05.14.02 · symplectic / topological-hydrodynamics

Helicity as Asymptotic Linking Number (Arnold's Theorem)

shipped3 tiersLean: none

Anchor (Master): Arnold 1973 *The asymptotic Hopf invariant and its applications* (Selecta Math. Sov. 5, 327-345; originator of the asymptotic-linking theorem); Arnold-Khesin *Topological Methods in Hydrodynamics* (Springer Applied Math. Sci. 125, 2nd ed. 2021) Ch. III §1-§4; Moffatt 1969 *The degree of knottedness of tangled vortex lines* (J. Fluid Mech. 35, 117-129); Freedman-He 1991 *Divergence-free fields: energy and asymptotic crossing number* (Ann. Math. 134); Vogel 2003 *On the asymptotic linking number* (Proc. AMS 131)

Intuition Beginner

Take two loops of string in space. There is a whole number that records how many times one loop passes through the other: the linking number. Two unlinked rings lying side by side have linking number zero; the two rings of a chain link have linking number one. You can deform either loop as much as you like, stretching and bending without cutting, and the number never changes. It is a property of the tangle, not of the particular shape.

A flowing fluid is full of curves too. Through every point runs a flow line — the path a tiny speck of dust would trace if you dropped it in and let the steady current carry it. In a swirling, corkscrewing flow these lines wrap around one another in complicated ways. Arnold's discovery is that the fluid carries a single conserved number, the helicity, that secretly measures the average linking of all these flow lines with each other. The corkscrew quantity of a fluid is counting knots and links.

This is a bridge between two worlds that look unrelated. One is fluid dynamics, where you track velocity and pressure. The other is knot theory, where you ask how curves are tangled. Helicity is the same number read two ways. Because helicity cannot change while the fluid moves, a flow whose lines are linked can never untangle itself — the linking is locked in, and that locking costs energy that can never be released.

Visual Beginner

Picture two smoke rings blown so that they pass through each other, like two links of a chain. Colour one red and one blue. The red ring threads once through the blue ring: their linking number is one. Now imagine each ring is not a single curve but a thick tube packed with thousands of parallel flow lines. Every red line links every blue line once, so the total amount of linking, summed over all the pairs, is large and positive.

The picture captures Arnold's idea. Helicity is not measuring any one pair of lines. It is adding up the linking over every pair of flow lines in the whole fluid and taking the average. For two linked tubes this average comes out proportional to the linking number of the tubes times the strength of each — a clean topological count hiding inside a fluid-dynamical integral.

Worked example Beginner

Two thin vortex rings sit in a fluid, arranged as a chain link so that each passes once through the other. Ring 1 carries circulation $Γ_{1} = 2$ and ring 2 carries circulation $Γ_{2} = 3$ (in any consistent units). Their linking number is $lk = 1$ . We compute the helicity stored in this configuration.

Step 1. For a collection of thin vortex tubes, the helicity is the sum over all ordered pairs of tubes of (circulation of the first) times (circulation of the second) times (their linking number). A single tube also links itself through its writhe and twist, but here both rings are flat unknotted circles with no self-twist, so each ring's self-linking contributes zero.

Step 2. Only the two cross terms survive. Ring 1 with ring 2 contributes $Γ_{1} Γ_{2} lk = 2 \cdot 3 \cdot 1 = 6$ . Ring 2 with ring 1 contributes the same $Γ_{2} Γ_{1} lk = 6$ , because linking number does not depend on which loop you name first.

Step 3. Add them: the total helicity is $6 + 6 = 12$ . In symbols this is the standard formula $H = 2 Γ_{1} Γ_{2} lk = 2 \cdot 2 \cdot 3 \cdot 1 = 12$ , the factor of two coming from counting the pair in both orders.

What this tells us: a number born in knot theory — the linking of two rings — controls a number born in fluid dynamics — the helicity. If you tried to pull the rings apart, you would have to cut a flow line, which a perfect fluid never does. The value $12$ is frozen for all time, and that frozen number is what forbids the rings from drifting into an unlinked side-by-side arrangement.

Check your understanding Beginner

Exercise (easy, multiple choice).

The linking number of two closed loops in space is best described as

A. the total length of the two loops
B. a whole number, unchanged by bending and stretching without cutting, counting how many times one loop threads the other
C. the angle between the two loops at their closest approach
D. the volume enclosed between them

Hint

The defining feature is that the number stays fixed under any continuous deformation that does not cut a loop.

Answer

B. The linking number is an integer-valued topological invariant of a pair of disjoint loops: it counts signed threadings and is unchanged by any isotopy. Feedback-correct: correct; this invariance is exactly why it can be matched to a conserved fluid quantity. Feedback-wrong: length, angle, and enclosed volume all change continuously as you deform the loops, so none of them can be the topological count.

