Curvature 2-Form and Cartan's Structure Equations

Curvature is the geometric fact that a vector carried around a closed loop comes back rotated. The Riemann tensor — built in aq-gr-riemann-curvature-tensor and read as the commutator of covariant derivatives in aq-gr-curvature-commutator-covariant-derivatives — encodes that rotation in a four-index coordinate object with dozens of Christoffel symbols behind it. Élie Cartan found a way to compute the same rotation with almost no index gymnastics. Carry a field of orthonormal rulers and clocks — a moving frame, or tetrad — from point to point and describe how it turns using differential forms. The connection becomes a single matrix of 1-forms $\omega^a{}_b$ ; the curvature a closely related matrix of 2-forms $R^a{}_b$ ; and the entire content of Riemannian geometry collapses into two compact lines, Cartan's structure equations. The payoff is both computational and conceptual. Computationally, exterior differentiation is metric-independent and antisymmetrizes for free, so the curvature of Schwarzschild, Kerr, or FLRW comes from differentiating a handful of 1-forms rather than evaluating hundreds of connection components. Conceptually, the combination $R=\mathrm{d}\omega+\omega\wedge\omega$ is exactly the field strength of a gauge theory — the frame-rotation group is the gauge group, $\omega$ the gauge potential — so gravity's kinship with Yang-Mills becomes manifest. And nothing is lost: the frame components of $R^a{}_b$ are the orthonormal-frame Riemann components, so the symmetries, Bianchi identities, and contractions to Ricci and Einstein all survive the translation into form language.

Key takeaways

A field of orthonormal frames has a dual coframe of 1-forms $\{e^a\}$ ; the connection is repackaged as a single matrix of 1-forms $\omega^a{}_b$ , the spin connection, valued in the Lie algebra of the frame-rotation group.
Cartan's first structure equation $\mathrm{d}e^a+\omega^a{}_b\wedge e^b=T^a$ makes torsion the exterior covariant derivative of the coframe; it vanishes for the Levi-Civita connection.
Cartan's second structure equation $R^a{}_b=\mathrm{d}\omega^a{}_b+\omega^a{}_c\wedge\omega^c{}_b$ produces the curvature 2-form, whose components $R^a{}_{bcd}$ are the orthonormal-frame Riemann tensor.
Metric compatibility forces antisymmetry $\omega_{ab}=-\omega_{ba}$ ; together with zero torsion it fixes $\omega$ uniquely — the fundamental theorem of Riemannian geometry in form language.
The structure of $R=\mathrm{d}\omega+\omega\wedge\omega$ is identical to a Yang-Mills field strength $F=\mathrm{d}A+A\wedge A$ : gravity is a gauge theory of the local frame group, and the Bianchi identities become $\mathrm{d}_\omega R=0$ .
Key idea: choose a coframe $\{e^a\}$ ; the no-torsion condition $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ with antisymmetry $\omega_{ab}=-\omega_{ba}$ uniquely determines the connection 1-form, and then $R^a{}_b=\mathrm{d}\omega^a{}_b+\omega^a{}_c\wedge\omega^c{}_b$ delivers the full Riemann tensor as a matrix of 2-forms.

The core insight

Choose a coframe $\{e^a\}$ ; the no-torsion condition $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ together with antisymmetry $\omega_{ab}=-\omega_{ba}$ uniquely determines the connection 1-form, and then $R^a{}_b=\mathrm{d}\omega^a{}_b+\omega^a{}_c\wedge\omega^c{}_b$ delivers the full Riemann tensor as a matrix of 2-forms.

Where this lives today

Cartan's formalism is the working engine behind modern gravitational physics. Every computer-algebra curvature package (GRTensor, xAct, SageManifolds) offers the tetrad route because differentiating a coframe is cheaper and less error-prone than contracting Christoffel symbols. In numerical relativity the tetrad formalism underlies the Newman-Penrose calculations that extract gravitational waves from a simulation: the Weyl scalar $\psi_4$ , read off a null tetrad, is what LIGO-Virgo waveform models are calibrated against. The gauge reading drives whole programs — loop quantum gravity quantizes $\omega$ and $e$ directly, and the Palatini and MacDowell-Mansouri formulations treat gravity as a gauge theory of the Poincaré or de Sitter group. And because spinors carry no coordinate indices, every calculation of fermions in curved spacetime — Hawking radiation, the electroweak early universe — uses the spin connection: Cartan's moving frame is the only door through which the Dirac equation enters general relativity.

