Advanced Mathematics

Topology: The Mathematics of Shape and Space

Topology is the branch of mathematics that studies properties preserved under continuous deformations. It abstracts the notion of nearness to its logical essence, revealing deep truths about shape, space, holes, and connectivity that apply across geometry, analysis, algebra, and modern data science.

Learning Objectives

After mastering this material, you will be able to:

  • ✓Define topological spaces, open and closed sets, and the standard constructions including product, subspace, and quotient topologies
  • ✓Characterize continuous functions between topological spaces and prove homeomorphism results
  • ✓State and apply the Heine-Borel and Tychonoff theorems on compactness
  • ✓Distinguish path-connectedness from connectedness with counterexamples
  • ✓Apply separation axioms T0 through T4 to classify topological spaces
  • ✓Compute the fundamental group for circles, spheres, tori, and graphs
  • ✓Compute simplicial homology groups and the Euler characteristic
  • ✓Describe covering spaces, deck transformations, and the universal cover

1. Topological Spaces: Open and Closed Sets

A topological space is a set X equipped with a topology — a collection of subsets designated as open — satisfying three axioms. This abstraction captures the intuitive idea of points being "close to" one another without requiring any measurement.

The Three Topology Axioms

Definition: Topology

A topology on a set X is a collection T of subsets of X satisfying:

  • (T1) The empty set and X itself belong to T
  • (T2) Arbitrary unions of members of T belong to T
  • (T3) Finite intersections of members of T belong to T

The pair (X, T) is called a topological space. Members of T are called open sets.

Standard Examples of Topologies

Discrete Topology

Every subset of X is open. This is the finest topology on X. In the discrete topology, every function out of X is continuous. A set with the discrete topology is homeomorphic to no non-discrete space unless the set has at most one point.

Indiscrete Topology

Only the empty set and X itself are open. This is the coarsest topology on X. Every function into an indiscrete space is continuous. The indiscrete topology makes all points topologically indistinguishable.

Standard Topology on R

Open sets are unions of open intervals (a, b). A subset U of R is open if for every point x in U, there exists epsilon greater than 0 such that the interval (x minus epsilon, x plus epsilon) is contained in U.

Zariski Topology

Used in algebraic geometry. Closed sets are zero loci of polynomials. On the real line, closed sets are finite sets plus the whole line. This topology is not Hausdorff and illustrates how topologies arise from algebraic structures.

Closed Sets, Interior, Closure, and Boundary

A set C is closed if its complement X minus C is open. Closed sets are closed under arbitrary intersections and finite unions. Note that sets can be both open and closed (clopen) or neither.

  • Interior int(A): The largest open set contained in A. Equivalently, the union of all open subsets of A. A point x is in int(A) if some open neighborhood of x is contained in A.
  • Closure cl(A): The smallest closed set containing A. Equivalently, the intersection of all closed sets containing A. A point x is in cl(A) if every open neighborhood of x intersects A.
  • Boundary bd(A): The set cl(A) minus int(A). A point x is a boundary point if every open neighborhood of x intersects both A and the complement of A.
  • Limit points: A point x is a limit point of A if every open neighborhood of x contains a point of A different from x. The set of all limit points of A is the derived set. We have cl(A) equal to A union its derived set.

Key Example

In the standard topology on R, consider A = (0, 1). Then int(A) = (0, 1), cl(A) = [0, 1], and bd(A) contains the points 0 and 1. The set A = Q (the rationals) has int(Q) = empty, cl(Q) = R, and bd(Q) = R, illustrating that a dense set with empty interior can still have closure equal to the whole space.

2. Bases, Subbases, and Topology Constructions

Rather than specifying all open sets explicitly, topologies are often defined by a simpler generating collection called a basis. Standard constructions then build new topological spaces from existing ones.

Basis for a Topology

A collection B of subsets of X is a basis for a topology if: (1) every point of X belongs to some element of B, and (2) for any B1, B2 in B and any point x in their intersection, there exists B3 in B with x in B3 and B3 contained in B1 intersected with B2. The topology generated by B consists of all arbitrary unions of basis elements.

Example: Basis for the Standard Topology on R

The collection of all open intervals (a, b) with a less than b forms a basis. So does the smaller collection of open intervals with rational endpoints — this shows the standard topology on R is second-countable (has a countable basis), a property with important consequences for metrizability and separability.

Subbasis

A subbasis S for a topology on X is any collection of subsets of X whose union equals X. The topology generated by S consists of all unions of finite intersections of elements of S. Every collection is a subbasis for some topology, making subbases very convenient for defining topologies.

