WorksPhysics and Astronomy
Photon sphere, shadow, and polarized images of thick accretion flows around a quantum Lee-Wick black hole
Abstract
We study the optical appearance of a static, spherically symmetric black hole (BH) in sixth-derivative Lee-Wick gravity with an exponential coupling S1 and an oscillatory coupling S2. From the null-geodesic effective potential we derive the photon sphere, critical impact parameter, and shadow radius. In the weak-coupling (Block-1) regime the shadow radius rapidly approaches the Schwarzschild value 3 sqrt(3) M, while in the strong-coupling (Block-2) regime the interplay of exponential damping and oscillatory modulation yields non-monotonic and, in some regions, strongly suppressed shadows. We classify the (S1, S2) plane by the number of horizons and unstable photon spheres: horizonless configurations retain a critical curve but lack a central brightness depression, and configurations with two unstable photon spheres show a shadow bounded by the inner one plus an outer photon ring. The 1-sigma/2-sigma EHT bounds for M87* and Sgr A* exclude certain strong-coupling bands. We develop a ray-tracing code that integrates exact Lee-Wick null geodesics, computes 230 GHz thermal-synchrotron emission from a geometrically thick, optically thin RIAF-like flow, and transports linear polarization via the Walker-Penrose constants. Compared with Schwarzschild, in the strong-coupling regime the beam-convolved ring diameter deviates by up to ~14%, the photon-ring Lyapunov exponent grows from pi to about 4-9, and the beta_2 amplitude rises by up to ~50% for a radial-toroidal field, while the phase of beta_2 is nearly insensitive to (S1, S2) at low inclination, mainly tracing the field geometry. A parametrized shock-cone model, motivated by earlier Bondi-Hoyle-Lyttleton simulations, shows that shock structures modulate the flux in a viewing-azimuth-dependent way without changing the critical-curve size. These are the first shadow, thick-disk, and polarized synthetic images of the Lee-Wick BH.
Introduction
Higher-derivative extensions of general relativity (GR) were originally introduced to cure the ultraviolet (UV) divergences of the gravitational path integral, at the cost of introducing massive ghost-like degrees of freedom. Lee and Wick proposed that such ghosts, when arranged in complex-conjugate pairs, need not spoil unitarity of the -matrix [1, 2, 3]. This mechanism, extended to gravity, yields super-renormalizable or finite quantum gravity theories [4, 5, 6, 7, 8, 9, 10], generically referred to as Lee-Wick gravity [11].
An effective, spherically symmetric black-hole solution sourced by a smeared, oscillatory energy density was constructed in Ref. [12] and has since been studied extensively: regularity and thermodynamics [13, 14, 15, 16], horizon multiplicity [14], analogies with noncommutative-geometry-inspired black holes [17, 18], gravitational lensing [19, 20], periastron precession [21], a rotating extension via the Newman–Janis algorithm together with its thermodynamics [22, 23, 24, 25], and quasinormal ringing [26]. Exact solutions of the underlying sixth-derivative theory have also been explored [27, 28].
Most recently, the orbital dynamics of test particles and the Bondi-Hoyle-Lyttleton (BHL) accretion morphology around a non-rotating Lee-Wick BH were analyzed in detail, including the innermost stable circular orbit (ISCO), the epicyclic frequencies, and the resulting quasi-periodic oscillations (QPOs) obtained from general relativistic hydrodynamic (GRHD) simulations of the shock cone [29]. That study showed that the weak-coupling regime (, , hereafter Block-1) reproduces Schwarzschild-like shock cones and low-frequency QPOs, whereas the strong-coupling regime (, , hereafter Block-2) produces broad, asymmetric bow-shock structures and a shift toward high-frequency QPOs, with frequency ratios (3 : 2, 2 : 1, 4 : 3) comparable to those observed in Galactic black-hole X-ray binaries [30, 31, 32, 33].
What remains unexplored — and what constitutes the primary aim of the present work — is the direct horizon-scale imaging signature of the Lee-Wick corrections : the photon sphere, the critical curve (shadow boundary), and the polarized brightness distribution produced by a thick, optically thin accretion flow illuminating the BH. This is precisely the observable measured by the EHT for M87* [34, 35, 36, 37, 38] and Sgr A* [39, 40], and it has become a standard diagnostic for constraining deviations from the Kerr/Schwarzschild geometry [41, 42, 43, 44]. The shadow formalism traces back to the classic works of Synge [45], Bardeen [46], and Luminet [47], and its use as a probe of alternative gravity and compact-object geometries is by now well established [48, 49, 50, 51, 52, 53, 54, 55, 56].
