Newman–Janis–Azreg-Aïnou Algorithm

• The Newman–Janis–Azreg-Aïnou algorithm is a procedure that transforms static, spherically symmetric metrics into stationary, axisymmetric rotating metrics without relying on ad hoc complexification.
• It replaces the traditional complex radial function transformation with real functions A(r,θ,a), B(r,θ,a), and Ψ(r,θ,a), constrained by circularity and matter assumptions.
• The method extends to various settings, including modified gravity and cosmological constants, highlighting both its practical applications and inherent ambiguities in the residual undetermined function Ψ.

The Newman–Janis–Azreg-Aïnou algorithm designates a family of rotation-generating procedures that start from a static, spherically symmetric seed metric and attempt to construct a stationary, axisymmetric rotating metric. In the narrow sense, it refers to Azreg-Aïnou’s “without complexification” modification of the Newman–Janis algorithm, in which the ad hoc complexification of radial functions is replaced by real functions $A(r,\theta,a)$, $B(r,\theta,a)$, and $\Psi(r,\theta,a)$ fixed by circularity and matter assumptions. In broader usage, the label is also applied to later modified Newman–Janis prescriptions that preserve some of this structure while still performing explicit complex coordinate transformations; the distinction is essential because formal overlap with Azreg-Aïnou does not, by itself, amount to a strict Azreg-Aïnou implementation (Junior et al., 2020, Li et al., 22 Jan 2025).

1. Classical Newman–Janis construction

The original Newman–Janis algorithm begins with a static seed such as Schwarzschild or Reissner–Nordström written in Eddington–Finkelstein-type coordinates, rewrites the inverse metric in null-tetrad form, allows $r$ and a null coordinate to become complex, performs a shift of the type

$$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$

and then reconstructs a real Lorentzian metric from the transformed tetrad. In the standard prescription, expressions such as $1/r$ and $r^2$ are replaced by real combinations involving $r$ and $\bar r$, leading to the familiar Kerr-type quantity $\Sigma=r^2+a^2\cos^2\theta$; a final Boyer–Lindquist transformation produces Kerr or Kerr–Newman (Brauer et al., 2014).

What makes the classical construction powerful is also what makes it controversial. A thesis-length analysis of the trick emphasizes that no fully satisfactory explanation is known for why it works, and identifies the central difficulty as the non-holomorphic and ambiguous complexification step. In that account, the usual derivation also contains a hidden tetrad-repair step: after the complex coordinate transformation, one must effectively restore $B(r,\theta,a)$0 and $B(r,\theta,a)$1 as complex conjugates before the final Kerr tetrad is obtained (Rajan, 2016).

A distinctive reinterpretation of the classical algorithm is that the complexification does more than generate rotation. It also naturally extends the original nonnegative radial coordinate to a real coordinate $B(r,\theta,a)$2, thereby introducing the negative-$B(r,\theta,a)$3 region of Kerr and Kerr–Newman directly within the algorithm rather than only through a later analytic extension. In this picture, the horizons and ergosurfaces remain confined to the $B(r,\theta,a)$4 sector, while the $B(r,\theta,a)$5 region is asymptotically flat but horizonless (Brauer et al., 2014).

2. Azreg-Aïnou’s “without complexification” reformulation

Azreg-Aïnou’s modification starts from a generic static spherical seed

$B(r,\theta,a)$6

with $B(r,\theta,a)$7, $B(r,\theta,a)$8, and $B(r,\theta,a)$9 left arbitrary rather than fixed to the usual gauge $\Psi(r,\theta,a)$0, $\Psi(r,\theta,a)$1 (Junior et al., 2020). A formal Newman–Janis-type complex shift is still written,

$\Psi(r,\theta,a)$2

but the crucial difference is that the radial seed functions are not complexified by ad hoc rules. Instead, they are replaced by independent real functions,

$\Psi(r,\theta,a)$3

subject only to the nonrotating limits

$\Psi(r,\theta,a)$4

This is the precise sense in which the method is “without complexification”: the formal complex shift remains, but the actual complexification rules for the seed functions are abandoned (Junior et al., 2020).

The defining structural requirement is circularity. One introduces a Boyer–Lindquist-like transformation

$\Psi(r,\theta,a)$5

and imposes $\Psi(r,\theta,a)$6. With

$\Psi(r,\theta,a)$7

the circularity-preserving choice is

$\Psi(r,\theta,a)$8

together with

$\Psi(r,\theta,a)$9

This guarantees the existence of Boyer–Lindquist-like coordinates by construction (Junior et al., 2020).