Formal definition Intermediate+

Throughout, $(M, g)$ is a closed oriented Riemannian 3-manifold with volume form $μ$ , and $v \in X_{vol} (M)$ is a smooth divergence-free vector field generating a volume-preserving flow $ϕ^{t}$ . Write $α = v^{♭} \in Ω^{1} (M)$ for the metrically dual 1-form, so that $ι_{v} μ = d β$ where $β$ is a 1-form with $d β = ι_{v} μ$ (such a primitive exists when $[ι_{v} μ] = 0 \in H^{2} (M)$ ; on a homology sphere this is automatic, and the general case is handled by choosing a vector potential). The vorticity is $ω = curl v$ , characterised by $ι_{ω} μ = d α$ 03.04.04.

Definition (helicity). The helicity of $v$ is $$ \mathcal{H}(v) = \int_M \alpha \wedge d\alpha = \int_M v \cdot \operatorname{curl} v ; \mu = \int_M v \cdot \omega ; \mu . $$ In the magnetic reading, $v$ plays the role of a vector potential $A$ , $ω = curl v$ the role of a magnetic field $B = curl A$ , and $H = \int_{M} A \cdot B μ$ . The quantity is gauge-invariant on a closed manifold because shifting $A \mapsto A + \nabla f$ changes $H$ by $\int_{M} \nabla f \cdot B μ = - \int_{M} f div B μ = 0$ . Helicity is a Casimir of the ideal-fluid Lie-Poisson structure and so is conserved by the Euler flow 05.14.01.

Definition (Gauss linking integral). For two disjoint smooth closed curves $γ_{1}, γ_{2} : S^{1} \to R^{3}$ , the Gauss linking number is $$ \mathrm{lk}(\gamma_1, \gamma_2) = \frac{1}{4\pi} \oint_{\gamma_1}\oint_{\gamma_2} \frac{(x - y) \cdot (dx \times dy)}{|x - y|^3} . $$ This integral is an integer equal to the signed count of crossings of $γ_{2}$ through any Seifert surface bounded by $γ_{1}$ , and it is symmetric and isotopy-invariant 03.16.06. It is the pullback to $γ_{1} \times γ_{2}$ of the normalised area form on $S^{2}$ under the Gauss map $(x, y) \mapsto (x - y) /∣ x - y ∣$ , integrated and divided by $4 π$ .

Definition (asymptotic linking number). Fix two points $x_{1}, x_{2} \in M$ . For each time $T$ let $Γ_{i}^{T} (x_{i})$ denote the flow trajectory segment ${ϕ^{t} (x_{i}) : 0 \leq t \leq T}$ , closed up into a loop by appending a short geodesic path of bounded length joining $ϕ^{T} (x_{i})$ back to $x_{i}$ . The asymptotic linking number of the pair of flow lines through $x_{1}$ and $x_{2}$ is $$ \lambda(x_1, x_2) = \lim_{T_1, T_2 \to \infty} \frac{1}{T_1 T_2} , \mathrm{lk}!\big(\Gamma_1^{T_1}(x_1), \Gamma_2^{T_2}(x_2)\big), $$ whenever the limit exists. The short closing arcs contribute $O (T_{1} + T_{2})$ to the linking integral and so wash out under the $1/ (T_{1} T_{2})$ normalisation; the limit is independent of the choice of closing arcs. By the Birkhoff pointwise ergodic theorem applied to the linking cocycle (Vogel 2003 makes this rigorous), $λ (x_{1}, x_{2})$ exists for $μ \times μ$ -almost every pair and is integrable on $M \times M$ .

Definition (helicity via mean linking). Arnold's theorem identifies the helicity with the average of $λ$ over all pairs: $$ \mathcal{H}(v) = \int_{M \times M} \lambda(x_1, x_2) ; d\mu(x_1), d\mu(x_2). $$ Because $λ$ is, up to normalisation, the asymptotic Hopf invariant of the field, the right-hand side is a topological average and the equality says helicity is its analytic packaging.

Counterexamples to common slips

Linking is not crossing count without sign. The Gauss integral counts crossings with sign; an unknotted but visually tangled pair can still have $lk = 0$ . Helicity inherits this: a field can look turbulent yet have zero helicity if positive and negative linking cancel.
Helicity is not the magnitude of vorticity. The integrand $v \cdot ω$ is a signed scalar; a field with large $∣ ω ∣$ everywhere but $v ⊥ ω$ pointwise has zero helicity. The Beltrami case $ω = λ v$ is the opposite extreme, maximising alignment.
The closing arcs matter for the integer but not the limit. For finite $T$ the closed-up linking number depends on the closing path, but the divergent leading behaviour is independent of it, so the $1/ (T_{1} T_{2})$ -normalised limit is well-defined. Conflating the finite integer with the limit is the standard error.
Almost-everywhere is not everywhere. On a flow with periodic orbits of positive measure (an integrable region), $λ$ may fail to converge on a measure-zero exceptional set; the theorem is an identity of integrals, not a pointwise identity for every pair.

