Symmetries, Lie Derivatives, and Killing Fields

Continuous symmetry turns geometric invariance into differential equations and then into conserved structure. This chapter begins with the connection-free Lie derivative, specializes it to metric-preserving flows, derives geodesic first integrals, and then separates ordinary isometries from conformal and hidden symmetries. Its final step replaces a list of generators by the geometry of group orbits, stabilizers, and fixed sets.

The chapter concerns exact differential geometry. It does not define gravitational charges, classify particular spacetimes, or infer a global group action from a merely local vector field. Those distinctions are essential whenever a formal symmetry is used in dynamics.

Required background. Tensor Fields and Natural Operations supplies tensor products, contractions, and symmetry operations; Affine Connections and Covariant Derivatives supplies the covariant-derivative framework used in Killing’s equation. Vector Fields, Flows, and Lie Brackets reviews the local flows behind the Lie derivative.

Helpful background. Calculus and Linear Algebra for Relativity supplies differentiation, linear maps, and bilinear forms; Affine and Variational Geodesics supplies the affine equation used to derive conserved quantities. The Levi-Civita Connection reviews the metric specialization, and Tensor and Geometry Bridge offers a shorter review of the notation.

Continuous symmetry as geometric structure

A vector field $X$ first supplies a local flow $\Phi_s$. Pulling a tensor back by that flow and differentiating at $s=0$ gives $\mathcal L_XT$, so preservation is expressed by $\mathcal L_XT=0$. No connection is required at this stage. For a metric and its Levi-Civita connection, the same statement becomes Killing’s equation,

$$\mathcal L_Xg_{ab}=2\nabla_{(a}X_{b)}=0.$$

The equation is local, linear, and overdetermined. Its solutions form a Lie algebra because commutators of symmetry-preserving derivatives again preserve the metric. A complete generator integrates to a global one-parameter isometry, but local solutions need not be complete. Carroll develops the flow, Lie derivative, and Killing equation in Carroll 1997, §5, pp. 135–141, eqs. (5.18)–(5.46).

Once a Killing field is known, contracting it with an affine geodesic tangent gives a constant. Conformal Killing fields retain this conclusion only for null tangents. Higher-rank Killing tensors yield polynomial constants that may exist without any corresponding flow of points. The sequence exposes a useful hierarchy: preserving the metric, preserving its conformal class, and preserving the geodesic Hamiltonian through a higher-degree integral are related but not interchangeable.

Preparation and shortest repairs

Readers should be comfortable with pointwise tensor algebra, smooth pullbacks, vector-field brackets, and the Levi-Civita derivative. Use Tensor Fields and Natural Operations for products, contractions, and symmetry operations; Smooth Maps, Pushforwards, and Pullbacks for pullbacks; and Vector Fields, Flows, and Lie Brackets for flows and brackets. If covariant derivatives are the obstacle, use Affine Connections and Covariant Derivatives for the general framework and The Levi-Civita Connection for metric compatibility. For conserved quantities along curves, Affine and Variational Geodesics supplies the affine geodesic equation.

Only elementary Lie-group language is assumed at the end. The orbit–stabilizer dimension formula follows from the rank of the evaluation map from the Lie algebra to the tangent space. Free and proper actions are stated with their hypotheses; no general course in Lie theory is presupposed.

Route through the six pages

Read Flows and Lie Derivatives first. It derives the flow definition, Cartan identity, commutator relation, and the tensorial Lie derivative of a connection. The distinction between local flow and complete flow begins there and remains active throughout the chapter.

Killing Fields and Isometry Groups specializes the construction to the metric, derives $\nabla_{(a}X_{b)}=0$, explains bracket closure, and obtains the upper bound $n(n+1)/2$ on isometry-algebra dimension. Killing Symmetries and Geodesic Constants then proves the linear first-integral theorem and checks its affine-parameter hypothesis.

