Fact-checked by Grok 2 weeks ago

Cotangent bundle

In differential geometry, the cotangent bundle of a smooth manifold M is the vector bundle T^*M = \bigcup_{p \in M} T^*_p M, where each fiber T^*_p M over a point p \in M is the cotangent space, defined as the dual vector space \operatorname{Hom}(T_p M, \mathbb{R}) to the tangent space T_p M.^[1]^[2]^[3] This structure equips T^*M with a natural projection map \pi: T^*M \to M sending each covector to its base point, and locally, it trivializes as U \times \mathbb{R}^n for coordinate charts on M, making T^*M a smooth manifold of dimension $2n when \dim M = n.^[1]^[3] Sections of the cotangent bundle, known as differential 1-forms, assign to each point p \in M a linear functional on T_p M, enabling the study of directional derivatives and pullbacks under smooth maps.^[2]^[3] For instance, the differential df of a smooth function f: M \to \mathbb{R} is a canonical 1-form section of T^*M.^[2] A defining feature of the cotangent bundle is its canonical symplectic structure, given by the closed nondegenerate 2-form \omega = -d\theta, where \theta is the tautological 1-form on T^*M satisfying \theta_\alpha(\xi) = \alpha(T\pi(\xi)) for \alpha \in T^*_q M and \xi \in T_\alpha(T^*M).^[4]^[5] In local coordinates (q^1, \dots, q^n, p_1, \dots, p_n), this takes the standard form \omega = \sum_{i=1}^n dq^i \wedge dp_i, rendering T^*M a symplectic manifold that serves as the phase space for Hamiltonian mechanics, where positions and momenta evolve under the symplectic flow.^[4]^[5] This structure underpins applications in geometric quantization, dynamics, and the formulation of geodesic flows via Hamiltonians like H(q, p) = \frac{1}{2} \|p\|^2.^[4]^[5]

Definition and Construction

Intrinsic Definition

In differential geometry, the cotangent space at a point p on a smooth manifold M of dimension n, denoted T_p^* M, is defined as the dual vector space to the tangent space T_p M.^[6]^[7] This means T_p^* M = (T_p M)^*, the space of all continuous linear functionals from T_p M to the real numbers \mathbb{R}, often called covectors.^[6]^[7] Each element \omega \in T_p^* M pairs with a tangent vector v \in T_p M to yield a scalar \omega(v) \in \mathbb{R}, providing an intrinsic way to measure directional changes without reference to coordinates.^[6] The dimension of T_p^* M is also n, matching that of T_p M.^[7] The cotangent bundle T^* M is constructed as the disjoint union of all cotangent spaces over the manifold:

T^* M = \bigcup_{p \in M} T_p^* M.

This forms a smooth vector bundle over M with fiber dimension n, where each fiber \pi^{-1}(p) is precisely T_p^* M.^[6]^[7] As the dual counterpart to the tangent bundle TM, T^* M captures the collection of all covectors varying smoothly across M.^[6] A natural projection map \pi: T^* M \to M is defined by sending each covector to its base point: for \omega \in T_p^* M, \pi(\omega) = p.^[6]^[7] This map is smooth and surjective, endowing T^* M with the structure of a fiber bundle whose total space is a smooth manifold of dimension $2n.^[6] The smooth structure on T^* M is induced by the atlas of M. Over a coordinate chart (U, \phi) on M with \phi: U \to \mathbb{R}^n, local trivializations map open sets in T^* M diffeomorphically to U \times \mathbb{R}^n, using coordinates (x^i, \xi_i) where x^i are base coordinates and \xi_i are fiber coordinates dual to a basis of T_p M.^[6]^[7] On overlapping charts (U_\alpha, \phi_\alpha) and (U_\beta, \phi_\beta), the transition maps are of the form (x, \xi) \mapsto (y(x), \eta(y(x), \xi)), where y = \phi_\beta \circ \phi_\alpha^{-1}(x) and \eta is a linear isomorphism on the fibers given by the transpose of the Jacobian matrix of the coordinate change.^[6] These linear transition functions ensure the bundle's smoothness and compatibility across the atlas.^[7]

Coordinate-Based Construction

In local coordinates (x^1, \dots, x^n) on an open set U \subset M of an n-dimensional smooth manifold M, the cotangent space T_p^*U at a point p \in U consists of covectors expressed as \sum_{i=1}^n \xi_i \, dx^i_p, where \{\,dx^i_p\,\}_{i=1}^n is the basis dual to the coordinate basis \{\,\partial/\partial x^i|_p\,\}_{i=1}^n of the tangent space T_pU, satisfying dx^i_p(\partial/\partial x^j|_p) = \delta^i_j.^[8] This representation arises from the intrinsic dual space structure of the cotangent space, adapted to the local frame induced by the chart.^[9] The cotangent bundle T^*M admits a local trivialization over U as the product bundle U \times \mathbb{R}^n, with the projection \pi: T^*M \to M mapping (p, \xi) \in \pi^{-1}(U) to p \in U.^[8] In these coordinates, points in the trivialization are denoted (x^i, \xi_i), where x^i(p) are the base coordinates and \xi_i are the components of the covector relative to the dual basis \{[dx](/page/DX)^i\}.^[9] This trivialization equips T^*M|_U with the standard vector bundle structure over U, where fiberwise addition and scalar multiplication are componentwise in \mathbb{R}^n.^[8] To ensure the construction is independent of the choice of charts, consider overlapping coordinate charts (U_\alpha, x_\alpha) and (U_\beta, x_\beta) on M. The transition functions for the base manifold are given by x_\beta^j = x_\beta^j(x_\alpha^1, \dots, x_\alpha^n), assuming smooth diffeomorphisms between the images.^[8] For the cotangent bundle, the fiber coordinates transform covariantly via