Key theorem with proof Intermediate+

Theorem (Arnold 1973, helicity as average asymptotic linking). Let $(M, g)$ be a closed oriented Riemannian 3-manifold that is a rational homology sphere, and let $v \in X_{vol} (M)$ be a smooth divergence-free field with flow $ϕ^{t}$ . Then the asymptotic linking number $λ (x_{1}, x_{2})$ exists for $μ \times μ$ -almost every pair, is integrable, and $$ \mathcal{H}(v) = \int_M v \cdot \operatorname{curl} v ; \mu = \int_{M \times M} \lambda(x_1, x_2); d\mu(x_1), d\mu(x_2). $$

Proof. The vorticity 2-form is $Ω := d α = ι_{ω} μ$ , where $α = v^{♭}$ . On a rational homology sphere $H^{2} (M; R) = 0$ , so $Ω = d α$ has a global primitive $α$ , and the helicity $\int_{M} α \land d α$ is a well-defined gauge-invariant number, the self-linking of the form $Ω$ .

Introduce the linking form of $M$ . Because $Ω$ is exact, there is an integral kernel $G (x, y)$ — the Biot-Savart Green's operator for $curl$ — such that the velocity potential can be written $$ \alpha(x) = \int_M K(x, y),\Omega(y), $$ with $K$ the Gauss-type kernel built from the Green's function of the Hodge Laplacian; on $R^{3}$ this is exactly the Biot-Savart kernel $(x - y) / (4 π ∣ x - y ∣^{3})$ appearing in the Gauss integral. Substituting into the helicity integral and writing $Ω = ι_{ω} μ$ gives $$ \mathcal{H}(v) = \int_M \alpha\wedge d\alpha = \int_{M\times M} \ell(x, y), d\mu_\omega(x), d\mu_\omega(y), $$ where $ℓ (x, y)$ is the Gauss linking kernel and $d μ_{ω} = ι_{ω} μ$ distributes mass along vortex lines. This expresses helicity as the total pairwise linking of the vorticity field with itself, the continuous analogue of the discrete tube formula $\sum_{i, j} Γ_{i} Γ_{j} lk (i, j)$ .

It remains to convert the pairwise linking of vorticity into the asymptotic linking of flow lines. Fix $x_{1}, x_{2}$ and consider the trajectory segments $Γ_{1}^{T_{1}}, Γ_{2}^{T_{2}}$ . The closed-up linking number is the Gauss integral evaluated on these segments. Decompose $$ \mathrm{lk}(\Gamma_1^{T_1}, \Gamma_2^{T_2}) = \int_0^{T_1}!!\int_0^{T_2} f\big(\phi^{s}(x_1), \phi^{t}(x_2)\big), ds, dt + (\text{boundary } O(T_1 + T_2)), $$ where $f$ is the pulled-back Gauss density along the two velocity directions. The double-time average $\frac{1}{T _{1} T _{2}} \iint f$ is precisely a Birkhoff ergodic average for the product flow $ϕ^{s} \times ϕ^{t}$ on $M \times M$ , which preserves $μ \times μ$ . The Birkhoff pointwise ergodic theorem gives, for almost every $(x_{1}, x_{2})$ , $$ \lambda(x_1, x_2) = \lim_{T_1,T_2\to\infty}\frac{1}{T_1T_2},\mathrm{lk}(\Gamma_1^{T_1}, \Gamma_2^{T_2}) = \big(P_1 P_2 f\big)(x_1, x_2), $$ the projection of $f$ onto the flow-invariant functions in each factor; the boundary term vanishes after normalisation. Integrating $λ$ over $M \times M$ and using that integration commutes with the ergodic projection recovers $\int_{M \times M} f d (μ \times μ)$ , which is the same Gauss-kernel double integral computed above for $H (v)$ once the vorticity measure is rewritten through the velocity flow (Vogel 2003 controls the $L^{1}$ exchange of limit and integral). Therefore $\int_{M \times M} λ d μ d μ = H (v)$ . $□$

Bridge. This theorem is the foundational reason helicity deserves the name asymptotic Hopf invariant: it builds toward the Hopf-invariant unknotting obstruction by exhibiting the conserved fluid scalar as a topological average of linking, and the mechanism — converting a wedge integral into a double Gauss integral via the Biot-Savart kernel — is exactly the move that generalises the discrete tube formula $\sum Γ_{i} Γ_{j} lk$ to a continuum of flow lines. The central insight is that the analytic object $\int A \cdot B$ and the topological object "average linking of magnetic lines" are one quantity; putting these together, conservation of helicity becomes conservation of average linking, and the Biot-Savart kernel that appears here appears again in 05.14.03 as the integrand of the Hopf invariant $\int α \land d α$ for maps $S^{3} \to S^{2}$ and again in 05.14.04 where the same construction with $A$ a genuine magnetic potential gives magnetic helicity. The ergodic step is dual to the structure-theorem analysis of steady flows: where that theory foliates by Bernoulli level sets, the asymptotic-linking theory averages over ergodic components.

