Riemannian Geometry

Riemannian geometry equips smooth manifolds with a metric tensor—an inner product on each tangent space that varies smoothly from point to point. This structure enables the measurement of lengths, angles, areas, and curvature, and provides the mathematical foundation for Einstein's general theory of relativity.

Historical Context

Bernhard Riemann introduced the concept of a metric tensor in his legendary 1854 Habilitationsschrift, delivered before Gauss at Göttingen. Riemann generalized Gauss's intrinsic geometry of surfaces to arbitrary dimensions, proposing that the geometry of a space should be determined by a quadratic form $ds^2 = g_{ij}\,dx^i\,dx^j$ on infinitesimal displacements. This was sixty years before Einstein used exactly this framework to formulate general relativity in 1915.

The Levi-Civita connection, developed by Tullio Levi-Civita in 1917, provided the canonical notion of parallel transport compatible with the Riemannian metric. Elwin Bruno Christoffel had earlier (1869) introduced the symbols that bear his name for expressing covariant differentiation in coordinates. Together, these tools form the computational backbone of modern differential geometry and general relativity.

The theory reached maturity through the work of Elie Cartan, who reformulated connections using moving frames, and Marcel Berger, who classified Riemannian holonomy groups.

Derivation 1: The Riemannian Metric Tensor

A Riemannian metric on a smooth manifold $M$ is a smooth assignment of an inner product $g_p$ on each tangent space $T_pM$. In local coordinates$\{x^i\}$, the metric is specified by a symmetric, positive-definite matrix:

$g = g_{ij}(x)\,dx^i \otimes dx^j$

The line element gives the infinitesimal distance:

$ds^2 = g_{ij}\,dx^i\,dx^j$

Transformation Law

Under a coordinate change $x^i \to \bar{x}^a$, the metric components transform as a (0,2)-tensor:

$\bar{g}_{ab} = \frac{\partial x^i}{\partial \bar{x}^a}\frac{\partial x^j}{\partial \bar{x}^b}\,g_{ij}$

Example: The 2-Sphere

In spherical coordinates $(\theta, \phi)$ on the unit sphere $S^2 \subset \mathbb{R}^3$:

$ds^2 = d\theta^2 + \sin^2\theta\,d\phi^2, \quad g = \begin{pmatrix} 1 & 0 \\ 0 & \sin^2\theta \end{pmatrix}$

General relativity: In GR, spacetime carries a pseudo-Riemannian (Lorentzian) metric of signature $(-,+,+,+)$. The Schwarzschild metric for a black hole of mass $M$ is $ds^2 = -(1-\frac{2GM}{rc^2})c^2dt^2 + (1-\frac{2GM}{rc^2})^{-1}dr^2 + r^2 d\Omega^2$.

Derivation 2: Christoffel Symbols

The Christoffel symbols of the second kind encode how basis vectors change from point to point. They are derived from the metric by requiring two conditions: metric compatibility and torsion-freedom.

The Koszul Formula

From the conditions $\nabla g = 0$ (metric compatible) and $\nabla_X Y - \nabla_Y X = [X,Y]$ (torsion-free), one derives the Koszul formula:

$2g(\nabla_X Y, Z) = X(g(Y,Z)) + Y(g(X,Z)) - Z(g(X,Y)) + g([X,Y],Z) - g([X,Z],Y) - g([Y,Z],X)$

Applying this to coordinate vector fields $X = \partial_i$, $Y = \partial_j$, $Z = \partial_k$(where $[\partial_i, \partial_j] = 0$):

$\boxed{\Gamma^k_{ij} = \frac{1}{2}g^{kl}\left(\frac{\partial g_{il}}{\partial x^j} + \frac{\partial g_{jl}}{\partial x^i} - \frac{\partial g_{ij}}{\partial x^l}\right)}$

Computation for $S^2$

With $g_{11} = 1$, $g_{22} = \sin^2\theta$, $g_{12} = 0$, the only non-vanishing partial derivative is $\partial_\theta g_{22} = 2\sin\theta\cos\theta$. This yields:

$\Gamma^\theta_{\phi\phi} = -\frac{1}{2}g^{\theta\theta}\partial_\theta g_{\phi\phi} = -\sin\theta\cos\theta$

$\Gamma^\phi_{\theta\phi} = \Gamma^\phi_{\phi\theta} = \frac{1}{2}g^{\phi\phi}\partial_\theta g_{\phi\phi} = \frac{\cos\theta}{\sin\theta} = \cot\theta$

Counting Christoffel symbols: In $n$ dimensions, there are $\frac{n^2(n+1)}{2}$ independent Christoffel symbols (symmetric in lower indices). In 4D spacetime, this gives 40 independent components.

