Manifold - Geometric Structures

Additional Structures on Manifolds Introduction A manifold is a topological space that locally looks like Euclidean space, but manifolds by themselves have very limited structure. To do meaningful mathematics on manifolds—whether studying physics, geometry, or analysis—we need to add extra layers of organization. This chapter focuses on several important structures that can be placed on manifolds, with special emphasis on Riemannian manifolds, which are perhaps the most commonly encountered in differential geometry and its applications. Differentiable Manifolds Before discussing Riemannian manifolds, we need to understand differentiable manifolds, which form the foundation for everything that follows. A differentiable manifold (or smooth manifold) is a manifold equipped with an atlas—a collection of coordinate charts—whose transition maps are differentiable (infinitely smooth). This is what allows us to do calculus on the manifold: we can talk about derivatives, smooth functions, and vector fields in a well-defined way that doesn't depend on which coordinate system we choose. The key insight is that differentiability is a property we can verify locally using coordinates, and the smoothness of transition maps ensures this property is globally well-defined. Other Important Structures on Manifolds Beyond differentiable manifolds, there are several other significant structures worth briefly understanding. Riemannian Manifolds carry a Riemannian metric—a smoothly varying inner product on tangent spaces. This is our main focus and will be explored thoroughly in the next section. <extrainfo> Symplectic Manifolds are equipped with a closed, non-degenerate 2-form (a type of differential form). These are the natural setting for Hamiltonian mechanics and classical mechanics in general, providing the mathematical formalism for phase spaces where position and momentum coordinates are treated symmetrically. Lorentzian Manifolds carry a metric of signature $(-,+,+,\ldots)$ (one negative, rest positive), rather than the all-positive signature of a Riemannian metric. These are the mathematical model for spacetime in general relativity, where the negative signature direction represents time. Complex Manifolds have transition maps that are holomorphic (complex-differentiable) functions rather than just real-differentiable. This gives them a locally complex-analytic structure, placing them at the intersection of differential geometry and complex analysis. </extrainfo> Riemannian Manifolds: A Comprehensive Guide Riemannian manifolds are the geometric spaces where we can measure distances and angles. They are central to differential geometry and appear throughout physics, engineering, and pure mathematics. What is a Riemannian Metric? A Riemannian manifold is a differentiable manifold $M$ equipped with a Riemannian metric—an inner product $\langle \cdot, \cdot \ranglep$ defined on the tangent space $TpM$ at each point $p \in M$, and these inner products vary smoothly as we move across the manifold. Why this matters: Without a Riemannian metric, a manifold has no notion of distance or angle. Once we add a metric, we can ask questions like: "What is the shortest path between two points?" or "What angle do two curves make at their intersection?" The Inner Product on Tangent Spaces At each point $p$ on the manifold, the tangent space $TpM$ is a real vector space. The Riemannian metric provides an inner product on this space, which means it's a bilinear, symmetric, positive-definite map. For tangent vectors $u, v \in TpM$: Bilinearity: The map is linear in each argument Symmetry: $\langle u, v \ranglep = \langle v, u \ranglep$ Positive-definiteness: $\langle v, v \ranglep \geq 0$, with equality only when $v = 0$ This inner product allows us to define: Length of a tangent vector: $\|v\|p = \sqrt{\langle v, v \ranglep}$ Angle between vectors: $\cos \theta = \frac{\langle u, v \ranglep}{\|u\|p \|v\|p}$ Perpendicularity: Vectors are orthogonal when $\langle u, v \ranglep = 0$ Local Representation In local coordinates $(x^1, \ldots, x^n)$ around a point, the Riemannian metric is written as: $$g = g{ij}(x) \, dx^i \otimes dx^j$$ where $g{ij}(x)$ are smooth functions. This is the metric tensor. For the coordinate basis vectors $\frac{\partial}{\partial x^i}$ and $\frac{\partial}{\partial x^j}$ of the tangent space, we have: $$\left\langle \frac{\partial}{\partial x^i}, \frac{\partial}{\partial x^j} \right\rangle = g{ij}(x)$$ The components $g{ij}$ form a symmetric, positive-definite matrix at each point. Important: Different coordinate systems will represent the metric with different components $g{ij}$, but the geometric object—the inner product itself—remains the same. This is why Riemannian geometry is an intrinsic geometry. Distances and Curve Length One of the