Knot Theory in Physics

Knot Theory in Physics studies how knotted and linked curves arise in physical systems and how topological invariants of these curves encode physical quantities. Key result: the Jones polynomial and other knot invariants emerge as expectation values of Wilson loops in Chern–Simons gauge theory,

⟨ W_{K} ⟩ \sim V_{K} (q)

, with

q = e^{2 π i / (k + 2)}

. Where it applies: topological quantum field theory, anyons and topological quantum computation, fluid dynamics, plasma physics, DNA topology, and condensed matter. Who/when: Lord Kelvin (1867), Tait, then revived by Witten, Jones, and Atiyah in 1984–1989.

Knot theory is the study of embeddings of the circle $S^{1}$ (a knot) or several circles (a link) into three-dimensional space $R^{3}$ or $S^{3}$ , considered up to ambient isotopy — continuous deformations that do not allow cutting. In physics, knots appear whenever one-dimensional objects (field lines, vortex cores, polymer chains, particle worldlines) wind around each other, and the topology of these configurations is conserved by the dynamics. What makes the subject central to modern theoretical physics is Witten's 1989 discovery that knot invariants — purely topological numbers — are computed by quantum field theories whose action depends only on topology, not metric. A surprising fact: the Jones polynomial, invented in 1984 from operator algebras in statistical mechanics, turned out to be a partition function of a three-dimensional gauge theory, linking subfactors, conformal field theory, quantum groups, and gauge theory into one structure.

The Core Insight

A knot is a topological object: no local measurement distinguishes a knotted curve from an unknotted one. Yet certain integrals and quantum expectation values of fields living along the curve are isotopy invariants — they change only when the curve passes through itself. In physics this means that conserved topological charges (helicity, linking number, anyonic monodromy) are computed by knot invariants, and conversely that knot invariants are realized as physical observables of gauge theories with no propagating degrees of freedom. The bridge is the metric-independence of the Chern–Simons action.

Formal Statement

Knot, Link, and Isotopy

A knot is a smooth (or piecewise-linear) embedding $K : S^{1} ↪ S^{3}$ . A link $L$ with $n$ components is an embedding of $n$ disjoint circles. Two knots are ambient isotopic if there is a continuous family of homeomorphisms $h_{t} : S^{3} \to S^{3}$ , $t \in [0, 1]$ , with $h_{0} = id$ and $h_{1} \circ K_{0} = K_{1}$ . The unknot is the standard round circle. A knot invariant is a map $f$ from knots to some set such that $f (K_{0}) = f (K_{1})$ whenever $K_{0} \sim K_{1}$ .

The most elementary physical invariant is the Gauss linking number of two disjoint closed curves $C_{1}, C_{2}$ :

Lk (C_{1}, C_{2}) = \frac{1}{4 π} \oint_{C_{1}} \oint_{C_{2}} \frac{( r _{1} - r _{2} ) \cdot ( d r _{1} \times d r _{2} )}{∣ r _{1} - r _{2} ∣ ^{3}}

⟳this is the degree of the Gauss map

T^{2} \to S^{2}

(s, t) \mapsto (r_{1} (s) - r_{2} (t)) /∣ \dots ∣

; equivalently it is

\frac{1}{4 π}

times the solid angle swept, an integer because the map has integer degree.

Helicity of a Vector Field

For a divergence-free field $B$ (e.g. magnetic field or vorticity) in a domain $Ω \subset R^{3}$ with $B \cdot \overset{n}{^} = 0$ on $\partial Ω$ , the helicity is

H = \int_{Ω} A \cdot B d^{3} x, B = \nabla \times A .

⟳gauge invariance follows because

A \to A + \nabla χ

changes the integrand by

\nabla χ \cdot B = \nabla \cdot (χ B)

, whose volume integral becomes a boundary flux that vanishes under the boundary condition.

Helicity as Total Linking — Moffatt 1969

If $B$ is concentrated in $N$ thin flux tubes of fluxes $Φ_{i}$ following closed curves $C_{i}$ , then

H = i \sum Φ_{i}^{2} Wr (C_{i})_{(twist+writhe)} + 2 i < j \sum Φ_{i} Φ_{j} Lk (C_{i}, C_{j}) .

