Differential Forms

Differential forms are the natural objects to integrate on manifolds. They encode antisymmetric multilinear functions on tangent vectors and unify the gradient, curl, and divergence of vector calculus into a single framework: the exterior derivative.

Historical Context

Elie Cartan developed the calculus of differential forms in the early 20th century, building on the work of Grassmann (exterior algebra, 1844) and Poincare (who discovered the relation $d^2 = 0$ in 1895). The Hodge star operator was introduced by W.V.D. Hodge in 1941 to study harmonic forms on Riemannian manifolds, leading to the celebrated Hodge decomposition theorem.

Derivation 1: The Exterior Algebra

A $k$-form at a point $p \in M$ is an alternating multilinear map:

$\omega_p: \underbrace{T_pM \times \cdots \times T_pM}_{k \text{ copies}} \to \mathbb{R}$

"Alternating" means that swapping any two arguments reverses the sign:$\omega(\ldots, v, \ldots, w, \ldots) = -\omega(\ldots, w, \ldots, v, \ldots)$.

The Wedge Product

Given a $k$-form $\alpha$ and an $\ell$-form $\beta$, their wedge product is a $(k+\ell)$-form defined by antisymmetrization:

$(\alpha \wedge \beta)(v_1, \ldots, v_{k+\ell}) = \frac{1}{k!\ell!} \sum_{\sigma \in S_{k+\ell}} \text{sgn}(\sigma) \, \alpha(v_{\sigma(1)}, \ldots, v_{\sigma(k)}) \, \beta(v_{\sigma(k+1)}, \ldots, v_{\sigma(k+\ell)})$

Key properties of the wedge product:

$\alpha \wedge \beta = (-1)^{k\ell} \beta \wedge \alpha$

Graded commutativity

$(\alpha \wedge \beta) \wedge \gamma = \alpha \wedge (\beta \wedge \gamma)$

Associativity

$dx^i \wedge dx^j = -dx^j \wedge dx^i, \quad dx^i \wedge dx^i = 0$

Anticommutativity of basis 1-forms

A general $k$-form on an $n$-manifold can be written:

$\boxed{\omega = \frac{1}{k!} \omega_{i_1 \cdots i_k}(x) \, dx^{i_1} \wedge \cdots \wedge dx^{i_k}}$

The space $\Lambda^k(T_p^*M)$ has dimension $\binom{n}{k}$. In particular,$\Lambda^n(T_p^*M)$ is 1-dimensional—the volume form direction—and$\Lambda^k = 0$ for $k > n$.

Derivation 2: The Exterior Derivative

The exterior derivative $d: \Omega^k(M) \to \Omega^{k+1}(M)$ is the unique linear map satisfying:

On 0-forms: $df = \frac{\partial f}{\partial x^i} dx^i$

Graded Leibniz rule: $d(\alpha \wedge \beta) = d\alpha \wedge \beta + (-1)^k \alpha \wedge d\beta$

Nilpotency: $d^2 = d \circ d = 0$

Proof of $d^2 = 0$

Step 1: For a 0-form $f$:

$d(df) = d\left(\frac{\partial f}{\partial x^i} dx^i\right) = \frac{\partial^2 f}{\partial x^j \partial x^i} dx^j \wedge dx^i$

Step 2: Since $\partial_j \partial_i f = \partial_i \partial_j f$ (symmetry) and $dx^j \wedge dx^i = -dx^i \wedge dx^j$ (antisymmetry):

$d^2f = \frac{1}{2}\left(\frac{\partial^2 f}{\partial x^j \partial x^i} - \frac{\partial^2 f}{\partial x^i \partial x^j}\right) dx^j \wedge dx^i = 0$

The general case follows by the Leibniz rule. This $d^2 = 0$ property is the foundation of de Rham cohomology.

Unification of Vector Calculus

In $\mathbb{R}^3$, the exterior derivative unifies:

$d$ on 0-forms = gradient, $d$ on 1-forms = curl, $d$ on 2-forms = divergence. The identities $\nabla \times (\nabla f) = 0$ and $\nabla \cdot (\nabla \times \mathbf{F}) = 0$ are both instances of $d^2 = 0$.

