← Back to Part VII: Perelman's Ricci Flow & Spin Memory

Chern--Simons Gravity in 2+1d: Spin Memory as Holonomy

1. From Palatini to Chern--Simons

In 2+1 dimensions the Riemann tensor has $6$ independent components, exactly matching the $6$ components of the Ricci tensor. Hence $R_{\mu\nu\rho\sigma}$ is algebraically determined by $R_{\mu\nu}$ and $g_{\mu\nu}$:

$$R_{\mu\nu\rho\sigma} = g_{\mu\rho}R_{\nu\sigma} - g_{\mu\sigma}R_{\nu\rho} - g_{\nu\rho}R_{\mu\sigma} + g_{\nu\sigma}R_{\mu\rho} - \frac{R}{2}(g_{\mu\rho}g_{\nu\sigma} - g_{\mu\sigma}g_{\nu\rho})$$

This means there are no local propagating degrees of freedom. Start from the first-order Palatini action with cosmological constant $\Lambda = -1/\ell^2$:

$$S_{\rm Pal} = \frac{1}{16\pi G}\int_{\mathcal{M}} \epsilon_{abc}\left(e^a \wedge R^{bc}[\omega] + \frac{1}{3\ell^2}\,e^a \wedge e^b \wedge e^c\right)$$

The dreibein $e^a = e^a_\mu dx^\mu$ and spin connection $\omega^a = \frac{1}{2}\epsilon^{abc}\omega_{bc\mu}dx^\mu$ are independent fields. The curvature 2-form is $R^{ab} = d\omega^{ab} + \omega^a{}_c \wedge \omega^{cb}$. Now define the $\mathfrak{sl}(2,\mathbb{R})$ connections:

$$A^a = \omega^a + \frac{e^a}{\ell}, \qquad \bar{A}^a = \omega^a - \frac{e^a}{\ell}$$

Inverting: $e^a = \frac{\ell}{2}(A^a - \bar{A}^a)$ and $\omega^a = \frac{1}{2}(A^a + \bar{A}^a)$. Substituting into $S_{\rm Pal}$ and using the trace form $\mathrm{tr}(J_a J_b) = \frac{1}{2}\eta_{ab}$, we obtain after collecting terms:

$$S_{\rm CS}[A] - S_{\rm CS}[\bar{A}] = \frac{k}{4\pi}\int_{\mathcal{M}} \mathrm{tr}\!\left(A \wedge dA + \tfrac{2}{3}A \wedge A \wedge A\right) - (A \to \bar{A})$$

where the Chern--Simons level is $k = \ell/(4G)$. The key identity used is:

$$\epsilon_{abc}\,e^a \wedge R^{bc} = \frac{\ell}{2}\left[\mathrm{tr}(A \wedge dA + \tfrac{2}{3}A^3) - \mathrm{tr}(\bar{A} \wedge d\bar{A} + \tfrac{2}{3}\bar{A}^3)\right]$$

This is the celebrated Witten--Achucarro--Townsend equivalence: 2+1d gravity with $\Lambda < 0$ is a difference of two Chern--Simons theories.

2. Flatness and the Einstein Equations

The Chern--Simons equations of motion are simply the flatness conditions:

$$F = dA + A \wedge A = 0, \qquad \bar{F} = d\bar{A} + \bar{A} \wedge \bar{A} = 0$$

To see equivalence with Einstein's equations, compute the field strengths component by component. Writing $F^a = d(\omega^a + e^a/\ell) + \frac{1}{2}\epsilon^a{}_{bc}(\omega^b + e^b/\ell)\wedge(\omega^c + e^c/\ell)$, we expand:

$$F^a = \underbrace{\left(R^a + \frac{1}{\ell^2}\,\frac{1}{2}\epsilon^a{}_{bc}\,e^b \wedge e^c\right)}_{\text{Einstein eq.}} + \frac{1}{\ell}\underbrace{\left(de^a + \epsilon^a{}_{bc}\,\omega^b \wedge e^c\right)}_{\text{torsion-free condition}}$$

Setting $F^a = 0$ yields two independent equations. The torsion-free condition $T^a = de^a + \omega^a{}_b \wedge e^b = 0$ determines $\omega$ in terms of $e$. The remaining equation gives:

$$R^a + \frac{1}{2\ell^2}\,\epsilon^a{}_{bc}\,e^b \wedge e^c = 0 \quad \Longleftrightarrow \quad R_{\mu\nu} - \frac{1}{2}g_{\mu\nu}R - \frac{1}{\ell^2}g_{\mu\nu} = 0$$

These are precisely the vacuum Einstein equations with $\Lambda = -1/\ell^2$. The flatness of the CS connection encodes the full content of 2+1d general relativity.

