Chapter 4 › 4.2 Cartan's Structural Equations › 4.2.6 Cartan's structural equation for surfaces
Subsection

4.2.6  Cartan's structural equation for surfaces

Now we are ready to state the main result of this chapter:

Theorem 4.14 (Cartan)

Let \(\Sigma^2\subset\mathbb{R}^3\) be a regular surface. Given an orthonormal frame \(\{e_1,e_2\}\) of tangent vectors, denote \(\{\eta^1,\eta^2\}\) to be the dual basis of \(\{e_1,e_2\}\). Then the connection 1-form \(\omega_{12}\) satisfies the equation: \[(4.4)\qquad d\omega_{12}=K\eta^1\wedge\eta^2,\] where \(K\) is the Gauss curvature.

Checkpoint
Theorem 4.14 (Cartan) states that for an orthonormal frame \(\{e_1,e_2\}\) on a regular surface, \(d\omega_{12}=K\,\eta^1\wedge\eta^2\). This theorem is significant because it expresses \(K\) in terms of what data?
WhyThe connection 1-form \(\omega_{12}\) is computed from the Christoffel symbols, which are determined entirely by the metric \(g_{ij}\) and its derivatives. Therefore \(K=d\omega_{12}/(\eta^1\wedge\eta^2)\) depends only on \(g\), not on how the surface is embedded — this is the Theorema Egregium of Gauss, recoverable from Cartan's equation.
Proof — revealed step by step

Define a differential 2-form \(\Omega_{12}\) by \(\Omega_{12}(X,Y):=\mathrm{Rm}(e_1,e_2,X,Y)\). It is a 2-form because \(\mathrm{Rm}\) is alternating in the last two slots. We aim to prove \(\Omega_{12}=-d\omega_{12}\), which for surfaces reduces to checking \(\Omega_{12}(e_1,e_2)=-d\omega_{12}(e_1,e_2)\).

We compute \(\Omega_{12}(e_1,e_2)=-\mathrm{Rm}(e_1,e_2,e_2,e_1)=-g(\nabla_{e_1}\nabla_{e_2}e_2-\nabla_{e_2}\nabla_{e_1}e_2-\nabla_{\nabla_{e_1}e_2-\nabla_{e_2}e_1}e_2,e_1)\). Since \(\{e_1,e_2\}\) are orthonormal, \(\nabla_{e_2}e_2=\omega_{12}(e_2)e_1\). Then \(\nabla_{e_1}\nabla_{e_2}e_2=D_{e_1}(\omega_{12}(e_2))e_1+\omega_{12}(e_2)\nabla_{e_1}e_1\).

Similarly, \(\nabla_{e_2}\nabla_{e_1}e_2=D_{e_2}(\omega_{12}(e_1))e_1+\omega_{12}(e_1)\nabla_{e_2}e_1\). Combining: \[\Omega_{12}(e_1,e_2)=-\bigl(D_{e_1}(\omega_{12}(e_2))-D_{e_2}(\omega_{12}(e_1))\bigr)+\omega_{12}(\nabla_{e_1}e_2-\nabla_{e_2}e_1)=-d\omega_{12}(e_1,e_2),\] where the last equality follows from (4.2). Thus \(\Omega_{12}=-d\omega_{12}\).

Now, \(K=\mathrm{Rm}(e_1,e_2,e_2,e_1)=\Omega_{12}(e_2,e_1)\), so \(K\eta^1\wedge\eta^2(e_1,e_2)=K=-\Omega_{12}(e_1,e_2)=d\omega_{12}(e_1,e_2)\). As both \(K\eta^1\wedge\eta^2\) and \(d\omega_{12}\) are 2-forms on surfaces, we have proved \(d\omega_{12}=K\eta^1\wedge\eta^2\). \(\square\)

Exercise 4.13

Complete the calculation that \(\nabla_{e_2}\nabla_{e_1}e_2=D_{e_2}(\omega_{12}(e_1))e_1+\omega_{12}(e_1)\nabla_{e_2}e_1\).

Need help?
Remark 4.15

The Cartan's structural equations can be generalized to hypersurfaces \(\Sigma^n\subset\mathbb{R}^{n+1}\). Let \(\{e_1,\ldots,e_n\}\) be an orthonormal frame. For each \(i\neq j\) and vector fields \(X,Y\), define the connection 1-form and curvature 2-form by \(\omega_{ij}(X):=g(e_i,\nabla_X e_j)\) and \(\Omega_{ij}(X,Y):=\mathrm{Rm}(e_i,e_j,X,Y)\). Then: \[-\Omega_{ij}=d\omega_{ij}+\sum_k\omega_{ik}\wedge\omega_{kj}.\] Note: Perhaps mistakes on the sign.

Checkpoint
Remark 4.15 gives the generalized Cartan structural equation \(-\Omega_{ij}=d\omega_{ij}+\sum_k\omega_{ik}\wedge\omega_{kj}\) for hypersurfaces \(\Sigma^n\subset\mathbb{R}^{n+1}\). How does this reduce to Theorem 4.14 (Cartan) for a surface \(\Sigma^2\)?
WhyFor a 2-dimensional surface, the only non-trivial pair is \((i,j)=(1,2)\). The sum \(\sum_k\omega_{1k}\wedge\omega_{k2}\) runs over \(k=1,2\); both terms vanish since \(\omega_{11}=\omega_{22}=0\) (antisymmetry of connection forms for an orthonormal frame). Thus the structural equation becomes \(-\Omega_{12}=d\omega_{12}\), and since \(\Omega_{12}(X,Y)=\mathrm{Rm}(e_1,e_2,X,Y)=K\,\eta^1\wedge\eta^2(X,Y)\), we recover Theorem 4.14 (Cartan).
Example 4.16

Recall from Example 4.10 that the connection 1-form on the unit sphere is \(\omega_{12}=\cos\varphi\,d\theta\), and so \(d\omega_{12}=-\sin\varphi\,d\varphi\wedge d\theta\). As \(e_1=\frac{1}{\sin\varphi}\frac{\partial F}{\partial\theta}\) and \(e_2=\frac{\partial F}{\partial\varphi}\), we take the dual basis \(\eta^1:=\sin\varphi\,d\theta\) and \(\eta^2=d\varphi\). It is easy to check \(\eta^i(e_j)=\delta_{ij}\). Then \[d\omega_{12}=-\eta^2\wedge\eta^1=\eta^1\wedge\eta^2.\] From (4.4), we can see that \(K=1\).

Checkpoint
In Example 4.16, using the result \(\omega_{12}=\cos\varphi\,d\theta\) from Example 4.10, one computes \(d\omega_{12}=\eta^1\wedge\eta^2\) and concludes \(K=1\). What conceptual point does this calculation illustrate?
WhyThe computation in Example 4.16 derives \(K=1\) for the unit sphere entirely from \(\omega_{12}=\cos\varphi\,d\theta\), which is built from the metric (Christoffel symbols). No reference to the second fundamental form is needed, illustrating that Gauss curvature is intrinsic — the core message of the Theorema Egregium.