Geodesics and Curvature in General Relativity

Newtonian gravity treats space as a fixed, flat stage on which objects move under the influence of a force that acts instantaneously across distance. This picture works extraordinarily well for predicting planetary orbits, tides, and the trajectories of spacecraft, but it carries a conceptual cost that becomes visible only under scrutiny. A force that propagates instantaneously violates the finite speed limit that special relativity establishes for any causal influence. Einstein's resolution was not to patch the force law but to discard the fixed stage entirely, replacing it with a dynamical geometric object, the metric tensor, whose curvature reproduces every prediction of Newtonian gravity in the weak-field limit while also explaining phenomena Newton's theory cannot touch.

The shift from force to geometry changes what counts as an explanation. In the Newtonian picture, an orbiting planet follows a curved path because a force continuously deflects it from straight-line motion. In the geometric picture, the planet follows the straightest possible path available to it, a geodesic, and that path only looks curved because the underlying space is curved. The famous comparison is to a sphere: two sailors starting at the equator and each traveling due north along their own meridian will find themselves converging toward each other even though neither one ever changes direction. Nothing pulled them together. The curvature of the sphere did that work.

Apply the same reasoning to two nearby test particles falling freely near a planet. Standard freshman physics attributes their gradual approach to tidal forces, treated as a residual effect of gravity varying with distance. The geometric treatment replaces this with the geodesic deviation equation, which expresses the relative acceleration of the two particles directly in terms of the Riemann curvature tensor contracted with their separation vector and common velocity. This is not a different way of describing the same force. It is a demonstration that what looks like a force is entirely an artifact of curvature acting on freely falling trajectories.

The practical payoff of this reframing shows up whenever a physical law needs to be written so it holds in any coordinate system, including the rotating, accelerating, or arbitrarily warped coordinates that curved spacetime forces upon you. Ordinary partial derivatives fail at this task because they do not transform as tensors under a general change of coordinates. A law written using ordinary derivatives might hold in one coordinate patch and silently break in another, producing spurious predictions that have nothing to do with physics and everything to do with a poor choice of coordinates. The covariant derivative, built from the affine connection, is the tool engineered specifically to repair this, and its construction is one of the central threads running through the chapters ahead.

Consider a worked contrast that makes the stakes concrete. Suppose you want to describe a particle moving at constant velocity using polar coordinates instead of Cartesian ones. In Cartesian coordinates the particle's velocity components are literally constant, and the ordinary derivative of the velocity vector vanishes, correctly signaling zero acceleration. Convert to polar coordinates and that same particle's coordinate velocity components change continuously, even though nothing physical has changed, because the basis vectors themselves rotate from point to point. An ordinary derivative applied naively would report a nonzero, physically meaningless acceleration. The covariant derivative corrects for exactly this basis rotation, and the rotation rate it needs to know is precisely what the Christoffel symbols encode.

This single example generalizes to the entire subject. Every physically meaningful law in general relativity, from charge conservation to the field equations themselves, must be expressible as a tensor equation, because a tensor equation that holds in one coordinate system automatically holds in all of them^[2]. This property, sometimes called general covariance, is not a mathematical nicety layered on top of the physics. It is the design requirement that forced Einstein toward Riemannian geometry in the first place, since Riemannian geometry was already the branch of mathematics built to describe curved spaces using quantities that transform consistently under arbitrary coordinate changes.

The geometric program also resolves a puzzle that troubled Einstein directly: why does gravitational mass, the quantity that determines how strongly an object is pulled, always equal inertial mass, the quantity that determines how strongly an object resists acceleration? In Newtonian mechanics this equality is an unexplained coincidence, verified to extraordinary precision by experiments but never derived from first principles. In the geometric picture, every object with zero proper acceleration follows the same geodesic regardless of its mass or composition, because the geodesic equation contains no reference to mass at all. The equivalence of gravitational and inertial mass stops being a coincidence and becomes a structural consequence of treating gravity as curvature rather than force.

None of this machinery is specific to astrophysical extremes. A GPS satellite experiences a weaker gravitational field than a particle falling into a black hole, yet the same geometric formalism computes the small time dilation correction that keeps GPS accurate to within meters. Ignoring general relativistic corrections entirely would cause positioning errors that accumulate at roughly ten kilometers per day, since the satellite clocks run fast relative to ground clocks by about 45 microseconds daily due to weaker gravitational time dilation at orbital altitude, partially offset by special relativistic time dilation from orbital velocity. The mathematics in this book is not reserved for black holes and the early universe. It underlies a piece of infrastructure in every smartphone.

