Navier–Stokes and Yang–Mills: comparison and bridges

Markdown source

The two problems share questions about nonlinear fields, scaling, constraints and uniform estimates. This comparison derives their scaling and dissipation identities, distinguishes regularity from relaxation and spectral separation, and specifies what each proposed bridge must transport.

First substantive milestone, 2026-09-05; exposition revised 2026-09-06.

The definitions preserve the official targets; the source catalogue provides PDFs and reading scope. Below, “derived” means an elementary calculation under the stated hypotheses, “cited” means a result imported from a named source, and “proposed” means a comparison whose transfer obligations remain to be specified. Source companions record the versions and passages used.

1. Exact targets and unlike meanings of existence

Sources: F, pp. 1–2 and JW, §§3–5. Fefferman’s existence alternatives A/B specify zero forcing; breakdown alternatives C/D permit admissible forcing. These quantifiers and forcing hypotheses determine the exact target.

The required evidence concerns smooth deterministic evolution for the fluid and quantum correlations, reconstruction and spectrum for the gauge theory. Each bridge must identify which object it transports and how it supplies the target’s remaining structure.

2. Equations and the several time variables

Let \(x\) denote fluid position and \(t\) fluid time. For a smooth unforced fluid, pressure elimination on whole space with suitable decay, or the standard periodic pressure convention, gives

\[ \partial_tu=\nu\Delta u-\mathbb P((u\cdot\nabla)u), \qquad \nabla\cdot u=0, \]

where \(\mathbb P\) is the orthogonal projection onto divergence-free fields. This pressure-eliminated representation supports the calculations below.

For gauge fields use \(y\in\mathbb R^d\), a Lie-algebra-valued connection \(A\), an invariant positive inner product, curvature

\[ F_{ij}=\partial_iA_j-\partial_jA_i+[A_i,A_j], \qquad D_i=\partial_i+[A_i,\,\cdot\,]. \]

The Euclidean Yang–Mills action in our normalisation is

\[ S(A)=\frac12\int|F_A|^2\,d^dy. \]

Coupling factors can be restored; this convention fixes the gradient identities below. Gauge heat flow has the form

\[ \partial_sA=-D_A^*F_A. \]

It is degenerate along gauge directions. A gauge term yields a parabolic representative. L, equations (1.1)–(1.2), (2.2)–(2.4) provides the concrete flow and gauge modification. The parameter \(s\) is auxiliary.

In a stochastic construction in \(d=4\), the noise lives over \((s,y)\): an auxiliary time and four Euclidean coordinates. The fluid has three spatial coordinates and its physical evolution time.

3. Scaling: a precise common language, unequal criticalities

The following calculations are derived for smooth fields on whole space. For periodic fields the rescaling changes the box size along with the field.

For fixed viscosity the unforced fluid scaling is

\[ u_\lambda(x,t)=\lambda u(\lambda x,\lambda^2t), \qquad p_\lambda(x,t)=\lambda^2p(\lambda x,\lambda^2t). \]

Each term of the velocity equation gains \(\lambda^3\). Changing variables in a spatial integral gives, in spatial dimension \(d\),

\[ \|u_\lambda(t)\|_{L^q}=\lambda^{1-d/q} \|u(\lambda^2t)\|_{L^q},\qquad E(u_\lambda(t))=\lambda^{2-d}E(u(\lambda^2t)), \quad E(u)=\tfrac12\|u\|_2^2. \]

In \(d=3\), the \(L^3\) norm is invariant, whereas energy scales as \(\lambda^{-1}\). Concentrating a profile can therefore increase pointwise size while decreasing its energy. This is the energy-supercritical scaling that a regularity estimate must address. For a fixed smooth profile also \(\|\nabla u_\lambda\|_2^2=\lambda\|\nabla u\|_2^2\) in three dimensions.

For a connection put \(A_\lambda(y)=\lambda A(\lambda y)\). Both the derivative and commutator contributions give

\[ F_{A_\lambda}(y)=\lambda^2F_A(\lambda y),\qquad S(A_\lambda)=\lambda^{4-d}S(A). \]

Thus the classical Euclidean action is scale-invariant in \(d=4\): action-critical in four Euclidean dimensions, alongside the fluid’s energy-supercritical scaling in three spatial dimensions. The effect of quantisation on scaling is a further question.

Proposed use: identify the norms or observables controlled at each scale, then derive estimates connecting the relevant solution spaces, measures or spectra.

4. Dissipation and curvature: useful structure with different roles

For smooth unforced fluid solutions with vanishing boundary terms, integration by parts gives

\[ \frac{dE}{dt}=-\nu\|\nabla u\|_2^2, \qquad \langle u,(u\cdot\nabla)u\rangle=0. \]

The cancellation follows by integrating \(u\cdot\nabla(|u|^2/2)\) and using \(\nabla\cdot u=0\). The pressure term vanishes under the same hypotheses. For smooth gauge heat flow the first variation is

\[ \delta S(A)[a]=\langle D_A^*F_A,a\rangle, \qquad \frac{dS}{ds}=-\|D_A^*F_A\|_2^2. \]

Both identities hold under the stated regularity and boundary assumptions. Here \(E\) is quadratic in velocity; \(S\) is quadratic in curvature, which contains derivatives and commutators. Navier–Stokes combines viscous diffusion and energy-conserving advection. For comparison, ordinary \(L^2\) gradient descent of kinetic energy would give \(\partial_tu=-u\).