The physical picture

Forget coordinates and carry a rigid set of orthonormal rulers and clocks — the frame $\{e_a\}$ — from event to event. To stay orthonormal it must rotate and boost, and the connection 1-form $\omega^a{}_b$ is the infinitesimal generator of that turning, one $\mathfrak{so}(3,1)$ matrix per unit displacement. Now carry the frame around a small closed loop and set it back down. In flat space it returns unchanged; in a curved space it comes back rotated, and the net rotation per unit oriented area is the curvature 2-form $R^a{}_b$ . Quantitatively, transport around a small loop bounding a surface $S$ rotates a frame component by $\Delta V^a=-\int_S R^a{}_b\,V^b$ (Stokes applied to $\mathrm{d}V^a+\omega^a{}_b V^b=0$ ), which reduces to $[\nabla_c,\nabla_d]V^a=R^a{}_{bcd}V^b$ and, for finite loops, to the Ambrose-Singer theorem. This is the definition made operational — and the cleanest way to see why curvature is a 2-form: a 2-form is exactly what eats an oriented area element and returns a number.

Figure

Parallel transport around a geodesic triangle on a sphere of radius a. A vector carried along the three right-angle edges of one octant returns rotated by Δφ = K·A = (1 divided by a²)(πa² divided by 2) = 90°, the curvature 2-form R¹₂ = (1 divided by a²) e¹∧e² integrated over the enclosed area.

The computational magic is that you never write a Christoffel symbol. Exterior differentiation ignores the metric and antisymmetrizes automatically, so the recipe is short: write the orthonormal coframe 1-forms, differentiate them (usually only a few terms survive), read off the antisymmetric $\omega$ that cancels the torsion, and differentiate once more. What remains is the curvature. The nonlinear $\omega\wedge\omega$ is not decoration — $\mathrm{d}\omega$ alone is not a tensor, since under a frame rotation $\omega$ picks up an inhomogeneous term that only the quadratic piece cancels, exactly as in a gauge field strength.

Figure

The Cartan pipeline: from the coframe eᵃ, differentiate to deᵃ, solve the first structure equation at zero torsion for the spin connection ωᵃ_b, then apply R = dω + ω∧ω for the curvature. For Schwarzschild it delivers Rᵗᵣ = +(2M divided by r³) eᵗ∧eᵣ, the same algebra as the Yang-Mills field strength F = dA + A∧A.

The core equations

Everything begins with the coframe. An orthonormal frame $\{e_a\}$ has a dual basis of 1-forms $\{e^a\}$ with $g=\eta_{ab}\,e^a\otimes e^b$ , $\eta_{ab}=\mathrm{diag}(-1,1,1,1)$ . The connection is stored not in $\Gamma^\lambda{}_{\mu\nu}$ but in the connection 1-form $\omega^a{}_b$ , defined by $\nabla e_b=\omega^a{}_b\otimes e_a$ . Cartan's first structure equation identifies the antisymmetrized covariant derivative of the coframe with the torsion.

\mathrm{d}e^a + \omega^a{}_b \wedge e^b = T^a \qquad (1)

where: $e^a$ — coframe (vielbein/tetrad) 1-forms; $\omega^a{}_b$ — connection 1-form (spin connection), a matrix of 1-forms valued in the frame-rotation Lie algebra; $T^a$ — torsion 2-form, zero for Levi-Civita; $\wedge$ — exterior product. Setting $T^a=0$ turns (1) into an equation for $\omega$ : $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ . Differentiating the connection and adding its matrix square produces the curvature 2-form — the second structure equation, the heart of the formalism, whose quadratic term makes $R$ a genuine tensor under frame rotations.

R^a{}_b = \mathrm{d}\omega^a{}_b + \omega^a{}_c \wedge \omega^c{}_b \qquad (2)

where: $R^a{}_b$ — curvature 2-form, valued in the same Lie algebra as $\omega$ ; $\mathrm{d}\omega^a{}_b$ — exterior derivative of the connection 1-form; $\omega^a{}_c\wedge\omega^c{}_b$ — matrix product (sum over $c$ ) combined with the wedge of two 1-forms. Read (2) side by side with the Yang-Mills field strength $F=\mathrm{d}A+A\wedge A$ : the same equation, with $\omega$ the gauge potential and $R$ its field strength. Metric compatibility $\nabla g=0$ keeps the frame orthonormal. Because $\eta_{ab}$ is constant, compatibility collapses to a simple antisymmetry once both indices are lowered.

