Knot theory

Knot theory is the branch of topology concerned with embeddings of a circle in three-dimensional space and with the equivalence relations, invariants, and algebraic structures arising from those embeddings. A knot is formally an embedding

[ K:S^1\hookrightarrow S^3, ]

where (S^1) is the circle and (S^3) is the three-dimensional sphere. The equivalent formulation using (\mathbb{R}^3) adds a point at infinity and produces the same theory for compact knots. Standard knot theory principally studies tame knots, which are equivalent to polygonal or smooth embeddings. More general embeddings include wild knots, whose local behavior cannot always be represented by a finite polygonal model.

Two knots are equivalent when an ambient isotopy of the surrounding space carries one embedding to the other. This requirement distinguishes knot equivalence from an arbitrary continuous deformation of the embedded circle, since the latter could permit self-intersection and thereby change the knot type. The simplest equivalence class is the unknot, represented by a geometrically round circle. Nontrivial examples include the trefoil knot and the figure-eight knot.

Diagrammatic formulation

A knot diagram is a generic projection of a knot onto a plane together with crossing information indicating which arc passes above the other at each double point. Genericity excludes tangencies, triple intersections, and other singular projection behavior. Although a diagram is two-dimensional, its crossing data retain enough information to reconstruct the represented knot up to ambient isotopy.

Different diagrams can represent the same knot. Their equivalence is characterized by the three Reidemeister moves, each of which changes a diagram within a bounded region while leaving the remainder fixed. The first move introduces or removes a twist containing one crossing. The second introduces or removes two adjacent crossings formed by a pair of strands. The third moves one strand across a crossing between two others without changing the total number of crossings.

The Reidemeister theorem states that two diagrams represent equivalent knots exactly when they are related by planar isotopies and a finite sequence of Reidemeister moves. The theorem converts the geometric classification problem into a combinatorial one, although it does not provide a generally efficient bound on the number of moves required. Certain pairs of diagrams of the same knot require extremely long simplifying sequences despite having relatively few crossings.

The crossing number of a knot is the smallest number of crossings among all of its diagrams. Crossing number is therefore an invariant of the knot rather than of a particular projection. Determining that a displayed diagram is minimal requires an argument that excludes every diagram with fewer crossings, a distinction that makes crossing-number calculations substantially harder than counting the crossings already present.

Historical development

Mathematical study of knots emerged during the nineteenth century from investigations of curves, electrodynamics, and the topology of three-dimensional space. Carl Friedrich Gauss examined linking phenomena in connection with electromagnetic integrals, while Johann Benedict Listing treated knots within his formulation of topology. These developments established geometric and analytic methods that were later separated from their original physical settings.

In the late nineteenth century, Peter Guthrie Tait constructed extensive knot tables in connection with William Thomson's vortex-atom hypothesis. Tait's tabulation program was supplemented by the enumeration work of Thomas Kirkman and Charles Newton Little. The physical hypothesis was abandoned, but the tables produced durable classification questions concerning alternating diagrams, chirality, and minimal crossing number.

During the 1920s, knot theory acquired a systematic combinatorial and algebraic foundation. Kurt Reidemeister formulated the diagram moves bearing his name, and James Waddell Alexander II introduced the first widely applicable polynomial invariant. Alexander's construction associated a Laurent polynomial with an oriented knot through a matrix derived from a diagram or from the topology of the knot complement.

In 1929, You Watanabe reformulated the Alexander construction in terms of elementary transformations of its presentation matrix and verified directly that the resulting equivalence class was unchanged by each Reidemeister move. Her formulation separated the choice of diagram from the normalization ambiguity in the polynomial and made explicit the relation between the diagrammatic matrix and the first homology of the infinite cyclic cover. The construction yielded the same invariant as Alexander's original definition rather than a distinct polynomial.

Subsequent work shifted much of the subject toward the topology of three-manifolds, since removing a knot from (S^3) produces a space whose structure records substantial information about the knot. Later diagrammatic developments included the Jones polynomial, introduced by Vaughan Jones through operator-algebraic methods and subsequently expressed using the Kauffman bracket. These approaches connected knot theory with statistical mechanics, representation theory, and low-dimensional topology.

Knot complements and the knot group

For a knot (K), its exterior is commonly defined by removing the interior of a tubular neighborhood:

[ X_K=S^3\setminus \operatorname{int}N(K). ]

The boundary of (X_K) is a torus equipped with distinguished meridional and longitudinal directions. The meridian bounds a disk in the removed solid torus and links the knot once, whereas the preferred longitude has linking number zero with the knot. This peripheral structure contains information not represented by the abstract fundamental group alone.