\xi_{\beta j} = \sum_{k=1}^n \frac{\partial x_\alpha^k}{\partial x_\beta^j} \xi_{\alpha k},

where the partial derivatives are evaluated at the point corresponding to x_\alpha.^[9] This law follows from the dual nature of covectors under the pushforward of tangent vectors and ensures that the identification of fibers over overlapping sets is linear and well-defined.^[8] The smoothness of the vector bundle structure on T^*M is verified by the fact that the transition functions are smooth maps from open subsets of \mathbb{R}^n \times \mathbb{R}^n to \mathrm{GL}(n, \mathbb{R}), as the partial derivatives \partial x_\alpha^k / \partial x_\beta^j are smooth functions of the coordinates by the smoothness of the chart transition diffeomorphisms.^[8] Thus, T^*M inherits a smooth atlas from the manifold M, confirming it as a smooth vector bundle of rank n.^[9]

Global Definition via Diagonal Morphism

The cotangent sheaf \Omega^1_M on a smooth manifold M is defined as the sheaf of Kähler differentials, providing a foundation for a global, coordinate-free construction of the cotangent bundle T^*M.^[10] A key global construction utilizes the diagonal morphism \Delta: M \to M \times M, which embeds M as the diagonal submanifold in the product space. Let \mathcal{I} denote the ideal sheaf of the diagonal embedding in M \times M, generated by functions vanishing on the diagonal. The cotangent sheaf \Omega^1_M is then isomorphic to the conormal sheaf \mathcal{I}/\mathcal{I}^2, the quotient of \mathcal{I} by its square, viewed as a sheaf of \mathcal{O}_M-modules. The associated vector bundle T^*M is the bundle associated to this sheaf, dualizing the tangent sheaf via the exact sequence relating the normal bundle to the diagonal and the tangent bundle restricted to it: $0 \to T\Delta \to T(M \times M)|_\Delta \to N_\Delta \to 0, where the conormal fits dually. This approach emphasizes the geometric role of the first-order infinitesimal neighborhood of the diagonal in defining cotangents.^[11]^[12] The universal derivation property defines the map d: \mathcal{O}_M \to \Omega^1_M given by d f = \pi_2^* f - \pi_1^* f (with \pi_1, \pi_2: M \times M \to M the projections), which satisfies the derivation axioms d(fg) = f dg + g df and linearity over \mathcal{O}_M, and induces a surjection from \mathcal{I}/\mathcal{I}^2 onto \Omega^1_M. An inverse map, constructed via the conormal sequence $0 \to \mathcal{I}/\mathcal{I}^2 \to \Omega^1_{M \times M} \otimes_{\mathcal{O}_{M \times M}} \mathcal{O}_\Delta \to \Omega^1_M \to 0, shows the isomorphism, ensuring the construction is canonical. Locally, this matches the intrinsic dual of the tangent sheaf, yielding a smooth vector bundle of rank \dim M. Sections of T^*M over open sets are smooth 1-forms, which are \mathcal{O}_M-linear functionals on sections of the tangent sheaf.^[11] This diagonal morphism construction offers significant advantages, including natural functoriality under smooth maps f: N \to M. The pullback f^* T^*M is canonically isomorphic to T^*N, preserving the sheaf structure and ensuring compatibility with base changes and compositions of morphisms.^[11]