Relation to previous work and novelty. The photon-sphere and shadow-radius formulas of § 1.2 follow the standard treatment of static, spherically symmetric metrics [43, 45], and the critical impact parameter of the Lee-Wick BH already enters, for selected couplings, the strong-deflection lensing analysis of Ref. [19]; light bending in quadratic and higher-derivative gravity was also discussed in Ref. [20]. We do not claim novelty for these formulas. What is new in the present work is: (i) a global map of the optical structure of the Lee-Wick spacetime over the plane, including the classification of horizonless configurations and of configurations with two unstable photon spheres, and the identification of the critical curve that actually bounds the shadow; (ii) the confrontation of the full shadow-radius map with the and EHT bounds for M87* and Sgr A*; (iii) the first ray-traced intensity and polarization images of the Lee-Wick BH illuminated by a thick, optically thin 230 GHz flow, computed with a dedicated code and compared quantitatively with Schwarzschild through image-domain observables (ring diameter, brightness asymmetry, net linear polarization, and the coefficient [57]); and (iv) an explicit, parametrized test of how shock-cone/bow-shock density structures of the type found in Ref. [29] imprint on the images.
In this paper we (i) derive the photon sphere and shadow-radius formulas for the Lee-Wick lapse function, present illustrative numerical results across the Block-1/Block-2 parameter space, and compare them with the Schwarzschild reference and with the angular sizes measured by the EHT for M87* and Sgr A*; and (ii) set up the general relativistic polarized radiative-transfer (GRRT) formalism for a geometrically thick, optically thin accretion flow around a Lee-Wick BH, using both a phenomenological RIAF-like prescription [58, 59, 60] and the analytic Hou disk model [61, 62], deliberately setting aside the geometrically thin Novikov-Thorne disk. We connect the expected disk-image morphology to the shock-cone/bow-shock structures found in Ref. [29], and we implement the RIAF-like framework in our own ray-tracing and polarized-transfer code (Sec. IV E), which produces synthetic 230 GHz images and polarization maps that can be compared with EHT-like data. Public GRRT codes such as GYOTO [63], ipole [64], grtrans [65], and Odyssey [66] were used as methodological references and for the conventions adopted, but none of them was used to generate the results presented here.
The paper is organized as follows. Section II reviews the Lee-Wick BH metric. § 1.2 derives the photon sphere and shadow radius and presents numerical results, including the classification of horizon and photon-sphere structures and the EHT constraints. Section IV develops the thick-disk radiative-transfer and polarization formalism and presents the numerical implementation and the ray-traced images. § 1.4 discusses the observational strategy and EHT constraints. Section VI concludes.
Throughout we use geometrized units and set the BH mass unless stated otherwise; carry dimensions of inverse length .
Lee-Wick Black Hole Spacetime
The non-rotating Lee-Wick BH is a solution of the effective field equations sourced by a smeared mass distribution with oscillatory-exponential profile [12], arising as an approximate solution of the sixth-derivative action
with and , following the notation of Ref. [29]. The line element is
with the lapse function
The Schwarzschild limit is recovered for . Near the center (e.g. for , i.e. the core is regular, and may even exceed unity in some intervals for large , corresponding to a locally negative effective mass function. In what follows, and as in Ref. [29], the parameters appearing in Eq. (3) are scanned as phenomenological deformation parameters of the lapse function; part of the Block-1 range (e.g. , ) lies outside the interval quoted above for the couplings of the action, and the corresponding results should be read as tests of the metric (3) rather than of a specific UV completion. Horizons are located at the zeros of ; depending on the spacetime can have zero, two, or four horizons over the parameter range studied here (§ 1.2 D) [14, 29]. As established in the companion hydrodynamical study, two representative parameter blocks are of physical interest:
Block-1 (weak coupling): , – exponential damping dominates, and the horizon structure are close to Schwarzschild;
Block-2 (strong coupling): , – the oscillatory term competes with the exponential term, producing multiple horizons or naked-singularity configurations and strongly modified curvature near .
