb{H}^n\) for hyperbolic structures), the moduli of \((G,X)\)-structures on \(M\) is parametrized (up to isotopy) by an open subset of \(\mathrm{Rep}(\pi_1 M,G)\) consisting of Anosov representations. This generalizes the classical parametrization of hyperbolic structures by convex cocompact Kleinian representations.

\(X\)	\(G\)	\((G,X)\)-structures	Parametrizing reps
\(\mathbb{H}^n\)	\(\mathrm{SO}_0(n,1)\)	hyperbolic structures	convex cocompact (= \(B\)-Anosov)
\(\mathbb{RP}^{n-1}\) (convex)	\(\mathrm{PSL}_n(\mathbb{R})\)	convex projective structures	\(P_1\)-Anosov + convex hull; Benoist
properly convex domains	\(\mathrm{PO}(p,q)\)	convex divisible domains	\(P\)-Anosov for appropriate \(P\)
flag structures	\(\mathrm{PSL}_n(\mathbb{R})\)	higher-rank geometric structures	Hitchin / Anosov

Dynamics on character varieties: entropy and the pressure metric¶

An Anosov representation \(\rho:\Gamma\to G\) has a critical exponent \(h(\rho)\): the exponential growth rate of \(\rho(\Gamma)\) in \(G/K\),

\[ h(\rho):=\lim_{T\to\infty}\frac{1}{T}\log\#\{\gamma\in\Gamma\mid d_{G/K}(o,\rho(\gamma)o)\le T\}. \]

For the Hitchin component, Bridgeman–Canary–Labourie–Sambarino construct a pressure metric \(P\) on \(\mathrm{Hit}_n(\Sigma)\) from the thermodynamic formalism (the variational principle applied to the geodesic flow): the pressure form is a non-negative 2-form whose zero directions are exactly the conformal deformations along the Fuchsian locus, and which restricts to a Riemannian metric on the Hitchin component modulo the Fuchsian directions. The entropy / critical exponent is the central dynamical invariant; the pressure metric generalizes the Weil–Petersson metric on Teichmüller space.

Connection to the Langlands program and the Hitchin system¶

The Higgs-bundle branch of higher Teichmüller theory identifies \(\mathrm{Hit}_n(\Sigma)\) with a component of the Hitchin system (the integrable system \(\mathcal{M}_{\mathrm{Higgs}}\to\bigoplus H^0(K^{\otimes k})\)). The generic fiber is the Prym variety of the spectral cover; the Hitchin system is the archetype of an algebraic integrable system and is the geometric side of:

Geometric Langlands (for curves over \(\mathbb{C}\)): the Hitchin system is the moduli of Higgs bundles and is mirror (in the sense of SYZ / Strominger–Yau–Zaslow) to the moduli of flat connections, the spectral decomposition side. Higher Teichmüller components are real slices of this complex integrable system.
Mirror symmetry: the Langlands dual group \({}^LG\) controls the cluster structure (Fock–Goncharov \(\mathcal{X}\)-moduli is a cluster variety of type \({}^LG\)), and the duality expected on the Langlands side appears as cluster duality on the higher Teichmüller side.

This makes higher Teichmüller theory a meeting point of Lie theory, dynamics, Higgs-bundle geometry, and the Langlands program.

See it¶

Fock–Goncharov positive coordinates on \(\mathrm{Hit}_3(\Sigma)\). For \(G=\mathrm{PSL}_3(\mathbb{R})\) and a triangulation \(\tau\) of \(\Sigma\), a Hitchin representation with its Frenet boundary map \(\xi:\partial\pi_1\Sigma\to\mathbb{RP}^2\) assigns to each edge \(e=[a,b]\) of \(\tau\) (with \(a,b\in\partial\pi_1\Sigma\)) a cross-ratio \(X_e\) built from the triple / quadruple ratios of \(\xi\) along the boundary — concretely, the \(\mathcal{X}\)-coordinate of \(e\) is the cluster coordinate obtained from the configuration \((\xi(a),\xi(b),\xi(c))\) at a triangle with vertices \(a,b,c\). Positivity says all these cross-ratios are \(>0\) (the matrices of transition between flags are totally positive in Lusztig's sense), and the cluster mutation under a flip of \(\tau\) is the standard cluster mutation of type \(\widetilde A_1\) (the golden ratio transformation \(x\mapsto(1+x)/x\) appears as the simplest mutation). The cluster variety \(\mathcal{X}_{\mathrm{PSL}_3,\Sigma}^+(\mathbb{R})\) is then identified with \(\mathrm{Hit}_3(\Sigma)\times(\mathbb{R}_{>0})^{\text{flags}}\), and the positive coordinates give a global parametrization of the Hitchin component by positive real numbers indexed by the edges of any triangulation, with mutation-invariant transition — a "golden ratio atlas" on \(\mathrm{Hit}_3(\Sigma)\). This is the prototypical Fock–Goncharov computation: a Hitchin representation is detected by the positivity of its boundary cross-ratios, and the coordinates form a cluster variety of Langlands-dual type.

Check¶

Explain why Fock–Goncharov positivity is stronger than being Hitchin for \(\mathrm{SL}_n(\mathbb{R})\), and identify the extra condition on the boundary curve.

