Weingarten Formula

How can we describe the shape of a curved surface without stepping outside of it? This fundamental question lies at the heart of differential geometry, a field that provides the mathematical language for understanding curvature. While our intuition about shape is often based on an external, three-dimensional perspective, the true genius of geometry lies in uncovering properties that are intrinsic to the surface itself. This article tackles this challenge by exploring the Weingarten formula, a powerful tool that, along with its counterpart the Gauss formula, builds a bridge between the extrinsic and intrinsic views of a surface.

The following chapters will guide you through this fascinating landscape. In "Principles and Mechanisms," we will dissect the Weingarten formula, introducing the shape operator and explaining how it quantifies the way a surface bends in space. We will see how this leads to fundamental concepts like principal, mean, and Gaussian curvature, culminating in Gauss's famous Theorema Egregium. Subsequently, in "Applications and Interdisciplinary Connections," we will witness these abstract principles in action, exploring how the Weingarten formula governs the shape of soap bubbles, guides engineering design, models the dynamic evolution of surfaces, and even helps determine the stability of objects in the curved spacetime of general relativity.

Imagine you are a tiny, two-dimensional creature living on the surface of a sphere. To you, your world seems vast and, on a small scale, flat. But if you were to embark on a long journey, or attempt to draw a very large triangle, you would begin to notice strange things. The angles of your triangle wouldn't add up to $180$ degrees. Straight lines, followed diligently, would somehow bring you back to where you started. You would discover that your world is curved. But how could you, a creature who has never seen a third dimension, describe this curvature? This is the central question that the beautiful machinery of differential geometry sets out to answer.

The secret lies in a brilliant strategy: understanding the surface by studying how it sits within the larger, three-dimensional space we inhabit. We, as three-dimensional observers, have a privileged view. We can see the sphere curving "outward". The genius of mathematicians like Carl Friedrich Gauss was to build a bridge between our extrinsic, higher-dimensional viewpoint and the intrinsic viewpoint of our tiny, surface-bound creature. The Gauss and Weingarten formulas are the twin pillars of this bridge.

How do we measure bending? In basic calculus, curvature is related to the second derivative. A straight line has a zero second derivative; a tight circle has a large one. Let's apply this idea to a surface. Imagine our surface is described by a position vector $X(u,v)$ that depends on two coordinates, $u$ and $v$. The first derivatives, $X_u = \frac{\partial X}{\partial u}$ and $X_v = \frac{\partial X}{\partial v}$, are tangent vectors that define the "floor" of our creature's world—the tangent plane.

To see how this floor curves, we must take another derivative, giving us vectors like $X_{uu} = \frac{\partial^2 X}{\partial u^2}$. This vector represents the "acceleration" of the surface. If the surface were a flat plane, this acceleration would be zero. On a curved surface, it's not. But where does this acceleration vector point? It might point in any direction in the ambient 3D space. The key insight is to break it down into two parts: a component that lies in the tangent plane and a component that points perpendicularly out of it, along the normal vector $N$.

This decomposition gives us the famous ​​Gauss formula​​:

$X_{ij} = \Gamma^k_{ij} X_k + h_{ij} N$

Let's unpack this. The equation says that the acceleration of the surface ($X_{ij}$) is the sum of two distinct effects. The term $h_{ij} N$ is the normal component. The scalar quantity $h_{ij}$ is a component of the ​​second fundamental form​​, and it measures precisely how much the surface is lifting off from its own tangent plane. This is the essence of ​​extrinsic curvature​​—the bending we can see from the outside.

The other term, $\Gamma^k_{ij} X_k$, is the tangential part of the acceleration. You might think this part also depends on the ambient space. But here comes the first surprise. The coefficients $\Gamma^k_{ij}$ are the ​​Christoffel symbols​​, and they are purely ​​intrinsic​​. Our two-dimensional creature, simply by measuring distances and angles on the surface (which is encoded in the metric, or first fundamental form), can calculate these symbols without any knowledge of the third dimension! They govern the rules of "straight-line" travel (geodesics) within the surface itself. So, the Gauss formula beautifully separates the change in the tangent vectors into an intrinsic part (how to stay on the surface) and an extrinsic part (how the surface bends away from its tangent plane).

Now let's turn our attention to the other main character in this story: the ​​unit normal vector​​ $N$, which points "up," perpendicular to the surface. As we move from point to point, the tangent plane tilts, and so does the normal vector. How can we describe the rate at which $N$ changes? Let's take its derivative, $N_i = \frac{\partial N}{\partial u^i}$.