Photon Sphere and Black Hole Shadow
Null geodesics and the photon sphere
For a photon moving in the equatorial plane of the metric (2), the conserved energy and angular momentum combine with the null condition to give the radial equation
The circular photon orbit (photon sphere) at satisfies , i.e.,
For Schwarzschild, Eq. (5) gives the familiar . Because in Eq. (3) is transcendental in , Eq. (5) must be solved numerically for general .
Each root of Eq. (5) outside the outer horizon (or anywhere, for horizonless configurations) is either an unstable photon sphere () or a stable one (); stable photon spheres always come paired with an additional unstable one [67]. Each unstable photon sphere defines a critical impact parameter
and a Lyapunov exponent, which for metrics with takes the form [68]
with for Schwarzschild. The quantity sets the flux ratio of successive higher-order (“”) photon-ring sub-images [56].
Shadow radius and celestial coordinates
For a static observer at radial coordinate , the apparent angular radius of the shadow is , so that the linear shadow radius measured on the observer’s celestial sphere is
Using standard celestial coordinates for a photon with impact parameters arriving at an observer at inclination [46, 63], the shadow boundary of the (non-rotating) Lee-Wick BH is the circle , since the spacetime is spherically symmetric. We quantify deviations from Schwarzschild using the fractional shadow deviation
with . For the rotating extension of the Lee-Wick BH [22], obtained via the Newman-Janis algorithm [23, 24, 25], the shadow boundary is no longer circular and is more conveniently characterized by the Hioki-Maeda observables [50] or by the endpoint asymmetry used in Refs. [49, 51]; we leave the fully spin-dependent shadow to future work and focus here on the non-rotating case, which already isolates the effect of .
Which critical curve bounds the shadow. When has more than one local maximum outside the horizon, a photon arriving from infinity with impact parameter is captured only if exceeds every potential barrier along its path, i.e. if . Rays with cross the lower barrier, are reflected by the higher one, and escape. Hence the shadow edge is set by the global maximum of (the smallest critical impact parameter), which we denote ; when this global maximum is the inner one, the outer unstable photon sphere is nevertheless visible as a second critical curve outside the shadow. This criterion is applied consistently in Table I and Figure 1–Figure 7.

Figure 1. Radius of the shadow-bounding (innermost-relevant, see text) unstable photon sphere of the non-rotating Lee-Wick spacetime as a function of (left panel, ) and of (right panel, ). Dotted segments denote horizonless configurations, for which the unstable photon sphere still exists but no horizon does. Dashed thin curves in panel (b) give the radius of a second, outer unstable photon sphere whenever it exists. The dotted vertical line at marks the horizon bifurcation of the sequence discussed in the text.

Figure 7. Fractional shadow deviation as a function of , for , , and , together with the EHT bounds on the shadow size of M87* (1, gold) and Sgr A* (1 and 2, cyan) quoted in Figure 3. Jumps correspond to the appearance of an inner, higher potential barrier or to the horizon bifurcation. Portions of the curves lying outside the shaded bands correspond to parameter values that are disfavored by current horizon-scale imaging data.
Numerical results
We solved Eqs. (5)–(8) numerically (bracketing of all sign changes of and of on a dense radial grid, followed by Brent root refinement, and classification of every extremum of ) over a broad grid of spanning both the Block-1 and Block-2 regimes identified in the hydrodynamical study [29]. Table I lists the number of horizons , outer horizon , photon sphere radius , shadow radius , Lyapunov exponent , and the fractional deviation from Schwarzschild for twenty-one representative parameter combinations; Figure 2–Figure 6 present the same information graphically, from several complementary viewpoints.

Figure 2. Shadow radius of the non-rotating Lee-Wick black hole as a function of the coupling parameter (left panel, with fixed) and as a function of (right panel, for , and ). The horizontal dashed line marks the Schwarzschild value , to which the shadow radius converges as grows large. Dotted segments: horizonless configurations, for which only a critical curve (no dark shadow) exists. Thin dashed curves in panel (b): critical impact parameter of the outer unstable photon sphere in configurations with two (or three) unstable photon spheres; this outer critical curve produces an additional bright photon ring outside the shadow (Figure 19). The thin dotted vertical line at on the curve marks the horizon-bifurcation jump discussed in the text.