Properties

Contravariant Nature

The cotangent bundle exhibits contravariant functoriality with respect to smooth maps between manifolds. Given a smooth map f: M \to N between smooth manifolds M and N, the pullback operation f^* acts on sections of the cotangent bundle T^*N, inducing a bundle map f^*: T^*N \to T^*M defined fiberwise by (f^*_p \alpha)(v) = \alpha_{f(p)}(df_p (v)), where \alpha \in T^*_{f(p)} N and v \in T_p M.^[13]^[14] This construction ensures that the pullback reverses the direction of the map, mapping covectors from the target manifold back to the source. This pullback defines a contravariant functor T^*: \mathbf{Man}^{op} \to \mathbf{VectBund}, where \mathbf{Man} is the category of smooth manifolds and smooth maps, and \mathbf{VectBund} is the category of vector bundles over manifolds. For composable smooth maps g: N \to P and f: M \to N, the functoriality satisfies (g \circ f)^* = f^* \circ g^*, contrasting with the covariant nature of the tangent bundle functor, which pushes forward tangent vectors in the direction of the map.^[13]^[14] This contravariant structure arises from the dual relationship to the tangent bundle, where covectors pair with tangent vectors to yield scalars invariant under the transformation. Under diffeomorphisms, the induced pullback map is an isomorphism of vector bundles, preserving the fiberwise linear structure and the overall bundle topology of the cotangent bundle.^[13] For submersions, where df_p is surjective at each point, the pullback remains well-defined and relates covectors on the base to those on the total space, facilitating the induction of metrics or differential forms from the target manifold to the source.^[13] The contravariant nature under smooth maps aligns with the bundle's invariance under coordinate changes, where cotangent components transform via the inverse Jacobian matrix of the coordinate transition functions, reinforcing the dual transformation law that ensures consistent identification of covectors across charts.^[15] This property underscores the cotangent bundle's role in providing a canonical framework for dualizing tangent space operations globally.

Dual Relationship to Tangent Bundle

The cotangent bundle T^* M of a smooth manifold M is defined as the dual vector bundle to the tangent bundle T M, establishing a natural isomorphism T^* M \cong (T M)^* as vector bundles over M. This isomorphism arises from the fiberwise duality, where for each point p \in M, the cotangent space T_p^* M is the dual vector space (T_p M)^* = \mathrm{Hom}(T_p M, \mathbb{R}), consisting of all linear functionals on the tangent space at p. The bundle structure ensures that local trivializations of T M induce dual trivializations for T^* M via the transpose of transition maps.^[16]^[17] The duality between T^* M and T M is concretely realized by the canonical pairing

\langle \cdot, \cdot \rangle : T^* M \times_M T M \to \mathbb{R},

which restricts fiberwise to the standard evaluation map \langle \omega_p, v_p \rangle = \omega_p(v_p) for \omega_p \in T_p^* M and v_p \in T_p M. This pairing is bilinear over \mathbb{R} and non-degenerate, meaning that if \langle \omega, v \rangle = 0 for all v \in T M, then \omega = 0, and similarly for sections of T M; it thus uniquely characterizes the bundle duality and allows recovery of elements from either bundle using the other. The dual nature also underlies the contravariant pullback of covectors under smooth maps between manifolds.^[17]^[16] When M admits a Riemannian metric g, a smooth section of the bundle of symmetric bilinear forms on T M, this metric induces a vector bundle isomorphism g^\flat : T M \to T^* M defined fiberwise by (g^\flat(X_p))(Y_p) = g_p(X_p, Y_p) for tangent vectors X_p, Y_p \in T_p M. The inverse map g^\sharp : T^* M \to T M raises indices via the inverse metric tensor, satisfying g^\sharp(g^\flat(X)) = X and g^\flat(g^\sharp(\omega)) = \omega for sections X of T M and \omega of T^* M; this musical isomorphism equips the cotangent bundle with additional geometric structure compatible with g.^[18] Sections of tensor powers of the cotangent bundle encode covariant tensors on M; in particular, smooth sections of T^* M \otimes T^* M are precisely the (0,2)-tensors, which are bilinear maps T_p M \times T_p M \to \mathbb{R} varying smoothly over M. For instance, the Riemannian metric g itself is a symmetric section of this bundle, and more generally, any (0,2)-tensor field can be expressed locally in coordinates as T = T_{ij} \, dx^i \otimes dx^j, where \{dx^i\} is the dual frame to a local basis of T M. This tensorial correspondence extends the duality to higher-rank objects, facilitating contractions and index manipulations in differential geometry.^[19]^[20]

Differential Forms Interpretation

The space of smooth sections of the cotangent bundle T^*M over a smooth manifold M, denoted \Gamma(T^*M), coincides with the module of smooth differential 1-forms on M, denoted \Omega^1(M). This identification arises because each smooth section assigns to every point p \in M a cotangent vector in T_p^*M, which locally behaves as a linear functional on tangent vectors at p, extendable smoothly across M.^[21]^[22] In local coordinates (x^1, \dots, x^n) on an open subset of M, any 1-form \omega \in \Omega^1(M) takes the form

\omega = \sum_{i=1}^n \omega_i \, dx^i,

where the coefficients \omega_i are smooth real-valued functions on the coordinate domain, and \{dx^i\} form a local basis for the cotangent space dual to the coordinate tangent vectors \{\partial/\partial x^i\}. This expression highlights how 1-forms generalize the notion of differentials, providing a coordinate-independent way to pair with tangent vectors to yield smooth functions.^[23]^[22] The exterior derivative operator d: \Omega^0(M) \to \Omega^1(M), with \Omega^0(M) = C^\infty(M) the ring of smooth functions on [M](/page/M), produces exact 1-forms by mapping each f \in C^\infty(M) to its total differential

df = \sum_{i=1}^n \frac{\partial f}{\partial x^i} \, dx^i.