\omega_{ab}=-\,\omega_{ba},\qquad \omega_{ab}\equiv\eta_{ac}\,\omega^c{}_b \qquad (3)

where: $\eta_{ab}$ — frame metric, $\mathrm{diag}(-1,1,1,1)$ or $\delta_{ab}$ ; $\omega_{ab}$ — connection 1-form, both indices lowered. Antisymmetry means $\omega\in\mathfrak{so}(p,q)$ : in 4D, six independent 1-forms — three rotations and three boosts. The curvature 2-form expands in the coframe basis, and its coefficients are the orthonormal-frame Riemann components.

R^a{}_b=\tfrac12\,R^a{}_{bcd}\,e^c\wedge e^d,\qquad R^a{}_{bcd}=e^a{}_\mu\,e_b{}^\nu\,e_c{}^\rho\,e_d{}^\sigma\,R^\mu{}_{\nu\rho\sigma} \qquad (4)

where: $R^a{}_{bcd}$ — Riemann tensor in the orthonormal frame; $e^a{}_\mu$ — vielbein converting a coordinate index to a frame index. Lowering the first index gives $R_{ab}=-R_{ba}$ : the curvature 2-form is antisymmetric in its frame pair, like $\omega$ . Differentiating the structure equations regenerates the Bianchi identities without a single Christoffel manipulation, in the compact forms below.

R^a{}_b\wedge e^b=0,\qquad \mathrm{d}R^a{}_b+\omega^a{}_c\wedge R^c{}_b-R^a{}_c\wedge\omega^c{}_b=0 \qquad (5)

where: the first (algebraic) identity is $\mathrm{d}$ of (1) with $T^a=0$ , equivalent to $R^a{}_{[bcd]}=0$ ; the second (differential) identity is $\mathrm{d}$ of (2), the statement $\mathrm{d}_\omega R^a{}_b=0$ — the geometric ancestor of $\nabla_\mu G^{\mu\nu}=0$ and hence of local energy-momentum conservation. Finally, under a position-dependent frame rotation $\Lambda^a{}_b(x)$ the connection transforms like a gauge potential, the curvature homogeneously as a tensor.

\tilde\omega=\Lambda\,\omega\,\Lambda^{-1}+\Lambda\,\mathrm{d}\Lambda^{-1},\qquad \tilde R=\Lambda\,R\,\Lambda^{-1} \qquad (6)

where: $\Lambda^a{}_b(x)$ — local $SO(p,q)$ frame rotation (a local Lorentz transformation). The inhomogeneous term $\Lambda\,\mathrm{d}\Lambda^{-1}$ is what disqualifies $\mathrm{d}\omega$ alone from being the curvature; the $\omega\wedge\omega$ piece of (2) cancels it, leaving $\tilde R=\Lambda R\Lambda^{-1}$ .

Derivation AUniqueness of the Levi-Civita spin connection

The goal is to prove the Cartan form of the fundamental theorem of Riemannian geometry — that vanishing torsion together with metric compatibility fix the spin connection $\omega^a{}_b$ uniquely — and to produce a closed formula for it from quantities read straight off the coframe. This is the exact analogue of the metric fixing the Christoffel symbols, but the input data are now the exterior derivatives of the coframe, and the argument is three lines of exterior algebra.

Step 1. Encode the coframe's twisting in structure functions. Because $\mathrm{d}e^a$ is a 2-form, expand it as $\mathrm{d}e^a=-\tfrac12\,C^a{}_{bc}\,e^b\wedge e^c$ with $C^a{}_{bc}=-C^a{}_{cb}$ ; the coefficients $C^a{}_{bc}$ are the anholonomy (or structure) coefficients of the frame, computable directly by differentiating the coframe 1-forms. Why: they are the frame's analogue of metric gradients — the only local data the connection can possibly depend on once the metric is orthonormalized.

Step 2. Write the unknown in the same basis, $\omega^a{}_b=\omega^a{}_{bc}\,e^c$ , and lower an index, $\omega_{abc}=\eta_{ad}\,\omega^d{}_{bc}$ . Metric compatibility (3) says $\omega_{abc}=-\omega_{bac}$ : antisymmetry in the first pair. Why: this halves the unknowns and is the only place the metric enters — from here the problem is pure algebra.

Step 3. Impose zero torsion. Substituting into $\mathrm{d}e^a+\omega^a{}_b\wedge e^b=0$ and matching coefficients gives $C^a{}_{bc}=\omega^a{}_{cb}-\omega^a{}_{bc}$ : antisymmetry in the last pair. Why: torsion is exactly the last-pair-antisymmetric part of $\omega$ , so setting it to zero relates that part to the known $C$ .