Derivation 3: The Hodge Star Operator

On an oriented Riemannian manifold $(M^n, g)$, the Hodge star is a linear isomorphism $*: \Lambda^k \to \Lambda^{n-k}$ defined by:

$\alpha \wedge *\beta = \langle \alpha, \beta \rangle \, \text{vol}_g$

where $\text{vol}_g = \sqrt{|\det g|} \, dx^1 \wedge \cdots \wedge dx^n$ and $\langle \cdot, \cdot \rangle$ is the inner product on forms induced by $g$.

Explicit Formula

For a basis $k$-form $dx^{i_1} \wedge \cdots \wedge dx^{i_k}$:

$*(dx^{i_1} \wedge \cdots \wedge dx^{i_k}) = \frac{\sqrt{|g|}}{(n-k)!} g^{i_1 j_1} \cdots g^{i_k j_k} \epsilon_{j_1 \cdots j_k j_{k+1} \cdots j_n} dx^{j_{k+1}} \wedge \cdots \wedge dx^{j_n}$

Key property:

$\boxed{*\,* = (-1)^{k(n-k)} s}$

where $s = \text{sgn}(\det g)$ ($+1$ for Riemannian, $-1$ for Lorentzian)

The Hodge Laplacian

The codifferential is $\delta = (-1)^{nk+n+1} *d*$ on $k$-forms. The Hodge Laplacian is:

$\Delta = d\delta + \delta d = (d + \delta)^2$

On 0-forms in $\mathbb{R}^n$, this reduces to $\Delta f = -\sum_i \partial^2 f / \partial (x^i)^2$, the usual Laplacian (with physicist sign convention).

Derivation 4: Interior Product and Cartan's Magic Formula

The interior product (contraction) with a vector field $X$ is$\iota_X: \Omega^k(M) \to \Omega^{k-1}(M)$:

$(\iota_X \omega)(v_1, \ldots, v_{k-1}) = \omega(X, v_1, \ldots, v_{k-1})$

Cartan's Magic Formula

The Lie derivative of a form along a vector field $X$ is given by Cartan's remarkable formula:

$\boxed{\mathcal{L}_X \omega = \iota_X (d\omega) + d(\iota_X \omega)}$

Proof sketch: Both sides are derivations of the exterior algebra that agree on 0-forms (where $\mathcal{L}_X f = Xf = \iota_X df + d(\iota_X f) = Xf + 0$) and on exact 1-forms (where $\mathcal{L}_X df = d(\mathcal{L}_X f) = d(Xf)$). Since derivations are determined by their values on generators, the formula holds for all forms.

Cartan's formula is the key computational tool in symplectic geometry, where$\mathcal{L}_X \omega = 0$ defines symplectomorphisms and $\iota_X \omega = dH$ defines Hamiltonian vector fields.

Derivation 5: Maxwell's Equations as Differential Forms

Electromagnetism is the paradigmatic example of differential forms in physics. The electromagnetic field strength is a 2-form on Minkowski spacetime:

$F = \frac{1}{2} F_{\mu\nu} \, dx^\mu \wedge dx^\nu = E_i \, dx^i \wedge dt + \frac{1}{2} \epsilon_{ijk} B^k \, dx^i \wedge dx^j$

All four Maxwell equations reduce to just two elegant equations:

$dF = 0$

Bianchi identity: $\nabla \cdot \mathbf{B} = 0$ and $\nabla \times \mathbf{E} + \partial_t \mathbf{B} = 0$

$d{*F} = {*J}$

Source equations: $\nabla \cdot \mathbf{E} = \rho$ and $\nabla \times \mathbf{B} - \partial_t \mathbf{E} = \mathbf{J}$

Since $dF = 0$ and spacetime is contractible, by the Poincare lemma there exists a 1-form $A$ with:

$\boxed{F = dA, \quad A = A_\mu \, dx^\mu = -\phi \, dt + A_i \, dx^i}$

The gauge freedom $A \to A + d\Lambda$ is automatic since $d(A + d\Lambda) = dA + d^2\Lambda = dA = F$. Current conservation follows from $d^2(*F) = 0 = d(*J)$, giving $\partial_\mu J^\mu = 0$.