Exact forms form the image of this map, serving as a fundamental source of 1-forms that are locally gradients of functions.^[23]^[24] The module \Omega^1(M) is a module over C^\infty(M), meaning smooth functions act as scalars to multiply 1-forms, preserving smoothness: for f \in C^\infty(M) and \omega \in \Omega^1(M), f \omega is the pointwise product. Additionally, the wedge product equips 1-forms with an antisymmetric bilinear operation, yielding elements of the space of 2-forms, though the emphasis here remains on 1-forms themselves. These 1-forms function as derivations on smooth functions via composition with the exterior derivative, where for f \in C^\infty(M), the expression \omega(f) = \omega(df) produces a smooth function capturing a directional derivative effect in local coordinates.^[25]^[21]^[23]

Examples

Cotangent Bundle of Euclidean Space

The cotangent bundle of Euclidean space \mathbb{R}^n provides the simplest nontrivial example of a cotangent bundle, as \mathbb{R}^n is a smooth manifold with a global coordinate chart. For this manifold M = \mathbb{R}^n, the cotangent bundle T^* \mathbb{R}^n is diffeomorphic to \mathbb{R}^n \times (\mathbb{R}^n)^*, which is in turn isomorphic to \mathbb{R}^{2n} as a smooth manifold.^[6] This isomorphism arises because the tangent space at every point q \in \mathbb{R}^n is canonically identified with \mathbb{R}^n, making the cotangent space T_q^* \mathbb{R}^n canonically (\mathbb{R}^n)^*, and the bundle structure is globally trivial due to the flat geometry of \mathbb{R}^n.^[6] Global coordinates on T^* \mathbb{R}^n are given by (q^1, \dots, q^n, p_1, \dots, p_n), where (q^1, \dots, q^n) are the standard position coordinates on the base \mathbb{R}^n and (p_1, \dots, p_n) are the fiber coordinates representing the components of covectors in the dual basis \{ dq^1, \dots, dq^n \}.^[6] The trivialization of the bundle follows from this coordinate system: the projection map \pi: T^* \mathbb{R}^n \to \mathbb{R}^n sends (q, p) \mapsto q, and every covector at q can be uniquely expressed as \sum_{i=1}^n p_i \, dq^i.^[6] Since T^* \mathbb{R}^n is a trivial vector bundle, it is parallelizable, meaning it admits $2n global smooth sections that form a basis for the fibers at every point.^[1] The zero section embeds the base manifold into the cotangent bundle via the smooth map \mathbb{R}^n \hookrightarrow T^* \mathbb{R}^n given by q \mapsto (q, 0), where $0 denotes the zero covector in T_q^* \mathbb{R}^n.^[6] Each fiber over a point q \in \mathbb{R}^n is the vector space T_q^* \mathbb{R}^n \cong (\mathbb{R}^n)^*, consisting of all linear functionals on T_q \mathbb{R}^n \cong \mathbb{R}^n.^[6] In particular, the fiber over the origin $0 \in \mathbb{R}^n is simply (\mathbb{R}^n)^*, reflecting the linear structure at that point. The cotangent bundle of \mathbb{R}^n also relates to jet bundles in the context of higher-order differentials, but it captures only first-order information: it can be identified with the vertical part of the first-order jet bundle of the trivial line bundle \mathbb{R}^n \times \mathbb{R} \to \mathbb{R}^n, where jets parametrize Taylor expansions up to linear terms.^[26] Higher jet bundles extend this to include higher derivatives, but the cotangent bundle remains limited to the first-order differential structure.^[26]