Geometry also supplies a vocabulary for distinguishing coordinate artifacts from genuine physical singularities, a distinction that trips up even careful students encountering the Schwarzschild solution for the first time. The coordinate singularity at the Schwarzschild radius looks catastrophic in the original coordinates Schwarzschild used, yet curvature invariants built from the Riemann tensor remain perfectly finite there, revealing that the apparent breakdown is an artifact of a poor coordinate choice rather than a physical pathology. Only at the true singularity at the center does a curvature invariant diverge. Telling these two situations apart requires exactly the tensor machinery this book develops, and the appendix returns to the Schwarzschild metric with worked calculations once that machinery is in place.

By the time the full arc from manifold to Einstein field equations is complete, the phrase spacetime tells matter how to move, matter tells spacetime how to curve stops being a slogan and becomes a literal summary of two coupled mathematical statements: the geodesic equation governing motion, and the Einstein field equations governing curvature. Holding both statements with full derivational confidence, rather than as inherited folklore, is the end state this book is built to produce.

This book assumes a working comfort with multivariable calculus, linear algebra, and ordinary differential equations at the level expected of an undergraduate physics or mathematics degree. It also assumes some prior exposure to special relativity, including four-vectors, Lorentz transformations, and the invariant interval, since these concepts motivate the generalization to curved spacetime without needing to be rebuilt from scratch. Given the stated background in physics and astronomy at the MSc level, these prerequisites should already feel familiar rather than new, which frees the chapters ahead to move directly into the differential-geometric content rather than spending pages re-deriving material a physics curriculum typically covers earlier.

What this book does not assume is prior formal training in differential geometry or tensor calculus as mathematical subjects in their own right. Many physics curricula introduce tensors operationally, as objects with indices that transform in a particular way under coordinate changes, without ever constructing them as multilinear maps on vector spaces or explaining why that construction matters. This book builds both pictures side by side, starting with the coordinate-based operational definition that will feel familiar and then connecting it to the coordinate-free multilinear-map definition that underlies modern differential geometry texts. Readers who already know one picture well will find the other filled in without redundant repetition.

The scope covers eleven chapters moving from foundational structures to advanced applications, organized so each chapter supplies tools the next chapter needs immediately. The following structure outlines that progression.

Deliberately out of scope is the full apparatus of quantum field theory in curved spacetime, the Hamiltonian (ADM) formulation of general relativity used in numerical relativity, and detailed cosmological model-building beyond the Friedmann-Robertson-Walker geodesics treated briefly in the appendix. These are natural next steps after this material is secure, and the conclusion points toward specific resources for pursuing them, but including them here would dilute the depth available for the core differential-geometric program. The aim is rigor and completeness within a bounded scope, not breadth across the entire field of relativistic physics.

Numerical and computational methods also fall outside this book's scope, by design rather than oversight. Modern research in general relativity frequently relies on numerical relativity codes to solve the field equations in situations, like binary black hole mergers, where no closed-form solution exists. Understanding those codes well enough to use or extend them requires first understanding exactly the analytic structures this book develops: without a secure grasp of the Riemann tensor's symmetries or the geodesic equation's derivation, a numerical relativity codebase is an opaque black box rather than a tool. This book supplies that analytic foundation and leaves numerical methods to dedicated texts.

One prerequisite deserves explicit emphasis because it is easy to underestimate: patience with index gymnastics. Tensor calculus expresses deep geometric content through expressions with multiple upper and lower indices, summation conventions, and symmetrization operations that can look intimidating on first encounter even to someone who understands the underlying geometry perfectly well. Chapter 2 builds fluency with this notation deliberately and slowly, through worked contractions and explicit index manipulations, because every later chapter depends on reading expressions like Rρσμν as naturally as one reads an ordinary algebraic formula.

A reasonable target pace is one chapter per week for a first pass, with the first four or five chapters likely moving faster given the stated background, and the chapters on curvature tensors and geodesic deviation warranting more time given their conceptual density. Re-deriving key results by hand rather than only reading the derivation is the single highest-value habit to bring to this material. Differential geometry rewards active reconstruction far more than passive recognition, and a derivation that looks clear on the page often reveals gaps in understanding only when attempted independently with the book closed.