Subspace Topology

If (X, T) is a topological space and A is a subset of X, the subspace topology on A consists of all sets of the form U intersected with A where U is open in X. With this topology, the inclusion map from A into X is continuous, and A is the coarsest topology making the inclusion continuous.

Product Topology

The product topology on a product X times Y is generated by the basis of sets U times V where U is open in X and V is open in Y. Equivalently, it is the coarsest topology making both projection maps from X times Y to X and from X times Y to Y continuous.

Infinite Products: Product vs. Box Topology

For infinite products, the product topology uses a subbasis of sets obtained by restricting only finitely many factors at a time. The box topology, by contrast, allows all factors to be simultaneously restricted. The product topology is coarser and has better properties — in particular, it makes the Tychonoff theorem hold, while the box topology does not.

Quotient Topology

Given a surjective map q from X to Y, the quotient topology on Y is the finest topology making q continuous: U is open in Y if and only if the preimage of U under q is open in X. This construction allows the formation of new spaces by identifying (gluing together) points.

Classic Quotient Constructions

  • Circle S^1: take [0,1] and identify 0 with 1
  • Torus T^2: take [0,1] times [0,1] and identify opposite edges in pairs
  • Mobius band: take [0,1] times [0,1] and identify one pair of opposite edges with a reversal (orientation flip)
  • Real projective plane RP^2: take S^2 and identify antipodal points

3. Continuous Functions and Homeomorphisms

Continuity in topology is defined in terms of open sets, not epsilon and delta. This topological definition captures the essential idea while working in full generality.

Topological Continuity

Definition: Continuous Function

A function f from X to Y between topological spaces is continuous if for every open set V in Y, the preimage of V under f is open in X. Equivalently, f is continuous if and only if the preimage of every closed set is closed, or if and only if for every x in X and every open neighborhood V of f(x), the preimage of V is an open neighborhood of x.

Homeomorphisms

Two topological spaces X and Y are homeomorphic if there exists a bijection f from X to Y such that both f and its inverse are continuous. Homeomorphisms are the isomorphisms of the category of topological spaces — they preserve all topological properties.

Examples and Non-Examples

  • The open interval (0, 1) is homeomorphic to all of R via the function f(x) = tan(pi times (x minus 1/2))
  • Any two closed intervals [a, b] and [c, d] are homeomorphic
  • The circle S^1 is NOT homeomorphic to R because removing any point from S^1 leaves a connected space, while removing a point from R leaves two disconnected components
  • A figure-8 (two loops joined at a point) is not homeomorphic to a circle because they have different fundamental groups

Topological Invariants

A topological invariant is any property preserved by homeomorphisms. To show two spaces are not homeomorphic, it suffices to find an invariant one has and the other lacks. Key invariants include:

  • →Compactness, connectedness, path-connectedness
  • →The fundamental group and higher homotopy groups
  • →Homology and cohomology groups
  • →The Euler characteristic
  • →Orientability (for manifolds)

Embeddings and Quotient Maps

An embedding is a continuous injective function f from X to Y such that f is a homeomorphism onto its image f(X) with the subspace topology. A quotient map is a surjection q from X to Y such that V is open in Y if and only if the preimage of V is open in X. Open maps carry open sets to open sets; closed maps carry closed sets to closed sets.

4. Connectedness and Path-Connectedness

Connectedness formalizes the idea that a space cannot be split into two separate pieces. There are several related but distinct notions, and understanding their differences is essential.

Connected Spaces

Definition: Connectedness

A topological space X is connected if it cannot be written as a union of two disjoint nonempty open sets. Equivalently, X is connected if the only clopen (simultaneously open and closed) subsets of X are the empty set and X itself.

Path-Connectedness

A space X is path-connected if for any two points x and y in X, there exists a continuous function p from [0,1] to X with p(0) = x and p(1) = y. Such a function is called a path from x to y. Path-connectedness implies connectedness, but the converse fails.

Classic Counterexample: The Topologist's Sine Curve

The set consisting of the graph of sin(1/x) for x greater than 0 together with the vertical segment at x = 0 is connected but not path-connected. No path can cross from the oscillating part to the vertical segment because the oscillations become infinitely rapid near x = 0. This is the canonical example distinguishing the two notions.

Connected Components and Path Components

Every topological space can be partitioned uniquely into its connected components — the maximal connected subspaces. Connected components are always closed (but need not be open in general). Similarly, path components are maximal path-connected subspaces. Path components are contained within connected components, but a connected component may split into multiple path components.

Local Connectedness and Local Path-Connectedness

A space is locally connected at a point x if every open neighborhood of x contains a connected open neighborhood. Local path-connectedness is defined analogously. For locally path-connected spaces, connectedness and path-connectedness coincide. These local properties are important in the theory of covering spaces.