Elegance of forms: The formulation $dF = 0$, $d*F = *J$ is manifestly coordinate-independent, works in any dimension, and immediately generalizes to non-Abelian gauge theories (Yang-Mills: $D_A F = 0$, $D_A *F = *J$).

Applications to Physics

General Relativity

The Riemann curvature 2-form $\Omega^a{}_b = \frac{1}{2} R^a{}_{b\mu\nu} dx^\mu \wedge dx^\nu$ and the torsion 2-form $T^a = de^a + \omega^a{}_b \wedge e^b$ are central to the Cartan formulation of GR, where the Einstein equations become equations on differential forms.

Gauge Theory

The gauge field $A$ is a Lie-algebra-valued 1-form, and $F = dA + A \wedge A$ is the curvature 2-form. The Chern-Simons 3-form $\text{CS}(A) = \text{tr}(A \wedge dA + \frac{2}{3} A \wedge A \wedge A)$ plays a key role in topological quantum field theory.

Condensed Matter

The Berry curvature is a 2-form on parameter space. Integration over the Brillouin zone torus gives topological invariants (Chern numbers) that classify quantum Hall states and topological insulators.

Interactive Simulation

This simulation visualizes 1-forms as covector fields on $\mathbb{R}^2$, computes wedge products and exterior derivatives explicitly, and demonstrates the Hodge star by verifying the Laplacian for harmonic and non-harmonic functions.

Differential Forms: Wedge Product, Exterior Derivative & Hodge Star

Python

script.py153 lines

#!/usr/bin/env python3
"""
differential_forms.py - Wedge Product, Exterior Derivative, and Hodge Star
Visualizes differential forms, their operations, and the Hodge dual
"""
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import base64
from io import BytesIO

# ============================================================
# 1. Visualize 1-forms and 2-forms on R^2
# ============================================================
x = np.linspace(-2, 2, 20)
y = np.linspace(-2, 2, 20)
X, Y = np.meshgrid(x, y)

# 1-form: omega = x dy - y dx (angular form)
# As a covector field: omega_x = -y, omega_y = x
U1 = -Y  # dx component
V1 = X   # dy component

# Another 1-form: eta = x dx + y dy (radial form)
U2 = X
V2 = Y

# Exterior derivative of 0-form f = x*y: df = y dx + x dy
U3 = Y
V3 = X

# Wedge product omega ^ eta = (-y*y - x*x) dx^dy = -(x^2+y^2) dx^dy
wedge_density = -(X**2 + Y**2)

# Exterior derivative of omega: d(omega) = dx^dy - (-dy^dx) = 2 dx^dy
d_omega = 2 * np.ones_like(X)

print("=" * 72)
print("  DIFFERENTIAL FORMS: WEDGE PRODUCT & EXTERIOR DERIVATIVE")
print("=" * 72)
print()
print("  SETUP:")
print("    omega = -y dx + x dy    (angular 1-form)")
print("    eta   =  x dx + y dy    (radial 1-form)")
print("    f     =  x * y          (0-form)")
print()
print("  EXTERIOR DERIVATIVE OF f:")
print("    df = (df/dx) dx + (df/dy) dy = y dx + x dy")
print()
print("  EXTERIOR DERIVATIVE OF omega:")
print("    d(omega) = d(-y) ^ dx + d(x) ^ dy")
print("             = -dy ^ dx + dx ^ dy")
print("             = dx ^ dy + dx ^ dy = 2 dx ^ dy")
print()
print("  WEDGE PRODUCT omega ^ eta:")
print("    = (-y dx + x dy) ^ (x dx + y dy)")
print("    = -y*y (dx ^ dy) + x*x (dy ^ dx)")
print("    ... wait: dy ^ dx = -dx ^ dy")
print("    = (-y^2 - x^2) dx ^ dy = -(x^2 + y^2) dx ^ dy")
print()
print("  HODGE STAR IN R^2 (Euclidean):")
print("    *1 = dx ^ dy")
print("    *(dx) = dy,  *(dy) = -dx")
print("    *(dx ^ dy) = 1")
print()
print("  HODGE STAR OF omega:")
print("    *(omega) = *(-y dx + x dy) = -y dy + x(-dx) = -x dx - y dy")
print("    = -(x dx + y dy) = -eta")
print()