Thus helicity measures the topological linking and self-linking of field lines and is conserved by ideal (zero-resistivity) magnetohydrodynamics.

⟳insert

B = \sum_{i} Φ_{i} δ_{C_{i}}

into [2], use

A (r) = \frac{1}{4 π} \sum_{j} Φ_{j} \oint_{C_{j}} \frac{d r _{j} \times ( r - r _{j} )}{∣ r - r _{j} ∣ ^{3}}

(Biot–Savart), and recognize the double integral as Gauss's formula [1]; the self-terms reduce by Călugăreanu–White to

Lk = Wr + Tw

Witten 1989 — Chern–Simons / Jones Polynomial

Let $G = S U (N)$ with Chern–Simons action at level $k \in Z$ ,

S_{C S} = \frac{k}{4 π} \int_{M^{3}} Tr (A \land d A + \frac{2}{3} A \land A \land A),

and let

W_{R} (K) = Tr_{R} P exp \oint_{K} A

be the Wilson loop in representation

R

. Then the normalized expectation value

⟨ W_{R} (K)⟩ = \frac{\int D A W _{R} ( K ) e ^{i S_{C S}}}{\int D A e ^{i S_{C S}}}

is an ambient isotopy invariant of

K

. For

G = S U (2)

R = 2

, it equals the Jones polynomial

V_{K} (q)

evaluated at

q = e^{2 π i / (k + 2)}

⟳

S_{C S}

contains no metric, so

⟨ W ⟩

depends only on the smooth structure of

M^{3}

and the isotopy class of

K

; canonical quantization on

Σ \times R

gives Hilbert spaces equal to conformal blocks of the

S U (N)_{k}

WZW model, and the skein relations of those blocks reproduce the Jones polynomial's defining relation.

The Jones polynomial itself is defined by the skein relation on three diagrams identical except inside a small disk:

t^{- 1} V_{L_{+}} - t V_{L_{-}} = (t^{1/2} - t^{- 1/2}) V_{L_{0}}, V_{unknot} = 1.

⟳

L_{+}, L_{-}, L_{0}

denote an overcrossing, undercrossing, and the smoothing that resolves the crossing; the relation follows from the Hecke/Temperley–Lieb algebra relations

e_{i}^{2} = δ e_{i}

that govern the braid group representation underlying the polynomial.

Why It Holds / Derivation

Invariance of the Linking Number under Reidemeister Moves

Any ambient isotopy of links in $S^{3}$ projects, generically, to a planar diagram, and two diagrams represent isotopic links iff they are related by the three Reidemeister moves (R1: add/remove a kink; R2: pass two strands over/apart; R3: slide a strand across a crossing). To prove $Lk$ is invariant:

Combinatorial form. Assign to each crossing between distinct components $ϵ = \pm 1$ (right/left handed). Define $Lk (C_{1}, C_{2}) = \frac{1}{2} \sum_{inter-component crossings} ϵ$ . This equals the integral [1] (Gauss map = signed count).

R1 invariance. An R1 kink involves a self-crossing of a single component; it contributes nothing to inter-component crossings. Invariant. ✓

R2 invariance. R2 creates or destroys two crossings of opposite sign, $ϵ_{+} + ϵ_{-} = 0$ . The sum is unchanged. ✓

R3 invariance. R3 permutes three crossings but neither creates nor changes signs. ✓

Conclusion. Since the half-signed sum is invariant under all generators of isotopy, $Lk \in Z$ is a link invariant. The factor $\frac{1}{2}$ pairs each crossing with its mirror occurrence in the symmetric sum. $■$

Well-Definedness of the Jones Polynomial via Skein

Setup. Define $V$ recursively by [6]. To show this gives a function of the link, not the diagram, check invariance under Reidemeister moves on $V$ .

R2. Apply [6] to one of the two crossings created by R2. Resolving shows the two new crossings' contributions cancel against the smoothed term, recovering the original $V$ . Algebraically this is the Temperley–Lieb relation $e_{i}^{2} = (t^{1/2} + t^{- 1/2}) e_{i}$ combined with the bracket normalization.