Key Theorems on Connectedness

  • →Intermediate Value Theorem: If f from X to R is continuous and X is connected, then the image f(X) is an interval (connected subset of R). In particular, if f(a) and f(b) have opposite signs, there exists c between a and b with f(c) = 0.
  • →The continuous image of a connected space is connected. Thus connectedness is a topological invariant.
  • →Products of connected spaces are connected in the product topology.

5. Compactness and the Heine-Borel Theorem

Compactness is one of the most powerful and widely used concepts in topology. It generalizes the properties of closed bounded intervals to abstract spaces, enabling many existence and extremum theorems.

Open Covers and Compactness

Definition: Compact Space

A topological space X is compact if every open cover of X has a finite subcover. An open cover is a collection of open sets whose union contains X. Compactness says: no matter how you cover the space with open sets, you can always extract a finite subcollection that still covers.

The Heine-Borel Theorem

In Euclidean space R^n, a subset K is compact if and only if K is both closed and bounded. This is the Heine-Borel theorem, and it is specific to finite-dimensional Euclidean space. In infinite- dimensional spaces (such as Hilbert spaces), closed and bounded sets need not be compact.

Proof Sketch: [0,1] is Compact

Let U be an open cover of [0, 1]. Define S to be the set of x in [0, 1] such that [0, x] can be covered by finitely many sets from U. The set S is nonempty (0 belongs to S) and bounded above by 1. Let s = sup(S). By assumption, s is in some open set from U containing an interval (s minus epsilon, s plus epsilon), so s minus epsilon is in S, meaning [0, s minus epsilon] has a finite subcover. Extending by one more set covers [0, s plus epsilon/2], so s cannot be the supremum unless s = 1, proving [0, 1] compact.

Consequences of Compactness

  • →Extreme Value Theorem: Every continuous function from a compact space to R attains its maximum and minimum values.
  • →Uniform Continuity: Every continuous function from a compact metric space to a metric space is uniformly continuous.
  • →Sequential Compactness: In metric spaces, compactness is equivalent to sequential compactness — every sequence has a convergent subsequence. This equivalence fails in general topological spaces.
  • →Closed subsets of compact spaces are compact. Compact subsets of Hausdorff spaces are closed.

Tychonoff's Theorem

The product of any collection of compact spaces is compact in the product topology. This is Tychonoff's theorem, and its proof for arbitrary (including uncountable) products requires the Axiom of Choice. For finite and countable products, the theorem follows more directly. The Tychonoff theorem has applications in functional analysis, number theory (via profinite completions), and model theory.

6. Separation Axioms: T0 Through T4

Separation axioms describe how richly a topology can distinguish points and sets from each other using open sets. The standard hierarchy runs from T0 (weakest) to T4 (strongest), with each axiom implying all weaker ones for the most natural formulations.

T0 — Kolmogorov Space

For any two distinct points x and y, there exists an open set containing one but not the other. T0 is the minimal requirement to distinguish points topologically. The Sierpinski space (two points, three open sets) is T0 but not T1.

T1 — Frechet Space

For any two distinct points x and y, there exist open sets U containing x but not y, and V containing y but not x. Equivalently, every singleton set is closed. The cofinite topology on an infinite set is T1 but not T2.

T2 — Hausdorff Space

Any two distinct points can be separated by disjoint open neighborhoods. Most spaces encountered in analysis and geometry are Hausdorff. Key fact: in a Hausdorff space, limits of sequences are unique and compact subsets are closed. Metric spaces are always Hausdorff.

T3 — Regular Hausdorff Space

A T1 space is regular if for any closed set C and point x not in C, there exist disjoint open sets separating x from C. A T1 regular space is called T3. Manifolds and metric spaces are regular. Urysohn's lemma characterizes T4 spaces via continuous functions.

T4 — Normal Hausdorff Space

A T1 space is normal if any two disjoint closed sets can be separated by disjoint open sets. Compact Hausdorff spaces are normal. The Tietze Extension Theorem states that in a normal space, every continuous function from a closed subset extends to a continuous function on the whole space.

Urysohn's Lemma

A topological space is normal if and only if for any two disjoint closed sets A and B, there exists a continuous function f from X to [0, 1] with f = 0 on A and f = 1 on B. Such a function is called a Urysohn function. This lemma is the key tool for constructing continuous functions in general topology and is used to prove the Tietze Extension Theorem and metrization theorems.

7. Metric Spaces, Completeness, and the Baire Category Theorem

Metric spaces provide a concrete setting for topology, bridging abstract topological concepts with analysis. Completeness in metric spaces is the condition that makes Cauchy sequences converge.

