Classification of Electromagnetic Fields¶

A pointwise relativistic taxonomy that uses the two Lorentz scalar invariants of the electromagnetic field tensor to distinguish null and non-null fields and determine which electric–magnetic simplifications some observer can realize.

Version: v1 · 2026-08-30 · History
Domain-specific #: 1477
Origin domain: theoretical physics
Subdomain: relativistic electromagnetism
Aliases: Algebraic classification of electromagnetic fields, Electromagnetic field classification

Core Idea¶

The Classification of Electromagnetic Fields is a pointwise relativistic taxonomy of the electromagnetic field tensor. An observer splits the tensor into electric and magnetic three-vectors, but another inertial observer generally obtains different vectors. Two scalar contractions of the field tensor are invariant under Lorentz transformations. Their values determine whether the field is null or non-null and which observer frames can make the magnetic field vanish, make the electric field vanish, or make the two fields parallel.^[1]

Using one common sign convention and units with explicit ©, the invariants can be written as P = |B|² - |E|²/c² and Q = E·B/c, up to conventional numerical factors and signs. The locked identity is field tensor at an event -> compute two Lorentz invariants -> classify their zero/sign pattern -> infer frame-independent type and permitted observer-frame simplifications. Authors using the opposite metric signature may reverse the electric-like/magnetic-like sign labels; vanishing and physical consequences, not a naked sign copied across conventions, carry the identity.

The classification turns an apparently observer-dependent pair of fields into an invariant statement. It distinguishes a plane-wave-like null field, whose electric and magnetic parts remain equal in magnitude and perpendicular for inertial observers, from non-null fields for which a suitable frame can simplify the electric–magnetic relation. It is local: a spacetime field may occupy different classes at different events.

Structural Signature¶

a Lorentzian spacetime event — the point at which the field is classified;
an electromagnetic field tensor — a real antisymmetric rank-two tensor or bivector (F_{ab});
a metric and orientation — structures needed to raise indices and form the Hodge dual;
an observer split — a timelike four-velocity defining measured electric and magnetic three-fields;
the first scalar invariant — a contraction proportional to (F_{ab}F^{ab}), comparing magnetic and electric magnitudes;
the second scalar invariant — a contraction proportional to (F_{ab}{}^*F), corresponding to the electric–magnetic dot product;
Lorentz invariance — both scalars retain their values under changes of inertial observer;
a null test — both fundamental invariants vanish;
a non-null test — at least one invariant does not vanish;
electric-like and magnetic-like branches — when the mixed invariant vanishes, the magnitude invariant identifies which component can be transformed away;
a mixed or impure branch — nonzero electric–magnetic contraction prevents either component from vanishing but permits a parallelizing frame;
canonical-frame inference — the invariant class constrains observers who measure a simplified field;
principal directions — null directions associated with the bivector provide an equivalent algebraic description;
pointwise scope — classification may change from event to event;
sign-convention control — formulas must declare metric signature, dual, units, and normalization.

Recognition requires classification of the relativistic electromagnetic bivector by invariant algebra. Sorting fields by source, frequency, polarization alone, or coordinate appearance is not this abstraction.

What It Is Not¶

Not classification generally. The categories follow from two Lorentz invariants and field-tensor algebra.
Not a choice of electromagnetic gauge. Gauge transformations alter potentials while leaving the field tensor and these invariants unchanged.
Not merely electric versus magnetic sources. The same field tensor may be split differently by different observers.
Not polarization classification. Linear, circular, and elliptical polarization describe wave behavior, not the complete invariant field taxonomy.
Not the Petrov classification. Petrov types classify the Weyl curvature tensor in general relativity.
Not a global label automatically. A field configuration can change algebraic type across spacetime.
Not a coordinate artifact. The invariants are scalars, although component formulas use a chosen frame.
Not tensor representation generally. Tensor is the carrier; the abstraction is the electromagnetic orbit classification under Lorentz transformations.
Not “null” meaning zero field. A nonzero plane electromagnetic wave is the standard null-field example.

Scope of Application¶

The taxonomy applies in special and general relativity, classical field theory, exact Einstein–Maxwell solutions, observer analysis, and differential geometry. At each event, the tangent space has Lorentzian structure, so the algebraic classification can be performed locally even in curved spacetime. Derivatives, field equations, and global topology are not needed for the pointwise type, though they determine how types vary and which configurations are physically realizable.

The simplest use starts from measured electric and magnetic vectors in one inertial frame, computes the scalar combinations, and asks what another observer can measure. If E·B = 0 and the electric magnitude dominates under the declared convention, a frame exists in which the magnetic part vanishes; if the magnetic magnitude dominates, a frame exists in which the electric part vanishes. If both invariants vanish and the field is nonzero, the field is null. If the dot-product invariant does not vanish, neither field can be transformed away, but a frame can be found in which they are parallel.^[2]

Care is required near zeros. Numerical data can make an exactly null field look weakly non-null, and measured fields always carry uncertainty. The classification should therefore report tolerance and units when applied computationally rather than treating floating-point equality as physical exactness.

Clarity¶

An observer’s E and B are not independently invariant objects; they are a decomposition of one field tensor relative to that observer’s motion. Lorentz boosts mix them. The two scalar contractions survive that mixing, so they label Lorentz-group orbits of bivectors. Classification says which qualitative relations are common to all observers and which simplification some observer can attain.

A static electric field has a frame with B = 0, so its invariants remain non-null and electric-like under every boost even though moving observers may measure both components. A plane wave has E·B = 0 and |E| = c|B|, so both invariants vanish; no inertial boost turns it into a purely electric or purely magnetic field. These are class statements rather than descriptions of whatever components happen to appear in the starting coordinates.