most powerful applications of a Riemannian metric is measuring distances. Length of a Curve Given a smooth curve $\gamma: [a,b] \to M$, the length of $\gamma$ is defined as: $$L(\gamma) = \inta^b \sqrt{\langle \dot{\gamma}(t), \dot{\gamma}(t) \rangle{\gamma(t)}} \, dt$$ Here, $\dot{\gamma}(t)$ is the tangent vector to the curve at time $t$. The integrand is the norm of this tangent vector, measuring how "fast" the curve is moving at each instant, and we integrate this speed over the entire curve. Example: On the Euclidean plane with the standard metric, if $\gamma(t) = (t, t^2)$ for $t \in [0,1]$, then $\dot{\gamma}(t) = (1, 2t)$, and the length is: $$L(\gamma) = \int0^1 \sqrt{1 + 4t^2} \, dt$$ Distance Between Points The distance $d(p,q)$ between two points $p$ and $q$ on a Riemannian manifold is defined as the infimum (greatest lower bound) of the lengths of all smooth curves joining them: $$d(p, q) = \inf \{ L(\gamma) : \gamma \text{ is a smooth curve from } p \text{ to } q \}$$ This definition captures the intuitive notion: the distance is determined by the "shortest path" between the points. On a Euclidean space, this recovers the usual straight-line distance. On a curved surface, the shortest path is generally not a straight line but a geodesic (discussed below). Curvature and Geometric Quantities Understanding how a Riemannian manifold curves is central to differential geometry. The key tool is the curvature tensor, from which we derive several curvature measures. The Levi-Civita Connection Before we can measure curvature, we need a notion of how vectors change as we move along curves. This is provided by the Levi-Civita connection, which is the unique connection on a Riemannian manifold that satisfies two properties: Torsion-free: It has no "twisting" component Compatible with the metric: Parallel transport preserves inner products The Levi-Civita connection defines how to differentiate vector fields along curves on the manifold, respecting the geometric structure. The Riemann Curvature Tensor The most complete measure of curvature is the Riemann curvature tensor, denoted $R$. Intuitively, it measures how parallel transport fails to return vectors to themselves when we move around an infinitesimal closed loop on the manifold. Specifically, if you take a vector field, parallel transport it around a small loop, and return to the starting point, the Riemann tensor tells you how much the vector has changed. On a flat space (like the Euclidean plane), the Riemann tensor is zero everywhere because parallel transport is perfect. Tricky point: The Riemann tensor is quite complex—it's a 4-index object with many components. In coordinates, it's written as $R^l{ijk}$ or $R{ijkl}$, and computing it directly is generally involved. But we don't often need all these components. Sectional Curvature Rather than dealing with the full Riemann tensor, we often use sectional curvature, which is simpler and more geometric. For any 2-dimensional subspace (a "2-plane") in the tangent space at a point, the sectional curvature measures how much that 2-dimensional slice of the manifold curves. The sectional curvature can be positive (like a sphere), negative (like a hyperbolic space), or zero (like Euclidean space). On surfaces, the sectional curvature is simply the Gaussian curvature you may already be familiar with. Ricci Curvature and Scalar Curvature From the Riemann tensor, we can derive simpler curvature measures by summing over certain indices: Ricci curvature: Obtained by contracting the Riemann tensor, giving a 2-index object (a symmetric tensor). It roughly measures how much volume contracts or expands when moving along geodesics. Scalar curvature: Obtained by further contraction of the Ricci tensor, giving a single number at each point. This is the most "coarse" curvature measure, but often the most important for global questions. In general relativity, the Ricci tensor and scalar curvature appear in Einstein's field equations, showing why understanding curvature is essential for physics. Geodesics A geodesic is a curve that locally minimizes distance on a Riemannian manifold—it's the analog of a straight line in Euclidean space. More precisely, geodesics are curves $\gamma(t)$ that satisfy the geodesic equation: $$\frac{D\dot{\gamma}}{dt} = 0$$ where $\frac{D}{dt}$ denotes the covariant derivative defined by the Levi-Civita connection. This equation says that the tangent vector to the curve doesn't change when we move along the curve (in the sense of parallel transport). In coordinates, this becomes a second-order differential equation. Geometric intuition: Imagine walking on a surface. If you move in such a way that your direction never changes relative to the surface itself (as opposed to relative to the ambient space), you're walking along a geodesic. For instance, great