# Check: d*d on 0-form = Laplacian
print("  LAPLACIAN VIA HODGE STAR:")
print("    For f = x*y:")
print("    df = y dx + x dy")
print("    *df = y dy - x dx")
print("    d(*df) = dy ^ dy - dx ^ dx = 0")
print("    *d(*df) = 0  (Laplacian of xy is 0: harmonic!)")
print()

# Verify for non-harmonic: g = x^2 + y^2
print("    For g = x^2 + y^2:")
print("    dg = 2x dx + 2y dy")
print("    *dg = 2x dy - 2y dx")
print("    d(*dg) = 2 dx ^ dy + 2 dy ^ (-dx) = 2 dx^dy + 2 dx^dy = 4 dx^dy")
print("    *d(*dg) = 4  (Laplacian of x^2+y^2 = 4, correct!)")
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=0.3)

# Plot 1: Angular 1-form
ax1 = axes[0, 0]
mag1 = np.sqrt(U1**2 + V1**2)
mag1[mag1 == 0] = 1
ax1.quiver(X, Y, U1/mag1, V1/mag1, mag1, cmap='winter', alpha=0.8, scale=25)
ax1.set_title('Angular 1-form: omega = -y dx + x dy', color='#99f6e4', fontsize=11, fontweight='bold')
ax1.set_xlabel('x', color='#5eead4')
ax1.set_ylabel('y', color='#5eead4')
ax1.set_aspect('equal')

# Plot 2: Exterior derivative df for f=xy
ax2 = axes[0, 1]
mag3 = np.sqrt(U3**2 + V3**2)
mag3[mag3 == 0] = 1
ax2.quiver(X, Y, U3/mag3, V3/mag3, mag3, cmap='autumn', alpha=0.8, scale=25)
ax2.set_title('df for f = xy: df = y dx + x dy', color='#99f6e4', fontsize=11, fontweight='bold')
ax2.set_xlabel('x', color='#5eead4')
ax2.set_ylabel('y', color='#5eead4')
ax2.set_aspect('equal')

# Plot 3: Wedge product density
ax3 = axes[1, 0]
c3 = ax3.contourf(X, Y, wedge_density, levels=20, cmap='RdBu_r')
plt.colorbar(c3, ax=ax3, label='Density')
ax3.set_title('omega ^ eta = -(x^2+y^2) dx^dy', color='#99f6e4', fontsize=11, fontweight='bold')
ax3.set_xlabel('x', color='#5eead4')
ax3.set_ylabel('y', color='#5eead4')
ax3.set_aspect('equal')

# Plot 4: Exterior derivative of omega (constant 2-form)
ax4 = axes[1, 1]
xf = np.linspace(-2, 2, 100)
yf = np.linspace(-2, 2, 100)
Xf, Yf = np.meshgrid(xf, yf)
d_omega_fine = 2 * np.ones_like(Xf)
c4 = ax4.contourf(Xf, Yf, d_omega_fine, levels=20, cmap='viridis')
plt.colorbar(c4, ax=ax4, label='d(omega) coefficient')
ax4.set_title('d(omega) = 2 dx ^ dy (constant)', color='#99f6e4', fontsize=11, fontweight='bold')
ax4.set_xlabel('x', color='#5eead4')
ax4.set_ylabel('y', color='#5eead4')
ax4.set_aspect('equal')

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="Differential Forms Visualization" />')
buf.close()
print()
print("Visualization complete.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

← Tangent Spaces Next: Integration on Manifolds →