Where does this new vector $N_i$ point? Here lies a wonderfully simple and profound piece of geometry. Since $N$ is a unit vector, its length is always 1, so $\langle N, N \rangle = 1$. If we differentiate this relationship, the product rule gives us:

$\langle N_i, N \rangle + \langle N, N_i \rangle = 2 \langle N_i, N \rangle = 0$

This means that the vector $N_i$ must be orthogonal to $N$. But the set of all vectors orthogonal to $N$ is none other than the tangent plane itself! So, the change in the normal vector is always a vector in the tangent plane. This is a spectacular constraint. The normal vector can't just flail about; its change is tied directly back to the surface itself.

This change is captured by the ​​Weingarten map​​, more commonly known as the ​​shape operator​​, $S$. It is a machine that takes a tangent vector $X_i$ (the direction you're moving in) and tells you how the normal vector is changing, $-N_i$. This relationship is the ​​Weingarten formula​​:

$\bar{\nabla}_X \nu = -S X$

Here we use the more general notation where $\bar{\nabla}_X \nu$ is the covariant derivative of the normal vector $\nu$ in the direction of the tangent vector $X$. This equation tells us that the rate of change of the normal is given by applying this new operator, $S$, to our direction vector. The minus sign is a historical convention, but it's the standard one. In more general situations, where the surface might be embedded in a higher-dimensional space, the derivative of the normal vector can also have a component that remains in the normal space. The Weingarten formula then elegantly isolates the part that gets projected back into the tangent space, which is governed by the shape operator, from the part that stays normal, which is governed by a "normal connection".

The shape operator $S$ is the star of the show. At each point on the surface, it's a linear transformation of the tangent plane to itself. By studying its properties, we unlock the secrets of curvature.

Like any linear operator, we can ask about its eigenvalues and eigenvectors. These are the ​​principal curvatures​​ and ​​principal directions​​. The principal directions are the directions on the surface where the bending is most extreme (either maximal or minimal). For instance, on a saddle point, one principal direction curves down, and the other curves up. The corresponding eigenvalues tell you the amount of curvature in those directions.

Now, one might ask if these special directions are always at a right angle to each other. The answer is yes, and the reason is a deep and beautiful property of the shape operator: it is ​​self-adjoint​​ with respect to the surface metric. This means that for any two tangent vectors $X$ and $Y$, the relation $\langle S(X), Y \rangle = \langle X, S(Y) \rangle$ holds. This property is not an accident; it's a direct consequence of the symmetry of second partial derivatives in the ambient space. This symmetry ensures that the principal directions are always orthogonal, giving us a perfect, natural grid at every point to measure curvature.

From the principal curvatures, we can compute the two most important measures of surface curvature. Their average gives the ​​mean curvature​​, $H$, which is important in the study of soap films and minimal surfaces. Their product gives the ​​Gaussian curvature​​, $K$.

To get a better feel for what is truly fundamental, consider a simple experiment. What happens if we reverse our definition of "up"? That is, we replace our normal vector $\nu$ with its opposite, $-\nu$. If we re-derive our formulas, we find that both the second fundamental form $h$ and the shape operator $S$ flip their signs. Consequently, the mean curvature $H$ (the sum of the eigenvalues of $S$) also flips its sign. However, the Gaussian curvature $K$ (the product of the eigenvalues of $S$) remains unchanged, as does any curvature term coming from the Gauss equation, because they involve products of two $h$'s or two $S$'s. This is a powerful clue: the Gaussian curvature seems to be a more robust, fundamental property of the surface, independent of our arbitrary choices.

We have two pictures of our surface: the Gauss formula, describing how tangent vectors change, and the Weingarten formula, describing how the normal vector changes. They both originate from the same process—taking derivatives in the ambient space—so they must be compatible. This compatibility condition is not just a consistency check; it is the source of one of the deepest results in all of geometry.

The condition boils down to the simple fact that for a smooth surface, mixed partial derivatives must be equal. For example, differentiating the surface twice with respect to $u$ and then once to $v$ must yield the same result as differentiating once to $u$, then $v$, then $u$ again. When we write this condition out using the Gauss and Weingarten formulas, the single identity of commuting derivatives in the ambient space splits into two separate equations on the surface.

One is the ​​Codazzi-Mainardi equation​​, which relates the change in the shape operator to the curvature of the ambient space. For a surface in flat Euclidean space $\mathbb{R}^3$, where the ambient curvature is zero, it takes a particularly simple form: $(\nabla_X S)(Y) = (\nabla_Y S)(X)$. It expresses a beautiful symmetry in how the shape operator varies across the surface.

The other, and more famous, equation is the ​​Gauss equation​​. It relates the intrinsic curvature of the surface (described by its Riemann curvature tensor $R$) to the extrinsic shape operator $S$:

$R(X,Y)Z = \langle SY,Z \rangle SX - \langle SX,Z \rangle SY$

Look at this equation carefully. It is a purely algebraic relationship. All the messy terms involving the derivatives of the shape operator have vanished—they have been neatly collected and isolated into the separate Codazzi equation!. This "separation of concerns" is the mathematical miracle at the heart of the matter.

