ConceptComplete

Tangent and Cotangent Bundles - Examples and Constructions

The pushforward and pullback operations transfer geometric data between manifolds via smooth maps. They provide the functorial properties that make differential geometry natural and coordinate-free.

DefinitionPushforward of Vectors

Let f:M→Nf: M \to N be a smooth map. The pushforward or differential of ff at p∈Mp \in M is the linear map f∗:TpM→Tf(p)Nf_*: T_pM \to T_{f(p)}N (also denoted dfpdf_p) defined by

(f∗v)(g)=v(g∘f)(f_*v)(g) = v(g \circ f)

for v∈TpMv \in T_pM and g∈C∞(N)g \in C^\infty(N). In coordinates, if v=vi∂∂xiv = v^i \frac{\partial}{\partial x^i}, then

f∗v=vi∂fj∂xi∂∂yjf_*v = v^i \frac{\partial f^j}{\partial x^i} \frac{\partial}{\partial y^j}

where (yj)(y^j) are coordinates on NN.

The pushforward encodes the linear approximation to ff at each point. It generalizes the Jacobian matrix from multivariable calculus to the manifold setting.

ExamplePushforward of Curves

If γ:I→M\gamma: I \to M is a smooth curve, its velocity vector at tt is γ′(t)∈Tγ(t)M\gamma'(t) \in T_{\gamma(t)}M. For a smooth map f:M→Nf: M \to N, the chain rule gives

(f∘γ)′(t)=f∗(γ′(t))(f \circ \gamma)'(t) = f_*(\gamma'(t))

This shows that the pushforward correctly encodes the chain rule.

DefinitionPullback of Forms

Let f:M→Nf: M \to N be smooth. The pullback of a covector ω∈Tf(p)∗N\omega \in T_{f(p)}^*N is the covector f∗ω∈Tp∗Mf^*\omega \in T_p^*M defined by

(f∗ω)(v)=ω(f∗v)(f^*\omega)(v) = \omega(f_*v)

for v∈TpMv \in T_pM. This extends to differential forms by f∗(ω∧η)=f∗ω∧f∗ηf^*(\omega \wedge \eta) = f^*\omega \wedge f^*\eta and f∗(dω)=d(f∗ω)f^*(d\omega) = d(f^*\omega).

Remark

The pullback is contravariant: (f∘g)∗=g∗∘f∗(f \circ g)^* = g^* \circ f^*. While vectors push forward along maps, covectors and differential forms pull back. This duality reflects the functorial nature of tangent and cotangent bundles.

ExamplePullback of Exact Forms

If f:M→Rnf: M \to \mathbb{R}^n is smooth and gg is a function on Rn\mathbb{R}^n, then

f∗(dg)=d(g∘f)f^*(dg) = d(g \circ f)

In coordinates, if g=g(y1,…,yn)g = g(y^1, \ldots, y^n), then

f∗(dg)=∂g∂yj∘f⋅d(fj)=∂g∂yj∘f⋅∂fj∂xidxif^*(dg) = \frac{\partial g}{\partial y^j} \circ f \cdot d(f^j) = \frac{\partial g}{\partial y^j} \circ f \cdot \frac{\partial f^j}{\partial x^i} dx^i
DefinitionRelated Vector Fields

Two vector fields X∈X(M)X \in \mathfrak{X}(M) and Y∈X(N)Y \in \mathfrak{X}(N) are ff-related for a smooth map f:M→Nf: M \to N if

f∗(Xp)=Yf(p)f_*(X_p) = Y_{f(p)}

for all p∈Mp \in M. Equivalently, Y∘f=f∗∘XY \circ f = f_* \circ X as maps M→TNM \to TN.

TheoremProperties of Pushforward

The pushforward satisfies:

  1. (g∘f)∗=g∗∘f∗(g \circ f)_* = g_* \circ f_* (functoriality)
  2. (idM)∗=idTM(\text{id}_M)_* = \text{id}_{TM} (identity)
  3. If XX and X′X' are ff-related to YY and Y′Y' respectively, then [X,X′][X, X'] is ff-related to [Y,Y′][Y, Y']

These properties ensure that geometric constructions behave well under smooth maps, forming the foundation for naturality in differential geometry.