Metric Spaces

Definition: Metric Space

A metric space is a set X with a function d from X times X to R (the metric or distance function) satisfying:

  • (M1) d(x, y) is at least 0, with equality iff x = y
  • (M2) d(x, y) = d(y, x) (symmetry)
  • (M3) d(x, z) is at most d(x, y) + d(y, z) (triangle inequality)

Completeness and Cauchy Sequences

A sequence (x_n) in a metric space is Cauchy if for every epsilon greater than 0, there exists N such that for all m, n greater than N, d(x_m, x_n) is less than epsilon. A metric space is complete if every Cauchy sequence converges to a limit in the space.

Complete Spaces

  • R with the standard metric
  • R^n with the Euclidean metric
  • The space C([0,1]) of continuous functions
  • Every compact metric space
  • The Baire space N^N (sequences of naturals)

Incomplete Spaces

  • Q with the standard metric (sqrt(2) is missing)
  • (0, 1) with the standard metric
  • Q^n in any standard metric
  • Dense proper subsets of complete spaces

Completion of a Metric Space

Every metric space X has a completion — a complete metric space containing X as a dense subset, unique up to isometry. The completion of Q is R. The completion of a normed vector space is a Banach space. Completions are constructed via equivalence classes of Cauchy sequences, a technique also used to construct R from Q.

The Baire Category Theorem

A set is nowhere dense if its closure has empty interior. A set is meager (or first category) if it is a countable union of nowhere dense sets. The Baire Category Theorem states: in a complete metric space (or a locally compact Hausdorff space), no nonempty open set is meager. Equivalently, a countable intersection of dense open sets is dense.

Applications of the Baire Category Theorem

  • R is uncountable: if R were countable, it would be meager as a union of singletons (nowhere dense), contradicting Baire
  • Existence of continuous nowhere differentiable functions: such functions are in fact generic (comeager) in C([0,1])
  • The open mapping theorem and closed graph theorem in functional analysis both rely on the Baire Category Theorem
  • In descriptive set theory, Baire category provides a notion of largeness complementary to measure theory

8. Homotopy and the Fundamental Group

Algebraic topology assigns algebraic structures to topological spaces to capture and classify their shape. The fundamental group is the first and most concrete such invariant, measuring the one-dimensional holes in a space.

Homotopy of Paths

Definition: Homotopy

Two continuous functions f and g from X to Y are homotopic if there exists a continuous function H from X times [0, 1] to Y with H(x, 0) = f(x) and H(x, 1) = g(x) for all x in X. H is called a homotopy from f to g. For paths (with X = [0, 1]), we require the endpoints to remain fixed: H(0, t) = f(0) = g(0) and H(1, t) = f(1) = g(1) for all t in [0, 1].

The Fundamental Group pi_1(X, x0)

Fix a basepoint x0 in X. A loop based at x0 is a continuous function p from [0, 1] to X with p(0) = p(1) = x0. Two loops are homotopic if there is a path-homotopy between them fixing x0. The fundamental group pi_1(X, x0) is the set of homotopy classes of loops based at x0, with group operation given by concatenation: the product [f][g] is the class of the loop obtained by traversing f then g.

Fundamental Groups of Standard Spaces

  • pi_1(R^n) = trivial (simply connected)
  • pi_1(S^1) = Z (the integers, one generator per winding)
  • pi_1(S^n) = trivial for n at least 2
  • pi_1(T^2) = Z times Z (the torus, two generators)
  • pi_1(RP^2) = Z/2Z (real projective plane)
  • pi_1(figure-eight) = free group on 2 generators
  • pi_1(Klein bottle) = presentation with generators a, b and relation aba = b

Deformation Retracts and Homotopy Equivalence

A subspace A of X is a deformation retract if there exists a continuous map r from X to A (the retraction) such that r restricted to A is the identity, and the inclusion of A into X composed with r is homotopic to the identity on X. Two spaces are homotopy equivalent if there exist continuous maps f from X to Y and g from Y to X such that g composed with f is homotopic to the identity on X, and f composed with g is homotopic to the identity on Y.

Van Kampen's Theorem

Van Kampen's theorem computes the fundamental group of a space from the fundamental groups of its open subsets. If X = U union V where U, V, and U intersected with V are path-connected and open, then pi_1(X) is the amalgamated free product of pi_1(U) and pi_1(V) over pi_1(U intersected with V). This theorem is the main tool for computing fundamental groups of CW complexes.

9. Covering Spaces and Deck Transformations

Covering spaces provide a powerful way to study topological spaces through their "unfolded" versions and connect topology to group theory via the Galois correspondence.