The price of this gain is that $r$0 remains undetermined at the purely algorithmic stage. It is fixed only after extra physical structure is imposed. In the imperfect-fluid interpretation associated with Azreg-Aïnou, $r$1 must satisfy nonlinear PDEs, so the residual freedom is a matter-model ambiguity rather than a coordinate-complexification ambiguity (Junior et al., 2020). Application papers therefore describe the method explicitly as the “Newman–Janis algorithm without complexification” and treat that label as synonymous with Azreg-Aïnou’s modification (Contreras et al., 2019).

3. Canonical metric form and separable dynamics

The output of the modified Newman–Janis algorithm is a circular rotating metric that is manifestly Boyer–Lindquist-like. In the notation of the general construction, the rotating line element depends on $r$2, $r$3, $r$4, and $r$5, with the Kerr-like combinations

$r$6

playing the rôle of the usual radial and angular functions. The decisive point is that the circularity problem of the original Newman–Janis algorithm is removed by construction, but the function $r$7 survives as an undetermined conformal factor of the rotating family (Junior et al., 2020).

That residual ambiguity is much less important for null dynamics than for timelike dynamics. For the entire class of metrics generated by the modified Newman–Janis algorithm, null Hamilton–Jacobi separability is automatic. In the null case, $r$8 cancels from the Hamilton–Jacobi equation, so the ambiguity carried by $r$9 does not affect photon trajectories as unparametrized curves. The same paper shows that the Carter constant survives in the usual separated form and that spherical photon orbits, shadows, and asymptotic lensing observables depend only on the seed data $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$0, $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$1, and $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$2, not on the unresolved function $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$3. By contrast, timelike geodesics do depend on $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$4, and in a plasma the separability condition is recovered only when

$$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$5

The conformal irrelevance of $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$6 is therefore specific to vacuum null optics (Junior et al., 2020).

A related result applies to a wider class of rotating spacetimes generated by successful Newman–Janis constructions. If the combinations

$$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$7

depend only on $$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$8, then the null Hamilton–Jacobi equation separates completely and one obtains analytic formulas for the shadow boundary. In that sense, the same condition that permits a clean Boyer–Lindquist reduction also organizes the separable null dynamics of the resulting rotating metric (Shaikh, 2019).

4. Extensions, applications, and borderline cases

One extension of the original Newman–Janis algorithm addresses the cosmological-constant sector directly. A unified replacement rule

$$u' = u - i a \cos\theta, \qquad r' = r + i a \cos\theta,$$9

reproduces the standard prescriptions for $1/r$0 and $1/r$1 and extends them to $1/r$2. In particular, it yields

$1/r$3

which is precisely what is needed to handle de Sitter and anti-de Sitter seed metrics and derive Kerr-$1/r$4 and Kerr–Newman-$1/r$5 line elements (Urreta et al., 2015).

Azreg-Aïnou’s no-complexification procedure has also been applied to nonvacuum seeds. For the static AdS “polytropic black hole”

$1/r$6

one has $1/r$7 and $1/r$8. In this sector,

$1/r$9

and one admissible solution of the $r^2$0-equations is

$r^2$1

The resulting rotating geometry is Kerr-like, with standard $r^2$2, $r^2$3, and $r^2$4 functions, and it is then used to study horizons, static limits, causality, shadows, Hawking temperature, and emission rates (Contreras et al., 2019).

Not every paper that introduces rotating deformations $r^2$5 and $r^2$6 is, however, a strict Azreg-Aïnou implementation. In the dRGT-Maxwell construction of rotating black holes in massive gravity, the authors derive the static seed

$r^2$7

rewrite it in Eddington–Finkelstein form, promote $r^2$8 and $r^2$9 to $r$0 and $r$1, and then perform explicit complexified differential transformations adapted to a cosmological-constant background. The resulting rotating metric is Kerr–Newman-$r$2-like, with the massive-gravity hair entering through the shift $r$3. The paper cites Azreg-Aïnou only for the idea that the seed functions become $r$4- and $r$5-dependent; since it still carries out explicit complexification, it is not Azreg-Aïnou’s non-complexification algorithm in the strict sense (Li et al., 22 Jan 2025).

A closely related 2025 construction uses the Newman–Janis algorithm together with non-complexification results to generate a rotating exact Einstein solution dressed by anisotropic matter. There the rotating metric takes the Kerr-like form

$r$6

with

$r$7

The residual freedom is fixed by the condition $r$8, and the resulting Einstein tensor is diagonal in an orthonormal frame, with $r$9 and matter hair controlled by the radial deformation $\bar r$0 (Kim et al., 10 Mar 2025).

