Riemannian geometry is the mathematics of measuring length, angle, area and curvature on smooth manifolds. Differential topology supplies the smooth manifold. A Riemannian metric adds an inner product to every tangent space, turning abstract smooth shape into measurable geometry.
Once a metric exists, we can ask for shortest paths, angles between tangent vectors, volume, curvature and how local geometric distortion controls global structure. This makes Riemannian geometry a central meeting point of analysis, topology, tensor calculus and mathematical physics.
Series route: Mathematics Learning Hub → How Mathematics Works → Riemannian Geometry. Useful foundations include Differential Topology, Differential Geometry and Tensor Analysis.
1. A Riemannian metric is a smoothly varying inner product
At each point p of a smooth manifold M, a Riemannian metric g_p assigns an inner product to tangent vectors in T_pM.
The metric varies smoothly with p and is positive definite, so nonzero tangent vectors have positive squared length.
2. Length becomes intrinsic
For a tangent vector v, its length is √g(v,v). For a smooth curve γ(t), its length is the integral of the speed induced by g.
The length is defined from the manifold’s own metric, not from an external Euclidean embedding.
3. Distance is the infimum of path lengths
The Riemannian distance between two points is the infimum of lengths over all smooth or piecewise smooth curves joining them.
This turns the manifold into a metric space while preserving its smooth structure.
4. Angles are determined by the metric
The inner product defines cosθ=g(u,v)/(|u||v|) for nonzero tangent vectors u and v.
Orthogonality is therefore a metric notion, not purely topological.
5. Coordinate components hide the invariant object
In coordinates, the metric is represented by a symmetric positive-definite matrix g_{ij}.
Under coordinate changes, these components transform tensorially so geometric lengths and angles remain unchanged.
6. Volume comes from the determinant of the metric
In local coordinates, the Riemannian volume element involves √det(g_{ij}).
This compensates for coordinate distortion and produces coordinate-independent integration.
7. The Levi-Civita connection differentiates compatibly
There is a unique torsion-free connection compatible with the metric.
This Levi-Civita connection preserves inner products under parallel transport and defines covariant derivatives of vector fields.
8. Christoffel symbols encode the connection in coordinates
The Levi-Civita connection has Christoffel symbols computed from first derivatives of the metric components.
They depend on coordinates and are not tensor components, even though they are essential in tensor equations.
9. Geodesics are locally straight according to the metric
A geodesic γ satisfies ∇_{γ′}γ′=0.
This means its velocity vector is parallel transported along itself. In Euclidean space, the equation reduces to straight-line motion.
10. Geodesics are also stationary paths of length or energy
Geodesics arise from a variational principle applied to the energy functional or, with suitable parametrisation, the length functional.
This creates a direct bridge to Calculus of Variations.
11. Locally shortest is not always globally shortest
A geodesic can minimise length locally yet cease to be globally shortest beyond a cut point.
On a sphere, many great-circle arcs are geodesics, but sufficiently long arcs are not the shortest route between their endpoints.
12. The exponential map sends tangent vectors to geodesics
Given p and tangent vector v, the exponential map exp_p(v) follows the geodesic starting at p with initial velocity v for unit parameter time when defined.
Near the origin of T_pM, exp_p provides a natural coordinate map.
13. Normal coordinates simplify the metric at one point
Using the exponential map, one can choose coordinates in which g_{ij}(p)=δ_{ij} and first derivatives of the metric vanish at p.
Curvature remains in second-order behaviour. Geometry can look Euclidean to first order without being globally flat.
14. Curvature measures failure of second-order Euclidean behaviour
The Riemann curvature tensor measures the failure of covariant derivatives to commute and the failure of parallel transport around small loops to return vectors unchanged.
Curvature is therefore an intrinsic tensorial property.
15. Sectional curvature measures curvature of two-dimensional directions
Given a two-dimensional tangent plane, sectional curvature records the Gaussian-like curvature of that infinitesimal plane.
Knowing all sectional curvatures determines the full Riemann curvature tensor.
16. Ricci curvature averages sectional curvature
The Ricci tensor is a contraction of the Riemann tensor and captures average curvature involving a chosen direction.
It controls volume distortion, geodesic focusing and appears centrally in geometric analysis and relativity.
17. Scalar curvature compresses curvature further
Contracting the Ricci tensor with the inverse metric produces scalar curvature.
It is one number at each point, useful but far from a complete description of curvature.
18. Constant-curvature spaces are model geometries
Euclidean space has zero sectional curvature. Spheres have positive constant curvature under the standard metric. Hyperbolic spaces have negative constant curvature.
These three model geometries organise many comparison theorems.
19. Geodesic triangles reveal curvature
In Euclidean geometry, triangle angles sum to π. On a positively curved sphere, geodesic triangles can have angle sum greater than π. On hyperbolic space, the sum is less than π.