circles on a sphere are geodesics—they're the "straightest" paths you can follow while staying on the sphere. Key property: Geodesics locally minimize distance. If you have two nearby points on a Riemannian manifold, the shortest curve joining them is a geodesic. However, globally, there may be multiple geodesics between two points, or geodesics may not be globally minimizing (this depends on the curvature of the manifold). <extrainfo> The Gauss-Bonnet Theorem One of the most beautiful results in differential geometry is the Gauss-Bonnet theorem for surfaces. It states that for a compact, oriented surface $S$: $$\intS K \, dA = 2\pi \chi(S)$$ where: $K$ is the Gaussian curvature $dA$ is the area element $\chi(S)$ is the Euler characteristic of $S$ This theorem shows a deep connection between local geometry (the curvature $K$) and global topology (the Euler characteristic). The integral of curvature over the entire surface is always an integer multiple of $2\pi$, determined solely by the topological type of the surface—a remarkable rigidity! </extrainfo> The Existence of Riemannian Metrics A fundamental question: Does every smooth manifold admit a Riemannian metric? Answer: Yes. Every smooth manifold admits at least one Riemannian metric. Why This Matters This might seem like a small detail, but it's profound. It means we can always equip any smooth manifold with a notion of distance and curvature. There's no topological obstruction—the metric structure and smooth structure are compatible. The proof uses partitions of unity, a powerful tool in differential geometry that allows us to patch together locally defined objects into global ones. Metrics on Submanifolds Many important Riemannian manifolds arise as submanifolds of Euclidean space. When a submanifold $S$ sits inside $\mathbb{R}^n$ (or any Euclidean space), it inherits a Riemannian metric by restricting the ambient inner product. Specifically, if $p \in S$ and $u, v \in Tp S$ (tangent vectors to $S$), we define: $$\langle u, v \rangleS = \langle u, v \rangle{\mathbb{R}^n}$$ where the right-hand side is the standard Euclidean inner product. Since $Tp S$ is a subspace of $Tp \mathbb{R}^n$, this makes sense. Example: A sphere $S^n$ in $\mathbb{R}^{n+1}$ inherits a metric where the distance between two points on the sphere is the arc length of the great circle joining them. This induces positive sectional curvature (the sphere is "curved outward"). In local coordinates, we can express this induced metric by pulling back the Euclidean metric through the inclusion map. Metric Choices and Geometry A crucial observation: different Riemannian metrics on the same manifold can produce drastically different geometric properties. Two metrics on the same smooth manifold might assign different distances between points, identify different curves as geodesics, or measure curvature differently. Yet the underlying smooth structure—the notion of differentiability—remains unchanged. This flexibility means that the same manifold as a topological/smooth object can be geometrized in multiple ways, leading to entirely different geometric theories. <extrainfo> Finsler Manifolds: A Generalization While Riemannian manifolds are the most common setting, there's a broader class called Finsler manifolds that are worth knowing about. A Finsler manifold equips each tangent space $TpM$ with a smoothly varying norm $Fp: TpM \to [0, \infty)$, without requiring an inner product structure. The crucial difference: A norm satisfies positive-definiteness and scaling properties, but it's more general than an inner product. In particular, a Finsler metric doesn't necessarily define angles between vectors in the usual way. However, Finsler metrics still allow us to define lengths of curves: $$L(\gamma) = \inta^b F{\gamma(t)}(\dot{\gamma}(t)) \, dt$$ and hence distances. So we retain the ability to study geodesics and metrical properties. Finsler geometry is useful in contexts where the geometry might not be "quadratic" (as Riemannian geometry is), or where the geometry might depend on the direction of motion. However, Riemannian manifolds remain the primary focus of most differential geometry courses and applications. </extrainfo> Summary A Riemannian manifold is a smooth manifold with an inner product on each tangent space, varying smoothly. This structure allows us to: Measure distances and angles through the metric tensor Define geodesics as curves that locally minimize distance Study curvature using the Riemann tensor and derived quantities Connect local geometry to global topology (e.g., through Gauss-Bonnet) Every smooth manifold admits a Riemannian metric, and many important Riemannian manifolds arise as submanifolds of Euclidean space. The choice of metric is flexible—different metrics on the same manifold produce different geometries.