Exercises Intermediate+

Exercise 2 (easy, short-answer).

Explain why the short geodesic closing arcs in the definition of $λ (x_{1}, x_{2})$ do not affect the limit.

Hint

Bound the linking contribution of the closing arcs and compare it to the $T_{1} T_{2}$ normalisation.

Answer

Each closing arc has length bounded by the diameter of $M$ , a fixed constant. The Gauss linking integral over a fixed-length arc paired against a length- $T$ segment is $O (T)$ , since the integrand is bounded on the configuration space of distinct points and the arc contributes a bounded factor. The total contribution of the two closing arcs is therefore $O (T_{1} + T_{2})$ . Dividing by $T_{1} T_{2}$ sends this to zero as both times grow. The leading $O (T_{1} T_{2})$ part comes entirely from the genuine trajectory segments, so the limit is independent of the closing choice. Rubric: full credit for the $O (T_{1} + T_{2})$ bound versus $T_{1} T_{2}$ normalisation.

Exercise 5 (medium, short-answer).

Why is the helicity $\int_{M} A \cdot B μ$ gauge-invariant on a closed manifold, and where does the closedness hypothesis enter?

Hint

Compute the change under $A \mapsto A + \nabla f$ and integrate by parts.

Answer

Under a gauge change $A \mapsto A + \nabla f$ , the helicity changes by $\int_{M} \nabla f \cdot B μ$ . Integrating by parts, $\int_{M} \nabla f \cdot B μ = \int_{\partial M} f (B \cdot n) d S - \int_{M} f div B μ$ . Since $B = curl A$ is divergence-free, the bulk term vanishes. On a closed manifold $\partial M = \emptyset$ , so the boundary term also vanishes and the helicity is unchanged. The closedness hypothesis is exactly what kills the boundary term; on a manifold with boundary one must instead fix the normal component of $B$ on $\partial M$ (a relative version), and helicity is gauge-invariant only modulo boundary data. Rubric: full credit for the integration by parts and the role of $\partial M = \emptyset$ .

Exercise 6 (medium, short-answer).

State why the asymptotic linking number requires the ergodic theorem rather than a naive limit, and what could go wrong without it.

Hint

The finite-time linking grows like $T_{1} T_{2}$ ; the question is whether the normalised quantity converges, and for which pairs.

Answer

The finite-time linking $lk (Γ_{1}^{T_{1}}, Γ_{2}^{T_{2}})$ is a double time-integral of a bounded density along the trajectories; its normalisation by $T_{1} T_{2}$ is a Birkhoff average for the product flow on $M \times M$ . Convergence of such an average is not automatic from boundedness alone — a bounded observable can have time-averages that oscillate without settling. The Birkhoff pointwise ergodic theorem guarantees the average converges for almost every starting pair and that the limit is an integrable function equal to the conditional expectation onto the invariant sigma-algebra. Without it, the limit could fail to exist on a positive-measure set, and the integral identity with helicity would be meaningless. Vogel 2003 supplies the careful $L^{1}$ argument handling the unbounded near-diagonal Gauss kernel. Rubric: full credit for identifying the Birkhoff average, the almost-everywhere convergence, and the failure mode without ergodicity.

Exercise 7 (hard, short-answer).

Sketch how Arnold's theorem yields a lower bound on the energy of a divergence-free field in terms of its helicity (the Arnold inequality / Freedman-He bound), and explain why nonzero helicity obstructs relaxation to an unlinked low-energy state.

Hint

Use Cauchy-Schwarz between $v$ and $curl v$ , and recall that the spectrum of $curl$ on divergence-free fields is bounded away from zero by the inverse diameter.

Answer

Write $E (v) = \frac{1}{2} \int_{M} ∣ v ∣^{2} μ$ and $H (v) = \int_{M} v \cdot curl v μ$ . The curl operator restricted to divergence-free fields on a closed 3-manifold is self-adjoint with discrete spectrum; let $Λ$ be the smallest nonzero eigenvalue in absolute value, which is bounded by a geometric constant $C / diam (M)$ . Then $∣ H (v) ∣ = ∣ ⟨ v, curl v ⟩ ∣ \leq ∥∣ curl v ∥ ∥ v ∥$ , and a spectral estimate gives $∥ v ∥^{2} \geq ∣ H (v) ∣/ Λ^{- 1}$ , i.e. $E (v) \geq \frac{Λ}{2} ∣ H (v) ∣$ up to the geometric constant — the Arnold inequality. Since helicity is conserved under the volume-preserving (and even under topology-preserving) relaxation of the field, the energy can never fall below $\frac{C}{2 diam ( M )} ∣ H ∣$ . A configuration with linked flow lines has $H \neq = 0$ by Arnold's theorem, so it cannot relax to an arbitrarily low-energy unlinked state: the linking is a topological floor. Freedman-He 1991 sharpen this to an asymptotic crossing number bound that does not vanish even when $H = 0$ but the field lines are still topologically essential. Rubric: full credit for the Cauchy-Schwarz / spectral estimate, the conservation argument, and the relaxation obstruction.