From this, Gauss's ​​Theorema Egregium​​ (Latin for "Remarkable Theorem") follows. The Gaussian curvature $K$, which we can get from the intrinsic curvature tensor $R$, is given by $K = \det(S)$. Even though $S$ is defined extrinsically, its determinant must be an intrinsic quantity, calculable by our two-dimensional creature who knows nothing of the third dimension. This is why you cannot wrap a sphere with a piece of paper without tearing or crumpling it. The paper is intrinsically flat ($K=0$), while the sphere has constant positive curvature ($K > 0$). No amount of bending in 3D space can change the intrinsic geometry. The Theorema Egregium proves that the Gaussian curvature is a true property of the surface, as real to its inhabitants as the laws of physics.

So far, we have mostly imagined our surface living in a "flat" Euclidean space. What if the ambient space itself is curved, as in Einstein's theory of General Relativity? The entire framework still holds, but the equations pick up new terms that reflect the geometry of the surrounding universe.

The Codazzi and Gauss equations are no longer zero on their right-hand sides. Instead, they are equal to components of the ambient space's curvature tensor, $\bar{R}$. Most tellingly, the Gauss equation becomes:

$K = \det(S) + \bar{K}$

where $\bar{K}$ is the sectional curvature of the ambient space on the surface's tangent plane. This equation is breathtakingly beautiful. It says that the curvature you feel intrinsically ($K$) is the sum of two effects: how you are bent extrinsically within your universe ($\det S$), and the curvature of the universe itself ($\bar{K}$). The geometry of the part is inextricably linked to the geometry of the whole. From a simple question about describing a surface, the Weingarten formula and its cousins have led us to a profound statement about the very fabric of space.

In our previous discussion, we became acquainted with the formal machinery of extrinsic geometry, culminating in the Gauss and Weingarten formulas. These equations might seem, at first glance, to be abstract mathematical statements. But to think of them that way is like looking at the score of a grand symphony and seeing only notes on a page. The truth is that the Weingarten formula is a key—a Rosetta Stone—that unlocks a universe of phenomena, from the shape of a soap bubble to the evolution of galaxies. It tells us how the notion of "up" changes as we move along a curved surface, and in that simple-sounding idea lies the secret to shape, stability, and change. Let us now embark on a journey to see how this one mathematical idea blossoms into a rich tapestry of applications across science and engineering.

The most immediate and fundamental application of the Weingarten formula is in the very definition and calculation of a surface's shape. The formula introduces the shape operator, or Weingarten map, which you can think of as the surface's "operator's manual for curvature." At any point, this operator tells you how the surface bends in every possible direction. Its eigenvalues, the principal curvatures $k_1$ and $k_2$, represent the maximum and minimum bending.

From these two numbers, we can distill the two most important local descriptions of curvature. One is the ​​Gaussian curvature​​, $K = k_1 k_2$, and the other is the ​​mean curvature​​, $H = \frac{1}{2}(k_1+k_2)$. The Gaussian curvature is an intrinsic property, a fact so profound it was dubbed the Theorema Egregium or "Remarkable Theorem" by Gauss himself. It tells us about the geometry that a two-dimensional inhabitant living on the surface would measure, without any knowledge of the outside world. A flat sheet, a cylinder, and a cone all have zero Gaussian curvature, which is why you can roll a sheet of paper into a cylinder without any stretching or tearing. The mean curvature, on the other hand, is purely extrinsic. It measures how the surface curves within the ambient space. A soap film, trying to minimize its area, will adjust its shape until its mean curvature is zero everywhere.

There is no better place to see these ideas in action than with the most perfect of shapes: the sphere. If you apply the Weingarten formula to a sphere of radius $R$, you find something remarkably simple. The shape operator at every point is just a scalar multiplication: it bends equally in all directions. The principal curvatures are everywhere equal, $k_1 = k_2 = 1/R$ (or $-1/R$ depending on which way your normal vector points). From this, the curvatures are immediate: the Gaussian curvature is a constant $K=1/R^2$, and the mean curvature is a constant $H=1/R$. This constant, positive Gaussian curvature is the sphere's geometric signature. It is why you cannot wrap a sphere with a piece of paper without wrinkling it, and it is why a sphere is such a strong structure, distributing stress evenly across its surface. This principle is at the heart of everything from the design of pressure vessels and submarine hulls to the focusing of light by a simple lens.

While the sphere is a perfect ideal, the world is filled with more complex shapes. How do we analyze the curvature of a trumpet bell, a vase, or an airplane fuselage? Many such objects are ​​surfaces of revolution​​, generated by rotating a 2D profile curve around an axis. The Weingarten formula provides a direct and powerful method to compute the curvature of such an object at any point, based solely on the properties of its generating curve. An engineer can design a 2D profile and, using this mathematical toolkit, know precisely the stresses and strains the final 3D object will experience.

