What is Topology? An Introduction to Rubber-Sheet Geometry

Topological invariants are properties unchanged under continuous deformation. The Euler characteristic is a key invariant: a sphere has Euler characteristic 2, while a torus has 0. This distinguishes topologically different objects. The number of holes is also an invariant - you cannot create or destroy holes through continuous deformation. These invariants help mathematicians determine whether objects are topologically equivalent without physically transforming them.

de Rham cohomology groups H^n(M) = Z^n(M)/B^n(M) capture topological information about M. Despite being defined using only the differentiable structure, de Rham's theorem states these groups depend only on topology. For example, H⁰(M) ≅ ℝ^k where k is the number of connected components. Higher cohomology groups vanish for Euclidean spaces (Poincaré lemma), but detect nontrivial topology elsewhere. This provides computationally accessible topological invariants, contrasting with fundamental groups which are harder to compute.

Topology provides a new classification scheme for quantum phases using invariants like the genus (integrated Gaussian curvature) and Berry curvature. The TKNN integer, obtained by integrating Berry curvature over the Brillouin zone, determines quantized Hall conductance as an integer. A fundamental principle emerges: gapless states must appear at interfaces between regions with different topological indices. This bulk-edge correspondence means the quantum Hall effect is simultaneously a bulk property (measurable via Faraday rotation) and an edge property (observable through edge currents), representing a unified framework for understanding topological phases.

The generalized Gauss-Bonnet formula extends the theorem to surfaces with handles: total curvature = -2π(2g - 2), where g is the genus (number of handles). For a sphere (g=0): total curvature = 4π. For a torus (g=1): total curvature = 0. For a three-hole pretzel (g=3): total curvature = -8π. This shows that total curvature is a topological invariant—it depends only on the number of handles, not on the specific geometry. This connects topology and geometry through a precise quantitative relationship.

A topological invariant is a quantitative discrete property that remains unchanged under continuous deformations (morphisms). Unlike order parameters that measure symmetry breaking, topological invariants classify phases where symmetries may be preserved. The Chern number, defined as the integral of Berry curvature over the Brillouin zone, acquires integer values and manifests as quantized Hall conductivity. The Z2 invariant provides binary classification for time-reversal symmetric systems: 0 for trivial, 1 for topological. In 3D, four Z2 invariants classify strong versus weak topological insulators. These invariants cannot be directly measured but lead to observable signatures like edge state crossings at the Fermi level.

A topological space is defined by a non-empty set X and a collection τ of subsets called open sets, satisfying three axioms: (1) X and the empty set are in τ; (2) arbitrary unions of open sets are open; (3) finite intersections of open sets are open. Closed sets are defined as complements of open sets, and they satisfy analogous properties: the empty set and X are closed, arbitrary intersections of closed sets are closed, and finite unions of closed sets are closed. Important examples include the discrete topology (all subsets are open) and the indiscrete topology (only X and ∅ are open). A T₀ space distinguishes any two distinct points with an open set, while a T₁ space requires every singleton set to be closed.

In point-set topology, a neighborhood N_r(p) is the set of all points within radius r of center p; a punctured neighborhood excludes the center point. A point p is a limit point of set E if every punctured neighborhood of p intersects E, and a set is closed if it contains all its limit points. A point p in E is isolated if it is not a limit point of E. A point p is an interior point of E if some neighborhood of p is entirely contained in E, and E is open if all its points are interior points. The complement of an open set is closed, and vice versa. Every neighborhood is open (proven via triangle inequality), and a point is a limit point of E if and only if every neighborhood contains infinitely many points of E.

This comprehensive section develops the core machinery of point-set topology through neighborhood-based definitions and set operations. It begins with Felix Hausdorff's 1914 definition of a neighborhood as a set containing an open ball around each of its points, extending beyond balls to irregular regions. The section systematically defines interior points (points with neighborhoods entirely contained in a set) and the interior operation, demonstrating that for any interval in ℝ, the interior corresponds to the open version of that interval. The concept of adherent points follows, characterized by the property that every neighborhood intersects the set, leading to the closure operation as the set of all adherent points. The boundary is defined as the difference between closure and interior, representing points that are neither interior nor exterior. The section then develops the theory of open and closed sets, proving that every open ball is an open set using the triangle inequality. Fundamental theorems establish that arbitrary unions and finite intersections of open sets remain open, while arbitrary intersections and finite unions of closed sets remain closed. The section culminates with the topological definition of continuity: a function is continuous if the preimage of every open set is open, unifying continuity across all metric spaces under a single elegant characterization.

