Petrov and Segre Classification

Petrov type captures the multiplicity structure of the Weyl tensor’s principal null directions in four-dimensional Lorentzian geometry; Segre type captures the eigenvalue and Jordan-block structure of the mixed Ricci operator $R^\mu{}_\nu$. Together they describe important algebraic information at a point, but neither classification determines a spacetime up to local isometry.

Required background. The Riemann Curvature Tensor supplies the Weyl–Ricci decomposition; Killing Fields and Isometry Groups supplies independent differential symmetry data; The Einstein Equation translates Ricci type into algebraic information about stress–energy.

Helpful background. Frames, Coframes, and Anholonomy supplies the null and orthonormal frames in which canonical algebraic forms are found.

Weyl curvature and Petrov type

Assume a real, oriented, four-dimensional Lorentzian vector space at a point. Complexifying the two-forms and restricting the Weyl operator to self-dual bivectors turns it into a trace-free $3\times3$ complex operator. Equivalently, its principal null directions are the roots of a quartic equation. Root multiplicities give the Petrov classes:

Type	Principal-null-direction pattern	Typical interpretation
I	four simple directions	algebraically general
II	one double and two simple directions	one repeated direction
D	two double directions	paired repeated directions
III	one triple and one simple direction	more degenerate null alignment
N	one quadruple direction	pure highest null degeneracy
O	Weyl tensor vanishes	conformally flat

The classification is invariant under changes of coordinates and null frame, although individual Newman–Penrose Weyl scalars are not. If Newman–Penrose components are used, one must first fix the null normalization, tetrad ordering, orientation, and Weyl-scalar sign. Here we avoid convention-dependent scalar factors and define type by the invariant multiplicity pattern.

Polynomial discriminants can screen algebraic speciality. With a fixed normalization, the two complex traces of the self-dual Weyl operator obey a discriminant relation on types II and D. That relation signals repeated eigenstructure; it does not by itself distinguish II from D or resolve all more-special cases. One must inspect eigenspaces, nilpotent structure, or an adapted frame.

Ricci curvature and Segre type

Form the real mixed operator

$$R^\mu{}_\nu=g^{\mu\alpha}R_{\alpha\nu}.$$

Its Segre symbol records real and complex eigenvalues, degeneracies, and Jordan blocks. Lorentzian self-adjointness is not the Euclidean spectral theorem: a $g$-self-adjoint operator need not possess four orthogonal real eigenvectors, and null Jordan chains are possible. This is why listing eigenvalues alone loses information.

Through the Einstein equation,

$$R^\mu{}_\nu =8\pi G\left(T^\mu{}_\nu-\frac12T\delta^\mu{}_\nu\right) +\Lambda\delta^\mu{}_\nu,$$

the Segre type also classifies the pointwise algebraic form of matter, subject to the stated theory. A perfect fluid with one timelike eigenvector and three equal spacelike eigenvalues differs algebraically from null radiation, whose stress tensor is rank one and nilpotent as a mixed operator. But identical Segre type does not imply identical density, pressure, derivatives, or dynamics.

What the pair can and cannot decide

Some reference cases calibrate the notation:

• Minkowski, de Sitter, and anti-de Sitter are Petrov O; their Ricci operators distinguish zero from nonzero constant curvature.
• Schwarzschild is Ricci flat and Petrov D wherever $m\ne0$ and $r>0$.
• A generic vacuum pp-wave is Ricci flat and Petrov N, while a pure-radiation pp-wave has a null Segre type.
• Generic Kerr is also vacuum Petrov D, so “vacuum, type D” plainly does not identify Schwarzschild.

Different Petrov or Segre types prove that two points cannot correspond under a local isometry. Matching types supply only zeroth-order algebraic data. They omit the functional variation of curvature, normalized derivatives, topology, and coordinate domain. Mergulhão and Batista use Petrov type as an efficient preliminary filter before the Cartan–Karlhede algorithm Mergulhão and Batista 2021, §IV, printed pp. 9–12, PDF. Coley, Hervik, and Pelavas show how curvature projectors constructed from the Ricci and Weyl operators contribute to invariant characterization, while exceptional Kundt branches require derivative information Coley, Hervik, and Pelavas 2009, §§3–6, printed pp. 5–21, PDF.

Algebraic type can also change on special surfaces or parameter limits. The classification should therefore be stated on connected regular strata, not attached indiscriminately to an entire coordinate formula. The $m\to0$ limit of Schwarzschild changes type D to O; that is a genuine degeneration, not a contradiction.

Transfer exercise

An Einstein space satisfies $R_{\mu\nu}=\lambda g_{\mu\nu}$ on a connected region. What does its Segre type tell you, and does it determine its Petrov type?

Solution

The mixed Ricci operator is $R^\mu{}_\nu=\lambda\delta^\mu{}_\nu$, so every tangent vector is an eigenvector with the same eigenvalue and there are no nontrivial Jordan blocks. The Ricci algebra is maximally degenerate and diagonalizable. It says nothing decisive about the Weyl tensor: de Sitter is Petrov O, Schwarzschild with $\lambda=0$ is Petrov D, and generic vacuum solutions may be type I. Hence the Segre class does not determine the Petrov class.

References

• Coley, Alan, Sigbjørn Hervik, and Nicos Pelavas. “Spacetimes Characterized by Their Scalar Curvature Invariants.” Classical and Quantum Gravity 26 (2009), 025013. doi:10.1088/0264-9381/26/2/025013.
• Mergulhão, Thiago M., and Carlos Batista. “On the Equivalence of Spacetimes, the Cartan–Karlhede Algorithm.” General Relativity and Gravitation 53 (2021), 28. doi:10.1007/s10714-021-02820-4.