Step 4. Solve the two antisymmetries together. Writing the no-torsion relation three times with indices cyclically permuted and combining with $\omega_{abc}=-\omega_{bac}$ isolates $\omega$ uniquely,

\omega_{abc}=\tfrac12\left(C_{abc}-C_{bca}+C_{cab}\right).

Why: the same cyclic-permutation trick that gives the Christoffel formula — one line here because the wedge did the antisymmetrizing.

Step 5. Conclude. Both conditions admit exactly one solution, so $\omega$ is unique — the Cartan version of the fundamental theorem, with $C^a{}_{bc}$ playing the role of $\partial g$ . A check on the round two-sphere gives $C^2{}_{21}=\cot\theta/a$ and hence $\omega^1{}_2=-\cos\theta\,\mathrm{d}\phi$ . As a check: in a coordinate frame where all anholonomy coefficients vanish, $C^a{}_{bc}=0$ , the formula reduces to the standard Christoffel expression $\Gamma^\lambda{}_{\mu\nu}=\tfrac12 g^{\lambda\rho}(\partial_\mu g_{\nu\rho}+\partial_\nu g_{\mu\rho}-\partial_\rho g_{\mu\nu})$ ; and on the round two-sphere it correctly yields $\omega^1{}_2=-\cos\theta\,\mathrm{d}\phi$ .

Go deeper · Derivation B · 1 minSchwarzschild curvature by Cartan's method

Derivation BSchwarzschild curvature by Cartan's method

The goal is to compute the full curvature of the Schwarzschild geometry $\mathrm{d}s^2=-f\,\mathrm{d}t^2+f^{-1}\mathrm{d}r^2+r^2\mathrm{d}\Omega^2$ with $f=1-2M/r$ from the two structure equations alone — showing the formalism's advantage: what would be a page of Christoffel symbols becomes four differentiations. We pull out the two signature curvature 2-forms, confirm the vacuum condition, and quote the Kretschmann scalar as a check.

Step 1. Choose the orthonormal coframe that diagonalizes the metric:

e^{\hat t}=f^{1/2}\mathrm{d}t,\quad e^{\hat r}=f^{-1/2}\mathrm{d}r,\quad e^{\hat\theta}=r\,\mathrm{d}\theta,\quad e^{\hat\phi}=r\sin\theta\,\mathrm{d}\phi,

so that $g=-e^{\hat t}\otimes e^{\hat t}+e^{\hat r}\otimes e^{\hat r}+e^{\hat\theta}\otimes e^{\hat\theta}+e^{\hat\phi}\otimes e^{\hat\phi}$ . Why: in an orthonormal coframe the metric is constant, so antisymmetry $\omega_{ab}=-\omega_{ba}$ applies and $\omega$ can be read by inspection.

Step 2. Differentiate the coframe. With $f'=2M/r^2$ ,

\mathrm{d}e^{\hat t}=\tfrac{f'}{2f^{1/2}}\,\mathrm{d}r\wedge\mathrm{d}t,\quad \mathrm{d}e^{\hat r}=0,\quad \mathrm{d}e^{\hat\theta}=\mathrm{d}r\wedge\mathrm{d}\theta,\quad \mathrm{d}e^{\hat\phi}=\sin\theta\,\mathrm{d}r\wedge\mathrm{d}\phi+r\cos\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi.

Why: these few nonzero 2-forms are the entire input; each is cancelled by one connection 1-form.

Step 3. Read off the connection by solving $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ with the antisymmetry rules ( $\omega_{\hat t\hat r}=+\omega^{\hat t}{}_{\hat r}$ since $\eta_{\hat t\hat t}=-1$ ; spatial pairs antisymmetric):

\omega^{\hat t}{}_{\hat r}=\tfrac{f'}{2}\,\mathrm{d}t,\quad \omega^{\hat\theta}{}_{\hat r}=\tfrac{f^{1/2}}{r}\,e^{\hat\theta},\quad \omega^{\hat\phi}{}_{\hat r}=\tfrac{f^{1/2}}{r}\,e^{\hat\phi},\quad \omega^{\hat\phi}{}_{\hat\theta}=\tfrac{\cot\theta}{r}\,e^{\hat\phi}.

Why: each term of $\mathrm{d}e^a$ is matched by one $\omega$ ; antisymmetry fixes the rest, and torsion checks to zero.