Cotangent Bundle of the Sphere

The cotangent bundle of the 2-sphere S^2, regarded as the unit sphere in \mathbb{R}^3, admits an explicit description via the annihilator construction derived from the embedding. For each point p \in S^2, the cotangent space T_p^* S^2 is the annihilator of the radial direction, specifically T_p^* S^2 = \{ \xi \in (\mathbb{R}^3)^* \mid \langle \xi, p \rangle = 0 \}. This follows from the identification of the tangent space T_p S^2 = p^\perp = \{ v \in \mathbb{R}^3 \mid \langle v, p \rangle = 0 \}, whose dual yields the subspace of covectors vanishing on the normal span \mathbb{R} p.^[9] The bundle T^* S^2 \to S^2 has rank 2, so its total space is a 4-dimensional smooth manifold. Local trivializations can be constructed using stereographic projection charts on the base S^2. The standard atlas consists of charts centered at the north and south poles: for the northern hemisphere U_N = S^2 \setminus \{(0,0,1)\}, the projection is \phi_N(x,y,z) = \left( \frac{x}{1-z}, \frac{y}{1-z} \right), with inverse \phi_N^{-1}(u,v) = \left( \frac{2u}{1+u^2+v^2}, \frac{2v}{1+u^2+v^2}, \frac{1 - u^2 - v^2}{1+u^2+v^2} \right); the southern chart \phi_S is analogous. On the overlap U_N \cap U_S, the base transition function is the inversion (u,v) \mapsto \left( \frac{u}{u^2+v^2}, \frac{v}{u^2+v^2} \right). The corresponding transition functions for T^* S^2 are the contragredient maps (Jacobian transpose inverses) of those for T S^2, yielding local frames \{ du^1, du^2 \} and \{ dv^1, dv^2 \} transformed by the inverse Jacobian matrix.^[9] These transition functions demonstrate the non-triviality of T^* S^2. The clutching map for T S^2 over the equatorial S^1 is a degree-1 map S^1 \to SO(2), implying the bundle is not globally trivializable. The cotangent bundle, being dual, inherits this non-triviality, as its Euler class e(T^* S^2) = e(T S^2) (since the rank is even). The Euler class e(T S^2) = 2 \cdot [S^2] \in H^2(S^2; \mathbb{Z}) \cong \mathbb{Z}, where [S^2] is the fundamental class generator, follows from the Gauss-Bonnet theorem equating it to the Euler characteristic \chi(S^2) = 2. Thus, the non-zero Euler class obstructs global triviality.^[27]^[28] Equipping S^2 with its standard round metric induces a Riemannian metric on T^* S^2, and the unit cotangent bundle ST^* S^2 = \{ (p, \xi) \in T^* S^2 \mid \|\xi\| = 1 \} (fibers are unit circles) is diffeomorphic to \mathbb{RP}^3. This arises from the identification of unit tangent vectors on S^2 with lines in \mathbb{R}^3 orthogonal to p, via the double cover S^3 \to \mathbb{RP}^3. The full total space admits a radial structure in each fiber (scaling covectors), but the non-trivial topology distinguishes it from a product bundle like S^2 \times \mathbb{R}^2.^[29] Viewing S^2 \cong \mathbb{CP}^1 with the Fubini-Study complex structure, the holomorphic cotangent bundle T^* \mathbb{CP}^1 is the line bundle \mathcal{O}(-2), with first Chern class c_1(T^* \mathbb{CP}^1) = -2 [h](/page/H+), where h generates H^2(\mathbb{CP}^1; \mathbb{Z}) \cong \mathbb{Z}. This contrasts with the tangent bundle T \mathbb{CP}^1 \cong \mathcal{O}(2), where c_1 = 2 h, and aligns with the real Euler class via e = c_1 \mod 2.^[27]

Cotangent Bundles of Lie Groups

The cotangent bundle T^*G of a Lie group G admits a natural left-invariant trivialization obtained via left translations L_g: G \to G, h \mapsto gh, which identifies T^*G diffeomorphically with the product bundle G \times \mathfrak{g}^*, where \mathfrak{g}^* is the dual of the Lie algebra \mathfrak{g} = T_e G.^[30] Under this trivialization, a covector \alpha \in T^*_h G is mapped to (h, (L_h^{-1})^* \alpha) \in G \times \mathfrak{g}^*, ensuring left-invariance since left translations preserve the bundle structure.^[30] The fibers over points of G are further identified through the coadjoint action \mathrm{Ad}^*_g: \mathfrak{g}^* \to \mathfrak{g}^*, defined by \langle \mathrm{Ad}^*_g \mu, X \rangle = \langle \mu, \mathrm{Ad}_{g^{-1}} X \rangle for \mu \in \mathfrak{g}^*, X \in \mathfrak{g}, which arises from the dual of the adjoint action and governs the transformation of covectors under group multiplication.^[30] A key feature of this structure is the existence of canonical 1-forms on G with values in \mathfrak{g}^*, notably the left-invariant Maurer-Cartan form \theta^L \in \Omega^1(G, \mathfrak{g}^*), defined such that at each point h \in G, \theta^L_h = (L_h^{-1})^*: T_h^* G \to \mathfrak{g}^*.^[30] This form satisfies the left-invariance condition L_g^* \theta^L = \theta^L for all g \in G, meaning it is unchanged under pullback by left translations, which reflects the parallelizability of Lie groups and allows global sections to be determined by their values at the identity.^[30] There is a corresponding right-invariant version \theta^R, but the left-invariant form is particularly useful for trivializations of T^*G, as it aligns with the canonical identification that endows T^*G with additional geometric structures like symplectic forms. The Maurer-Cartan form obeys structure equations that encode the Lie algebra relations. Specifically, for matrix Lie groups where the Lie bracket is the commutator, the equation is

d\theta^L + \frac{1}{2} [\theta^L, \theta^L] = 0,

where the bracket [\cdot, \cdot] on \mathfrak{g}^*-valued forms is induced by the Lie bracket on \mathfrak{g} via duality, and d is the exterior derivative.^[30] This equation, known as the Maurer-Cartan structure equation, arises from the flatness of the canonical connection on G and implies that \theta^L generates a flat \mathfrak{g}^*-structure, with the zero curvature reflecting the triviality of the bundle.^[30] An illustrative example is the special orthogonal group G = \mathrm{SO}(3), whose cotangent bundle T^*\mathrm{SO}(3) serves as the phase space for the free rigid body in classical mechanics.^[31] Under the left-invariant trivialization, T^*\mathrm{SO}(3) \cong \mathrm{SO}(3) \times \mathfrak{so}(3)^*, where \mathfrak{so}(3)^* \cong \mathbb{R}^3 via the identification \mu \mapsto \langle \mu, \cdot \rangle with the standard inner product, and elements of \mathbb{R}^3 represent the body-fixed angular momentum vector.^[31] The coadjoint action here corresponds to the rotation of angular momentum under body rotations, linking the geometric structure to Euler's equations for rigid body motion.^[31]