Manages Complexity¶

Six observer-dependent components of an antisymmetric field tensor appear to vary continuously under boosts and rotations. Two scalar invariants compress this orbit structure into a small diagnostic taxonomy. The classification then answers existence questions without searching every possible frame: can one eliminate B, eliminate E, or align them?

This also prevents coordinate reasoning from masquerading as physics. A surprising electric or magnetic component seen after a boost does not imply a new physical field; the invariant class shows whether the change is merely observer decomposition. In general relativity, the same local machinery can organize complex exact solutions before global dynamics are considered.

Abstract Reasoning¶

If both invariants vanish but the tensor is nonzero, the field is null rather than absent.
If the mixed invariant is nonzero, no Lorentz frame can make either E or B vanish, because their invariant dot product would then be zero.
If the mixed invariant vanishes and the magnitude invariant is electric-like, a purely electric frame exists under standard regularity conditions.
The corresponding magnetic-like sign permits a purely magnetic frame.
A Lorentz transformation can change component magnitudes while never changing the class.
A gauge transformation of the potential cannot change the class because it preserves the field tensor.
A type change across spacetime occurs where invariant values cross their defining zero surfaces.
Metric-sign conventions can reverse a printed inequality, so physical frame consequences must accompany formulas.
Classification at one event does not establish that Maxwell’s equations hold or determine sources.
Near-null numerical classifications require tolerance analysis because tiny invariant residuals may be discretization error.

Knowledge Transfer¶

Exact transfer occurs among flat-spacetime electrodynamics, local frames in curved spacetime, exact-solution analysis, and numerical relativistic field calculations. The tensor, dual, invariants, observer split, and canonical-frame consequences remain literal.

Other algebraic classifications—Petrov classification of curvature or canonical forms of antisymmetric matrices—share group-orbit and invariant reasoning but have different objects and categories. The portable skeleton belongs to Classification, Invariant, Equivalence Class, and Frame of Reference. The electromagnetic taxonomy remains domain-specific.

Examples¶

plane wave: nonzero field with perpendicular E and B, equal magnitudes in compatible units, and both invariants zero;
electrostatic field: a frame has B = 0, identifying an electric-like non-null class;
magnetostatic field: a frame has E = 0, identifying a magnetic-like non-null class;
parallelizable mixed field: nonzero E·B prevents elimination but permits an observer measuring parallel electric and magnetic vectors;
curved-spacetime event: a local tetrad supplies components, while scalar contractions determine the class independent of tetrad choice;
non-example—polarization: circular polarization categorizes a wave’s oscillation, not the general bivector orbit;
non-example—gauge: changing vector potential representation leaves the class fixed.

Structural Tensions¶

observer-dependent components vs. invariant type — measurements change while the algebraic class does not;
local classification vs. global solution — pointwise type is simple while its spacetime distribution may be complex;
compact invariants vs. geometric interpretation — two numbers classify, but principal directions and canonical frames explain;
exact zeros vs. measured tolerances — theory has sharp classes while data are noisy;
formula convention vs. physical consequence — signs vary across texts while eliminability statements remain stable;
algebraic possibility vs. dynamical realization — the class permits a canonical frame but does not solve Maxwell’s equations.

Structural–Framed Character¶

This abstraction is structural. Tensor contractions and Lorentz-group action fix the result. Human conventions affect metric signature, factors of ©, normalization, and names, but consistent translations preserve the classification.

Structural Core vs. Domain Accent¶

The core is group action + complete invariants -> orbit class + canonical representative. The domain accent is the electromagnetic bivector on Lorentzian spacetime, its Hodge dual, the observer-dependent electric–magnetic split, and physical frame inferences. Remove those and the result is Classification or Invariant.

Classification — explicit invariant rules map a field tensor to a finite type.
Invariant — scalar contractions survive Lorentz transformations.
Frame of Reference — the class governs which observer decompositions are possible.
Equivalence Class — tensors related by Lorentz transformations share a type.

The prospective DAG uses strict subsumption under prime:classification.

Relationships to Other Abstractions¶

Current abstraction Classification of Electromagnetic Fields Domain-specific

Parents (1) — more general patterns this builds on

Classification of Electromagnetic Fields is a kind of Classification Prime

explicit invariant rules map a field tensor to a finite type.

Hierarchy path (1) — routes to 1 parentless root

Classification of Electromagnetic Fields → Classification

Neighborhood in Abstraction Space¶

Classification of Electromagnetic Fields sits in a sparse region of the domain-specific corpus (85^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

gauge choice;
electromagnetic polarization;
Petrov classification;
source classification;
tensor representation;
coordinate components;
the zero field;
classification as a general prime.

Notes¶

[n1] L. D. Landau and E. M. Lifshitz, The Classical Theory of Fields, section on electromagnetic-field invariants.

References¶

[1] Nikolai V. Mitskievich, “Classification of Electromagnetic Fields in General Relativity and Its Physical Applications,” arXiv:0802.3474, https://arxiv.org/abs/0802.3474. registry ↩

[2] M. Vargas-Rodríguez et al., “Vanishing Poynting Observers and Electromagnetic Field Classification in Kerr and Kerr-Newman Spacetimes,” Advances in High Energy Physics (2022), https://doi.org/10.1155/2022/1066886. registry ↩

[3] “Classification of electromagnetic fields,” Wikipedia, frozen evidence packet, https://en.wikipedia.org/wiki/Classification_of_electromagnetic_fields. registry