Definition of Covering Spaces

Covering Map

A continuous surjection p from E to B is a covering map if for every point b in B, there exists an open neighborhood U of b such that the preimage of U is a disjoint union of open sets, each mapped homeomorphically to U by p. The open set U is called evenly covered, and the space E is called a covering space of B.

Examples of Covering Spaces

R covers S^1

The map p from R to S^1 defined by p(t) = (cos(2*pi*t), sin(2*pi*t)) is a covering map. Each point of S^1 has countably many preimages, spaced 1 apart along R. This covering corresponds to the fact that pi_1(S^1) = Z.

S^n covers RP^n

The antipodal identification makes S^n a 2-sheeted cover of real projective space RP^n. Since S^n is simply connected for n at least 2, this is the universal cover of RP^n in those dimensions, reflecting pi_1(RP^n) = Z/2Z.

Lifting Properties

The unique path lifting property states: for any path f from [0, 1] to B and any preimage point e0 of f(0), there is a unique lifted path in E starting at e0. Similarly, homotopies of paths lift uniquely. These properties make covering spaces the fundamental tool for computing pi_1.

Deck Transformations and the Galois Correspondence

A deck transformation of a covering p from E to B is a homeomorphism phi from E to E satisfying p composed with phi = p. The group of deck transformations acts freely and properly discontinuously on E. For a universal cover (simply connected covering space), the group of deck transformations is isomorphic to pi_1(B).

The Galois Correspondence

For a path-connected, locally path-connected, and semi-locally simply connected base space B, there is a bijection between connected covering spaces of B (up to isomorphism) and conjugacy classes of subgroups of pi_1(B). The universal cover corresponds to the trivial subgroup. This mirrors the Galois correspondence between field extensions and subgroups of the Galois group.

10. CW Complexes, Simplicial Complexes, and Euler Characteristic

CW complexes provide a flexible combinatorial way to build topological spaces cell by cell. They are the preferred setting for algebraic topology computations, balancing generality with computability.

CW Complexes

A CW complex X is built inductively. Start with a discrete set X^0 (the 0-cells or vertices). To form the n-skeleton X^n, attach n-cells (copies of the closed n-disk D^n) to X^(n-1) via continuous attaching maps from the boundary sphere S^(n-1) to X^(n-1). The letters CW stand for closure-finite and weak topology.

CW Structures for Standard Spaces

  • S^n: one 0-cell and one n-cell (the attaching map is constant)
  • RP^n: one cell in each dimension 0, 1, ..., n
  • CP^n: one cell in each even dimension 0, 2, 4, ..., 2n
  • The torus T^2: one 0-cell, two 1-cells, and one 2-cell
  • A graph: vertices are 0-cells, edges are 1-cells

Simplicial Complexes

A simplicial complex is a combinatorial structure built from simplices (points, line segments, triangles, tetrahedra, etc.) glued along faces. Formally, a simplicial complex K is a collection of finite nonempty sets (simplices) such that every nonempty subset of a simplex also belongs to K. The geometric realization of K is the corresponding topological space.

The Euler Characteristic

For a CW complex X with c_k cells of dimension k, the Euler characteristic is chi(X) = c_0 minus c_1 plus c_2 minus c_3 plus the rest of the alternating sum. This is a topological invariant: homeomorphic spaces have the same Euler characteristic.

Euler Characteristics of Standard Spaces

  • Point: chi = 1
  • Circle S^1: chi = 0 (one 0-cell, one 1-cell: 1 minus 1)
  • Sphere S^2: chi = 2 (one 0-cell, one 2-cell: 1 plus 1)
  • Torus T^2: chi = 0 (one 0-cell, two 1-cells, one 2-cell: 1 minus 2 plus 1)
  • Genus-g surface: chi = 2 minus 2g
  • RP^2 (projective plane): chi = 1
  • Klein bottle: chi = 0

Euler's Formula for Polyhedra

For any convex polyhedron: V minus E plus F = 2, where V is the number of vertices, E the number of edges, and F the number of faces. This is because any convex polyhedron is homeomorphic to S^2, which has Euler characteristic 2. The formula generalizes: for any triangulation of a surface of genus g, V minus E plus F = 2 minus 2g.

11. Homology Groups

Homology groups provide a systematic way to count holes in a topological space using algebra. They are computable, homotopy invariant, and are among the most powerful tools in algebraic topology.

Simplicial Homology

For a simplicial complex K, the n-th chain group C_n(K) is the free abelian group generated by the n-simplices. The boundary map d_n from C_n to C_(n-1) sends each n-simplex to its oriented boundary (the alternating sum of its (n-1)-dimensional faces). The key identity: d_(n-1) composed with d_n = 0. That is, the boundary of a boundary is zero.