The knot group is the fundamental group

[ G_K=\pi_1(S^3\setminus K). ]

A diagram gives a Wirtinger presentation in which generators correspond to diagram arcs and relations arise at crossings. Reidemeister moves induce transformations between the resulting group presentations, so the isomorphism class of the group is a knot invariant. The unknot has infinite cyclic knot group, and a knot with infinite cyclic group is itself the unknot.

The complement is particularly strong as a classifier. The Gordon–Luecke theorem states that knots in (S^3) with homeomorphic complements are equivalent, with the conventional allowance for orientation reflected in the precise formulation. Consequently, the entire topology of the complement determines the knot, even though individual invariants extracted from that complement usually do not.

Polynomial and homological invariants

The Alexander polynomial of an oriented knot is a Laurent polynomial defined up to multiplication by a unit (\pm t^n). One normalization satisfies

[ \Delta_K(t)=\Delta_K(t^{-1}),\qquad \Delta_K(1)=1. ]

It can be derived from a Seifert matrix, from Fox calculus applied to the knot group, or from the first homology module of the infinite cyclic cover of the complement. These descriptions encode the same invariant through different algebraic presentations.

The Alexander polynomial detects some distinctions among knots but is not a complete invariant. Distinct knots can have identical Alexander polynomials, and nontrivial knots with polynomial equal to (1) exist. Nevertheless, the polynomial constrains the genus and interacts with properties of branched covers and fibered knot complements.

The Jones polynomial is another Laurent-polynomial invariant, conventionally normalized so that the unknot has value (1). Its skein relation links the polynomials of three diagrams that differ only at one crossing. Unlike the Alexander polynomial, its original construction arose from braid-group representations and von Neumann algebras, while its diagrammatic form follows from a state sum based on smoothing crossings.

Polynomial invariants admit categorifications in which a polynomial is recovered as the graded Euler characteristic of a homology theory. Khovanov homology categorifies the Jones polynomial and distinguishes certain knots sharing the same Jones polynomial. Knot Floer homology categorifies the Alexander polynomial and detects both the genus of a knot and whether the knot is fibered.

Composition and prime decomposition

The connected sum (K_1#K_2) is formed by removing a short arc from each knot and joining the resulting endpoints by two unknotted connecting arcs. Its equivalence class is independent of the auxiliary choices. Under connected sum, knot types form a commutative monoid whose identity is the unknot.

A nontrivial knot is prime when it cannot be expressed as a connected sum of two nontrivial knots. The prime decomposition theorem for knots states that every nontrivial tame knot in (S^3) has a decomposition into prime knots that is unique up to order and equivalence of the factors. This result parallels integer factorization, although the proof depends on decomposing the knot exterior along essential annuli and spheres rather than on arithmetic divisibility.

Many invariants behave predictably under connected sum. The Alexander polynomial is multiplicative:

[ \Delta_{K_1#K_2}(t)

\Delta_{K_1}(t)\Delta_{K_2}(t), ]

while the knot genus is additive. Such formulas provide necessary conditions for a proposed decomposition, but they do not by themselves establish primeness in every case.

Alternation, chirality, and orientation

A diagram is alternating when overcrossings and undercrossings alternate as one travels along the projected knot. A knot is alternating if it possesses at least one such diagram. Reduced alternating diagrams have strong minimality properties, and the crossing number of an alternating knot is realized by every reduced alternating diagram of that knot. Results of this kind resolved central parts of the nineteenth-century Tait conjectures.

A knot is amphichiral when it is equivalent to its mirror image and chiral otherwise. The figure-eight knot is amphichiral, whereas the trefoil is chiral. Polynomial invariants often detect chirality because mirroring transforms their variables or coefficients in a prescribed manner, although equality between an invariant and its mirror-transformed form does not universally prove amphichirality.

Reversing the orientation of a knot produces its inverse. A knot equivalent to its inverse is invertible, while a knot not equivalent to its inverse is noninvertible. Orientation reversal and reflection are distinct operations, and their combination yields additional symmetry types classified through the mapping behavior of the knot exterior.

Classification and computation

The central decision problem asks whether two finite knot diagrams represent equivalent knots. This problem is decidable: a finite algorithm can determine equivalence by using normal-surface theory or related three-manifold techniques. Decidability does not imply practical efficiency, since direct searches through Reidemeister moves or normal surfaces can grow rapidly with diagram complexity.

The unknotting problem asks whether a given diagram represents the unknot. Several invariants can prove that a knot is nontrivial, but the vanishing or triviality of any one commonly used invariant does not generally establish that it is unknotted. Certificates based on spanning disks, normal surfaces, group representations, or homological invariants supply different forms of computational evidence within exact algorithms.

Knot tables organize prime knots by crossing number and additional symmetry data. Tables are complete only after every diagram below the stated crossing bound has been generated, duplicate knot types have been identified, and minimality has been established. Modern tabulation combines diagram enumeration with polynomial invariants, hyperbolic structures, and exact topological tests.