R3. Iterated application of the skein relation on the movable strand reduces both sides to the same combination of smoothings; the identity follows from $e_{i} e_{i \pm 1} e_{i} = e_{i}$ .

R1. The Kauffman bracket $⟨ L ⟩$ is not R1-invariant — a kink multiplies it by $- A^{\pm 3}$ . Multiplying by the writhe correction $(- A^{3})^{- w (L)}$ restores invariance, where $w (L) = \sum_{all crossings} ϵ$ . Substituting $A = t^{- 1/4}$ yields $V_{K} (t)$ . ✓

Conclusion. $V_{K}$ is a Laurent polynomial in $t^{1/2}$ that is an ambient isotopy invariant. $■$

Edward Witten | b. 1951

The only physicist to receive the Fields Medal (1990), awarded largely for the work that connected quantum field theory to knot invariants. In his 1989 paper Quantum Field Theory and the Jones Polynomial, Witten showed that the entire family of polynomial invariants (Jones, HOMFLY, Kauffman) arise as correlation functions of Wilson loops in Chern–Simons theory, a theory with no local degrees of freedom and a metric-independent action. This reframed Vaughan Jones's purely algebraic 1984 discovery — born from von Neumann subfactor theory — as a statement about three-dimensional topological quantum field theory, and it tied together conformal field theory, quantum groups, and the geometry of moduli spaces. Witten's approach also produced new three-manifold invariants (Witten–Reshetikhin–Turaev invariants) by evaluating the partition function $Z (M^{3})$ itself.

Building Intuition

Picture two magnetic field loops threaded through each other like links in a chain. You can stretch, bend, and slide them, but you can never separate them without cutting — the linking number counts how many times one threads the other and is a hard topological constraint. In ideal fluids and plasmas, field lines are "frozen in" the medium (Alfvén's theorem), so this linking is conserved exactly: helicity is a topological charge.

Now replace "field lines" with "particle worldlines." In two spatial dimensions, the worldlines of identical particles braid around each other as time advances, tracing out a knot/braid in 2+1 dimensions. The quantum amplitude picks up a phase or matrix that depends only on the braid's topology — these are anyons. The Chern–Simons theory is precisely the field theory whose Wilson loops compute these braiding statistics; the Jones polynomial is the bookkeeping of how the braid was tied. Crossing a strand over (vs. under) flips $t \to t^{- 1}$ in [6], encoding the chirality that mirror-symmetric physics would forbid but that handedness in 3D allows.

A trefoil knot and its three crossings, with overcrossing/undercrossing skein resolutions

Worked Examples

Linking Number of the Hopf Link

Given: Two circles forming the Hopf link, the simplest nontrivial 2-component link, with one positive-handed clasp (two inter-component crossings of the same sign). Steps:

Draw the standard diagram: the two components cross twice, both with $ϵ = + 1$ .
Apply the combinatorial formula $Lk = \frac{1}{2} \sum ϵ = \frac{1}{2} (+ 1 + 1)$ .
Compute: $Lk = \frac{1}{2} (2) = 1$ .
Cross-check via Jones: the positive Hopf link has $V = - t^{- 5/2} - t^{- 1/2}$ , whose lowest exponent reflects nontrivial linking, confirming non-splittability.

Answer:

Lk (Hopf^{+}) = + 1

(dimensionless integer). For the mirror image,

Lk = - 1

Jones Polynomial of the Right-Handed Trefoil

Given: The right-handed trefoil

3_{1}

, writhe

w = + 3

, three positive crossings. Use the skein relation [6] starting from the unknot normalization

V_{◯} = 1

and a chain to the Hopf link. Steps:

Apply [6] at one trefoil crossing: $L_{+} =$ trefoil, $L_{-} =$ unknot, $L_{0} =$ Hopf link.
$t^{- 1} V_{3_{1}} - t V_{◯} = (t^{1/2} - t^{- 1/2}) V_{Hopf}$ .
Insert $V_{◯} = 1$ and the positive Hopf value $V_{Hopf} = - t^{1/2} - t^{5/2}$ (with this normalization/orientation).
Solve: $t^{- 1} V_{3_{1}} = t + (t^{1/2} - t^{- 1/2}) (- t^{1/2} - t^{5/2})$ .
Expand the product: $(t^{1/2} - t^{- 1/2}) (- t^{1/2} - t^{5/2}) = - t - t^{3} + 1 + t^{2}$ .
So $t^{- 1} V_{3_{1}} = t + (- t - t^{3} + 1 + t^{2}) = - t^{3} + t^{2} + 1$ .
Multiply by $t$ : $V_{3_{1}} = - t^{4} + t^{3} + t$ .