Cycles, Boundaries, and Homology

An n-cycle is a chain c with d_n(c) = 0 (no boundary). An n-boundary is a chain c that is the boundary of some (n+1)-chain. Since every boundary is a cycle, we can form the quotient. The n-th homology group H_n is the group of n-cycles modulo n-boundaries: ker(d_n) divided by im(d_(n+1)).

Singular Homology

Singular homology extends the simplicial construction to all topological spaces. A singular n-simplex in X is any continuous function from the standard n-simplex Delta^n to X. The singular chain group C_n(X) is the free abelian group on all singular n-simplices. The resulting homology groups are the same as simplicial homology when both apply, but singular homology is defined for all spaces.

Computing Homology Groups

Homology of Standard Spaces

  • H_n(point) = Z if n = 0, else 0
  • H_n(S^k) = Z if n = 0 or n = k (with k at least 1), else 0
  • H_0(T^2) = Z, H_1(T^2) = Z times Z, H_2(T^2) = Z
  • H_0(RP^2) = Z, H_1(RP^2) = Z/2Z, H_2(RP^2) = 0
  • chi(X) = alternating sum of ranks of H_n(X)

The Long Exact Sequence and Mayer-Vietoris

The Mayer-Vietoris sequence is the homological analog of Van Kampen's theorem. If X = U union V with U, V, and U intersected with V open, there is a long exact sequence connecting the homology groups of the pieces to those of X. This is the main computational tool for computing homology of spaces built from simpler pieces.

The Hurewicz Theorem

The Hurewicz theorem connects homotopy and homology: if X is (n-1)-connected (meaning pi_k(X) = 0 for k less than n), then H_k(X) = 0 for 0 less than k less than n, and H_n(X) is isomorphic to pi_n(X) (abelianized when n = 1). This theorem shows that for simply connected spaces, the first nonzero homotopy and homology groups agree.

12. Manifolds, Knot Theory, and Topological Data Analysis

Topology finds deep applications in geometry (manifolds), pure mathematics (knot theory), and modern data science (topological data analysis). These applications demonstrate the power of abstract topological ideas.

Topological Manifolds

An n-dimensional topological manifold (without boundary) is a second-countable Hausdorff space in which every point has an open neighborhood homeomorphic to R^n. Manifolds are the natural setting for physics, differential geometry, and geometric topology.

Classification of Compact Surfaces

Every compact connected 2-manifold (surface) without boundary is homeomorphic to exactly one of:

  • Orientable surfaces: the sphere S^2, or a connected sum of g tori (genus-g surface). These have Euler characteristic 2 minus 2g.
  • Non-orientable surfaces: connected sums of copies of the real projective plane RP^2. These include RP^2 (Euler characteristic 1) and the Klein bottle (Euler characteristic 0).

Manifolds with Boundary

An n-manifold with boundary is a space where each point has a neighborhood homeomorphic to either R^n or the upper half-space R^(n-1) times [0, infinity). Points of the latter type form the boundary of M, a manifold of dimension n minus 1. The Mobius band and the disk D^2 are manifolds with boundary.

Knot Theory

A knot is a smooth embedding of S^1 into S^3 (or R^3). Two knots are equivalent if there is an ambient isotopy taking one to the other. Knot theory classifies knots up to equivalence using topological invariants.

Knot Invariants

  • Knot group: The fundamental group of the complement S^3 minus K. For the unknot, this is Z. For the trefoil, it has presentation with generators a and b satisfying a^2 = b^3.
  • Alexander polynomial: A polynomial invariant derived from the homology of the knot complement. Distinguishes many pairs of knots.
  • Jones polynomial: A more powerful polynomial invariant defined using the Kauffman bracket, connected to quantum groups and statistical mechanics.
  • Crossing number: The minimum number of crossings in any diagram of the knot. The trefoil has crossing number 3.

Topological Data Analysis

Topological data analysis applies topology to study the shape of data. The key tool is persistent homology, which tracks how topological features (connected components, loops, voids) appear and disappear as a scale parameter varies.

Persistent Homology

Given a point cloud, build a growing sequence of simplicial complexes (the Vietoris-Rips filtration) by connecting points within distance epsilon of each other. As epsilon increases, topological features are born and die. The persistence diagram records the birth-death pairs, providing a summary of the data's shape that is robust to noise. Applications include analyzing brain networks, protein structure, material science, and cancer genomics.

Practice Problems with Solutions

Problem 1: Proving a Set is Open

Let X = R with the standard topology and let U = (1, 3) union (5, 7). Prove that U is open using the definition of the standard topology.