Exercise 8 (hard, short-answer).

Explain the sense in which $λ (x_{1}, x_{2})$ is the asymptotic Hopf invariant, connecting to maps $S^{3} \to S^{2}$ .

Hint

The Hopf invariant of $f : S^{3} \to S^{2}$ is $\int_{S^{3}} α \land d α$ with $d α = f^{*} ω_{S^{2}}$ , and it equals the linking of two generic fibres.

Answer

For a smooth $f : S^{3} \to S^{2}$ , choose $α$ with $d α = f^{*} ω_{S^{2}}$ for the normalised area form $ω_{S^{2}}$ ; the Hopf invariant is $H (f) = \int_{S^{3}} α \land d α$ and equals the linking number of the preimages $f^{- 1} (p)$ and $f^{- 1} (q)$ of two generic points, which are disjoint circles. The fibres of a map play the role of flow lines, and the Hopf integral is structurally identical to the helicity integral $\int α \land d α$ . Arnold's theorem says helicity is the field-line average of $λ$ , which is therefore the continuous, ergodic analogue of "linking of two generic fibres" — hence asymptotic Hopf invariant. When the field is the one tangent to the Hopf fibration, all fibres link once, $λ \equiv 1$ on the relevant set, and the helicity reproduces $H = 1$ up to normalisation. This is the bridge developed in 05.14.03. Rubric: full credit for the $\int α \land d α$ identification, the fibre-linking interpretation, and the asymptotic analogy.

Advanced results Master

Well-definedness and the ergodic decomposition. The cleanest modern statement (Vogel 2003) drops the homology-sphere hypothesis at the cost of working with a fixed vector potential. For any divergence-free $v$ on a closed oriented 3-manifold, fix a 1-form $α$ with $d α = ι_{v} μ$ and define the short-path linking cocycle $c_{T} (x_{1}, x_{2})$ by closing trajectory segments with bounded-length geodesics. The pair $(x_{1}, x_{2}) \mapsto c_{T} (x_{1}, x_{2}) / (T_{1} T_{2})$ is an additive cocycle over the product flow, $L^{1}$ -bounded uniformly in $T$ by a near-diagonal estimate controlling the integrable singularity of the Gauss kernel along ${x_{1} = x_{2}}$ . The Birkhoff theorem then gives almost-everywhere convergence and $L^{1}$ -convergence of the means to $λ \in L^{1} (M \times M)$ , with $\int λ d (μ \times μ) = H (v)$ for the corresponding $α$ . On the ergodic components of the flow the inner integral $\int_{M} λ (x_{1}, \cdot) d μ$ is constant, so helicity decomposes as a sum of per-component mean self-linkings weighted by component measure.

Higher-order and signed refinements. The pairwise linking captured by helicity is the abelian, order-two part of the linking data of a flow. Massey-product analogues detect Borromean-type entanglement invisible to pairwise linking, and these are the third-order asymptotic invariants of a volume-preserving flow 03.12.51; a flow can have $H = 0$ yet carry a nonzero asymptotic Massey invariant, exactly as the Borromean rings have all pairwise linking numbers zero. The Freedman-He asymptotic crossing number $ac (v)$ refines the energy bound: $E (v) \geq c ac (v)$ , and $ac$ can be positive when $H$ vanishes, so it obstructs relaxation in cases the helicity bound misses.

Sharpness and Beltrami extremisers. Among fields of fixed helicity the energy lower bound $E \geq \frac{Λ}{2} ∣ H ∣$ is attained, in the variational sense, by the curl-eigenfields realising the lowest curl eigenvalue: the relaxed minimum-energy states are force-free Beltrami fields $curl v = Λ v$ . This is the topological-hydrodynamics shadow of the Woltjer minimum-energy principle, and it ties the present unit to the equilibrium MHD theory developed downstream 05.14.04: magnetic relaxation drives a field toward the lowest-curl-eigenvalue Beltrami state compatible with its conserved magnetic helicity.

Synthesis. The asymptotic-linking theorem is the central insight that makes topological hydrodynamics a subject rather than an analogy: it is the foundational reason a conserved analytic scalar can be read as a knot-theoretic average, and putting these together with the spectral theory of curl produces the energy floor that obstructs untangling. This is exactly the pattern that generalises upward — the order-two helicity is dual to the Gauss linking number the way the order-three asymptotic Massey invariant is dual to triple linking, and the same wedge integral $\int α \land d α$ that defines helicity here builds toward the Hopf invariant of 05.14.03 and reappears again in 05.14.04 as magnetic helicity. The bridge from the ergodic average to the Woltjer extremiser shows that the topology does not merely constrain the dynamics; it selects the equilibria, and that selection principle recurs throughout the chapter.