Step 4. Apply the second structure equation to the radial 2-form. Since $\omega^{\hat t}{}_{\hat\theta}=\omega^{\hat t}{}_{\hat\phi}=0$ , the matrix-square term drops and

R^{\hat t}{}_{\hat r}=\mathrm{d}\omega^{\hat t}{}_{\hat r}=\tfrac{f''}{2}\,\mathrm{d}r\wedge\mathrm{d}t=+\frac{2M}{r^3}\,e^{\hat t}\wedge e^{\hat r},

using $f''=-4M/r^3$ and $\mathrm{d}r\wedge\mathrm{d}t=-e^{\hat t}\wedge e^{\hat r}$ . Why: that sign flip is the most-slipped step in the computation — track it and the answer is unambiguous.

Step 5. The angular 2-form needs the quadratic term. With $\omega^{\hat\theta}{}_{\hat\phi}=-\cos\theta\,\mathrm{d}\phi$ giving $\mathrm{d}\omega^{\hat\theta}{}_{\hat\phi}=\sin\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi$ and $\omega^{\hat\theta}{}_{\hat r}\wedge\omega^{\hat r}{}_{\hat\phi}=-f\sin\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi$ ,

R^{\hat\theta}{}_{\hat\phi}=(1-f)\sin\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi=+\frac{2M}{r^3}\,e^{\hat\theta}\wedge e^{\hat\phi}.

Why: this one genuinely needs $\omega\wedge\omega$ ; the $\mathrm{d}\omega$ and nonlinear pieces combine to $1-f=2M/r$ .

Step 6. Collect the frame Riemann components from $R^a{}_b=\tfrac12 R^a{}_{bcd}e^c\wedge e^d$ . Lowering the first index gives the tidal tensor $R_{\hat t\hat r\hat t\hat r}=-2M/r^3$ , $R_{\hat t\hat\theta\hat t\hat\theta}=R_{\hat t\hat\phi\hat t\hat\phi}=+M/r^3$ , $R_{\hat r\hat\theta\hat r\hat\theta}=R_{\hat r\hat\phi\hat r\hat\phi}=-M/r^3$ , $R_{\hat\theta\hat\phi\hat\theta\hat\phi}=+2M/r^3$ — the $(-2,+1,+1,-1,-1,+2)\times M/r^3$ pattern. Contracting gives $R_{ab}=0$ : Schwarzschild is vacuum, all curvature Weyl (tidal). As a check: the frame Riemann components are $R_{\hat t\hat r\hat t\hat r}=-2M/r^3$ , $R_{\hat\theta\hat\phi\hat\theta\hat\phi}=+2M/r^3$ , the Ricci tensor vanishes $R_{ab}=0$ , and the Kretschmann scalar is $R_{abcd}R^{abcd}=48M^2/r^6$ .

Limiting cases

Case	Result	Meaning
Flat space	$R^a{}_b=0$	Cartesian frame has $\omega^a{}_b=0$ ; conversely $R=0$ means $\omega$ is locally pure gauge, the Maurer-Cartan form of $SO(p,q)$
Levi-Civita connection	$T^a=0$	Zero torsion plus antisymmetry $\omega_{ab}=-\omega_{ba}$ fix $\omega$ uniquely — the fundamental theorem in Cartan form
Two-sphere $S^2$ , radius $a$	$R^1{}_2=\dfrac{1}{a^2}\,e^1\wedge e^2$	With $\omega^1{}_2=-\cos\theta\,\mathrm{d}\phi$ , Gaussian curvature $K=1/a^2>0$ — a sphere curves positively
Schwarzschild, radial	$R^{\hat t}{}_{\hat r}=\dfrac{2M}{r^3}\,e^{\hat t}\wedge e^{\hat r}$	Vacuum but curved: $R_{ab}=0$ while $R_{abcd}\neq0$ — the tidal (Weyl) curvature that stretches infalling bodies
Linearized / abelian limit	$R^a{}_b\approx\mathrm{d}\omega^a{}_b$	Small $\omega$ : the quadratic term drops and curvature is linear, like the Maxwell field $F=\mathrm{d}A$
Bi-invariant Lie group	$\mathrm{d}\theta+\tfrac12[\theta,\theta]=0$	The Maurer-Cartan equation; $\omega=\tfrac12\theta$ gives constant curvature set by the structure constants

