Second Variation of L and the Jacobi Equation

The L-Jacobi Field Equation

To understand the second-order behaviour of the L-length functional, we study its second variation along a family of curves. Let $\gamma$ be an L-geodesic and $J$ a variation field along $\gamma$. The L-Jacobi field equation governs infinitesimal variations of L-geodesics and takes the form:

$$\nabla_{\dot\gamma}\nabla_{\dot\gamma}J + R(\dot\gamma,J)\dot\gamma + \frac{1}{2}\nabla_J(\nabla R) - 2\,\mathrm{Ric}(\dot\gamma,\cdot)^\sharp \cdot J - \frac{1}{2\tau}\nabla_{\dot\gamma}J = 0$$

Here $R(\dot\gamma,J)\dot\gamma$ is the curvature term, $\nabla_J(\nabla R)$ captures the variation of the scalar curvature gradient, and the Ricci term $\mathrm{Ric}(\dot\gamma,\cdot)^\sharp$ acts on $J$ through the evolving metric. The factor $\frac{1}{2\tau}$ reflects the singular weighting $\sqrt{\tau}$ in the L-length integrand. Solutions $J$ satisfying this ODE along $\gamma$ are called L-Jacobi fields.

A key subtlety is that L-Jacobi fields must satisfy initial conditions compatible with the singularity of the L-length integrand at $\tau = 0$. Specifically, one requires$\lim_{\tau \to 0} \sqrt{\tau}\,J(\tau) = 0$ and a prescribed asymptotic rate for the derivative.

Hessian of the Reduced Distance

The second variation of L-length yields the Hessian of the reduced distance $\ell$. For an L-Jacobi field $J$ along a minimising L-geodesic $\gamma$ ending at time $\bar\tau$:

$$\mathrm{Hess}(\ell)(J,J) = \frac{1}{2\sqrt{\bar\tau}}\left[\langle\nabla_{\dot\gamma}J,\,J\rangle\Big|_{\bar\tau} - \int_0^{\bar\tau}\sqrt{\tau}\,\mathcal{K}(J)\,d\tau\right]$$

The curvature integrand $\mathcal{K}(J)$ appearing in this formula is:

$$\mathcal{K}(J) = -R(\dot\gamma,J,\dot\gamma,J) - \langle\nabla_J\nabla R,J\rangle + 2\,\mathrm{Ric}(\nabla_{\dot\gamma}J,J) - 2\,\mathrm{Ric}(J,J)_t + |\nabla_{\dot\gamma}J|^2$$

This integrand $\mathcal{K}$ encodes the full curvature information of the Ricci flow space-time along the L-geodesic. Its sign and magnitude control whether nearby L-geodesics focus or spread, analogous to the role of sectional curvature in classical Riemannian comparison geometry.

Laplacian and Time-Derivative Comparison

By summing the Hessian over an orthonormal frame of L-Jacobi fields, one obtains the crucial Laplacian comparison inequality for the reduced distance:

$$\Delta\ell \;\leq\; \frac{n}{2\bar\tau} - R + \frac{\ell}{2\bar\tau}$$

where $n$ is the dimension of the manifold. Complementing this, the time derivative of $\ell$ satisfies:

$$\frac{\partial\ell}{\partial\tau} \;\leq\; \frac{n}{2\bar\tau} - \frac{\ell}{\bar\tau} + R$$

These two inequalities are the Ricci-flow analogues of the classical Laplacian comparison theorem in Riemannian geometry. They encode the geometric information needed to control volumes under the flow.

Monotonicity of the Reduced Volume

Recall Perelman's reduced volume $\widetilde{V}(\bar\tau) = \int_M (4\pi\bar\tau)^{-n/2}\,e^{-\ell(q,\bar\tau)}\,d\mu_{g(\bar\tau)}(q)$. Combining the Laplacian comparison and the time-derivative estimate for $\ell$, one computes:

$$\frac{d\widetilde{V}}{d\bar\tau} \;\leq\; 0$$

Key Consequence

The reduced volume is monotonically non-increasing in $\bar\tau$. Since$\widetilde{V}(\bar\tau) \to (4\pi)^{n/2}$ as $\bar\tau \to 0$ (the Euclidean value), this provides a universal upper bound. The monotonicity of $\widetilde{V}$ is the engine behind Perelman's non-collapsing theorem: if the reduced volume stays bounded away from zero, the manifold cannot collapse at any scale.

← L-Geodesics Next: Pseudolocality →