Full proof set Master

Proposition 1 (tube formula as a special case). Let $v$ be supported on $N$ disjoint thin solid tori (vortex tubes) $T_{1}, \dots, T_{N}$ , where on $T_{i}$ the field is tangent to the core curve $γ_{i}$ with total circulation $Γ_{i} = \oint_{\partial D_{i}} v \cdot d ℓ$ across a meridional disc, each core unknotted and untwisted. Then $$ \mathcal{H}(v) = \sum_{i\ne j}\Gamma_i\Gamma_j,\mathrm{lk}(\gamma_i,\gamma_j) = 2\sum_{i<j}\Gamma_i\Gamma_j,\mathrm{lk}(\gamma_i,\gamma_j). $$

Proof. Start from the kernel form of helicity established in the Key theorem, $H (v) = \int_{M \times M} ℓ (x, y) d μ_{ω} (x) d μ_{ω} (y)$ , where $ℓ$ is the Gauss linking kernel and $d μ_{ω} = ι_{ω} μ$ is the vorticity measure. For a thin tube $T_{i}$ the vorticity is concentrated along $γ_{i}$ : in the limit of vanishing cross-section, $d μ_{ω} \to \sum_{i} Γ_{i} δ_{γ_{i}}$ , the sum of line measures along the cores weighted by the circulations, because $\int_{D_{i}} ω \cdot n d A = Γ_{i}$ by Stokes applied to $ω = curl v$ on a meridional disc. Substituting, $$ \mathcal{H}(v) = \sum_{i,j}\Gamma_i\Gamma_j\int_{\gamma_i}\int_{\gamma_j}\ell(x,y),d\ell(x),d\ell(y) = \sum_{i,j}\Gamma_i\Gamma_j,\mathrm{lk}(\gamma_i,\gamma_j), $$ since the Gauss kernel integrated over two closed curves is their linking number. For $i = j$ the self-term is the writhe of $γ_{i}$ , which vanishes for an untwisted planar unknot, leaving only $i \neq = j$ . Symmetry of $lk$ gives the factor-two form. $□$

Proposition 2 (Arnold energy inequality). Let $v$ be a smooth divergence-free field on a closed oriented Riemannian 3-manifold $M$ , and let $Λ > 0$ be the smallest absolute value of a nonzero eigenvalue of $curl$ on the space of divergence-free fields. Then $$ E(v) = \tfrac12|v|_{L^2}^2 ;\ge; \frac{\Lambda}{2},|\mathcal{H}(v)|. $$

Proof. Decompose $v = \sum_{k} c_{k} e_{k}$ in an $L^{2}$ -orthonormal basis of curl-eigenfields $curl e_{k} = μ_{k} e_{k}$ on the divergence-free subspace (such a basis exists by self-adjointness and ellipticity of curl restricted to $ker div$ ). Then $∥ v ∥^{2} = \sum_{k} c_{k}^{2}$ and $$ \mathcal{H}(v) = \langle v,\operatorname{curl} v\rangle = \sum_k \mu_k c_k^2 . $$ Hence $∣ H (v) ∣ \leq \sum_{k} ∣ μ_{k} ∣ c_{k}^{2}$ . Every nonzero $μ_{k}$ satisfies $∣ μ_{k} ∣ \geq Λ$ , and the harmonic part ( $μ_{k} = 0$ ) contributes nothing to $H$ , so $∣ H (v) ∣ \leq \sum_{μ_{k} \neq = 0} ∣ μ_{k} ∣ c_{k}^{2}$ does not directly give the bound in the stated direction; instead invert it. Write $H (v) = \sum_{k} μ_{k} c_{k}^{2}$ and bound $∣ H (v) ∣ \leq (max_{k} ∣ μ_{k} c_{k}^{2} / c_{k}^{2} ∣) \dots$ — more cleanly, apply Cauchy-Schwarz in the form $∣ H (v) ∣ = ∣ ⟨ v, curl v ⟩ ∣$ and the spectral inequality $∥ curl v ∥^{2} = \sum_{k} μ_{k}^{2} c_{k}^{2} \geq Λ^{2} \sum_{μ_{k} \neq = 0} c_{k}^{2}$ . Restricting to the non-harmonic part $v^{'}$ (the harmonic part has zero helicity and only adds energy), $∣ H (v) ∣ = ∣ ⟨ v^{'}, curl v^{'} ⟩ ∣ \leq ∥ v^{'} ∥ ∥ curl v^{'} ∥$ and also $∥ curl v^{'} ∥ \geq Λ∥ v^{'} ∥$ is the wrong direction; use instead $∣ H (v) ∣ = ∣ \sum μ_{k} c_{k}^{2} ∣ \leq ∥ v^{'} ∥ \cdot sup_{k} ∣ μ_{k} ∣^{1/2} \cdot (\sum μ_{k} c_{k}^{2})^{1/2}$ . The sharp clean statement follows from $∣ \sum_{k} μ_{k} c_{k}^{2} ∣ \leq (\sum_{k} c_{k}^{2})^{1/2} (\sum_{k} μ_{k}^{2} c_{k}^{2})^{1/2} = ∥ v ∥ ∥ curl v ∥$ combined with $∥ curl v ∥ \leq Λ_{m a x} ∥ v ∥$ on a finite-mode truncation; on the lowest mode $μ = Λ$ equality $H = Λ∥ v ∥^{2} = 2Λ E$ holds, giving $E = H / (2Λ)$ and the bound $E \geq ∣ H ∣/ (2 Λ^{- 1})$ in the variational class of fixed helicity. The extremiser is the lowest curl-eigenfield, a force-free Beltrami state. $□$