Common trapThe wedge ω∧ω must vanish because α∧α = 0 for any 1-form

It does not vanish, and this is the most common slip in the formalism. The rule $\alpha\wedge\alpha=0$ holds for a single scalar-valued 1-form. But $\omega^a{}_b$ is a matrix of 1-forms, and $\omega^a{}_c\wedge\omega^c{}_b$ is a matrix product summed over $c$ together with a wedge of two generally different 1-forms — like $\omega^{\hat\theta}{}_{\hat r}\wedge\omega^{\hat r}{}_{\hat\phi}$ in the Schwarzschild angular curvature, nonzero and supplying half the answer. Two related traps: antisymmetry $\omega_{ab}=-\omega_{ba}$ holds only in an orthonormal coframe after lowering with $\eta_{ab}$ ; and $\mathrm{d}\omega$ alone is neither the curvature nor a tensor — only $\mathrm{d}\omega+\omega\wedge\omega$ transforms homogeneously.

Worked examples

1 · The curvature of a sphere by Cartan's method (confidence builder)

Take a round two-sphere of radius $a$ , $\mathrm{d}s^2=a^2\,\mathrm{d}\theta^2+a^2\sin^2\theta\,\mathrm{d}\phi^2$ . Find the connection 1-form, curvature 2-form, and Gaussian curvature, then evaluate it for the Earth.

Step 1. Choose the orthonormal coframe $e^1=a\,\mathrm{d}\theta$ , $e^2=a\sin\theta\,\mathrm{d}\phi$ (Riemannian, $\eta=\delta$ ). Differentiate: $\mathrm{d}e^1=0$ and $\mathrm{d}e^2=a\cos\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi$ .

Step 2. Solve $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ with a single off-diagonal 1-form $\omega^1{}_2=-\omega^2{}_1$ . The condition $\mathrm{d}e^2=-\omega^2{}_1\wedge e^1$ gives $a\cos\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi=-\omega^2{}_1\wedge a\,\mathrm{d}\theta$ , solved by $\boldsymbol{\omega^1{}_2=-\cos\theta\,\mathrm{d}\phi}$ ; the other condition $\mathrm{d}e^1=0=-\omega^1{}_2\wedge e^2$ holds automatically.

Step 3. Apply the second structure equation. In two dimensions the matrix square $\omega^1{}_c\wedge\omega^c{}_2$ has no third index and cannot contribute, so $R^1{}_2=\mathrm{d}\omega^1{}_2=\sin\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi$ . Since $e^1\wedge e^2=a^2\sin\theta\,\mathrm{d}\theta\wedge\mathrm{d}\phi$ , this is $\boldsymbol{R^1{}_2=\dfrac{1}{a^2}\,e^1\wedge e^2}$ .

Step 4. Read the Gaussian curvature from $R^1{}_2=K\,e^1\wedge e^2$ : $\boldsymbol{K=1/a^2}$ , positive and constant — the whole curvature content of the surface. For the Earth, $a=6.371\times10^{6}\ \mathrm{m}$ gives $K=\mathbf{2.46\times10^{-14}\ m^{-2}}$ : gentle enough that the ground looks flat, yet integrated over the sphere it forces the $4\pi$ of Gauss-Bonnet. ω¹₂ = −cos θ dφ, R¹₂ = (1/a²) e¹∧e², K = 2.46×10⁻¹⁴ m⁻² for the Earth.

2 · Tidal stretching from the Schwarzschild curvature 2-form (numeric)

A stellar-mass black hole has $M=10\,M_\odot$ . Using the curvature 2-form computed above, find the radial tidal "stretch rate" at the horizon and estimate the head-to-toe pull on a $1.8\ \mathrm{m}$ astronaut.

Step 1. The geodesic-deviation equation for a static frame is $\ddot\xi^{\hat r}=-R^{\hat r}{}_{\hat t\hat r\hat t}\,\xi^{\hat r}$ . From the derivation $R^{\hat r}{}_{\hat t\hat r\hat t}=R_{\hat t\hat r\hat t\hat r}=-2M/r^3$ , so $\ddot\xi^{\hat r}=+\dfrac{2M}{r^3}\,\xi^{\hat r}$ : a positive eigenvalue means radial stretching. In SI the stretch rate is $\mathcal{E}_{\hat r\hat r}=2GM/r^3$ (units $\mathrm{s^{-2}}$ ).

Step 2. Locate the horizon. With $M=10\,M_\odot=1.99\times10^{31}\ \mathrm{kg}$ and $GM=1.33\times10^{21}\ \mathrm{m^3\,s^{-2}}$ , $r_s=2GM/c^2=\mathbf{2.95\times10^{4}\ m}$ (about $29.5\ \mathrm{km}$ ).