3. Point Particle Holonomy

A point particle of mass $M$ and spin $J$ at the origin creates a conical metric. In 2+1d flat space the metric around the particle is:

$$ds^2 = -dt^2 + dr^2 + r^2\,d\phi^2, \qquad \phi \sim \phi + 2\pi - \Delta\alpha$$

The holonomy around a loop $\mathcal{C}$ encircling the particle is the path-ordered exponential:

$$\mathrm{Hol}_{\mathcal{C}}(A) = \mathcal{P}\exp\!\left(-\oint_{\mathcal{C}} A\right) = \mathcal{P}\exp\!\left(-\oint_{\mathcal{C}} \left(\omega^a + \frac{e^a}{\ell}\right)J_a\,d\phi\right)$$

For the conical metric, the spin connection has a single nonzero component $\omega^\phi{}_{r\phi} = 1 - 4GM/\ell$. Evaluating the path-ordered exponential on a circle of constant $r$:

$$\mathrm{Hol}_{\mathcal{C}}(A) = \exp\!\left(-\left(2\pi - \frac{8\pi GM}{\ell}\right)J_0 - \frac{4\pi J}{\ell^2}J_2\right)$$

The rotation part of this $SL(2,\mathbb{R})$ matrix encodes the mass via the deficit angle, while the boost part encodes the spin. For a spinless particle ($J = 0$):

$$\mathrm{Hol}_{\mathcal{C}} = \begin{pmatrix} \cos(\pi - \Delta\alpha/2) & -\sin(\pi - \Delta\alpha/2) \\ \sin(\pi - \Delta\alpha/2) & \cos(\pi - \Delta\alpha/2) \end{pmatrix}$$

4. Deficit Angle from Holonomy

From the holonomy matrix we read off the deficit angle. The trace of the holonomy for a rotation by angle $\alpha$ in $SL(2,\mathbb{R})$ is$\mathrm{Tr}(\mathrm{Hol}) = 2\cos(\alpha/2)$. Comparing with the explicit computation:

$$\Delta\alpha = 2\pi - \alpha = \frac{8\pi GM}{\ell}$$

Physical interpretation: a gyroscope parallel-transported around the particle returns rotated by $\Delta\alpha$. In the flat-space limit $\ell \to \infty$ one recovers $\Delta\alpha = 8\pi GM$ (in units $c = 1$). The particle removes a wedge of angle $\Delta\alpha$ from the plane and identifies the edges. This is spin memory in its purest form: a topological observable detected by holonomy alone, with no local curvature.

5. Ricci--Hamilton Flow on Surfaces

On a closed surface $(\Sigma^2, g)$, the normalised Ricci flow is:

$$\partial_t g_{ij} = -(R - \bar{R})\,g_{ij}, \qquad \bar{R} = \frac{\int_\Sigma R\,d\mu}{\int_\Sigma d\mu}$$

Write $g(t) = e^{2u(t)}g_0$ for a conformal factor $u$. The scalar curvature transforms as $R = e^{-2u}(R_0 - 2\Delta_0 u)$ where $\Delta_0$ is the Laplacian of $g_0$. The flow equation becomes a nonlinear PDE for $u$:

$$\partial_t u = e^{-2u}\Delta_0 u - \frac{1}{2}R_0 e^{-2u} + \frac{\bar{R}}{2}$$

For $\Sigma = S^2$ with $\chi(S^2) = 2$, the Gauss--Bonnet theorem gives$\bar{R} = 4\pi\chi/\mathrm{Area} = 8\pi/\mathrm{Area}$. Hamilton proved convergence: linearising around the round sphere $u = 0$, the perturbation $\delta u$ satisfies:

$$\partial_t(\delta u) = \Delta_{S^2}(\delta u) + 2\,\delta u$$

Expanding in spherical harmonics $Y_\ell^m$ with eigenvalues $-\ell(\ell+1)$, each mode decays as $e^{-[\ell(\ell+1)-2]t}$. The $\ell = 1$ modes ($\lambda = 0$) are gauge (diffeomorphisms), while the first physical mode$\ell = 2$ decays as $e^{-4t}$. This provesexponential convergence to the round metric.

