---
title: T-Manifolds in Geometry and Physics
url: https://www.emergentmind.com/topics/t-manifolds
type: topic
---

# T-Manifolds in Geometry and Physics

In the literature represented here, the expression **T-manifold** is context-dependent rather than uniform. In equivariant topology it often means a smooth manifold equipped with an effective torus action; in Hamiltonian geometry it typically means a symplectic or Kähler manifold with an effective Hamiltonian torus action and moment map; in Hamiltonian dynamics it can denote local manifolds of \(T\)-relative equilibria; and in string theory it may refer to non-geometric torus fibrations, often called T-folds when T-duality is the only duality used. A further doubled-geometric usage identifies the total space of \(TM\oplus T^*M\) as a big-tangent manifold endowed with canonical tensorial structures. These usages are related by their reliance on torus symmetry, torus fibrations, or \(T\)-duality, but they are not equivalent notions [1909.00994] [2301.01011] [1803.00550] [1303.0658].

## 1. Core meanings and terminological range

In equivariant topology, a \(T\)-manifold means a smooth manifold \(M\) equipped with an **effective** smooth action of a torus \(T\) [1909.00994]. In the closely related language of locally standard \(T^k\)-manifolds, a smooth \(T^k\)-manifold is again a smooth manifold with a smooth effective \(T^k\)-action [2011.10460]. In Hamiltonian Kähler geometry, the basic setting is a compact Kähler manifold \((M,\omega,J)\) with an effective Hamiltonian \(T\)-action by isometries and a moment map \(\mu:M\to\mathfrak t^*\) [2301.01011]. In string theory, by contrast, T-manifolds are often called T-folds and are defined by allowing transition functions in the T-duality group \(O(d,d;\mathbb Z)\) rather than only geometric transition functions [1803.00550].

| Context | Meaning of “T-manifold” | Hallmark data |
|---|---|---|
| Equivariant topology | Smooth manifold with effective torus action | \(T\curvearrowright M\) [1909.00994] |
| Hamiltonian/Kähler geometry | Compact Kähler or symplectic manifold with Hamiltonian torus action | \((M,\omega,J,\mu)\) [2301.01011] |
| Hamiltonian dynamics | Local manifold of \(T\)-relative equilibria | \(V_0=\ker d^2(h-\mathbf J^\xi)(0)\) [2011.04350] |
| String theory | T-fold / non-geometric torus-fibered background | \(O(d,d;\mathbb Z)\)-valued monodromy [1803.00550] |
| Generalized geometry | Big-tangent manifold \(\mathfrak TM=TM\oplus T^*M\) | canonical \((P,Q,S)\) [1303.0658] |

This range already indicates that the letter \(T\) may denote a torus, a maximal torus, the T-duality group, or the doubled tangent object \(T\oplus T^*\). Accordingly, the term is best read relative to its research program.

## 2. Hamiltonian and Kähler \(T\)-manifolds

A Hamiltonian \(T\)-manifold in the Kähler setting consists of a compact Kähler manifold \((M,\omega,J)\), an effective Hamiltonian torus action
\[
\rho:T\to \mathrm{Diff}(M,\omega,J),
\]
and a moment map \(\mu:M\to\mathfrak t^*\) satisfying
\[
\iota_{\xi^\#}\omega=-d\langle\mu,\xi\rangle .
\]
If \((L,\nabla,h)\) is a \(T\)-invariant pre-quantum line bundle with \(F_\nabla=-i\,\omega\), the induced infinitesimal action on sections is
\[
\xi\cdot s=\nabla_{\xi^\#}s+i\,\mu_\xi s,\qquad \mu_\xi=\langle\mu,\xi\rangle .
\]
This is the basic Hamiltonian \(T\)-manifold framework used by Leung–Wang [2301.01011].

A central construction in that setting is the mixed polarization
\[
\mathcal P_{\mathrm{mix}}=(\mathcal P_J\cap \mathcal D_{\mathbb C})\oplus \mathcal I_{\mathbb C},
\]
where \(\mathcal P_J=T^{0,1}M\), \(\mathcal I_{\mathbb R}=\mathrm{Im}\,d\rho\), and \(\mathcal D_{\mathbb R}=\ker d\mu\). On toric manifolds, where \(n=m\), this reduces to the singular real polarization generated by the torus fibers of the moment map; for \(1\le n<m\), it is genuinely mixed [2301.01011]. The associated quantum space is defined distributionally by
\[
\mathcal H_P=\{\delta\in\Gamma'(M,L^{-1})\mid \nabla_X\delta=0\text{ for all }X\in\Gamma(M,P)\},
\]
and for \(P=\mathcal P_{\mathrm{mix}}\) one obtains a decomposition
\[
\mathcal H_{\mathrm{mix}}=\bigoplus_{\lambda\in\mathfrak t^*_{\mathbb Z}}\mathcal H_{\mathrm{mix},\lambda},
\]
with supports inside \(\mu^{-1}(\mathfrak t^*_{\mathbb Z})\). For regular integral \(\lambda\),
\[
\mathcal H_{\mathrm{mix},\lambda}\cong H^0(M//_\lambda T,L//_\lambda T),
\]
so geometric quantization commutes with symplectic reduction in this mixed-polarization picture [2301.01011].