5. Validity, ambiguity, and failure modes

The central ambiguity of the original Newman–Janis algorithm is operational as well as conceptual. A symbolic implementation in REDUCE makes this explicit: the code is intended for spherically symmetric or perturbed spherically symmetric seeds, but its applicability is restricted to metrics containing potentials of the form $\bar r$1, and the crucial complexification step must still be done by hand. In other words, even a successful computer implementation of the original workflow does not remove the need for a user-supplied complexification rule (Gutierrez-Chavez et al., 2014).

Azreg-Aïnou’s reformulation changes the locus of ambiguity rather than eliminating all ambiguity. The original problem is the ad hoc replacement of seed functions under $\bar r$2-complexification and the possible failure of circularity. The modified Newman–Janis algorithm fixes circularity by construction, but leaves a free real function $\bar r$3 whose determination requires additional assumptions about the matter content. The procedure is therefore more rigid geometrically and less rigid materially (Junior et al., 2020).

More serious objections arise in modified gravity. A direct analysis in quadratic gravity shows that applying the standard Newman–Janis algorithm to a non-GR spherically symmetric seed does not, in general, produce the correct rotating black-hole solution. In the explicit example studied there, the Newman–Janis-rotated metric fails the modified vacuum field equations, has $\bar r$4, and develops a new naked curvature singularity at $\bar r$5. The paper’s conclusion is that the Newman–Janis algorithm should not, in general, be used to construct rotating black holes outside general relativity without independent verification (Hansen et al., 2013).

A related caution appears in scalar backgrounds. Applying the Newman–Janis algorithm to the Papapetrou antiscalar spacetime yields a rotating metric whose scalar sector is exact but whose full Einstein–matter system is satisfied only asymptotically. The mismatch is tied in that analysis to the Hawking–Ellis classification of the scalar energy-momentum tensor and to the fact that the rotating matter sector is not preserved by the formal geometric rotation (Makukov et al., 2023).

A recurring misconception is that any later modified Newman–Janis derivation that introduces rotating deformations $\bar r$6 and $\bar r$7 is thereby “Azreg-Aïnou.” That is not correct. If a paper still performs explicit complexified transformations of $\bar r$8, $\bar r$9, or $\Sigma=r^2+a^2\cos^2\theta$0, then the implementation is better described as a modified Newman–Janis algorithm with one formal ingredient reminiscent of Azreg-Aïnou, not as Azreg-Aïnou’s non-complexification method itself (Li et al., 22 Jan 2025).

6. Interpretive frameworks and historical status

The historical status of the Newman–Janis construction remains unsettled. Rajan and Visser present a Kerr–Schild, Cartesian, metric-level reformulation in which the familiar four-step Newman–Janis procedure is reduced to a cleaner two-step process acting directly on the Kerr–Schild metric. They explicitly refuse to upgrade the method to the status of a genuine algorithm, however, emphasizing that it still requires motivated guesswork and remains, at best, a trick (Rajan et al., 2016).

Other analyses have tried to isolate the source of that “trick” character. A detailed treatment of complex spacetimes argues that the real mystery is not the complex shift by itself but the ambiguous non-holomorphic complexification that precedes it, together with the hidden tetrad-repair step needed to recover the final Kerr tetrad. The same work shows that Giampieri’s metric-level derivation is equivalent to the standard null-tetrad derivation once that hidden tetrad is identified, thereby clarifying why metric-level and tetrad-level versions of the trick can agree even when their intermediate steps look different (Rajan, 2016).

A more recent interpretive proposal goes further and claims to explain why the classical algorithm works for the Kerr family. In that account, Kerr is the nonlinear Kerr–Schild superposition of a self-dual and an anti-self-dual Taub–NUT instanton. The Newman–Janis algorithm is then not a coordinate transformation but the metric-level reflection of separating those chiral constituents to complex locations $\Sigma=r^2+a^2\cos^2\theta$1 and $\Sigma=r^2+a^2\cos^2\theta$2. On that reading, the familiar factors $\Sigma=r^2+a^2\cos^2\theta$3 are distances to distinct complex centers, not arbitrary substitutions (Kim, 2024).

These reinterpretations suggest that the Newman–Janis–Azreg-Aïnou algorithm is less a single closed prescription than a line of related constructions. Its core problem is always the same: how to deform a spherical seed into a Kerr-like axisymmetric geometry while keeping control over circularity, matter content, and field equations. The original Newman–Janis algorithm solves that problem by an obscure complex trick; Azreg-Aïnou’s reformulation solves it by replacing complexification ambiguity with real functional freedom constrained by geometry and matter; later modified algorithms occupy intermediate positions between those two poles.