Derivation 3: The Levi-Civita Connection

The fundamental theorem of Riemannian geometry states that on every Riemannian manifold $(M, g)$, there exists a unique affine connection $\nabla$ that is:

$\nabla_X g = 0 \quad \text{(metric compatible: inner products are preserved)}$

$T(X,Y) = \nabla_X Y - \nabla_Y X - [X,Y] = 0 \quad \text{(torsion-free)}$

Proof of Uniqueness

Suppose $\nabla$ and $\bar{\nabla}$ are both metric-compatible and torsion-free. Their difference $D(X,Y) = \nabla_X Y - \bar{\nabla}_X Y$ is a tensor (the non-tensorial parts cancel). From metric compatibility:

$g(D(X,Y), Z) + g(Y, D(X,Z)) = 0$

From the torsion-free condition, $D(X,Y) = D(Y,X)$. Cycling the metric compatibility equation over$(X,Y,Z)$ and using symmetry, one obtains $g(D(X,Y),Z) = 0$ for all $Z$. Since $g$ is non-degenerate, $D = 0$, proving uniqueness.

Covariant Derivative

The covariant derivative of a vector field $V = V^i \partial_i$ along a direction $\partial_j$ is:

$\nabla_j V^i = \partial_j V^i + \Gamma^i_{jk} V^k$

This extends to arbitrary tensors. For a (0,2)-tensor such as the metric:$\nabla_k g_{ij} = \partial_k g_{ij} - \Gamma^l_{ki} g_{lj} - \Gamma^l_{kj} g_{il} = 0$, which is precisely the metric compatibility condition.

Derivation 4: Parallel Transport and Holonomy

A vector field $V$ along a curve $\gamma(t)$ is parallel transported if its covariant derivative along the curve vanishes:

$\frac{DV^i}{dt} = \frac{dV^i}{dt} + \Gamma^i_{jk}\frac{dx^j}{dt}V^k = 0$

This is a system of first-order ODEs. Given an initial vector $V(0) \in T_{\gamma(0)}M$, there exists a unique parallel transport along $\gamma$. The resulting map:

$P_\gamma: T_{\gamma(0)}M \to T_{\gamma(1)}M$

is a linear isomorphism. For the Levi-Civita connection, it is also an isometry (preserves inner products).

Holonomy on $S^2$

When a vector is parallel-transported around a closed loop, it generally returns rotated. For a loop enclosing solid angle $\Omega$ on $S^2$, the rotation angle equals $\Omega$. For a latitude circle at colatitude $\theta$:

$\boxed{\Delta\alpha = \oint K\,dA = 2\pi(1 - \cos\theta)}$

Physical realization: The Foucault pendulum demonstrates holonomy on Earth. At latitude $\lambda$, the pendulum's plane rotates by $2\pi\sin\lambda$ per sidereal day—precisely the holonomy of parallel transport around the latitude circle on the sphere.

Derivation 5: Index Raising, Lowering, and Musical Isomorphisms

The metric provides a canonical isomorphism between vectors and covectors. The flat ($\flat$) and sharp ($\sharp$) operations (named for musical notation) allow index manipulation:

$\flat: TM \to T^*M, \quad V \mapsto V^\flat, \quad V^\flat_i = g_{ij}V^j \quad \text{(lowering)}$

$\sharp: T^*M \to TM, \quad \omega \mapsto \omega^\sharp, \quad \omega^{\sharp i} = g^{ij}\omega_j \quad \text{(raising)}$

Here $g^{ij}$ is the inverse metric satisfying $g^{ik}g_{kj} = \delta^i_j$. These operations extend to arbitrary tensors, enabling free movement between covariant and contravariant descriptions.

The Gradient as a Sharp

The gradient of a function $f$ is the sharp of its differential:

$(\text{grad}\, f)^i = g^{ij}\partial_j f = (df)^{\sharp i}$

In flat Cartesian coordinates, $g^{ij} = \delta^{ij}$ and this reduces to the familiar$\nabla f = (\partial_x f, \partial_y f, \partial_z f)$. But in curvilinear coordinates (spherical, cylindrical), the metric factors produce the well-known correction terms.

In GR: Raising and lowering indices with the spacetime metric$g_{\mu\nu}$ is fundamental. The Einstein equation $G_{\mu\nu} = 8\pi G\,T_{\mu\nu}$ involves the Ricci tensor and scalar obtained by contracting the Riemann tensor using the inverse metric.

Interactive Simulation

This simulation computes the metric tensor, Christoffel symbols, and parallel transport on the 2-sphere. It demonstrates how vectors rotate during parallel transport around latitude circles and verifies the holonomy formula.