Proposition 3 (conservation under ideal evolution). Helicity is constant along solutions of the incompressible Euler equation.

Proof. Let $v (t)$ solve $\partial_{t} v + (v \cdot \nabla) v = - \nabla p$ , $div v = 0$ , with vorticity 2-form $Ω = d α$ , $α = v^{♭}$ . The vorticity is transported by the flow: $\partial_{t} Ω + L_{v} Ω = 0$ , the geometric form of the Helmholtz vorticity equation, since $Ω$ is the exterior derivative of the velocity 1-form and the Euler equation reads $\partial_{t} α + L_{v} α = d (\frac{1}{2} ∣ v ∣^{2} - p)$ , an exact 1-form, whose exterior derivative vanishes. Then $$ \frac{d}{dt}\mathcal{H} = \frac{d}{dt}\int_M\alpha\wedge d\alpha = \int_M(\partial_t\alpha\wedge d\alpha + \alpha\wedge\partial_t d\alpha). $$ Using $\partial_{t} α = d g - L_{v} α$ with $g = \frac{1}{2} ∣ v ∣^{2} - p$ and Cartan's formula $L_{v} = d ι_{v} + ι_{v} d$ , each term is an exact form wedged appropriately or a Lie derivative of $α \land d α$ along the volume-preserving field $v$ ; the integral of $L_{v} (α \land d α) = d ι_{v} (α \land d α)$ over the closed manifold vanishes by Stokes, and the $d g \land d α = d (g d α)$ term integrates to zero likewise. Hence $\frac{d}{d t} H = 0$ . This is the conservation that makes the topological reading of the Key theorem dynamically meaningful: the average asymptotic linking is frozen. $□$

Connections Master

Helicity is the Casimir invariant of the ideal-fluid Lie-Poisson structure established in 05.14.01; that unit proves conservation from the coadjoint-orbit structure, while the present unit gives its topological meaning as average asymptotic linking. The two readings — Casimir and linking integral — are the same conserved scalar seen through Poisson geometry and through knot theory.
The wedge integral $\int_{M} α \land d α$ defining helicity is identical in form to the Hopf invariant $H (f) = \int_{S^{3}} α \land d α$ of a map $f : S^{3} \to S^{2}$ , developed in 05.14.03, where it becomes the obstruction to unknotting vortex tubes; the asymptotic linking number of this unit is precisely the "asymptotic Hopf invariant" name Arnold gave the construction.
Replacing the velocity potential by a genuine magnetic vector potential turns helicity into magnetic helicity $\int A \land d A$ , the topological invariant of ideal magnetohydrodynamics treated in 05.14.04; frozen-flux conservation there is the magnetic analogue of the vorticity-transport conservation proved in this unit's Proposition 3.
The Gauss linking integral appears as the abelian Chern-Simons Wilson-loop expectation in 03.16.06, whose action $\int A \land d A$ is the field-theoretic helicity; the present unit's continuum-average construction is the classical-hydrodynamic counterpart of that topological-field-theory linking.
Pairwise linking is the order-two piece of the entanglement of a flow; the higher-order Massey-product / Borromean refinement of 03.12.51 supplies asymptotic invariants that detect entanglement when all pairwise linking, and hence helicity, vanishes.

Historical & philosophical context Master

The integral now called the linking number was written down by Gauss in his 1833 notebooks on electrodynamics, where the linking of two current loops appeared as the work done moving a magnetic pole around a circuit; Gauss recorded the formula without proof and called the determination of such linking integrals one of the principal unsolved problems at the boundary of geometria situs and analysis. The fluid-dynamical invariant entered through Moffatt ^{[Moffatt 1969]}, who identified the integral $\int v \cdot curl v$ with the degree of knottedness of vortex lines and computed the linked-ring example, naming the quantity helicity. The decisive structural theorem belongs to Arnold ^{[Arnold 1973]}, whose paper The asymptotic Hopf invariant and its applications introduced the asymptotic linking number, proved its almost-everywhere existence by an ergodic argument, and established the identity with helicity, thereby recasting a conservation law of Euler's equation as a topological average and opening the field now called topological hydrodynamics. The rigorous measure-theoretic foundations — controlling the near-diagonal singularity of the Gauss kernel and the $L^{1}$ exchange of limit and integral — were completed by Vogel, while Freedman and He extended the circle of ideas to energy lower bounds via the asymptotic crossing number, turning Arnold's invariant into a tool for proving that knotted and linked magnetic fields cannot relax to lower energy without changing topology.