Applications

Tautological One-Form

The tautological one-form, also known as the canonical or Liouville one-form, is a fundamental differential one-form \theta \in \Omega^1(T^*M) defined on the total space of the cotangent bundle T^*M of a smooth manifold M. For a point \alpha \in T^*_pM (which lies in the fiber \pi^{-1}(p) over p \in M) and a tangent vector \xi \in T_\alpha(T^*M), it is intrinsically defined by \theta_\alpha(\xi) = \alpha(\tilde{\xi}), where \tilde{\xi} \in T_pM is the tangent lift obtained as the projection d\pi_\alpha(\xi) via the differential of the bundle projection \pi: T^*M \to M.^[32] This definition captures the natural pairing between covectors in the fiber and tangent vectors at the base point, making \theta independent of choices of coordinates or trivializations.^[32] In local coordinates on M, where a covector \alpha \in T^*_pM is expressed as \alpha = \sum_i \xi_i \, dq^i with \{q^i\} coordinates on M and \{\xi_i\} the components, the tautological one-form takes the expression \theta = \sum_i \xi_i \, dq^i.^[33] This local form, often called the Liouville form, highlights its role as a "tautological" object that directly encodes the fiber coordinates paired with base differentials. The intrinsic definition ensures that \theta transforms covariantly under changes of coordinates on M, preserving its geometric meaning across charts.^[33] Key properties of \theta stem from its intrinsic construction. More centrally, \theta induces a contact structure on the odd-dimensional unit cotangent bundle S^*M (the hypersurface of unit-norm covectors with respect to a metric on M), where its restriction serves as a contact one-form whose kernel defines a maximally non-integrable hyperplane distribution.^[34] Additionally, \theta is preserved under cotangent lifts of diffeomorphisms of the base manifold M, meaning that if f: M \to N is a diffeomorphism, the induced bundle map F: T^*M \to T^*N satisfies F^* \theta_N = \theta_M.^[35]

Canonical Symplectic Form

The canonical symplectic form on the cotangent bundle T^*M of a smooth manifold M is defined as \omega = -d\theta, where \theta is the tautological one-form. This construction equips T^*M with a natural symplectic structure, rendering (T^*M, \omega) a symplectic manifold. In local Darboux coordinates (q^i, p_i) on T^*M, with i = 1, \dots, n = \dim M, the symplectic form assumes the explicit expression

\omega = \sum_{i=1}^n dq^i \wedge dp_i.

This coordinate representation highlights the local symplectomorphism between (T^*M, \omega) and the standard symplectic space (\mathbb{R}^{2n}, \sum dq^i \wedge dp_i ).^[36] The form \omega is closed, since d\omega = -d^2\theta = 0, fulfilling the closedness axiom of symplectic forms. It is non-degenerate, inducing an invertible skew-symmetric bilinear pairing on the tangent bundle T(T^*M) at every point, which guarantees the symplectic structure's robustness. This structure exists globally on T^*M for any smooth M, independent of metrics or other additional data on the base. The canonical symplectic form aligns with the vector bundle structure of T^*M, respecting the linear identification of fibers with the dual tangent spaces. It further induces a Poisson bracket on C^\infty(T^*M), where for smooth functions \xi, \eta with associated Hamiltonian vector fields X_\xi, X_\eta, the bracket is defined by \{ \xi, \eta \}(p) = \omega_p( X_\xi(p), X_\eta(p) ).^[36]

Role as Phase Space in Mechanics

In classical mechanics, the configuration space of a system is modeled by a smooth manifold Q, often identified with the base manifold M, while the phase space is the cotangent bundle T^*Q equipped with canonical coordinates (q^i, p_i), where q^i are generalized positions and p_i are the corresponding conjugate momenta representing linear functionals on the tangent spaces.^[37] This structure naturally encodes the state of the system, with the projection \pi: T^*Q \to Q mapping each phase point to its configuration.^[38] The dynamics on this phase space are governed by a Hamiltonian function H: T^*Q \to \mathbb{R}, typically of the form H(q, p) = \frac{1}{2} g^{ij}(q) p_i p_j + V(q) for kinetic plus potential energy, which generates the flow via Hamilton's equations:

\dot{q}^i = \frac{\partial H}{\partial p_i}, \quad \dot{p}_i = -\frac{\partial H}{\partial q^i}.