For readers whose linear algebra is rusty specifically around dual spaces, eigenvalues of symmetric matrices, and the distinction between a vector space and its dual, a brief refresher before beginning Chapter 2 is worth the investment. The metric tensor's role in identifying a vector space with its dual, discussed in Chapter 3, depends on this background directly, and skipping that refresher tends to surface confusion later at a point where it is harder to isolate and resolve.

Each chapter builds directly on definitions and results established earlier, so working through the chapters in sequence on a first read is strongly recommended over skipping ahead to a topic of particular interest. The temptation to jump straight to the Einstein field equations or to geodesic deviation, since these are often the results that motivated picking up the book in the first place, is understandable but counterproductive. Those results are stated precisely and proven rigorously only after the connection, the covariant derivative, and the Riemann tensor have each been constructed from first principles, and reading them in isolation tends to produce the same superficial familiarity this book is designed to move beyond.

That said, the book is structured so a second pass can be nonlinear. Once the full arc from manifold to field equations has been read once, individual chapters function reasonably well as standalone references, particularly Chapter 3 on the metric tensor, Chapter 6 on the Riemann tensor, and Chapter 9 on the Ricci tensor and Einstein tensor, since these are the chapters most likely to be revisited when checking a specific calculation or recalling a specific symmetry property. The chapter on advanced topics (sectional curvature, holonomy, conformal geometry) is also written to be approachable somewhat independently, since it extends the core material in three directions that do not depend heavily on each other.

Working through the embedded examples by hand, rather than reading them as finished derivations, is where the real learning happens. A derivation of the Christoffel symbols for the Schwarzschild metric, for instance, is far more valuable attempted with pen and paper before reading the worked version in the appendix, even if the first attempt produces errors or dead ends. The appendix's worked examples on the Schwarzschild metric, Minkowski spacetime, and Friedmann-Robertson-Walker geodesics are placed at the end specifically so they can serve as checks against independent work rather than as the first encounter with those calculations.

Each chapter's subsections are also built to isolate a single derivational step, which makes it practical to pause at a subsection boundary, close the book, and attempt to reconstruct the step from memory before continuing. This is a more demanding way to read than continuous linear reading, but it mirrors how the material is actually used in research and problem-solving: not as a stream of prose to be absorbed, but as a sequence of specific techniques to be reconstructed on demand when a calculation requires them.

Footnotes throughout the book point to the primary sources behind major results, including Einstein's original field equation papers and foundational differential geometry references. These are included for readers who want to trace a result back to its origin or compare the notation used here against the notation of a classic paper, which often differs in sign conventions or index placement in ways worth knowing about before reading further in the literature. Treat these footnotes as optional deepening rather than required reading on a first pass.

The glossary at the end of the book is designed for quick lookup rather than sequential reading, and it is worth skimming once near the start simply to get a sense of which terms it covers, so that a precise definition is easy to locate quickly later without searching back through earlier chapters. Terms are defined there exactly as they are used in the main text, with the same symbol conventions, so there should be no friction translating between the glossary entry and the chapter where a term was first introduced.

A recurring structural device worth noting in advance is the callout box, used sparingly throughout the book to flag a definition, a worked example, a common error, or a key takeaway that deserves to stand apart from the surrounding paragraph. These are not asides to be skipped. They often contain the single most quotable statement of a result, condensed from a longer derivation, and are worth returning to when reviewing a chapter after finishing it.

Exercises embedded within chapters, where present, are designed to be attempted before moving to the next subsection, not collected and done in a batch at the end of a chapter. The reasoning behind this is pedagogical: an exercise placed directly after a derivation tests whether that specific derivation has actually been internalized, while a batch of exercises at a chapter's end tests a blurred mixture of everything in the chapter at once, which is a much less precise diagnostic of where understanding is solid and where it is not.

Finally, resist the urge to treat index notation as a notational inconvenience to be tolerated rather than a tool to be mastered. Fluency with Einstein summation, raised and lowered indices, and symmetrization brackets is not separate from understanding the geometry. The notation is doing real conceptual work, encoding transformation properties and symmetries directly in its structure, and a reader who learns to read ∇μTμν = 0 as a geometric statement about conservation, rather than as an opaque string of symbols, has already absorbed a meaningful fraction of what tensor calculus is for.