Conformal and Homothetic Killing Fields compares exact, constant-scale, and position-dependent scale symmetry. Its null-versus-timelike calculation is the quickest way to see what the conformal class knows. Killing Tensors and Hidden Symmetry replaces a linear integral by a polynomial one and distinguishes reducible products from genuinely hidden tensors. The broader relationship to integrability is developed in Frolov, Krtouš, and Kubizňák 2017, §§2.1–2.5, pp. 14–31.

Finish with Symmetry Algebras, Orbits, and Fixed Sets. It explains why the same group can generate orbits of different dimensions, why a generator can vanish smoothly at a fixed point, and why a free-and-proper hypothesis is stronger than properness alone. The relevant action and quotient theorems are stated in Meinrenken 2003, §§1.1–1.4, pp. 4–11.

Reader routes and recurring checks

For geodesic motion, take the direct route through flows, Killing’s equation, linear geodesic constants, conformal symmetry, and hidden symmetry. At each step, differentiate the proposed integral along an affine geodesic. This single check distinguishes antisymmetric derivatives, trace terms, higher-rank symmetrization, and nonaffine artifacts.

For group geometry, read flows, Killing fields, and the orbit page. Check brackets directly on generators, evaluate their rank at representative points, and compute stabilizers. The $SO(2)$ action on the round sphere is the recurring model: generic orbits are circles, the poles are fixed, and the quotient is an interval with special endpoints.

For separation methods, read the geodesic and hidden-symmetry pages after the Killing equation. Conservation is only the first test. Functional independence and mutual Poisson commutation are additional hypotheses, so the existence of one quadratic integral does not establish complete integrability.

Across every route, ask four questions: Is the statement local or global? Is the curve affinely parametrized? Is a proposed higher-rank integral reducible to known linear ones and the metric? Is a quotient action free as well as proper? These questions prevent most conceptual errors without requiring coordinate-heavy calculation.

Boundaries and onward connection

The geometric scalar $X_ap^a$ is not automatically a physical charge. A charge construction needs dynamics, boundary conditions, a normalization, and often asymptotic rather than exact symmetry. Likewise, orbit degeneration does not by itself identify an axis, horizon, or singularity in a particular spacetime.

The next chapter, Frames, Tetrads, Spinors, and Clifford Geometry, uses symmetry differently: tensors are resolved into local measured components, and local Lorentz transformations act as frame changes that preserve the metric. This is an internal frame freedom, not generally an isometry moving points of the manifold.

Chapter diagnostic

In two-dimensional Minkowski space, consider the translation $P=\partial_t$, boost $B=x\partial_t+t\partial_x$, and dilation $D=t\partial_t+x\partial_x$. Determine which are Killing, compute $[B,P]$, identify the fixed point of the boost, and state what $D_au^a$ does along affine null and unit timelike geodesics.

Solution

The Cartesian metric is constant. The derivative matrices of $P$ and $B$ are antisymmetric after lowering an index with $\mathrm{diag}(-1,1)$, so $\mathcal L_Pg=\mathcal L_Bg=0$. For $D$, direct substitution gives $\mathcal L_Dg=2g$, so it is homothetic with $\psi=1$, not Killing.

Since $P$ has constant components,

$$[B,P]=-\partial_tB=-\partial_x.$$

The boost generator vanishes only where $t=x=0$, which is its fixed point in this two-dimensional example; neighboring points still transform nontrivially. Finally,

$$\frac{d}{d\lambda}(D_au^a)=g(u,u).$$

The contraction is constant on an affine null geodesic and decreases at unit rate on a future-directed unit timelike geodesic. All four conclusions follow from different parts of the same infinitesimal-symmetry structure.

References

• Carroll, Sean M. Lecture Notes on General Relativity. arXiv:gr-qc/9712019 (1997), arXiv.
• Frolov, Valeri P., Pavel Krtouš, and David Kubizňák. “Black Holes, Hidden Symmetries, and Complete Integrability.” Living Reviews in Relativity 20, 6 (2017), arXiv.
• Meinrenken, Eckhard. Group Actions on Manifolds. University of Toronto lecture notes (2003), PDF.