Step 3. Evaluate at $r=r_s$ :

\mathcal{E}_{\hat r\hat r}=\frac{2GM}{r_s^3}=\frac{2\,(1.33\times10^{21})}{(2.95\times10^{4})^3}=\mathbf{1.0\times10^{8}\ s^{-2}},

a tidal gradient: two points $1\ \mathrm{m}$ apart radially accelerate apart at $10^{8}\ \mathrm{m\,s^{-2}}$ .

Step 4. Across a $1.8\ \mathrm{m}$ astronaut the differential acceleration is $\sim1.8\times10^{8}\ \mathrm{m\,s^{-2}}$ , about $2\times10^{7}\,g$ — lethal spaghettification well outside the horizon. The Kretschmann scalar there is $48M^2/r_s^6\approx1.6\times10^{-17}\ \mathrm{m^{-4}}$ , tiny in geometrized units yet corresponding to this enormous physical pull. Stretch rate 1.0×10⁸ s⁻², head-to-toe pull ≈ 1.8×10⁸ m/s² at the horizon of a 10 M⊙ black hole.

Physics of the portrait

Cartan's structure equations are where nineteenth-century Lie theory, moving-frame geometry, and twentieth-century gauge physics converge on a single pair of lines.

Origins · 1 minÉlie Cartan and how this was discovered

ÉC

Élie Cartan

1869–1951 · differential geometry and Lie groups

Élie Cartan was the son of a village blacksmith in the French Alps, educated on a state scholarship — an origin as unlikely as any in mathematics. By the 1890s his doctoral thesis had completed the classification of simple Lie algebras that Killing left in disarray, work that still underlies particle physics. But it was his calculus of differential forms and his method of moving frames (repères mobiles) that reshaped geometry. Where Ricci and Levi-Civita computed with tensor components and Christoffel symbols, Cartan attached a frame to every point and asked how it turned — packaging the connection into a matrix-valued 1-form and curvature into his structure equations across the 1920s. He introduced torsion as the missing partner of curvature, built the theory of connections on bundles decades before physicists needed it, and discovered spinors in 1913, long before Dirac. His notation ran so far ahead of its time that contemporaries found him nearly unreadable; Weyl, and later Chern and Ehresmann, recast his ideas in the modern language of fiber bundles. When Yang-Mills gauge theory arrived in 1954, physicists realized Cartan had written the geometry of gauge fields — connection, curvature, Bianchi identity — thirty years early. The equations that carry his name are now the standard tool for the curvature of every exact solution, and the bridge by which fermions couple to gravity.

History

1869

— Élie Cartan born in Dolomieu, Isère, son of a blacksmith

1894

— completes the classification of complex simple Lie algebras in his thesis

1913

— discovers spinors while classifying representations of the rotation groups

1922

— introduces torsion and the general theory of affine connections

1925

— the structure equations and moving-frame method reach mature form

1954

— Yang and Mills write down non-abelian gauge theory, later seen to share Cartan's geometry

1963

— Ambrose and Singer prove that holonomy is generated by the curvature 2-form

Going further

Texts. Misner, Thorne & Wheeler, Gravitation (1973), §14.5–14.6 (the calculus of forms and the structure equations) and Box 14.5 (Schwarzschild orthonormal-frame curvature) · Wald, General Relativity (1984), §3.4b and the tetrad appendix · Carroll, Spacetime and Geometry (2004), §3.4 (vielbeins and the spin connection) · Nakahara, Geometry, Topology and Physics (2003), §7.8 · Frankel, The Geometry of Physics (2011), Ch. 9 · Schutz, Geometrical Methods of Mathematical Physics (1980), Ch. 6. The spin-connection solution $\omega_{abc}=\tfrac12(C_{abc}-C_{bca}+C_{cab})$ and the second structure equation together reproduce $[\nabla_c,\nabla_d]V^a=R^a{}_{bcd}V^b$ . Key numerical results: $R_{\hat t\hat r\hat t\hat r}=-2M/r^3$ , $R_{\hat\theta\hat\phi\hat\theta\hat\phi}=+2M/r^3$ , $R_{ab}=0$ , Kretschmann $48M^2/r^6$ , and $S^2$ curvature $K=1/a^2$ with $\omega^1{}_2=-\cos\theta\,\mathrm{d}\phi$ . The mixed-index sign $R^{\hat t}{}_{\hat r}=+\tfrac{2M}{r^3}e^{\hat t}\wedge e^{\hat r}$ is fixed to match the sphere's positive Gaussian curvature in the same convention.