Solution

We must show that for every x in U, there exists epsilon greater than 0 such that (x minus epsilon, x plus epsilon) is contained in U. Case 1: x is in (1, 3). Let epsilon = min(x minus 1, 3 minus x), which is positive since 1 is less than x is less than 3. Then the open ball around x of radius epsilon is contained in (1, 3), which is contained in U. Case 2: x is in (5, 7). Let epsilon = min(x minus 5, 7 minus x), positive since 5 is less than x is less than 7. Then the open ball is contained in (5, 7) which is contained in U. Since every point of U has an open ball contained in U, U is open. Alternatively, U is a union of two open sets, hence open by the second topology axiom.

Problem 2: Computing Closure and Interior

In R with the standard topology, find the interior, closure, and boundary of A = [1, 2) together with the singleton set containing 3.

Solution

Interior: int(A) = (1, 2). The point 1 has every neighborhood intersecting the complement, so 1 is not interior. The isolated point 3 also fails the interior condition. Closure: cl(A) = [1, 2] together with the singleton containing 3. The point 2 is a limit point of [1, 2) since every open neighborhood of 2 intersects [1, 2). The point 3 is in A but is not a limit point of A, so no new limit points arise near 3. Boundary: bd(A) = cl(A) minus int(A) = the set containing the three points 1, 2, and 3.

Problem 3: Showing Spaces are not Homeomorphic

Show that the closed interval [0, 1] is not homeomorphic to the circle S^1.

Solution

We use a topological invariant. If [0, 1] and S^1 were homeomorphic via f, then removing any point from [0, 1] and the corresponding point from S^1 would give homeomorphic spaces. But: removing any interior point of [0, 1] gives a disconnected space (two open intervals), while removing any point of S^1 gives a connected space (homeomorphic to an open interval). Since connectedness is a topological invariant, removing a single interior point gives different connectivity, so [0, 1] and S^1 are not homeomorphic.

Problem 4: Fundamental Group Computation

Compute the fundamental group of the wedge sum S^1 wedge S^1 (the figure-eight space formed by joining two circles at a point).

Solution

Apply Van Kampen's theorem. Let X = S^1 wedge S^1. Write X = U union V where U is the first circle with a small arc of the second, V is the second circle with a small arc of the first, and U intersected with V is contractible (a small arc around the wedge point). Since U is homotopy equivalent to S^1, pi_1(U) = Z with generator a. Similarly pi_1(V) = Z with generator b. Since U intersected with V is contractible, pi_1 of the intersection is trivial. Van Kampen gives pi_1(X) = Z * Z (the free product), which is the free group on two generators a and b. Every loop in the figure-eight is a unique reduced word in a, b, a^(-1), and b^(-1).

Problem 5: Euler Characteristic of a Surface

Using the CW structure for the torus T^2, compute its Euler characteristic and verify it agrees with the genus formula.

Solution

The torus has a CW structure with one 0-cell (V = 1), two 1-cells called a and b (E = 2), and one 2-cell (F = 1). So chi(T^2) = 1 minus 2 plus 1 = 0. The genus formula gives chi = 2 minus 2g. Since the torus has genus g = 1, chi = 2 minus 2 = 0. Confirmed. The homology groups H_0(T^2) = Z, H_1(T^2) = Z times Z, H_2(T^2) = Z give ranks 1, 2, 1, and chi = 1 minus 2 plus 1 = 0 by the alternating rank formula. All three computations agree.

Problem 6: Compactness via Open Covers

Show directly (using the definition of compactness) that the open interval (0, 1) is not compact in the standard topology on R.

Solution

Exhibit an open cover with no finite subcover. Consider the collection U_n = (1/n, 1) for n = 2, 3, 4, ... Each U_n is open. Their union covers all of (0, 1): for any x in (0, 1), by the Archimedean property there exists N with 1/N less than x, so x belongs to U_N. Now suppose a finite subcollection covers (0, 1). Taking the maximum index N among the finitely many sets, the union is just U_N = (1/N, 1), which does not contain the point 1/(2N), which is in (0, 1). This contradiction shows no finite subcover exists, so (0, 1) is not compact.

Exam Tips and Common Mistakes

Mistake 1: Confusing Closed-and-Bounded with Compact

Heine-Borel says compact is equivalent to closed and bounded in R^n. This fails in general metric spaces. The unit ball in an infinite-dimensional Banach space is closed and bounded but not compact. Always specify the space and verify the theorem's hypotheses apply before invoking Heine-Borel.