Figure 6. Shadow silhouettes (celestial circles ) of the non-rotating Lee-Wick black hole for seven representative parameter combinations, compared with the Schwarzschild silhouette (gray dashed circle). Because the non-rotating metric is spherically symmetric, each shadow boundary is exactly circular, with its radius set entirely by the value of for that choice of and . For , both the inner (shadow-bounding) and the outer critical curves are shown.
Several trends emerge from Table I and Figure 2–Figure 7. First, in the Block-1 regime (, ), converges monotonically to the Schwarzschild value as increases (Figure 2, left panel; Figure 3), mirroring the convergence of the ISCO and of the shock-cone opening angle reported in Ref. [29]: the exponential suppression term dominates and washes out the oscillatory Lee-Wick correction, and already by the shadow is indistinguishable from Schwarzschild to three significant figures. Second, for fixed small (Block-2, e.g. ), the shadow is largest immediately after the horizon forms (, , i.e. ) and then shrinks monotonically with increasing , crossing the Schwarzschild value near and reaching at (Table I, Figure 7). For a second unstable photon sphere appears at larger radius (Figure 5b); since the inner barrier is the higher one, continues to decrease (to at ), while an outer critical curve with – appears outside the shadow. Third, for and the configuration is horizonless, consistent with Table I of Ref. [29]; however, it is not devoid of photon spheres. It has an unstable photon sphere at () and a stable one at , and its core is regular ( as ), so it is better described as a horizonless ultracompact object than as a naked singularity. Its critical curve exists, but rays with traverse the core and escape, so that no dark shadow is formed (§ 1.2 D, Figure 19). This non-monotonic behavior is the direct optical counterpart of the widened, asymmetric bow-shock morphology found for Block-2 accretion flows: a smaller/distorted photon sphere is consistent with deeper, oscillatory effective potential wells identified via in Ref. [29].

Figure 3. (a) Shadow radius over the plane, evaluated at the global maximum of outside the outer horizon (smallest critical impact parameter, see text). The hatched region is horizonless (no shadow), and the cyan dash-dotted contour encloses configurations with two unstable photon spheres. (b) Consistency of with the EHT shadow-size bounds: M87*, at [36, 41, 69]; Sgr A*, () and (), from the Keck- and VLTI-averaged EHT calibration [40, 42]. The Block-1 region is consistent with all bounds for every , while two diagonal strong-coupling bands (gray) are excluded.

Figure 5. Photon effective potential for several representative pairs of the coupling parameters . (a) Configurations with a single photon sphere, spanning both the weak-coupling regime (black, green, and blue curves) and the strong-coupling regime (red and magenta curves), compared with the Schwarzschild potential (gray dashed curve, recovered in the limit of large ). (b) A horizonless configuration (potential extending to , with a minimum corresponding to a stable photon sphere) and configurations with two unstable photon spheres, including a four-horizon case. The radial location and height of each peak fix the photon sphere radius and shadow radius for the corresponding choice of .

Figure 19. 230 GHz images at (toroidal field) of (a) Schwarzschild, (b) a configuration with two unstable photon spheres (, ; inner critical curve , outer one ), and (c) a horizonless configuration (, ). Cyan dotted circles: critical curves. Panels (d)–(f): the same images convolved with a beam. In (b) the dark region is bounded by the inner critical curve while the bright ring follows the outer one; in (c) rays crossing the regular core produce nested inner rings and no central brightness depression survives the convolution.