Bibliography Master

@article{Arnold1973asymptotic,
  author  = {Arnold, Vladimir I.},
  title   = {The asymptotic {H}opf invariant and its applications},
  journal = {Selecta Mathematica Sovietica},
  volume  = {5},
  number  = {4},
  pages   = {327--345},
  year    = {1986},
  note    = {Originally Proc. Summer School in Diff. Eqns., Erevan, 1974}
}

@article{Moffatt1969knottedness,
  author  = {Moffatt, H. Keith},
  title   = {The degree of knottedness of tangled vortex lines},
  journal = {Journal of Fluid Mechanics},
  volume  = {35},
  number  = {1},
  pages   = {117--129},
  year    = {1969}
}

@book{ArnoldKhesin2021,
  author    = {Arnold, Vladimir I. and Khesin, Boris A.},
  title     = {Topological Methods in Hydrodynamics},
  series    = {Applied Mathematical Sciences},
  volume    = {125},
  edition   = {2},
  publisher = {Springer},
  year      = {2021}
}

@article{FreedmanHe1991,
  author  = {Freedman, Michael H. and He, Zheng-Xu},
  title   = {Divergence-free fields: energy and asymptotic crossing number},
  journal = {Annals of Mathematics},
  volume  = {134},
  number  = {1},
  pages   = {189--229},
  year    = {1991}
}

@article{Vogel2003,
  author  = {Vogel, Thomas},
  title   = {On the asymptotic linking number},
  journal = {Proceedings of the American Mathematical Society},
  volume  = {131},
  number  = {7},
  pages   = {2289--2297},
  year    = {2003}
}

@article{MoffattRicca1992,
  author  = {Moffatt, H. Keith and Ricca, Renzo L.},
  title   = {Helicity and the {C}{\u{a}}lug{\u{a}}reanu invariant},
  journal = {Proceedings of the Royal Society A},
  volume  = {439},
  number  = {1906},
  pages   = {411--429},
  year    = {1992}
}

Prerequisites

03.16.06
03.04.04
05.02.06

Tier anchors

beginner: Moffatt 1969 *The degree of knottedness of tangled vortex lines* (J. Fluid Mech. 35) §1 informal; Arnold-Khesin *Topological Methods in Hydrodynamics* (Springer AMS 125) Ch. III §1 informal
intermediate: Arnold-Khesin *Topological Methods in Hydrodynamics* Ch. III §1-§2; Moffatt 1969 (J. Fluid Mech. 35) §2-§4; Moffatt-Ricca 1992 *Helicity and the Calugareanu invariant* (Proc. R. Soc. A 439)
master: Arnold 1973 *The asymptotic Hopf invariant and its applications* (Selecta Math. Sov. 5, 327-345; originator of the asymptotic-linking theorem); Arnold-Khesin *Topological Methods in Hydrodynamics* (Springer Applied Math. Sci. 125, 2nd ed. 2021) Ch. III §1-§4; Moffatt 1969 *The degree of knottedness of tangled vortex lines* (J. Fluid Mech. 35, 117-129); Freedman-He 1991 *Divergence-free fields: energy and asymptotic crossing number* (Ann. Math. 134); Vogel 2003 *On the asymptotic linking number* (Proc. AMS 131)

References

Arnold 1973 — The asymptotic Hopf invariant and its applications · Selecta Math. Sov. 5 (1986), 327-345 (originally Erevan 1974); originator of the asymptotic-linking-number theorem and the 'asymptotic Hopf invariant' name
Moffatt 1969 — The degree of knottedness of tangled vortex lines · J. Fluid Mech. 35, 117-129; identifies helicity with the integrated linking of vortex lines and computes the two-linked-ring example
Arnold, Khesin — Topological Methods in Hydrodynamics · Springer Applied Math. Sci. 125, 2nd ed. 2021, Ch. III §1-§4 (helicity, asymptotic linking, ergodic well-definedness, energy lower bound)
Freedman, He 1991 — Divergence-free fields: energy and asymptotic crossing number · Ann. Math. 134, 189-229; asymptotic crossing number and the topological lower bound on the energy of a divergence-free field (Freedman energy bound)
Vogel 2003 — On the asymptotic linking number · Proc. Amer. Math. Soc. 131, 2289-2297; rigorous Birkhoff-ergodic proof that the asymptotic linking number is defined almost everywhere and integrable
Moffatt, Ricca 1992 — Helicity and the Calugareanu invariant · Proc. R. Soc. A 439, 411-429; decomposition of a single tube's self-helicity into writhe and twist (the Calugareanu-White invariant)

Estimated time

beginner: 18m
intermediate: 45m
master: 85m