Riemannian Geometry: Metric, Christoffel Symbols & Parallel Transport

Python

script.py180 lines

#!/usr/bin/env python3
"""
riemannian_geometry.py - Metric Tensors, Levi-Civita Connection, Parallel Transport
Computes Christoffel symbols, parallel transport on curved surfaces, and metric properties
"""
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import base64
from io import BytesIO

# ============================================================
# 1. Metric tensor on the 2-sphere
# ============================================================
print("=" * 72)
print("  RIEMANNIAN GEOMETRY: METRIC TENSORS AND CONNECTIONS")
print("=" * 72)
print()
print("  The 2-sphere with radius R in spherical coordinates (theta, phi):")
print("    ds^2 = R^2 d(theta)^2 + R^2 sin^2(theta) d(phi)^2")
print()

R = 1.0
theta_vals = np.linspace(0.1, np.pi - 0.1, 200)

# Metric components
g_theta_theta = R**2 * np.ones_like(theta_vals)
g_phi_phi = R**2 * np.sin(theta_vals)**2
det_g = g_theta_theta * g_phi_phi

print("  Metric components g_ij for S^2 (R = 1):")
print("    g_11 = R^2 = {:.4f}".format(R**2))
print("    g_22 = R^2 sin^2(theta)")
print("    g_12 = g_21 = 0")
print()

for th in [np.pi/6, np.pi/4, np.pi/3, np.pi/2]:
    print("    theta = {:.4f}: g_22 = {:.6f}, det(g) = {:.6f}".format(
        th, R**2 * np.sin(th)**2, R**4 * np.sin(th)**2))
print()

# ============================================================
# 2. Christoffel symbols for S^2
# ============================================================
print("  CHRISTOFFEL SYMBOLS (Levi-Civita Connection)")
print("  " + "-" * 50)
print()
print("  Non-vanishing Christoffel symbols for S^2:")
print("    Gamma^theta_{phi phi} = -sin(theta) cos(theta)")
print("    Gamma^phi_{theta phi} = Gamma^phi_{phi theta} = cos(theta)/sin(theta)")
print()

for th in [np.pi/6, np.pi/4, np.pi/3, np.pi/2]:
    g1 = -np.sin(th) * np.cos(th)
    g2 = np.cos(th) / np.sin(th)
    print("    theta = {:.4f}: Gamma^th_{{pp}} = {:.6f}, Gamma^ph_{{tp}} = {:.6f}".format(
        th, g1, g2))
print()

# ============================================================
# 3. Parallel transport around a latitude circle
# ============================================================
print("  PARALLEL TRANSPORT AROUND LATITUDE CIRCLE")
print("  " + "-" * 50)
print()

theta0 = np.pi / 4  # latitude
n_steps = 500
dphi = 2 * np.pi / n_steps

# Initial vector: pointing along theta direction
v_theta = np.zeros(n_steps + 1)
v_phi = np.zeros(n_steps + 1)
v_theta[0] = 1.0
v_phi[0] = 0.0
phi_path = np.linspace(0, 2 * np.pi, n_steps + 1)

for i in range(n_steps):
    # Parallel transport equations on S^2 along phi
    dv_theta = np.sin(theta0) * np.cos(theta0) * v_phi[i] * dphi
    dv_phi = -np.cos(theta0) / np.sin(theta0) * v_theta[i] * dphi
    v_theta[i+1] = v_theta[i] + dv_theta
    v_phi[i+1] = v_phi[i] + dv_phi

angle_deficit = np.arctan2(v_phi[-1] * np.sin(theta0), v_theta[-1])
holonomy = 2 * np.pi * (1 - np.cos(theta0))

print("  Latitude: theta = pi/4 = {:.6f} rad".format(theta0))
print("  Initial vector: (v^theta, v^phi) = (1, 0)")
print("  After full circuit:")
print("    v^theta = {:.6f}".format(v_theta[-1]))
print("    v^phi   = {:.6f}".format(v_phi[-1]))
print("    Rotation angle (numerical)  = {:.6f} rad".format(angle_deficit))
print("    Holonomy angle (exact)       = {:.6f} rad".format(holonomy))
print("    = 2*pi*(1 - cos(theta))")
print()

# ============================================================
# 4. Levi-Civita uniqueness verification
# ============================================================
print("  LEVI-CIVITA CONNECTION PROPERTIES")
print("  " + "-" * 50)
print()
print("  Metric compatibility: nabla_k g_ij = 0")
print("  Torsion-free: Gamma^k_ij = Gamma^k_ji")
print()
print("  Verification for S^2:")
print("    Gamma^phi_{theta phi} = {:.6f}".format(np.cos(theta0)/np.sin(theta0)))
print("    Gamma^phi_{phi theta} = {:.6f}".format(np.cos(theta0)/np.sin(theta0)))
print("    Symmetric: confirmed (torsion-free)")
print()