| () | remarks | ||||||
| 1.0 | 1.0 | 0 | — | 3.806 (stable: 2.237) | 3.13 | horizonless, regular core | |
| 1.0 | 1.1 | 2 | 2.359 | 3.945 | 5.957 (+14.6%) | 3.86 | |
| 1.0 | 1.2 | 2 | 2.612 | 3.954 | 5.837 (+12.3%) | 4.40 | |
| 1.0 | 1.3 | 2 | 2.702 | 3.893 | 5.669 (+9.1%) | 4.86 | |
| 1.0 | 1.4 | 2 | 2.724 | 3.792 | 5.469 (+5.3%) | 5.31 | |
| 1.0 | 1.5 | 2 | 2.706 | 3.667 | 5.247 (+1.0%) | 5.77 | |
| 1.0 | 2.0 | 2 | 2.372 | 2.959 | 3.946 () | 8.70 | |
| 1.0 | 2.5 | 2 | 1.977 | 2.387, 4.640 | 2.650 () | 12.8, 7.9 | outer ring |
| 1.0 | 3.5 | 4 | 3.045 | 3.513, 5.177 | 3.916 () | 18.7, 11.7 | outer ring |
| 1.5 | 1.0 | 0 | — | 3.108 (stable: 1.899) | 2.55 | horizonless, regular core | |
| 1.5 | 1.5 | 2 | 2.330 | 3.364 | 5.354 (+3.0%) | 4.12 | |
| 1.5 | 4.0 | 4 | 2.583 | 3.082 | 4.522 () | 10.96 | |
| 2.0 | 1.0 | 2 | 1.686 | 3.008 | 5.242 (+0.9%) | 2.85 | |
| 2.0 | 1.3 | 2 | 2.038 | 3.144 | 5.291 (+1.8%) | 3.34 | |
| 2.0 | 2.0 | 2 | 2.160 | 3.008 | 5.132 () | 3.78 | |
| 2.0 | 4.0 | 2 | 1.416 | 1.596, 3.065 | 3.374 () | 10.3, 5.7 | outer ring |
| 4.0 | 1.0 | 2 | 1.989 | 3.000 | 5.196 (0.0%) | 3.14 | |
| 5.0 | 1.0 | 2 | 1.998 | 3.000 | 5.196 (0.0%) | 3.14 | |
| 6.0 | 1.0 | 2 | 2.000 | 3.000 | 5.196 (0.0%) | 3.14 | |
| 6.0 | 3.0 | 2 | 2.000 | 3.000 | 5.196 (0.0%) | 3.14 | |
| 6.0 | 6.0 | 2 | 2.000 | 3.000 | 5.196 (0.0%) | 3.14 |
Table 1. Number of horizons , outer horizon radius , radii of the unstable photon spheres , shadow radius (all in units of ), fractional deviation , and Lyapunov exponent [Eq. (7)] for representative combinations of . When two unstable photon spheres exist, both are listed; the shadow is bounded by the one with the smaller critical impact parameter, and the second value of (“outer ring”) gives the additional critical curve outside the shadow. For horizonless configurations no dark shadow forms, but a critical curve with impact parameter exists (the radius of the accompanying stable photon sphere is given in parentheses). Schwarzschild: , , , .
We note in addition a genuine, sharp discontinuity in the outer-horizon radius along the curve, visible in Figure 1(b) and Figure 2(b) (dotted vertical lines): at a local minimum of near touches zero and splits into a new pair of roots, so that jumps abruptly from to once this second, outer horizon becomes the largest root of . This is not a numerical artifact but a horizon-bifurcation of the type generic to multi-horizon Lee-Wick configurations [14]. At the bifurcation the two inner photon spheres are swallowed by the new outer horizon, so that the shadow radius jumps from to ; the earlier outer critical curve then becomes the shadow boundary (Figure 2(b)). An analogous bifurcation occurs at for . These jumps must be kept in mind when interpolating across the Block-2 region. All curves in Figure 1–Figure 7 are computed with the criterion of Sec. III B, which matters in the double-photon-sphere corner of the parameter plane.
Horizon and photon-sphere structure, and implications for higher-order images
Figure 4 summarizes the optical structure of the spacetime. Four qualitatively different cases occur in : (i) two horizons and a single unstable photon sphere (the generic, Schwarzschild-like case, which includes all of Block-1); (ii) two or four horizons and two unstable photon spheres separated by a stable one (a diagonal band at small and ); (iii) the four-horizon region beyond the bifurcation; and (iv) horizonless configurations at , . In the horizonless case the configuration should not be regarded as a naked singularity: remains positive everywhere and tends to unity at the center, the photon sphere pair always exists in this corner, and the configuration is an ultracompact, regular, horizonless object.

Figure 4. Classification of the Lee-Wick spacetime over the plane according to the number of horizons and of unstable photon spheres outside the outer horizon. Horizonless configurations (red) possess a pair of stable and unstable photon spheres and a regular core; the green and light-blue regions have two unstable photon spheres and therefore two critical curves; the boundary between two- and four-horizon regions is the horizon bifurcation discussed in the text.