A second major Hamiltonian usage appears in complexity one geometry. A Hamiltonian \(T\)-manifold \((M,\omega,\phi)\) has complexity one if
\[
\frac{1}{2}\dim M-\dim T=1.
\]
For compact, connected, Kähler complexity one Hamiltonian \(T\)-manifolds, every painting is trivial; as a corollary, two tall compact, connected Kähler complexity one Hamiltonian \(T\)-manifolds are symplectomorphic exactly if they have the same genus, Duistermaat–Heckman measure, and skeleton [2603.12404]. This places a strong rigidity constraint on the Kähler subclass inside the broader Karshon–Tolman framework.

## 3. Topological and toric-topological frameworks

In equivariant topology, a broad class of \(T\)-manifolds is provided by **locally \(k\)-standard \(T^{n+k}\)-manifolds**. These are \((2n+k)\)-dimensional smooth manifolds with effective \(T^{n+k}\)-action that are locally modeled on \(\mathbb C^n\times T^k\) with the \(k\)-standard action. Their orbit spaces are manifolds with corners of dimension \(n\), and when the orbit space is a simple polytope \(P\), the action is encoded by a hyper characteristic function
\[
\xi:\mathcal F(P)\to\mathbb Z^{n+k}
\]
satisfying the direct-summand condition at each vertex [1909.00994]. The associated model space is
\[
M(P,\xi):=(T^{n+k}\times P)/\sim,
\]
and every locally \(k\)-standard \(T^{n+k}\)-manifold with simple polytope orbit is equivariantly homeomorphic to some \(M(P,\Lambda)\) [1909.00994].

This framework includes quasitoric manifolds (\(k=0\)), moment-angle manifolds, good contact toric manifolds (\(k=1\)), and hyperplane cuts of quasitoric manifolds [1909.00994]. Under maximal-rank hypotheses, the equivariant cohomology ring is
\[
H_T^*(M)\cong \mathrm{SR}(P),
\]
and the \(H^*(BT)\)-algebra structure recovers the hyper characteristic vectors \(\xi_j\). Sarkar and Song prove an equivariant cohomological rigidity theorem: under the stated direct-summand condition, weak equivariant homeomorphism is equivalent to weak isomorphism of equivariant cohomology algebras as \(H^*(BT)\)-algebras [1909.00994].

A related smooth-classification result is due to Wiemeler for locally standard \(T^k\)-manifolds with a section of the orbit map \(\pi:M\to M/T^k\). In this setting, \(M/T^k\) is a smooth manifold with corners, the isotropy data define a characteristic function \(\lambda\) on faces, and the pair \((P,\lambda)\) determines the \(T^k\)-manifold up to equivariant diffeomorphism once a section exists [2011.10460].

Buchstaber–Terzić introduce a further axiomatic package for \((2n,k)\)-manifolds: a smooth compact \(2n\)-manifold with effective \(T^k\)-action, an open almost moment map \(\mu:M^{2n}\to\mathbb R^k\) whose image is a convex polytope \(P^k\), and six axioms governing strata, admissible polytopes, parameter spaces, and gluing maps. The complexity is
\[
d=n-k.
\]
From these data they construct a model space \(\mathfrak E\) with a \(T^k\)-action and a \(T^k\)-equivariant homeomorphism \(\mathfrak E\to M^{2n}\), inducing \(\mathfrak E/T^k\cong M^{2n}/T^k\). Their theory contains toric geometry and toric topology when \(d=0\), and also covers positive-complexity examples such as complex Grassmann manifolds \(G_{k+1,q}\) [1803.05766].

## 4. \(T\)-manifolds of relative equilibria

In equivariant Hamiltonian dynamics, the phrase denotes a different object. Let \(G\) be a compact connected Lie group, \(T\subset G\) a maximal torus, and \(V\) a symplectic \(G\)-representation modeling a neighborhood of a completely symmetric equilibrium. For \(\xi\in\mathfrak t\), define
\[
V_0(\xi):=\ker d^2(h-\mathbf J^\xi)(0)\subset V^c.
\]
Under the generic conditions denoted (GC) and (NR′), and when the weights in \(V_0\) are linearly independent and satisfy the maximality condition of Theorem 2.15, there exists a local \(T\)-invariant manifold of \(T\)-relative equilibria tangent to \(V_0\) at the origin [2011.04350].

The main result of that paper is an isotropy-preserving normal form: for sufficiently small \(U\subset V_0\), the map
\[
m_{V_0}:U\to V
\]
defines a manifold \(M_0=m_{V_0}(U)\) of \(T\)-relative equilibria with \(T_0M_0=V_0\) and
\[
G_{m_{V_0}(x)}=G_x\qquad \text{for all }x\in U.
\]
Thus the local \(G\)-isotropy structure of the nonlinear manifold is modeled exactly by the linear \(G\)-action on \(V_0\) [2011.04350].

