Yang-Mills Flow: The SM Analogue of Ricci Flow
Gradient flow of the Yang–Mills functional, its monotonicity, fixed points at flat connections, and the deep parallel with Perelman's Ricci flow program
1. The Yang-Mills Flow Equation
The Yang–Mills flow is the gradient flow of the Yang–Mills functional$S_{\mathrm{YM}}[A] = \frac{1}{2}\int|F_A|^2$. The flow equation is:
$$\frac{\partial A_\mu}{\partial t} = -D^\nu F_{\nu\mu}$$
Compare with Ricci flow, which is the gradient flow of the Einstein–Hilbert functional (or more precisely, Perelman's $\mathcal{F}$-functional):
$$\frac{\partial g_{ij}}{\partial t} = -2R_{ij}$$
In both cases, the right-hand side is a second-order differential operator acting on the geometric variable (connection $A$ or metric $g$), driving the system toward a minimum of the action functional.
2. Monotonicity of the YM Action
Under Yang–Mills flow, the Yang–Mills action decreases monotonically:
$$\frac{d}{dt}S_{\mathrm{YM}}[A(t)] = \frac{d}{dt}\frac{1}{2}\int|F|^2 = -2\int|D^\nu F_{\nu\mu}|^2 \leq 0$$
Equality holds if and only if $D^\nu F_{\nu\mu} = 0$, i.e., the connection satisfies the (vacuum) Yang–Mills equations. Compare with Perelman's monotonicity:
$$\frac{d}{dt}\mathcal{F}(g, f) = 2\int|R_{ij} + \nabla_i\nabla_j f|^2\,e^{-f}\,dV \geq 0$$
Both are dissipation inequalities. The YM action decreases (the connection becomes flatter), while Perelman's $\mathcal{F}$-functional increases (the metric becomes more Einstein). The sign difference reflects the convention: YM flow minimizes its action, while Perelman's flow is defined as an increasing entropy.
3. Fixed Points and Convergence
The fixed points of Yang–Mills flow are Yang–Mills connections, satisfying $D^\nu F_{\nu\mu} = 0$. The absolute minima are flat connections:
$$F_{\mu\nu} = 0 \quad \Longleftrightarrow \quad S_{\mathrm{YM}} = 0$$
Compare: the fixed points of Ricci flow are Ricci solitons (including Einstein metrics), and the simplest fixed point is the flat metric $R_{ij} = 0$.
Donaldson proved that in dimension 4, Yang–Mills flow converges (outside a singular set) to a Yang–Mills connection, possibly with bubbling of instantons. The instanton bubbling is analogous to the neck-pinch singularities in Ricci flow that Perelman resolved with surgery.
4. Gauge Fixing and the DeTurck Trick
The naive Yang–Mills flow is degenerate due to gauge invariance: the operator$A \mapsto -D^\nu F_{\nu\mu}$ has a kernel consisting of pure gauge deformations$\delta A_\mu = D_\mu \epsilon$. This is exactly the same issue that DeTurck resolved for Ricci flow, where the operator $g \mapsto -2R_{ij}$ has a kernel from diffeomorphisms.
The gauge-fixed Yang–Mills flow adds a DeTurck-type term:
$$\frac{\partial A_\mu}{\partial t} = -D^\nu F_{\nu\mu} + D_\mu(D^\nu A_\nu)$$
The additional term $D_\mu(D^\nu A_\nu)$ is a pure gauge transformation that makes the flow equation strictly parabolic. The resulting flow is equivalent to the original YM flow modulo gauge transformations, just as DeTurck–Ricci flow is equivalent to Ricci flow modulo diffeomorphisms.
5. Coupled Yang-Mills and Ricci Flow
When the Yang–Mills bundle lives over a Riemannian manifold evolving by Ricci flow, the coupled system becomes:
$$\frac{\partial g_{ij}}{\partial t} = -2R_{ij} + \frac{1}{2}\mathrm{tr}(F_{ik}F_j^{\ k})$$
$$\frac{\partial A_\mu}{\partial t} = -D^\nu F_{\nu\mu}$$
The energy-momentum of the gauge field feeds back into the metric evolution through the$\mathrm{tr}(F_{ik}F_j^{\ k})$ term. This coupled system has its own monotonicity formula generalizing Perelman's:
$$\mathcal{F}_{\mathrm{coupled}} = \int\left(R + |\nabla f|^2 + \frac{1}{4}|F|^2\right)e^{-f}\,dV$$
This unified functional encodes both geometric (Ricci) and gauge (Yang–Mills) degrees of freedom in a single variational principle, providing a concrete realization of the Perelman-to-Standard-Model bridge.
Simulation: Yang-Mills Flow in 1+1D
PythonClick Run to execute the Python code
Code will be executed with Python 3 on the server