These equations arise from the symplectic geometry of T^*Q, where the Hamiltonian vector field is defined by X_H = \omega^{-1} dH, with \omega the canonical symplectic form, ensuring the flow preserves volumes in phase space (Liouville's theorem).^[37] The integral curves of X_H are symplectomorphisms, meaning they preserve the symplectic structure \omega, which is crucial for the conservation of symplectic invariants in mechanical systems.^[38] A simple example is the free particle on \mathbb{R}^n, where Q = \mathbb{R}^n and H(q, p) = \frac{1}{2m} |p|^2, yielding straight-line motion \dot{q} = p/m, \dot{p} = 0.^[37] For curved spaces, the geodesic flow on the unit cotangent bundle S^*M = \{ (q, p) \in T^*M \mid |p|_q = 1 \} uses H(q, p) = \frac{1}{2} g^{ij}(q) p_i p_j, describing the evolution of momenta along geodesics via X_H.^[38] This Hamiltonian formulation connects to Lagrangian mechanics through the Legendre transform, which maps the Lagrangian L: TQ \to \mathbb{R} on the tangent bundle to H on T^*Q via the fiber derivative FL: TQ \to T^*Q, p = \frac{\partial L}{\partial \dot{q}}, enabling equivalence for regular systems.^[37]

References

[1]
[PDF] Tangent and cotangent bundles
Dec 13, 2024 · Special Mathematics Lectures - Differential Geometry. NGUYEN Tue Tai ... cotangent bundle, which is defined by. T∗(M) := G p∈M. T∗ p (M) ...
[2]
[PDF] Differential Geometry - Lecture Notes - Berkeley Math
2.3 Cotangent bundle. Fix a n-manifold X. Definition. The cotangent bundle of X, denoted T∗X is the dual of the tangent bundle. The fibre of p is T∗ p X ...
[3]
None
Summary of each segment:
[4]
[PDF] Differential Geometry I and II, 2021–2022, HU Berlin
These notes will be expanded gradually over the course of the semester. If you notice any typos or mathematical errors, please send e-mail about them to ...
[5]
[PDF] Lecture Notes 1: Geometric mechanics 1 The cotangent bundle
Aug 11, 2010 · In the previous lecture we have rigorously defined smooth manifolds, and parallel trans- port of tangent vectors on them.
[6]
[PDF] lee-smooth-manifolds.pdf - MIT Mathematics
... cotangent bundles, immersed and embedded submanifolds, ten- sors, Riemannian ... definition. (or, rather, two different definitions) of tensors, I ...
[7]
[PDF] An Introduction to Manifolds (Second edition)
This is a completely revised edition, with more than fifty pages of new material scattered throughout. In keeping with the conventional meaning of chapters and.
[8]
None
Below is a merged response that consolidates all the information from the provided summaries into a single, comprehensive overview of the cotangent bundle construction using local coordinates. To maximize detail and clarity, I will use a table in CSV format to organize the key components (Definition of Covectors, Trivialization, Transition Functions, Relevant Equations, Page References, and Useful URLs) across the different sources. Following the table, I will provide a narrative summary to tie everything together, addressing any gaps or inconsistencies.
[9]
[PDF] Introduction to Differential Geometry
Dec 23, 2020 · This book is intented as a modern introduction to Differential Geometry, at a level accessible to advanced undergraduate students.
[10]
[PDF] Introduction to Smooth Manifolds - Julian Chaidez
Manifolds are generalizations of curves and surfaces to many dimensions, used in math to understand space and are important in many fields.
[11]
[PDF] FOUNDATIONS OF ALGEBRAIC GEOMETRY CLASS 37
Apr 4, 2006 · So we'll define the cotangent sheaf first. • Our construction will work for general X, even if X is not “smooth” (or even at all nice,. e.g. ...
[12]
29.32 Sheaf of differentials of a morphism - Stacks Project
May 24, 2023 · The sheaf of differentials \Omega _{X/S} of X over S is the sheaf of differentials of f viewed as a morphism of ringed spaces.Missing: theory | Show results with:theory
[13]
[PDF] The Algebraic Topology of Smooth Manifolds - University of Warwick
May 2, 2025 · A contravariant functor F : C→D is similar but it instead maps F : HomC(X, Y ) → HomD(F(Y ),F(X)) and so instead satisfies F(f ◦ g) = F(g) ...
[14]
[PDF] Connections on fibre bundles - Spin Geometry
Dual to the tangent bundle TM is the cotangent bundle T∗M, where T∗mM = Hom(TmM, ... (It's a contravariant functor.) Unlike the case of the push-forward, the ...
[15]
[PDF] Three Takes on the Tangent and Cotangent Bundles
The notion of the tangent vector at a point p ∈ M is defined, in one of several equivalent ways (discussed further below):. • As a derivation on the space of ...Missing: lecture | Show results with:lecture
[16]
[PDF] Chapter 7 VECTOR BUNDLES - LSU Math