Every Fock–Goncharov positive representation is Hitchin (Fock–Goncharov), but the converse requires the boundary curve \(\xi:\partial\pi_1\Sigma\to\mathbb{RP}^{n-1}\) to be positive in Lusztig's sense: for any cyclically ordered tuple \((x_1,\dots,x_N)\) of boundary points, the transition matrices between consecutive flags are totally positive (in the appropriate Lusztig total positivity). A general Hitchin representation has a Frenet boundary curve (transverse, dynamics-preserving) but the transition matrices need not be totally positive — they can have mixed signs while preserving transversality. Hence the positive locus is an open subset (indeed a connected component) of the Hitchin component with the positivity condition imposed on the framing, and for \(n=2\) positivity is automatic (the cross-ratios of four cyclically ordered points on \(\mathbb{RP}^1\) are always positive for a Fuchsian representation), so the two notions coincide; for \(n\ge 3\) positivity is a genuine extra condition. The clean statement: positivity = Hitchin + total positivity of the boundary framing; it is the cluster-algebraic shadow of the Hitchin component.

State in one line why the pressure metric degenerates along the Fuchsian locus, and why this is the higher-rank analogue of the Weil–Petersson degeneration along the Fuchsian slice.

The pressure form \(P\) on \(\mathrm{Hit}_n(\Sigma)\) is built from the Hessian of the pressure / entropy functional \(h(\rho)\); along the Fuchsian locus the entropy is the topological entropy of the Fuchsian geodesic flow, which is constant under conformal deformations (changing the conformal structure on \(\Sigma\) does not change the hyperbolic volume / topological entropy), so \(P\) vanishes on conformal directions — exactly as the Weil–Petersson metric on \(\mathcal{T}(\Sigma)\) has the conformal directions as the non-degenerate part and degenerates along the holomorphic-differential deformations (the higher \(q_k\)). The pressure metric is the higher-rank analogue of Weil–Petersson, and the Fuchsian slice is the locus where the two agree; in rank \(n\ge 3\) the extra Hitchin directions (\(q_3,\dots,q_n\)) acquire non-degenerate pressure length, giving a genuinely higher-rank Riemannian metric on \(\mathrm{Hit}_n(\Sigma)\) modulo the Fuchsian conformal directions.

Pitfalls¶

Positivity (Fock–Goncharov) \(\neq\) Hitchin for \(n\ge 3\): positivity is the extra total-positivity condition on the boundary framing; Hitchin alone gives Frenet but not positivity. Confusing the two gives wrong cluster-coordinate formulas.
Anosov is open, not closed: the Anosov locus is open in \(\mathrm{Rep}(\Gamma,G)\) but not closed; the boundary of a higher Teichmüller component contains non-Anosov reps, and the pressure metric / entropy degenerates there. Compactification of the Hitchin component is via non-Anosov limits (the "boundary" of the Anosov locus), not by Anosov reps.
Hitchin components require split real \(G\); non-split real forms have no Hitchin component in this sense — one uses maximal components (Hermitian) or other paradigms (e.g. positive representations for non-split but "positive" \(G\) in the sense of Fock–Goncharov, which is a finer condition).
Maximal \(\neq\) Hitchin: the two paradigms give genuinely different higher Teichmüller spaces even when both apply (e.g. \(\mathrm{Sp}(2n,\mathbb{R})\), \(n\ge 2\)); see Hitchin & Maximal.
Relative Anosov needs relatively hyperbolic \(\Gamma\), not arbitrary: applying the relative Anosov definitions to a group without a specified peripheral structure / relative hyperbolicity is a category error; the Bowditch boundary is not defined without the peripheral data.
Pressure metric is defined on the Hitchin component, not on all Anosov reps: the Bridgeman–Canary–Labourie–Sambarino construction uses the thermodynamic formalism of the geodesic flow, which requires the Anosov representation to be into a split real \(G\) with a uniform spectral gap; the construction for general Anosov (e.g. into Hermitian \(G\)) is subtler and partly open.

Refs¶

Fock–Goncharov, Moduli spaces of local systems and higher Teichmüller theory (Publ. Math. IHÉS 2006)
Guichard–Wienhard, Anosov representations: characterizations and applications; Convexity of generalized cusps in rank-one symmetric spaces (\(G,X\)-structures)
Kapovich–Leeb–Porti, Morse actions of word-hyperbolic groups on symmetric spaces; Schrader–Wyatt, Anosov representations and the Morse property
Bridgeman–Canary–Labourie–Sambarino, The pressure metric for Anosov representations (pressure metric, entropy)
Benoist, Convexes divisibles (convex projective structures / \(P_1\)-Anosov)

Next¶

Higher Teichmüller index · Foundations · Anosov Representations · Hitchin & Maximal Components · Representation theory · Complex geometry / Kähler · Differential geometry · Algebraic geometry · Geometric Langlands — the Hitchin system / non-abelian Hodge bridge to opers and eigensheaves · Hitchin & GL mirror — the Hitchin fibration mirror symmetry / the Fock–Goncharov cluster · TG / scattering & mirror — the cluster structure of the scattering / the theta functions