Tangent space

Tangent vectors map to manifold points via the exponential map

The exponential map is a function that takes a tangent vector at a point on a manifold and maps it to a point on the manifold itself. This mapping is crucial for understanding the local structure of the manifold and how it relates to its tangent space. Essentially, it provides a way to move from the abstract concept of a tangent vector to a concrete point on the manifold.

Example

Consider a 2-dimensional sphere (S^2). If you have a tangent vector at the north pole, the exponential map will take this vector and map it to a point on the sphere, such as the equator.

The exponential map is fundamental in differential geometry and physics for connecting local properties (tangent vectors) with global properties (points on the manifold).

Related concepts

Riemannian manifold

Riemannian manifolds generalize Euclidean space concepts like distance and curvature

saddle points are more common than local minima in high dimensions

Saddle points arise due to mixed partial derivatives being zero, leading to more complex curvature in high dimensions

non-convex loss landscapes are hard: many local minima and saddle points

Non-convex loss landscapes are hard due to many local minima and saddle points

Lyapunov exponents measure: rate of divergence of nearby trajectories in a dynamical system

Lyapunov exponents measure the rate of divergence of nearby trajectories in a dynamical system

Manifold hypothesis

High-dimensional data lies on lower-dimensional manifolds

the do-calculus does: computes interventional probabilities from observational data

Do-calculus computes interventional probabilities from observational data

One email a day: 5 concepts + the 5 stories that matter →

Swipe through 100 ML concepts daily

Open TickerNews