Lecture 6

MTH 868 — Topology / Smooth Manifolds
Date: Jan 26, 2026
Topic: Tangent Vectors, Pushforwards, Curves, Immersions, Submersions, Critical Points


1. Review from Last Time (Page 1)

Smooth Manifolds

We are working with a smooth manifold ( M ).

  • Intuition: Locally, ( M ) looks like ( \mathbb{R}^n ).
  • Charts allow us to do calculus even though globally the space may be curved or weird.

2. Tangent Space via Derivations (Key Idea)

Definition (Very Important)

The tangent space at a point ( p \in M ), denoted ( T_p M ), is defined as:

The set of derivations at ( p ), i.e., linear maps
[ X : C^\infty(M) \to \mathbb{R} ] satisfying the Leibniz rule: [ X(fg) = f(p)X(g) + g(p)X(f) ]

Stats intuition:
This is an abstract way of defining a directional derivative without coordinates.
Think of it as “how functions change at point ( p ) along some direction”.


Coordinate Description (What Makes This Concrete)

If
[ (U, \phi = (x^1, \dots, x^n)) ] is a chart around ( p ), then the vectors [ \left{ \frac{\partial}{\partial x^i}\bigg|p \right}{i=1}^n ] form a basis of ( T_p M ).

Professor emphasis:

“This is why tangent spaces are ( n )-dimensional, not infinite-dimensional.”

Stats parallel:
This is exactly the same reason gradients live in ( \mathbb{R}^n ), even though functions live in infinite-dimensional spaces.


3. Pushforward (Differential of a Map)

Let ( F : M \to N ) be smooth.

Definition

The pushforward (also called the differential) [ F_|_p : T_p M \to T_{F(p)} N ] is defined by [ (F_ X)(f) = X(f \circ F) ]

Interpretation:

  • Vectors act on functions
  • Pushforward tells you how a vector on ( M ) acts on functions on ( N )

Coordinate Formula (Jacobians Appear!)

If ( F : \mathbb{R}^n \to \mathbb{R}^m ), then [ F_* \left( \frac{\partial}{\partial x^i} \right) = \sum_{j=1}^m \frac{\partial F^j}{\partial x^i} \frac{\partial}{\partial y^j} ]

Professor explicitly said:

“This is the Jacobian. Nothing mysterious is happening.”

Stats connection:
This is exactly the linearization you use in:

  • Delta method
  • Multivariate Taylor expansions
  • Sensitivity analysis

4. Chain Rule for Pushforwards (Proposition)

If [ M \xrightarrow{F} N \xrightarrow{G} P ] are smooth, then [ (G \circ F)*|_p = G|_{F(p)} \circ F_|_p ]

Proof Sketch (from lecture)

For ( X \in T_p M ): [ ((G \circ F)* X)(f) = X(f \circ G \circ F) ] and [ (G* (F_* X))(f) = (F_* X)(f \circ G) = X(f \circ G \circ F) ]

Thus they agree.

Stats analogy:
Linear approximations compose exactly the way you expect.


5. Curves and Velocity Vectors (Page 2)

Definition

A curve in ( M ) is a smooth map [ c : (a,b) \to M ]

Velocity Vector

The velocity at ( t_0 ) is [ c’(t_0) = c_* \left( \frac{d}{dt}\bigg|{t_0} \right) \in T{c(t_0)} M ]

Key intuition:
Tangent vectors are velocities of curves.


Fundamental Proposition

Given any ( p \in M ) and any ( X \in T_p M ),
there exists a curve ( c ) such that [ c(0) = p, \quad c’(0) = X ]

Professor emphasis:

“This is why curves are not just examples, they characterize tangent vectors.”


Proof Idea (Very Important Construction)

  • Choose coordinates ( \phi(p) = 0 )
  • Write [ X = \sum a^i \frac{\partial}{\partial x^i} ]
  • Define [ c(t) = \phi^{-1}(a^1 t, \dots, a^n t) ]

This is the manifold version of a straight line.


6. Immersions and Submersions (Page 2–3)

Let ( F : M \to N ).

Immersion

( F ) is an immersion if [ F_*|_p \text{ is injective for all } p ]

Meaning:
Locally, ( M ) does not collapse dimensions.


Submersion

( F ) is a submersion if [ F_*|_p \text{ is surjective for all } p ]

Meaning:
Locally, ( F ) hits all directions in ( N ).


Examples

  • ( F(t) = (\cos t, \sin t) ) is an immersion
  • ( F(t) = (t^2, t^3) ) is not an immersion at ( t=0 )

Why:
Derivative vanishes, rank drops.


7. Rank, Critical Points, Regular Values (Page 3)

Rank

[ \text{rank } F(p) = \dim(\text{image of } F_*|_p) ]


Critical Point

( p \in M ) is a critical point if [ F_*|_p \text{ is not surjective} ]

Regular Point

Otherwise, ( p ) is a regular point.


Regular Value

( y \in N ) is a regular value if [ \forall p \in F^{-1}(y), \; p \text{ is a regular point} ]


Example

[ F(x,y) = x^2 - y^2 ]

  • Gradient: [ \nabla F = (2x, -2y) ]
  • Only critical point is ( (0,0) )
  • Hence:
    • Only critical value is ( 0 )
    • All other values are regular

Stats analogy:
Critical points are where the Jacobian loses rank.
Regular values are “well-behaved” outputs.


8. Big Picture (Professor’s Framing)

“Everything we do from here on depends on understanding when maps behave like linear maps locally.”

  • Immersions → embedded submanifolds
  • Submersions → level sets are manifolds
  • Regular values → preimages are smooth manifolds

This is the foundation for:

  • Implicit Function Theorem
  • Sard’s Theorem
  • Transversality
  • Differential topology

9. Notation Appendix (For Sanity)

Symbol Meaning
( T_p M ) Tangent space at point ( p )
( F_* ) Pushforward / differential
( \partial / \partial x^i ) Coordinate basis vector
Immersion Injective differential
Submersion Surjective differential
Regular value All preimages are regular points

Comments