This comprehensive section covers the foundational concepts of point set topology. An interior point is where some neighborhood is entirely contained within the set, and the interior is the largest open set contained within it. A boundary point is where every neighborhood contains points both in the set and its complement. An exterior point belongs to the interior of the complement. Every point belongs to exactly one category: interior, exterior, or boundary. A dense set has closure equal to the entire space. A perfect set is closed and contains all its limit points with no isolated points. An isolated point belongs to the set but is not a limit point. A discrete set has every point isolated. The closure of a set is the union of the set and all its limit points. These concepts are essential for understanding the structure and behavior of sets in topology.

Separation axioms (T0 through T4) classify topological spaces by their ability to separate points and closed sets using open sets; starting from T0 (least restrictive) to T4 (normal spaces), these axioms describe increasingly stringent conditions on space structure, with key implications including the Tietze Extension Theorem for normal spaces and the Stone-Čech compactification for completely regular spaces.

A surface is non-orientable if there is no consistent notion of right-handed versus left-handed orientation on it; the Möbius band, created by gluing a rectangular strip with a half-twist, is the simplest example of a non-orientable surface with only one edge and one side, and it demonstrates non-orientability because a creature traveling around it returns reversed (left becomes right), unlike orientable surfaces where orientation remains consistent. The Möbius band serves as a fundamental building block for constructing more complex non-orientable surfaces like the projective plane and Klein bottle, and every compact 2-dimensional surface can be classified as either a sphere with handles (orientable) or a sphere with cross caps (non-orientable).

A surface is orientable if a continuous field of unit normal vectors can be assigned to it; for example, a sphere and cylinder are orientable as they have two distinct sides with consistent normal vector assignments, while a Mobius strip is non-orientable because any attempt to continuously assign normal vectors results in a contradiction where the vector points in the opposite direction after traversing the surface.

The video introduces non-orientable surfaces, using the Klein bottle as an example. A Klein bottle is created by taking a sphere, cutting it, and gluing the edges in a way that reverses orientation. Unlike orientable surfaces, a Klein bottle has the property that if you walk along one side, you eventually arrive on the other side without crossing an edge. This is related to the Möbius strip, which also has this non-orientable property. The video presents the general formula for Euler's characteristic of any orientable surface without boundary: χ = 2 - 2g, where g is the genus. This formula generalizes the sphere (g=0, χ=2) and torus (g=1, χ=0).

A non-orientable surface is one where an object can return to its starting position after a journey while being flipped (as if passing through a mirror). The Möbius strip and Klein bottle are examples. On such surfaces, an elastic band can become "trapped" in a way that prevents it from being contracted to a point.

A Möbius strip is a one-sided surface created by giving a strip of paper a half-twist and joining the ends, while a Klein bottle is a four-dimensional non-orientable surface that has only one continuous surface (allowing one to draw from inside to outside without crossing an edge); both demonstrate how topology challenges our intuitive understanding of dimensions and surfaces.

Topological Data Analysis (TDA) is a data science methodology that focuses on extracting the underlying shape or structure from complex, high-dimensional data, making it more robust to noise and perturbations; inspired by Euler's solution to the Seven Bridges of Königsberg problem, TDA translates data into geometric structures like graphs, allowing practitioners to analyze patterns that would be difficult to detect using traditional statistical methods alone.

Topological Data Analysis (TDA) addresses challenges where traditional statistical methods fail: data with auto-correlations across time/space, wide data with many variables but few observations, small data with many predictors, high-dimensional data difficult to visualize, and data breaking machine learning assumptions. TDA tools include persistent homology, Mapper, homotopy continuation, Morse functions, Euler calculus, and Ricci curvature. Data is represented as point clouds where each variable is a dimension, and relationships are understood through distance metrics.

Topological Data Analysis (TDA) is an operator that transforms data into simpler data while respecting symmetries. It uses persistence diagrams as its primary tool, transforming infinite-dimensional data (like distance from centroid) into collections of points representing topological features. TDA has three fundamental properties: invariance under homeomorphism (stretching doesn't change the diagram), computational efficiency, and stability (small perturbations cause small changes). The pseudodistance natural compares data by considering distances between functions composed with group elements, with the choice of group determining what equivalences are considered.

Topological Data Analysis (TDA) is a mathematical approach that studies the shape and structure of data by analyzing topological invariants called Betti numbers, which count connected components, loops, and higher-dimensional holes; the method uses homology persistent to track how these topological features appear and disappear as a parameter (epsilon) varies, producing persistence diagrams that reveal the underlying structure of data while distinguishing meaningful features from noise.

Topological Data Analysis (TDA) emerged ~20-25 years ago from observing that data carries underlying topological structure. TDA addresses two main scenarios: (1) point clouds in high-dimensional spaces where data concentrates around geometric/topological structures, and (2) complex data objects themselves carrying topology. The core challenge is extracting meaningful topological information from discrete data samples.