Curvature Invariants and Spacetime Identification

Curvature invariants let a computation ask geometric questions without attaching meaning to a particular coordinate component. They can confirm that two coordinate descriptions share necessary local properties, locate invariantly distinguished regions, and organize spacetime classification. Yet no short list of scalar contractions identifies every Lorentzian geometry. A sound method combines scalar polynomial invariants with frame information, derivatives, and explicit checks of the exceptional classes where scalars are incomplete.

Required background. Connections and Covariant Derivatives provides covariant derivatives of curvature, while Curvature defines the tensors and algebraic symmetries from which invariants are built.

Helpful background. Formulations of General Relativity supplies equations restricting admissible metrics; Ordinary Differential Equations helps evaluate invariant profiles; and Spacetime Symmetries and Killing Fields relates functional dependence among invariants to local symmetry.

Scalar and Cartan invariants

A scalar polynomial invariant is a complete contraction of the Riemann tensor, its dual, and covariant derivatives. Familiar examples are

$$R, \qquad R_{ab}R^{ab}, \qquad K=R_{abcd}R^{abcd}.$$

Because every index is contracted, the resulting function is independent of the chart used to calculate it. Equality of a chosen list is necessary for local isometry, but need not be sufficient. Cartan invariants instead fix curvature and its derivatives in a frame normalized by their algebraic structure. They retain directional information that full contractions can erase.

In four-dimensional Lorentzian geometry, the generic metric is I-nondegenerate: it admits no nontrivial continuous local deformation preserving all scalar polynomial curvature invariants. The exceptional continuous degeneracy lies in the degenerate Kundt class. Coley, Hervik, and Pelavas 2009, § 2, Theorems 2.1–2.3 assumes a neighborhood where the relevant algebraic curvature types remain fixed, or real analyticity as indicated there. Discrete ambiguities are a separate issue, illustrated already in their § 1, Eqs. 1–2. Thus “characterized by invariants” has a precise local sense; it neither identifies a metric from $R$ or $K$ alone nor licenses crossing a change of algebraic type without further analysis.

Schwarzschild as a positive benchmark

Let $M$ be the physical mass, so $GM$ is its geometrized length in the inherited $c=1$ units. Construct Schwarzschild independently in the exterior Schwarzschild chart and in ingoing Eddington–Finkelstein coordinates. In both representations the Ricci invariants vanish and

$$K=\frac{48(GM)^2}{r^6}.$$

The equality tests chart transformation and full contraction. The radial dependence distinguishes finite-radius curvature from the coordinate problem at $r=2GM$: $K$ remains finite there and diverges only as $r\rightarrow0$. Compute one Riemann component directly in each chart, transform it across the overlap, and confirm the scalar by a second contraction path. This avoids a circular check in which both claimed invariants come from the same cached expression.

For Kerr, complex Weyl invariants can be combined into dimensionless coordinates and geometric diagnostics, but zeros and level sets must be interpreted with their assumptions. A concrete construction and its invariant relations are given in Abdelqader and Lake 2015, § II, Eqs. 1–7, and § III, Eqs. 11–12. The computational lesson is broader: record normalization, dual convention, parameter domain, and every denominator before turning an invariant combination into a locator.

A counterexample to scalar-only identification

Minkowski spacetime has every curvature invariant equal to zero. Certain nonflat vacuum plane-fronted waves also have all scalar polynomial curvature invariants equal to zero; they are often called VSI spacetimes. Thus an all-zero scalar list does not imply flatness. The Riemann tensor, geodesic deviation, or appropriately normalized frame components distinguish the wave from Minkowski even when every scalar contraction vanishes. The degenerate Kundt limitation and the role of derivative invariants are discussed in Coley, Hervik, and Pelavas 2009, §§ 6 and 8.

This negative benchmark is as important as Schwarzschild. Feed both Minkowski and a declared vacuum pp-wave into the same scalar pipeline. It should return zero scalar invariants for both, then refuse the conclusion “locally isometric” and request frame or Cartan data. A system that labels both metrics flat has confused an inconclusive observable with a classification theorem.

A concrete input is the plane wave

$$ds^2=-2\,du\,dv+dx^2+dy^2+A(x^2-y^2)\,du^2, \qquad A\ne0,$$

where $A$ is a real constant of dimensions inverse length squared and all four coordinates range locally over real intervals. Its determinant is $-1$ and its Lorentzian signature is unchanged for any $x,y$. Direct differentiation in the global curvature convention gives

$$R_{uxux}=-A, \qquad R_{uyuy}=A, \qquad R_{uu}=-A+A=0.$$

All other independent curvature components vanish. Raising a lower $u$ index requires a lower $v$ component because $g^{uu}=0$ and $g^{uv}=-1$; no curvature component has a $v$ index. Consequently every complete algebraic curvature contraction, including $K$, vanishes despite the nonzero tidal tensor. For this constant-profile plane wave $\nabla R=0$ as well, so derivative scalar invariants add no information. The component calculation exhibits exactly what the scalar-only classifier misses, rather than relying on an unnamed exceptional metric.

Calculation steps and independent checks

The symbolic environment record pins the algebra system, tensor and invariant packages, curvature and dual signs, dimension, coordinate charts, assumptions, simplification settings, and working precision for numerical evaluations. Inputs include the exact metric, chart domains, candidate invariant basis, maximum derivative order, and any frame-normalization choices. Store expressions before factor cancellation so singular and zero sets can be checked against excluded denominators.

A reproducible sequence constructs curvature from each metric, canonicalizes it, computes the declared scalar basis, finds algebraic syzygies, evaluates functional rank on generic points, and only then attempts frame normalization. It exports exact expressions, package versions, selected component checks, chart-transition residuals, and a classification outcome that can be “inconclusive.” For Schwarzschild, vary chart and simplification route; for numerical spot checks, increase precision and demand convergence below a stated tolerance. For a sampled numerical metric, add grid refinement and compare invariant error norms away from nonsmooth boundaries.

Principal failure modes include division by an invariant that vanishes on part of the domain, branch changes hidden by radicals, accidental evaluation on an algebraically special surface, and uncontrolled expression growth. Equality at finitely many random points is evidence against a discrepancy, not proof of an identity. Conversely, a nonzero component difference does not disprove isometry until both tensors have been compared in related frames or charts. Independent coordinate and frame calculations should agree on each scalar while retaining the information needed to diagnose a mismatch.

Check your understanding

Two vacuum metrics have $R=0$, $R_{ab}R^{ab}=0$, and $K=0$ everywhere. May a program report that they are locally isometric?

Solution

No. Those three equalities are only necessary evidence. In particular, Minkowski spacetime and a nonflat VSI pp-wave share them. The program should compute additional derivative and Cartan invariants, compare normalized frame components, or return an explicitly inconclusive result. Reporting isometry from the three scalars would fail on the degenerate Kundt class.

References

• Abdelqader, M., and Lake, K. (2015), “Invariant Characterization of the Kerr Spacetime: Locating the Horizon and Measuring the Mass and Spin of Rotating Black Holes Using Curvature Invariants,” Physical Review D 91, 084017. arXiv.
• Coley, A., Hervik, S., and Pelavas, N. (2009), “Spacetimes Characterized by Their Scalar Curvature Invariants,” Classical and Quantum Gravity 26, 025013. arXiv.

Further reading

• MacCallum, M. A. H. (2018), “Computer Algebra in Gravity Research,” Living Reviews in Relativity 21, 6. DOI.