The model distinction is essential. T, Theorem 1.5 and p. 10 supplies a cited example with averaged nonlinearity, energy cancellation and finite-time blowup. A regularity argument must therefore exploit structure beyond the properties shared with that averaged equation.

There is also a geometric analogy. With \(u^\flat\) the velocity one-form, \(\omega=*du^\flat\) is vorticity in three Euclidean dimensions, while gauge curvature is \(F=dA+A\wedge A\). The unforced vorticity equation, obtained by taking the curl, is

\[ \partial_t\omega+(u\cdot\nabla)\omega =(\omega\cdot\nabla)u+\nu\Delta\omega. \]

The stretching term makes nonlinear derivative amplification explicit. Curvature likewise exposes derivatives and self-interaction, with the additional nonabelian term \(A\wedge A\). A map between the two descriptions must account for that term and respect the resulting dynamics.

Projection restricts the fluid to a physical incompressibility constraint; a gauge transformation changes the representative of a gauge orbit. This gives two analytic tasks to compare: constrained evolution and evolution modulo redundancy.

5. Stochastic quantisation: a concrete but conditional bridge

The schematic auxiliary equation is

\[ \partial_sA=-D_A^*F_A+\text{gauge term}+\text{noise} +\text{renormalisation terms}. \]

This schematic display identifies the terms a precise SPDE construction must define. C, §1.2, Theorem 1.6, Remark 1.9 and §4 describes rigorous finite-volume dynamics in dimensions two and three. The solution framework allows a cemetery state for possible explosion. Invariant measures and infinite volume appear as further questions in that 2022 review.

The proposed connection is through methods for singular parabolic equations. A route from auxiliary dynamics to the Millennium target would still have to supply:

1. A well-defined process and an appropriate invariant law.
2. Identification of that law with the intended gauge theory.
3. The required continuum and infinite-volume control, with correct observables.
4. Quantum reconstruction with the required axioms.
5. A gap for the reconstructed physical Hamiltonian.

In this chain, stationarity, reflection positivity and the physical spectral bound each have their own proof obligation.

6. Three different gaps

6.1 A finite-box fluid decay rate

Derived diagnostic: take the standard unforced periodic fluid on a box of side \(L\), with periodic pressure and zero mean velocity. Fourier modes have wavevectors \(k=2\pi n/L\). Therefore

\[ \|\nabla u\|_2^2\geq(2\pi/L)^2\|u\|_2^2, \qquad E(t)\leq E(0)e^{-2\nu(2\pi/L)^2t} \]

for as long as the solution is smooth. The Stokes decay-rate gap is \(\gamma_L=\nu(2\pi/L)^2\). It vanishes when \(L\to\infty\); constants are zero modes unless the mean is removed. Whole-space heat flow supplies a useful comparison: globally smooth evolution with nonzero decay rates accumulating at zero. The displayed fluid estimate controls low-frequency decay; regularity also requires control of high-frequency concentration.

6.2 An auxiliary mixing gap

Derived distinction: suppose an auxiliary Markov generator \(\mathcal L\) has invariant law \(\mu\) and an \(L^2(\mu)\) relaxation bound. Replacing it by \(c\mathcal L\), for any \(c>0\), keeps \(\mu\) invariant because \(\int c\mathcal Lf\,d\mu=0\), while multiplying its relaxation rate by \(c\). Equivalently its semigroup is \(P_{cs}\).

The sampling clock can therefore rescale a mixing gap while preserving equilibrium correlations. Extracting a physical mass requires a physical-time operator identification. A reversible diffusion conjugate to a Schrödinger-type operator still requires that identification with the field-theory Hamiltonian.

6.3 The quantum gap and physical correlations

Derived spectral consequence: assume a self-adjoint physical \(H\geq0\), a unique vacuum \(\Omega\), and a gap \(m>0\). Let \(\psi\) be a well-defined state orthogonal to \(\Omega\), for example one created by a suitably smeared centred observable. The spectral theorem gives

\[ \langle\psi,e^{-rH}\psi\rangle =\int_{[m,\infty)}e^{-rE}\,d\rho_\psi(E) \leq e^{-mr}\|\psi\|^2,\quad r\geq0. \]

This is decay in Euclidean physical-time separation \(r\), as distinct from auxiliary relaxation time \(s\). An observable constrains the energies to which it couples. Proving a full spectral gap requires a decay bound for a class of observables that detects all states under consideration.

In units \(\hbar=c=1\), mass has inverse-length units. A fluid decay rate has inverse-time units, with \(\nu\) carrying length-squared/time. Equating their numerical values requires a physically justified conversion.

7. Bridge assessment

For any regularised spectral calculation use a cutoff label \(a\) and box size \(L\). The statement \(m_{a,L}>0\) at each fixed pair is weaker than a positive bound in physical units along a justified continuum and infinite-volume construction. For example, a dimensionless lattice gap may scale as \(a m\) and tend to zero while a physical \(m\) stays positive. Conversely a box-induced gap can disappear. The construction must specify the order of limits and justify any interchange.

8. Applying the comparison

For each toy model, specify its variables and symmetries, its trajectory or state space, its clock and generator, its scale transformation and its limiting procedures. Then state the desired estimate and identify which structures it shares with the companion field problem.

The present mechanics programme starts with action variations and operational resolution, then moves to operator spectra. This comparison supplies the transfer questions for that sequence. Its proposed bridges remain conditional on the listed constructions and estimates; the Millennium targets remain the full problems in the official definitions.