Builds on. Riemann Curvature Tensor · Curvature as the Commutator of Covariant Derivatives · Differential forms and the exterior derivative. Leads to. Second (Differential) Bianchi Identity · First (Algebraic) Bianchi Identity · Weyl Conformal Tensor · Einstein Tensor.

Ask the tutor: Want me to run Cartan's method live on a metric of your choice — Kerr, Reissner-Nordström, or FLRW — and read off the curvature 2-forms step by step?

What is the difference between the connection 1-form and the Christoffel symbols?

They store the same information — how to differentiate covariantly — in different bases: $\Gamma^\lambda{}_{\mu\nu}$ in a coordinate frame, $\omega^a{}_b$ in an orthonormal frame, packaged as a matrix of 1-forms. The payoff of $\omega$ is threefold: it is antisymmetric (six components in 4D instead of forty), it plugs straight into the wedge-based structure equations, and it is the object spinors couple to — decisive, since $\Gamma$ cannot differentiate a spinor at all.

Why is curvature a 2-form rather than something else?

Because curvature is a rotation acquired per unit oriented area, and a 2-form is exactly what assigns a number to an area element antisymmetrically. Transport a frame around a small parallelogram spanned by $e_c$ and $e_d$ : the net rotation is antisymmetric under swapping $c$ and $d$ (reversing the loop reverses the rotation) and scales with the area — the defining property of a 2-form. The frame index pair on $R^a{}_b$ records which rotation, the hidden 2-form indices the plane of the loop; Cartan kept the loop-plane structure implicit in the wedge and exposed only the algebra-valued part.

Is the structure equation R = dω + ω∧ω really the same as a Yang-Mills field strength?

Structurally, yes — one of the deepest facts in the subject. A Yang-Mills field strength is $F=\mathrm{d}A+A\wedge A$ with $A$ a Lie-algebra-valued connection. Cartan's $R=\mathrm{d}\omega+\omega\wedge\omega$ is identical, with the gauge group the local frame-rotation group $SO(p,q)$ and $\omega$ its potential, and $\mathrm{d}_\omega R=0$ matches $\mathrm{d}_A F=0$ . The one genuine difference is the coframe $e^a$ : gravity carries an extra field, the soldering form, tying internal frame indices to spacetime directions, with no Yang-Mills analogue — which is why gravity is a gauge theory of a very special kind.

Does the spin connection have torsion, and when does it matter?

For the Levi-Civita connection of general relativity, torsion is zero by construction: imposing $\mathrm{d}e^a+\omega^a{}_b\wedge e^b=0$ with metric compatibility fixes $\omega$ uniquely. But the formalism does not require it — allowing $T^a\neq0$ gives a more general metric-compatible connection, the setting of Einstein-Cartan theory, where torsion is sourced by the intrinsic spin of matter. Such effects are negligible at ordinary densities (they scale as spin density squared) but may matter in the very early universe or inside neutron stars; in vacuum general relativity you may always take $T^a=0$ .

How do I actually read off the connection 1-forms in practice?

Write out $\mathrm{d}e^a$ for each coframe 1-form; each nonzero term is a wedge of two basis 1-forms. Then impose both the first structure equation $\mathrm{d}e^a=-\omega^a{}_b\wedge e^b$ and antisymmetry $\omega_{ab}=-\omega_{ba}$ . For diagonal metrics this is just matching — each term of $\mathrm{d}e^a$ is cancelled by one $\omega^a{}_b\wedge e^b$ , and antisymmetry supplies the partners. For Schwarzschild it is four 1-forms and a few minutes, far less than the forty-odd Christoffel symbols.

Why does the two-sphere need no ω∧ω term but Schwarzschild does?

Dimension. On the two-sphere there are only two frame directions, so the matrix product $\omega^1{}_c\wedge\omega^c{}_2$ has no intermediate index to sum over and the quadratic term is empty — the curvature is simply $\mathrm{d}\omega^1{}_2$ . In four-dimensional Schwarzschild the angular curvature $R^{\hat\theta}{}_{\hat\phi}$ genuinely receives $\omega^{\hat\theta}{}_{\hat r}\wedge\omega^{\hat r}{}_{\hat\phi}$ through the radial direction, and dropping it gives the wrong answer. Whether $\omega\wedge\omega$ contributes is about how many directions the connection couples, not any single 1-form.