6. Gauss--Bonnet, Holonomy, and Topological Constraints

For a compact surface $\Sigma$ with boundary $\partial\Sigma$ and corners with exterior angles $\delta_i$, the Gauss--Bonnet theorem reads:

$$\int_\Sigma R\,d\mu + \int_{\partial\Sigma} \kappa_g\,ds + \sum_i \delta_i = 4\pi\chi(\Sigma)$$

The boundary holonomy is the total geodesic curvature: $\Delta\alpha_{\partial\Sigma} = \int_{\partial\Sigma}\kappa_g\,ds$. Therefore the deficit angle for the full boundary is:

$$\Delta\alpha_{\partial\Sigma} = 4\pi\chi(\Sigma) - \int_\Sigma R\,d\mu - \sum_i \delta_i$$

On a closed surface ($\partial\Sigma = \emptyset$) this reduces to $\int R\,d\mu = 4\pi\chi$. For $S^2$ punctured by $n$ conical singularities of deficit $\Delta\alpha_i$:

$$\int_{\Sigma\setminus\{p_i\}} R\,d\mu = 4\pi\chi(S^2) - \sum_{i=1}^n \Delta\alpha_i = 8\pi - \sum_{i=1}^n \frac{8\pi GM_i}{\ell}$$

7. Spin Memory = Change in Holonomy

The connection is now exact. Under Ricci flow, the geometry evolves from a bumpy metric $g_{\rm init}$ toward the round metric $g_{\rm round}$. The total holonomy around a loop $\mathcal{C}$ changes from its initial value to the topological value:

$$\Delta\Psi_{\rm spin} = \mathrm{Hol}_{\mathcal{C}}(A_{\rm final}) - \mathrm{Hol}_{\mathcal{C}}(A_{\rm init}) = \int_0^\infty \frac{d}{dt}\mathrm{Hol}_{\mathcal{C}}(A(t))\,dt$$

Using the variation formula for holonomy under a connection change $\delta A$:

$$\frac{d}{dt}\mathrm{Hol}_{\mathcal{C}}(A) = -\oint_{\mathcal{C}} \mathrm{Hol}_{\mathcal{C}}^{s}\,\dot{A}(s)\,(\mathrm{Hol}_{\mathcal{C}}^{s})^{-1}\,ds \;\cdot\; \mathrm{Hol}_{\mathcal{C}}$$

The spin memory measures the total deviation from the topological baseline $4\pi\chi(\Sigma)$accumulated during the flow. As Ricci flow drives $g \to g_{\rm round}$, all non-topological holonomy decays exponentially, and the accumulated change is the spin memory. This is the 2+1d prototype of the 4d BMS spin memory effect.

2+1d Chern--Simons Gravity: Conical Deficit, Holonomy, and Ricci Flow

Python

script.py93 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# --- Panel 1: Conical deficit geometry around a point mass in 2+1d ---
fig, axes = plt.subplots(2, 2, figsize=(14, 12))
fig.patch.set_facecolor('#0a0a0a')
for ax in axes.flat:
    ax.set_facecolor('#0f0f1a')
    ax.tick_params(colors='#d4d4d4')
    for spine in ax.spines.values():
        spine.set_color('#444')

# Mass values (in units of 1/(8*pi*G/ell))
masses = np.linspace(0, 0.45, 200)
deficit_angle = 8 * np.pi * masses  # Delta alpha = 8*pi*G*M/ell

ax0 = axes[0, 0]
ax0.plot(masses, np.degrees(deficit_angle), color='#fbbf24', linewidth=2.5)
ax0.axhline(y=360, color='#ef4444', linestyle='--', alpha=0.5, label='Full cone (360 deg)')
ax0.set_xlabel('Mass M (units of ell/(8piG))', color='#e5e5e5', fontsize=11)
ax0.set_ylabel('Deficit angle (degrees)', color='#e5e5e5', fontsize=11)
ax0.set_title('Deficit Angle vs Point Mass', color='#fbbf24', fontsize=13, fontweight='bold')
ax0.legend(facecolor='#1a1a2e', edgecolor='#555', labelcolor='#e5e5e5')
ax0.grid(True, alpha=0.15)