The \(G\)-orbit of the union of these \(T\)-manifolds is Whitney-stratified by isotropy type. For isotropy type \((H)\), the corresponding stratum has dimension
\[
\dim G-\dim H+\dim(\mathfrak t')^L,
\]
where \(\mathfrak t'\subset\mathfrak t\) is the orthogonal complement of \(\mathfrak h\cap\mathfrak t\) and \(L\) is the minimal adjoint isotropy subgroup of an element of \(\mathfrak t\) containing \(H\) [2011.04350]. This usage is local, representation-theoretic, and bifurcation-theoretic rather than global and topological.

## 5. T-duality, T-folds, and doubled geometry

In string theory, T-manifolds are often called **T-folds** when T-duality is the only duality used. They are torus-fibered backgrounds whose transition functions are allowed to lie in the T-duality group
\[
\mathcal G_d=O(d,d;\mathbb Z)
\]
rather than only in the mapping class group of the torus fiber [1803.00550]. For \(d=3\), the relevant group is \(O(3,3;\mathbb Z)\), and the paper studies \(T^3\)-fibered T-folds together with the map
\[
SL(4;\mathbb Z)\to SO(3,3;\mathbb Z)^+.
\]
This realizes certain \(T^3\)-fibered T-folds as geometric \(T^4\)-fibrations in a dual picture. The monodromies around duality defects split into geometric diffeomorphisms, \(B\)-shifts, and genuinely non-geometric \(\beta\)-transformations [1803.00550].

A dg-geometric formulation of T-duality is developed by Lupercio–Rengifo–Uribe. They consider dg-manifolds that are \(\mathbb R[2]\)-bundles over \(\mathbb R[1]\)-bundles over manifolds, define T-dual dg-manifolds, construct the T-duality map as a degree \(-1\) map between the cohomologies of T-dual dg-manifolds, and prove an explicit isomorphism between the differential graded algebras of their symmetries [1208.6048]. In the self T-dual case, the derived dg-Leibniz algebra of the fixed points of the T-dual automorphism recovers the algebraic structure underlying \(B_n\) generalized geometry, and higher dg-manifolds similarly yield structures associated with exceptional generalized geometry [1208.6048].

A related doubled-geometric construction is Vaisman’s **big-tangent manifold**
\[
\mathfrak TM:=TM\oplus T^*M,
\]
the total space of the bundle \(TM\oplus T^*M\) over \(M\) [1303.0658]. Its vertical leaves are para-Hermitian vector spaces, and \(\mathfrak TM\) carries canonical tensor fields
\[
P=\frac{\partial}{\partial y^i}\wedge \frac{\partial}{\partial z_i},\qquad
Q=\frac{\partial}{\partial y^i}\odot \frac{\partial}{\partial z_i},\qquad
S=dx^i\otimes \frac{\partial}{\partial y^i},
\]
together with a canonical presymplectic form. From the viewpoint of \(G\)-structures, big-tangent geometry is equivalent to a suitable triple \((P,Q,S)\) satisfying \([P,P]=0\), \(\mathcal N_S=0\), and compatibility conditions linking the three tensors. Vaisman then defines horizontal bundles, compatible vertical metrics, canonical double-metric connections, and an action functional for such double fields [1303.0658]. In this sense, the term points not to torus actions but to doubled tangent–cotangent geometry.

## 6. Torus-bundle special holonomy manifolds

A broader torus-geometric usage arises in special holonomy. The paper on complete non-compact \(\mathrm{Spin}(7)\)-manifolds constructs torsion-free \(\mathrm{Spin}(7)\)-structures on the total spaces of principal \(T^2\)-bundles over asymptotically conical Calabi–Yau 3-folds [2407.19486]. The central ansatz is a \(T^2\)-invariant 4-form
\[
\Phi=\eta\wedge\theta\wedge\omega+\eta\wedge p\,\mathrm{Re}\,\Omega-\theta\wedge(r\,\mathrm{Re}\,\Omega+q\,\mathrm{Im}\,\Omega)+\tfrac12\,pq\,\omega^2,
\]
where \((\eta,\theta)\) is a principal \(T^2\)-connection and \((\omega,\Omega)\) is an \(SU(3)\)-structure on the base. The resulting metrics are complete, Ricci-flat, and asymptotically \(T^2\)-fibred conical, abbreviated \(AT^2C\) [2407.19486].

This construction produces infinitely many diffeomorphism types of \(AT^2C\) \(\mathrm{Spin}(7)\)-manifolds and, notably, the first known examples of complete toric \(\mathrm{Spin}(7)\)-manifolds [2407.19486]. Here the torus is not merely an acting symmetry group; it is built into the geometry both as a principal bundle and, in the toric examples, as an effective multi-Hamiltonian \(T^4\)-action.

Taken together, these usages show that **T-manifold** is a family resemblance term rather than a single definition. Its common thread is the structural role of a torus or of \(T\)-duality: as an acting compact group, as a reduction datum in Hamiltonian geometry, as a source of local bifurcation manifolds, as a non-geometric transition group in string theory, or as the doubled \(T\oplus T^*\) object of generalized geometry.

Source: https://www.emergentmind.com/topics/t-manifolds