# ============================================================
# Visualization
# ============================================================
fig, axes = plt.subplots(2, 2, figsize=(14, 11))
fig.patch.set_facecolor('#0a0a0a')

for ax in axes.flat:
    ax.set_facecolor('#111111')
    for spine in ax.spines.values():
        spine.set_color('#2dd4bf')
    ax.tick_params(colors='#99f6e4')
    ax.grid(True, color='#134e4a', alpha=0.4)

# Plot 1: Metric determinant
ax1 = axes[0, 0]
ax1.plot(theta_vals, det_g, color='#2dd4bf', linewidth=2.5, label='det(g)')
ax1.fill_between(theta_vals, det_g, alpha=0.15, color='#2dd4bf')
ax1.axvline(x=np.pi/2, color='#f59e0b', linestyle='--', alpha=0.6, label='equator')
ax1.set_xlabel('theta', color='#5eead4')
ax1.set_ylabel('det(g)', color='#5eead4')
ax1.set_title('Metric Determinant on S^2', color='#99f6e4', fontsize=12, fontweight='bold')
ax1.legend(facecolor='#111111', edgecolor='#134e4a', labelcolor='#99f6e4')

# Plot 2: Christoffel symbols
ax2 = axes[0, 1]
gamma1 = -np.sin(theta_vals) * np.cos(theta_vals)
gamma2 = np.cos(theta_vals) / np.sin(theta_vals)
ax2.plot(theta_vals, gamma1, color='#f472b6', linewidth=2, label='Gamma^th_{ph ph}')
ax2.plot(theta_vals, gamma2, color='#a78bfa', linewidth=2, label='Gamma^ph_{th ph}')
ax2.set_xlabel('theta', color='#5eead4')
ax2.set_ylabel('Christoffel symbol', color='#5eead4')
ax2.set_title('Christoffel Symbols on S^2', color='#99f6e4', fontsize=12, fontweight='bold')
ax2.legend(facecolor='#111111', edgecolor='#134e4a', labelcolor='#99f6e4')
ax2.set_ylim(-3, 3)

# Plot 3: Parallel transport vector components
ax3 = axes[1, 0]
ax3.plot(phi_path, v_theta, color='#34d399', linewidth=2, label='v^theta')
ax3.plot(phi_path, v_phi * np.sin(theta0), color='#f59e0b', linewidth=2, label='v^phi sin(theta)')
ax3.set_xlabel('phi (azimuthal angle)', color='#5eead4')
ax3.set_ylabel('Vector component', color='#5eead4')
ax3.set_title('Parallel Transport Around Latitude', color='#99f6e4', fontsize=12, fontweight='bold')
ax3.legend(facecolor='#111111', edgecolor='#134e4a', labelcolor='#99f6e4')

# Plot 4: Holonomy angle vs latitude
ax4 = axes[1, 1]
th_range = np.linspace(0.01, np.pi - 0.01, 300)
holonomy_angle = 2 * np.pi * (1 - np.cos(th_range))
ax4.plot(th_range, holonomy_angle, color='#2dd4bf', linewidth=2.5)
ax4.fill_between(th_range, holonomy_angle, alpha=0.1, color='#2dd4bf')
ax4.axhline(y=2*np.pi, color='#6b7280', linestyle='--', alpha=0.5)
ax4.axvline(x=np.pi/4, color='#f59e0b', linestyle='--', alpha=0.6, label='theta=pi/4')
ax4.set_xlabel('theta (latitude)', color='#5eead4')
ax4.set_ylabel('Holonomy angle (rad)', color='#5eead4')
ax4.set_title('Holonomy vs Latitude on S^2', color='#99f6e4', fontsize=12, fontweight='bold')
ax4.legend(facecolor='#111111', edgecolor='#134e4a', labelcolor='#99f6e4')

plt.tight_layout(pad=2.0)
buf = BytesIO()
plt.savefig(buf, format='png', dpi=150, bbox_inches='tight', facecolor='#0a0a0a')
buf.seek(0)
img_b64 = base64.b64encode(buf.read()).decode()
print('<img src="data:image/png;base64,' + img_b64 + '" alt="Riemannian Geometry Visualization" />')
buf.close()
print()
print("Visualization complete.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

Summary

Metric Tensor

The symmetric positive-definite tensor $g_{ij}$ defines lengths, angles, and volumes on a Riemannian manifold. It transforms as a (0,2)-tensor under coordinate changes.

Levi-Civita Connection

The unique torsion-free, metric-compatible connection. Its coefficients are the Christoffel symbols, determined entirely by the metric and its first derivatives.

Parallel Transport

Parallel transport along curves preserves inner products (for Levi-Civita). On curved spaces, transport around closed loops produces a rotation (holonomy) that encodes curvature.

Musical Isomorphisms

The metric provides canonical identifications between vectors and covectors via index raising and lowering, unifying contravariant and covariant descriptions.

← Integration on Manifolds Next: Connections & Curvature →