# --- Panel 2: Conical geometry embedding ---
ax1 = fig.add_subplot(2, 2, 2, projection='3d')
ax1.set_facecolor('#0f0f1a')
for mass_val, color, label in [(0.05, '#22d3ee', 'M=0.05'), (0.15, '#fbbf24', 'M=0.15'), (0.3, '#ef4444', 'M=0.30')]:
    alpha_def = 8 * np.pi * mass_val
    phi_max = 2 * np.pi - alpha_def
    phi = np.linspace(0, phi_max, 200)
    r = np.linspace(0.01, 2.0, 50)
    PHI, R = np.meshgrid(phi, r)
    X = R * np.cos(PHI)
    Y = R * np.sin(PHI)
    Z = R * 0.3 * mass_val  # slight lift to visualize cone
    ax1.plot_surface(X, Y, Z, alpha=0.25, color=color)
    ax1.plot(r[-1]*np.cos(phi), r[-1]*np.sin(phi), r[-1]*0.3*mass_val*np.ones_like(phi),
             color=color, linewidth=1.5, label=label)
ax1.set_title('Conical Geometry (wedge removed)', color='#fbbf24', fontsize=13, fontweight='bold', pad=15)
ax1.legend(facecolor='#1a1a2e', edgecolor='#555', labelcolor='#e5e5e5', fontsize=9)

# --- Panel 3: Holonomy matrix trace vs mass ---
ax2 = axes[1, 0]
masses2 = np.linspace(0, 0.48, 300)
# Holonomy matrix for rotation: Tr(Hol) = 2*cos(Delta alpha / 2)
hol_trace = 2 * np.cos(4 * np.pi * masses2)
ax2.plot(masses2, hol_trace, color='#a78bfa', linewidth=2.5, label='Tr(Hol)')
ax2.axhline(y=2, color='#22d3ee', linestyle=':', alpha=0.5, label='Tr = 2 (no deficit)')
ax2.axhline(y=-2, color='#ef4444', linestyle=':', alpha=0.5, label='Tr = -2 (max rotation)')
ax2.fill_between(masses2, -2, 2, alpha=0.05, color='#a78bfa')
ax2.set_xlabel('Mass M', color='#e5e5e5', fontsize=11)
ax2.set_ylabel('Tr(Holonomy)', color='#e5e5e5', fontsize=11)
ax2.set_title('Holonomy Trace vs Mass', color='#a78bfa', fontsize=13, fontweight='bold')
ax2.legend(facecolor='#1a1a2e', edgecolor='#555', labelcolor='#e5e5e5')
ax2.grid(True, alpha=0.15)

# --- Panel 4: Ricci flow on bumpy S^2 converging to round ---
ax3 = axes[1, 1]
theta = np.linspace(0.01, np.pi, 200)
# Initial bumpy curvature on S^2 (conformal factor perturbation)
R_round = 2.0  # constant curvature of round S^2 with radius 1
# Perturbed curvature: add spherical harmonic-like bumps
bump = 1.5 * np.sin(2*theta) * np.cos(3*theta) + 0.8 * np.sin(5*theta)
times = [0, 0.2, 0.5, 1.0, 2.0, 5.0]
colors_flow = plt.cm.plasma(np.linspace(0.1, 0.95, len(times)))
for i, t in enumerate(times):
    # Under normalized Ricci flow, perturbations decay as exp(-lambda*t)
    # lambda_1 = 2 for l=1 mode on S^2 (but that's a gauge mode)
    # lambda_2 = 6 for l=2, lambda_3 = 12 for l=3, etc.
    R_t = R_round + bump * (0.8*np.exp(-6*t)*np.sin(2*theta)*np.cos(3*theta)/
          (np.sin(2*theta)*np.cos(3*theta)+1e-10) +
          0.5*np.exp(-12*t)*np.sin(5*theta)/(np.sin(5*theta)+1e-10))
    # Simplified: just decay the bump
    R_t_simple = R_round + bump * np.exp(-2*t)
    ax3.plot(np.degrees(theta), R_t_simple, color=colors_flow[i], linewidth=1.8,
             label=f't = {t:.1f}')
ax3.axhline(y=R_round, color='#22d3ee', linestyle='--', alpha=0.6, label='Round S^2 (R=2)')
ax3.set_xlabel('Polar angle theta (deg)', color='#e5e5e5', fontsize=11)
ax3.set_ylabel('Scalar curvature R(theta)', color='#e5e5e5', fontsize=11)
ax3.set_title('Ricci Flow: Curvature Convergence on S^2', color='#fbbf24', fontsize=13, fontweight='bold')
ax3.legend(facecolor='#1a1a2e', edgecolor='#555', labelcolor='#e5e5e5', fontsize=8, ncol=2)
ax3.grid(True, alpha=0.15)

plt.tight_layout()
plt.savefig('output.png', dpi=130, bbox_inches='tight', facecolor='#0a0a0a')
plt.close()
print("Chern-Simons gravity simulation complete: conical deficit, holonomy trace, and Ricci flow convergence plotted.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

← Holonomy & L-Geometry Next: Perelman Entropy & c-Theorem →