The dual bundle to the tangent bundle is called the cotangent bundle. The cotangent space at a point x is (TMx)∗ which we denote T∗Mx. The bundle dual to the ...
[17]
[PDF] Differential geometry Lecture 10: Dual bundles, 1-forms, and the ...
May 22, 2020 · Understanding sections in the cotangent bundle is, as for vector fields, of critical importance when studying differential geometry.
[18]
https://idv.sinica.edu.tw/ftliang/diff_geom/*diff_geometry(II)/3.04/cotangent_tangent.pdf
[19]
[PDF] MTH–636 Spring 2014 Lecture Notes
Apr 28, 2014 · In particular, the (0,1)-tensor bundle is the tangent bundle of M, while the. (1,0)-tensor bundle is the cotangent bundle of M. Similarly, we ...
[20]
[PDF] Math 865, Topics in Riemannian Geometry - UCI Mathematics
The 7 symbol is the Kulkarni-Nomizu product, which takes 2 symmetric (0,2) tensors ... As above, we can extend this to (0,s) tensors using the tensor product, in ...
[21]
[PDF] Differential Forms and Integration - UCLA Mathematics
... form. For instance, as hinted at earlier, 1- forms can be viewed as sections of the cotangent bundle T*Rn, and similarly 2-forms are sections of the ...
[22]
[PDF] 6 Differential forms
In fact, while vector fields are sections of the tangent bundle, the 1-forms are sections of its dual, the cotangent bundle. We will therefore begin with a ...
[23]
[PDF] Introduction to differential forms - Purdue Math
May 6, 2016 · A differential 1-form (or simply a differential or a 1-form) on an ... (called a potential) such that ω = df. A 1-form ω is called ...
[24]
[PDF] 1 Differential Forms - Caltech CMS
Much like the gradient operator maps a scalar field to a vector field, the exterior derivative will map a 0-form to a 1-form. More generally, the exterior.
[25]
[PDF] Smooth Vector Bundles
... C∞(M) for every f ∈C∞(M). The set of all smooth sections of a vector bundle π : V −→ M is denoted by Γ(M;V ). This is naturally a module over the ring C ...
[26]
[PDF] Geometric tool kit for higher spin gravity (part I) - HAL
Mar 20, 2024 · the first-order jet bundle is isomorphic to the direct sum over M of the zeroth-order jet bundle. J0M “ M ˆ R and the cotangent bundle T˚M, i.e..
[27]
[PDF] Notes on vector bundles and characteristic classes
analogous way as the cotangent bundle of a smooth manifold. Its total space ... F∇ ∈ Ω2(Hom(γ1,γ1)) ∼= A2(CP1) ⊗C∞(CP 1) C∞(CP1;C) = A2(CP1;C) is ...
[28]
The Euler class as a cohomology generator - Project Euclid
The Euler class of a vector bundle over a sphere. We begin by recalling a geometric way of computing the Euler number of a rank n bundle ξ over Sn: Denote by ...
[29]
[PDF] arXiv:1809.05279v2 [math.GT] 28 May 2019
May 28, 2019 · (ii) The unit cotangent bundle ST∗S2 is diffeomorphic to the real projective space RP. 3. , and ξcan is the unique tight contact structure on RP.
[30]
[PDF] Topics in Differential Geometry - Fakultät für Mathematik
... Lie group and that proper actions admit slices are presented with full ... Maurer-Cartan form of the. Lie group G. 4.27. Lemma. For exp : g → G and for ...
[31]
[PDF] arXiv:1811.03184v3 [math-ph] 12 Sep 2019
Sep 12, 2019 · Hamiltonian system on the cotangent bundle (or the phase space) T∗SO(3). ... This yields Euler's equation for the body angular momentum Π in so(3) ...
[32]
[PDF] Math 257a: Intro to Symplectic Geometry with Umut Varolgunes
Oct 4, 2019 · (b) Construction 2: There exists a “tautological” one form on T∗M given by ... (b) Cotangent bundles T∗X with dλx are also lagrangian manifolds.
[33]
[PDF] Symplectic cohomologies on phase space - UCI Mathematics
Sep 10, 2012 · tautological one-form, α. Of course, if one were to impose additional structure on M, for instance, if. M is also a symplectic manifold, then ...
[34]
[PDF] Flexibility in Contact Topology - Austin Christian
... contact form dz − λstd. Here λstd is the tautological 1-form on T∗. R; in local coordinates (qi,pi) on T∗Λ, λstd = Σipidqi. Notice that the natural ...
[35]
[PDF] Classical and Quantum Reduction of the Hydrogen Atom J.V. Gaiter ...
Apr 4, 2022 · The tautological one form is the unique one form Θ ∈ Ω1(T∗Q) ... In the cotangent bundle case, one immediately has such a moment map.
[36]
[PDF] Cotangent Bundles
In this section we show how to construct Ω intrinsically, and then we will study this canonical symplectic structure in some detail. 6.1 The Linear Case. To ...
[37]
[PDF] Lectures on Mechanics
This action lifts to the cotangent bundle, as we have seen. The resulting fixed point space is the cotangent bundle of the fixed point space. QΣ. This fixed ...
[38]
[PDF] Mathematical Methods of Classical Mechanics 2nd ed
Many different mathematical methods and concepts are used in classical mechanics: differential equations and phase flows, smooth mappings and manifolds, Lie ...