Beyond analyzing given shapes, we can turn the question around: can we design surfaces that obey a specific "law of curvature"? This leads us to the fascinating world of ​​Weingarten surfaces​​ (or W-surfaces), which are defined by having a fixed relationship between their mean and Gaussian curvatures, such as $aH + bK = c$ for some constants $a, b, c$. A surface with constant mean curvature ($H = \text{constant}$) is a simple but profoundly important example. These are the shapes taken by soap bubbles, which enclose a volume with the minimum possible surface area. They appear as the interface between two immiscible fluids, in the study of capillarity, and even in models of cell membranes. The Weingarten formula is the essential tool for studying these surfaces and understanding the physical laws that give rise to their beautiful and efficient forms.

So far, we have viewed surfaces as static objects. But what if they change over time? What if a shape evolves? This is where the Weingarten formula reveals its deepest power, connecting geometry to dynamics. Consider the process of a rough crystal smoothing its sharp corners as it is heated, or a soap film collapsing into a point. Many such phenomena are modeled by ​​Mean Curvature Flow (MCF)​​, a process where every point on a surface moves in the normal direction with a speed equal to its mean curvature.

Think about what this means. Regions with high mean curvature—sharp points and edges—move fastest, while flatter regions move slowly. The result is a flow that relentlessly seeks to reduce surface area, smoothing out irregularities in the most efficient way possible. This geometric evolution equation has found stunning applications far beyond physics. In medical imaging and computer graphics, MCF is used as a sophisticated algorithm to denoise images and reconstruct 3D shapes from scattered data, effectively "ironing out" the noise while preserving significant features.

The Weingarten formalism gives us the equations of motion for this dance of geometry. With it, we can ask extraordinarily detailed questions. For instance, how does the intrinsic geometry—the very ruler used to measure distances on the surface—change during the flow? The answer is a beautiful equation: $\partial_t g_{ij} = -2H h_{ij}$. The rate of change of the metric ($g_{ij}$) at a point is directly proportional to the second fundamental form ($h_{ij}$). The fabric of the surface stretches or shrinks in direct response to how it is bent in space.

We can go even deeper. How does the curvature itself evolve? One of the pinnacle results of the theory, derived from the fundamental formulas, shows that the mean curvature $H$ obeys an evolution equation of its own: $\partial_t H = \Delta H + |S|^2 H$. This is a "reaction-diffusion" equation. The $\Delta H$ term, the Laplacian, tells us that curvature tends to spread out and average itself, like heat diffusing through a metal plate. The $|S|^2 H$ term, however, is a "reaction" term where curvature amplifies itself. It is the battle between these two effects that determines the fate of the surface—whether it smooths out into a perfect sphere or develops a "singularity" and pinches off, like a droplet breaking from a faucet.

Finally, we arrive at one of the most profound connections, linking the geometry of surfaces to the calculus of variations and the stability of physical systems. Minimal surfaces, like soap films, are defined by having zero mean curvature. They are solutions to the problem of finding the surface of least area spanning a given boundary, a direct echo of the Principle of Least Action that governs so much of physics.

But finding a solution is one thing; knowing if it is stable is another. A pencil balanced on its tip is in equilibrium, but it is not stable. Likewise, a soap film might be a minimal surface, but will it survive a small poke, or will it collapse? The Weingarten formula is the key to the answer. The analysis, known as the second variation of area, yields a "stability operator" that determines the surface's fate. The terms in this operator tell a fantastic story about the interplay of forces.

One dominant term is the squared norm of the shape operator, $|S|^2$. This term always acts to stabilize the surface. It can be thought of as a kind of geometric "stiffness" arising from the surface's own curvature. Its origin lies in the subtle way the surface's tangent vectors must twist and turn as the surface is deformed—a motion described by the Weingarten map.

Even more astonishing is the appearance of another term: $\mathrm{Ric}_M(\nu, \nu)$, the Ricci curvature of the ambient space in the normal direction. This means that the stability of a minimal surface depends not only on its own shape, but on the curvature of the universe it inhabits! A soap film in our nearly-flat room behaves differently than one near a black hole, where spacetime is strongly curved. The study of minimal surfaces in general relativity is an active area of research, where these geometric stability questions have direct consequences for our understanding of black hole physics and the large-scale structure of the cosmos.

From the simple turning of a normal vector, we have journeyed to the stability of objects in a curved universe. The Weingarten formula, initially a tool for describing shape, has become a dynamic principle that predicts evolution and determines stability. It is a testament to the remarkable unity of mathematics and physics, where a single, elegant idea can provide the language to describe the world on scales from a soap film to the cosmos itself.