Answer:

V_{3_{1}} (t) = - t^{4} + t^{3} + t

(dimensionless Laurent polynomial). Because

V_{3_{1}} (t) \neq = V_{3_{1}} (t^{- 1})

, the trefoil is chiral — distinct from its mirror image, a fact the Jones polynomial detects but the Alexander polynomial does not.

Common Misconceptions

The Jones polynomial does not classify knots. Distinct knots can share the same Jones polynomial; it is an invariant, not a complete invariant. It is famously unknown whether any nontrivial knot has trivial Jones polynomial

V = 1

(see open problem). Conversely, equal Jones polynomials do not prove two knots are equal. Linking number is not the same as "being linked." The Whitehead link has

Lk = 0

yet its components cannot be separated. Gauss linking detects only the first-order homological linking; higher-order linking requires Milnor invariants or the Massey products. Knot invariants are about embeddings, not abstract circles. Every knot is intrinsically a circle

S^{1}

; the topology that matters is the embedding in

S^{3}

, i.e. the complement

S^{3} ∖ K

Linking Number Fails for the Whitehead Link

Take the Whitehead link, a 2-component link with four crossings between components arranged so two are $+ 1$ and two are $- 1$ . The combinatorial sum gives $Lk = \frac{1}{2} (+ 1 + 1 - 1 - 1) = 0$ . Yet the components are demonstrably inseparable: the link is nonsplit, as shown by its nontrivial Jones (multivariable) polynomial and by the fact that one component is a nontrivial element of $π_{1}$ of the other's complement. This proves the Gauss linking number is not a faithful detector of linking — topology hides in higher invariants.

Connections

Knot theory in physics sits at the crossroads of several deep structures:

Chern–Simons Theory — the topological gauge theory whose Wilson-loop correlators are the knot polynomials, via [4]–[5].
Conformal Field Theory — the Hilbert spaces of Chern–Simons on $Σ \times R$ are spaces of conformal blocks of the WZW model; braiding equals the monodromy of the Knizhnik–Zamolodchikov equation.
Quantum Groups — the quantized universal enveloping algebra $U_{q} (g)$ provides $R$ -matrices solving the Yang–Baxter equation, giving an algebraic construction (Reshetikhin–Turaev) of the same invariants.
Topological Quantum Computation — anyon braiding realizes unitary representations of the braid group; non-Abelian anyons (e.g. Fibonacci) give universal quantum gates whose amplitudes are Jones polynomial evaluations.
Statistical Mechanics — the Temperley–Lieb algebra underlying the Jones polynomial first arose in the Potts and ice-type lattice models; Jones's discovery came from von Neumann algebras modeling these.
Magnetohydrodynamics — helicity [2]–[3] is the fluid/plasma incarnation of linking, conserved in the ideal limit.

Real-World Applications

Topological quantum computers (Microsoft Station Q, Majorana platforms). Information is stored in the braiding of non-Abelian anyons; logical gates are knots/braids, and fault tolerance comes from topological protection. Jones polynomial evaluation at roots of unity is the natural computational primitive (BQP-complete via Freedman–Kitaev–Larsen–Wang).

DNA topology and topoisomerase enzymes. Circular DNA gets supercoiled, knotted, and catenated; the linking number $Lk = Tw + Wr$ governs replication. Topoisomerase enzymes change $Lk$ in controlled steps; gel electrophoresis separates topoisomers by writhe.

Solar and astrophysical plasmas. Magnetic helicity conservation constrains coronal loops, solar flares, and the solar dynamo; helicity injection drives coronal mass ejections. The Taylor relaxation hypothesis says plasmas relax to minimum energy at fixed helicity.

Knotted optical and quantum fields. "Knotted light" — electromagnetic fields whose field lines form Hopf links — and knotted vortices in Bose–Einstein condensates and superfluid helium realize topological field configurations in the lab.