Mistake 2: Path-Connected Implies Connected, Not Vice Versa

The topologist's sine curve is connected but not path-connected. Always state which notion you are proving and verify you have the right definitions. Exams frequently test the distinction between these two properties with precisely this counterexample.

Mistake 3: Forgetting that Sets Can Be Neither Open Nor Closed

The interval [0, 1) in R is neither open nor closed. Do not assume that a set failing to be open must be closed. Also remember that a set can be both open and closed (clopen) — in a connected space, only the empty set and the whole space are clopen.

Mistake 4: Applying Van Kampen Without Checking Openness

Van Kampen's theorem requires U, V, and U intersected with V to be open (and path-connected). If you write X = U union V with U and V closed, Van Kampen does not apply directly. Always check the hypotheses carefully before applying the theorem.

Tip 1: Use Invariants to Distinguish Spaces

When asked to show two spaces are not homeomorphic, always reach for an invariant: connectedness after removing a point, compactness, fundamental group, homology groups, or Euler characteristic. Match the invariant to what is easiest to compute for the spaces at hand.

Tip 2: Draw Pictures for Quotient Spaces

For quotient topology problems, draw the square [0,1] times [0,1] and label which edges and corners are identified and in which direction. This visualization makes it easy to determine the resulting space (torus, Klein bottle, projective plane, sphere, Mobius band) from the identification data.

Tip 3: Memorize Key Fundamental Groups

Know by heart: pi_1(S^1) = Z, pi_1(S^n) = 0 for n at least 2, pi_1(T^n) = Z^n, pi_1(RP^2) = Z/2Z, and pi_1 of the figure-eight = free group on 2 generators. These come up constantly and memorizing them saves valuable time on exams.

Tip 4: For Homology, Use the Long Exact Sequence

Most homology computations on exams reduce to applying Mayer-Vietoris or the long exact sequence of a pair. Identify a decomposition X = U union V where each piece and their intersection have known homology, then chase the exact sequence. Keep track of whether maps are zero, injective, or surjective at each step.

Related Topics

Topology connects deeply to every area of advanced mathematics. These related subjects build directly on topological foundations:

Real Analysis

Metric spaces, completeness, and continuity in analysis are the concrete foundations underlying abstract topology. The Baire category theorem and Cauchy sequences connect both fields.

Abstract Algebra

The fundamental group, homology groups, and covering space theory all produce groups. The Galois correspondence in covering spaces mirrors the Galois correspondence in field extensions.

Measure Theory

Measure theory and topology share the study of sigma-algebras and Borel sets. Locally compact Hausdorff spaces provide the setting for Radon measures and the Riesz representation theorem.

Functional Analysis

Normed spaces, Banach spaces, and Hilbert spaces are special topological spaces. The weak and weak-star topologies on infinite- dimensional spaces are important examples of non-metrizable topologies.

Frequently Asked Questions

Is a topological space the same as a metric space?

No. Every metric space gives a topological space (via the topology of open balls), but not every topological space comes from a metric. Topological spaces satisfying Urysohn's metrization theorem (second- countable, regular Hausdorff) can be metrized. The Zariski topology and the cofinite topology on infinite sets are examples of topological spaces that cannot be given by any metric.

What is the difference between homotopy equivalence and homeomorphism?

Homeomorphism is stronger: a homeomorphism is a bijective bicontinuous map, giving a point-by-point correspondence. Homotopy equivalence only requires continuous maps in both directions whose compositions are homotopic to the identities. R^n is homotopy equivalent to a point, but is not homeomorphic to a point (unless n = 0). Algebraic topology invariants (fundamental group, homology) are homotopy invariants, not just homeomorphism invariants.

How is topology used in data science?

Topological data analysis (TDA) uses persistent homology to extract shape information from high-dimensional data. By building filtrations of simplicial complexes on point clouds, TDA identifies topological features (clusters, loops, voids) that persist across scales. These are more robust to noise than traditional statistical methods. TDA has been applied to cancer genomics, neuroscience, materials science, and image recognition.

What prerequisites do I need for topology?

For point-set topology, mathematical maturity and familiarity with set theory and proof-writing are the main prerequisites. For algebraic topology, you should be comfortable with group theory (from abstract algebra) and real analysis at the level of metric spaces. Differential topology additionally requires multivariable calculus and linear algebra.

Why is the Hausdorff condition so important?

The Hausdorff (T2) condition ensures that limits are unique, that compact subsets are closed, and that the diagonal in X times X is closed. Most theorems in analysis and geometry require Hausdorff as a baseline assumption. Non-Hausdorff spaces do arise naturally in algebraic geometry (Zariski topology) and logic (Scott topology), but they require different intuitions and techniques.