Treat the first read-through as building the complete scaffolding and the second read-through, chapter by chapter in reference mode, as the point where that scaffolding becomes fluent working knowledge rather than material recently seen. Both passes matter, and neither substitutes for the other.

This book uses the mostly-plus Lorentzian signature, written (-,+,+,+), for the metric tensor throughout, meaning a timelike interval has negative squared length and spacelike intervals have positive squared length. This choice is a convention, not a physical statement, and texts using the opposite (+,-,-,-) signature express identical physics with certain signs flipped in intermediate expressions. Being explicit about this choice up front matters because comparing a derivation in this book against a derivation in a paper using the opposite convention, without accounting for the signature difference, can make two correct results look contradictory when they are not.

Greek indices (μ, ν, ρ, σ and similar) range over all four spacetime coordinates, typically labeled 0 through 3 with index 0 reserved for the time coordinate. Latin indices (i, j, k) are reserved for purely spatial components, ranging over 1 through 3, and appear specifically in contexts like the appendix's Friedmann-Robertson-Walker treatment where separating time and space components clarifies the physics. This split follows standard convention in the general relativity literature and is worth internalizing early, since misreading a Latin index as ranging over all four coordinates is a common source of off-by-one errors in hand calculations.

The Einstein summation convention is used throughout without exception: whenever an index appears once as a superscript and once as a subscript in the same term, summation over that index's full range is implied, and the explicit summation symbol Σ is omitted. So AμBμ means Σμ=03 AμBμ, and this convention applies even inside more complex expressions with several repeated index pairs, such as the Ricci scalar R = gμνRμν, where both the metric and the Ricci tensor carry indices contracted according to this same rule. Chapter 2 develops this convention carefully with multiple worked examples, but it is used freely starting from Chapter 3 onward without re-explanation.

Partial derivatives are written using the shorthand ∂μ for ∂/∂xμ, and covariant derivatives are written using ∇μ, a distinction that matters enormously since the two operations coincide only for a scalar field and diverge for any tensor of rank one or higher once the connection is nonzero. An expression like ∇μVν expands explicitly as ∂μVν + ΓνμλVλ, and this expansion, derived carefully in the chapter on covariant derivatives, is referenced repeatedly enough in later chapters that it is worth committing to memory rather than looking up each time.

The Christoffel symbols themselves are written Γλμν, with the convention that they are symmetric in their two lower indices, Γλμν = Γλνμ, a property that follows from the Levi-Civita connection being torsion-free. This symmetry is proven in Chapter 4 rather than simply asserted, since the proof clarifies exactly why it fails for connections with nonzero torsion, a point that matters when the Riemann tensor's symmetries are derived in Chapter 6.

The Riemann curvature tensor is written with the index convention Rρσμν, defined through its action on a vector field via the commutator of covariant derivatives, [∇μ, ∇ν]Vρ = RρσμνVσ. Different textbooks place the four indices in different orders or define the tensor with an overall sign flip relative to this convention, and this is one of the most common sources of confusion when cross-referencing multiple sources, so the specific convention adopted here is stated explicitly again at the point of first use in Chapter 6 rather than relying on this introduction alone.

The Ricci tensor is obtained by contracting the first and third indices of the Riemann tensor, Rμν = Rρμρν, and the Ricci scalar by a further contraction with the inverse metric, R = gμνRμν. The Einstein tensor is then Gμν = Rμν - (1/2)gμνR, and the Einstein field equations, derived and discussed fully in Chapter 9, take the form Gμν + Λgμν = (8πG/c4)Tμν, where Λ is the cosmological constant, G is Newton's gravitational constant, and Tμν is the stress-energy tensor.

Units in this book follow standard practice for a rigorous treatment rather than the geometrized convention (G = c = 1) common in more terse research literature, meaning Newton's constant G and the speed of light c appear explicitly in physical formulas like the field equations above. This choice trades some notational economy for clarity about where physical constants enter a given expression, which matters when connecting an abstract geometric result back to a measurable quantity like an orbital precession rate or a light-bending angle.

Finally, a geodesic's affine parameter is written λ in purely geometric contexts and τ (proper time) specifically when the geodesic represents a massive particle's worldline, following the convention that proper time is only a valid affine parameter for timelike curves and some other parameter must be used for null geodesics representing light rays. This distinction, elaborated fully in the chapter on geodesic equations, is easy to blur in casual treatments but matters precisely at the points where this book aims for full rigor rather than physical intuition alone.