Global geometric behaviour therefore reflects local curvature.
20. Jacobi fields measure variation of geodesics
A Jacobi field describes how a family of nearby geodesics separates or focuses.
The Jacobi equation contains the curvature tensor explicitly.
21. Conjugate points mark loss of local optimality
Two points along a geodesic are conjugate when a nontrivial Jacobi field vanishes at both endpoints.
Beyond conjugate points, a geodesic can fail to remain locally length-minimising.
22. Completeness asks whether geodesics can continue forever
A Riemannian manifold is geodesically complete when every geodesic extends for all real parameter values.
Metric completeness and geodesic completeness coincide for connected Riemannian manifolds by Hopf–Rinow.
23. Hopf–Rinow turns completeness into existence of shortest geodesics
Under completeness, any two points in a connected Riemannian manifold can be joined by a length-minimising geodesic.
Analytic completeness becomes a global geometric existence theorem.
24. Comparison geometry bounds one space using model spaces
Curvature bounds can constrain distances, volumes and geodesic behaviour by comparison with constant-curvature spaces.
Theorems such as Rauch, Bishop–Gromov and Toponogov translate local curvature inequalities into global consequences.
25. Positive Ricci curvature can force compactness
Bonnet–Myers shows that a complete manifold with Ricci curvature bounded below by a positive constant has bounded diameter and is compact.
Local curvature restrictions therefore constrain the entire size and topology of the manifold.
26. The Laplace–Beltrami operator generalises the Laplacian
The metric and volume form define a coordinate-independent Laplacian on functions.
Its eigenvalues and eigenfunctions connect geometry to harmonic analysis, heat flow and spectral theory.
27. Heat flow probes geometry
The heat equation on a manifold uses the Laplace–Beltrami operator.
Short-time heat-kernel behaviour contains curvature information; long-time behaviour reflects global geometry and topology.
28. Spectral geometry asks whether shape can be heard
The spectrum of the Laplacian is invariant under isometry and contains geometric information.
But different manifolds can share the same spectrum, so spectral data does not always determine geometry completely.
29. Riemannian geometry enters optimisation on manifolds
When constraints place variables on spheres, rotation groups or positive-definite matrices, Euclidean gradient descent can leave the feasible space.
Riemannian optimisation uses tangent-space gradients and retractions or exponential maps to move intrinsically along the manifold.
30. A worked mechanism: great-circle distance
On a unit sphere, the shortest route between non-antipodal points follows the shorter great-circle arc.
- Represent the points by unit vectors u and v in R³.
- The central angle θ satisfies cosθ=u·v.
- For the unit sphere, geodesic distance is θ.
- For radius R, the distance becomes Rθ.
The ambient dot product helps compute the answer, but the geodesic distance belongs intrinsically to the spherical metric.
31. Common Riemannian-geometry failure modes
- Intrinsic/extrinsic confusion: treating curvature caused by embedding as the same as intrinsic Riemann curvature.
- Geodesic=global shortest confusion: ignoring cut points and conjugate points.
- Coordinate fixation: treating metric components as the geometry itself.
- Scalar-curvature overreach: assuming one contraction captures full curvature.
- Completeness blindness: using global geodesic existence without checking hypotheses.
- Euclidean intuition: importing flat-space triangle or distance properties into curved manifolds.
32. Riemannian geometry as a mathematical machine
Smooth Manifold → Riemannian Metric → Length/Angle/Volume → Levi-Civita Connection → Geodesics → Curvature → Comparison/Spectral Analysis → Global Geometric Conclusions.
33. What mastery looks like
- treat the metric as a smooth inner-product field;
- derive intrinsic length, distance, angle and volume;
- use the Levi-Civita connection and geodesic equation;
- distinguish local geodesics from global minimisers;
- interpret Riemann, sectional, Ricci and scalar curvature at their proper information levels;
- use completeness and comparison theorems with explicit hypotheses;
- connect Laplace–Beltrami spectra to geometry;
- return coordinate calculations to invariant geometric meaning.
34. Conclusion
Riemannian geometry works by adding metric measurement to a smooth manifold. The metric creates length and angle. The Levi-Civita connection compares tangent directions at different points. Geodesics define intrinsic straightness. Curvature records how geometry departs from Euclidean behaviour. Completeness and comparison theorems convert local conditions into global results.
Differential topology gives smooth shape. Riemannian geometry gives that shape a measurable internal geometry.
How Mathematics Works | Batch 11
- How Mathematics Works | Differential Topology
- Riemannian Geometry — this article
- How Mathematics Works | Operator Theory
- How Mathematics Works | Ergodic Theory
Return to the Mathematics Learning Hub or the How Mathematics Works root.