Polymer physics and gel rheology. Knotting and entanglement of long polymer chains determine viscosity, reptation dynamics, and mechanical strength; knot probability grows with chain length.

Tied vortex knots in fluids (Kleckner–Irvine 2013). Experimentally generated knotted and linked vortex loops in water demonstrated that fluid helicity is not perfectly conserved — knots untie via reconnection, a key test of topological fluid mechanics.

High-temperature superconductors and the fractional quantum Hall effect. Quasiparticle excitations in FQH states at filling $ν = 1/3$ (Laughlin) and $ν = 5/2$ (possibly Moore–Read non-Abelian) are described by effective Chern–Simons theories; their braiding statistics are knot-theoretic.

Historical Development

1833 — Carl Friedrich Gauss writes down the linking-number integral [1] while studying electromagnetism and the orbits of asteroids; arguably the birth of topology in physics.
1867 — William Thomson (Lord Kelvin) proposes the vortex atom theory: atoms are knotted vortices in the luminiferous aether, distinct elements being distinct knots. Though physically wrong, it motivates the first systematic knot tabulation.
1877–1885 — Peter Guthrie Tait, aided by Thomas Kirkman and C.N. Little, compiles the first knot tables up to ~10 crossings and conjectures the Tait conjectures on alternating knots.
1926–1928 — James Waddell Alexander defines the Alexander polynomial from the knot group / Seifert matrix, the first polynomial invariant.
1961 — Kurt Reidemeister's moves and the foundations of combinatorial knot theory become standard; Călugăreanu's $Lk = Tw + Wr$ theorem (1959–61).
1969 — H. Keith Moffatt proves helicity equals total linking [3], founding topological fluid dynamics.
1984 — Vaughan Jones, studying von Neumann subfactors and the Temperley–Lieb algebra, discovers the Jones polynomial [6]; it detects chirality the Alexander polynomial misses.
1985–1987 — The HOMFLY and Kauffman polynomials generalize Jones; the Yang–Baxter / quantum-group connection emerges (Drinfeld, Jimbo).
1989 — Edward Witten's Quantum Field Theory and the Jones Polynomial recasts all these invariants as Chern–Simons Wilson loops [4]–[5]; Reshetikhin–Turaev give the rigorous quantum-group construction, yielding 3-manifold invariants. Jones (1990) and Witten share major prizes.
2000s–present — Khovanov homology (2000) categorifies the Jones polynomial; Freedman–Kitaev–Larsen–Wang (2002) prove anyon braiding is universal for quantum computation.

Subtleties & Limits

1. The Jones unknotting conjecture. Is the unknot the only knot with trivial Jones polynomial

V_{K} (t) = 1

? Despite decades of effort and large computer searches, this remains open. (Khovanov homology does detect the unknot — Kronheimer–Mrowka 2011 — but the Jones polynomial alone is unresolved.) 2. Volume conjecture. Conjectured relation between the asymptotics of the colored Jones polynomials

J_{N} (K; e^{2 π i / N})

and the hyperbolic volume of the knot complement:

lim_{N \to \infty} \frac{2 π}{N} lo g ∣ J_{N} ∣ = Vol (S^{3} ∖ K)

. Proven only for special cases; ties quantum invariants to geometry. 3. Helicity in resistive/viscous flows. [3] is exactly conserved only in the ideal limit. Real plasmas and fluids reconnect; how helicity is dissipated, transferred between scales, and bounds energy (the "topological lower bound" on magnetic energy) is an active research frontier — and Kleckner–Irvine's experiments show knots untie faster than naive conservation predicts. 4. Categorification and physics. Khovanov homology and knot Floer homology assign graded vector spaces (not just polynomials) to knots. Witten and Gaiotto reinterpret Khovanov homology via 4d and 5d gauge theory (the "Kapustin–Witten" and "

Ω

-deformation" programs), but a complete physical understanding is incomplete. 5. Non-Abelian helicity / higher invariants. Gauss linking captures only first-order linking; the physical role of Milnor invariants, Massey products, and Vassiliev invariants in real fields is largely unexplored beyond formal analogies.