These structures have direct consequences for the image. (a) In case (ii) the shadow is bounded by the inner critical curve , but the brightest ring of a geometrically thick, optically thin flow tracks the outer critical curve (Figure 19b), and lensed emission fills the annulus . A measurement of the ring diameter therefore constrains , not , and the ring-to-shadow calibration that underlies EHT shadow inferences fails for such configurations. Similar double-ring structures were found for hairy black holes with two photon spheres [70]. (b) In the horizonless case (iv), rays with pass through the regular core and escape, which produces a sequence of bright inner rings and removes the central brightness depression altogether (Figure 19c,Figure 19f); a resolved central depression, as observed for M87* and Sgr A* [34, 39], thus disfavors this corner independently of any size argument. (c) The Lyapunov exponent of the shadow-bounding photon sphere, Eq. (7), increases from (Schwarzschild, and Block-1 for ) to – for , – and at (Table I). The flux ratio of successive photon-ring sub-images, , drops from 0.043 (Schwarzschild) to for and for ; the sub-rings of Block-2 black holes are
therefore much fainter and thinner, which is a distinctive target for future photon-ring interferometry [56]. Conversely, for one finds , i.e. slightly brighter sub-rings than in Schwarzschild.
Comparison with EHT-scale constraints
The EHT has measured the angular diameter of the M87* emission ring to be µas [36] and that of Sgr A* to be µas [39]. Translating these into bounds on the shadow size requires (i) independent mass-to-distance priors and (ii) a calibration factor between the emission-ring and critical-curve diameters, derived from GRMHD libraries of Kerr black holes [36, 40, 41]. Following the standard procedure [41, 42, 69], we adopt for M87* the fractional deviation , i.e. at , and for Sgr A* the Keck/VLTI-averaged bounds () and () [40, 42]. We then require to lie within these intervals.
Overlaying these bounds on the full plane (Figure 3b) yields considerably more restrictive information than the three one-dimensional cuts of Figure 7. (1) For every is consistent with both M87* and Sgr A* at , and for with M87* () and Sgr A* (). (2) Two diagonal bands at small are excluded by both objects: the first follows the decrease of along the single-photon-sphere branch [e.g. at , at , at for M87*], and the second lies just before the horizon bifurcation, inside the double-photon-sphere region. (3) Narrow re-entrant allowed strips exist just beyond the bifurcation (e.g. at ). (4) The upper bound of Sgr A*, , lies only 0.5% above the Schwarzschild value, so that the mild shadow enlargement for , (e.g. at and at ) is in tension at but allowed at . Over the BH part of the scanned plane, 82% of the area is consistent with both bounds and 7% is inconsistent with both M87* () and Sgr A* ().
These constraints should be interpreted with care. First, they treat as directly measured; our ray-traced images (Sec. IV E) show that the ratio between the beam-convolved ring diameter and varies from (Schwarzschild, Block-1) to for , so that the Kerr-calibrated conversion can bias by up to in the strong-coupling regime, and fails altogether in the double-photon-sphere and horizonless regions (§ 1.2 D). Second, spin is neglected; for the rotating Lee-Wick BH [22] the shadow size depends mildly on spin and inclination, which would broaden the allowed bands. Third, the bounds inherit the systematic uncertainties of the mass and distance priors. With these caveats, the shadow size provides a concrete, falsifiable target for horizon-scale tests of Lee-Wick gravity.
Thick Accretion Flow Imaging and Polarization
Motivated by the shock-cone/bow-shock structures obtained hydrodynamically in Ref. [29], and by the fact that EHT targets (M87*, Sgr A*) are surrounded by hot, optically thin, geometrically thick flows rather than thin Novikov-Thorne disks [58, 59], we deliberately construct the imaging pipeline using two thick-disk prescriptions and do not consider the thin-disk limit.
Polarized radiative transfer
Photon trajectories are obtained by backward ray tracing from the observer’s image plane, integrating the null geodesic equations in the metric (2) (and its rotating extension [22] when spin is included); the specific implementation used in this work is described in Sec. IV E. Along each ray, the invariant polarized radiative-transfer equation for the Stokes vector is integrated,
where is an affine parameter, is the synchrotron emissivity vector, and is the Mueller matrix encoding absorption, Faraday rotation, and Faraday conversion [64, 65]. The polarization vector is parallel-transported along the null geodesic using the invariant Walker-Penrose constant [71, 72], and the local emission/absorption coefficients are evaluated in the fluid rest frame and boosted to the observer frame via the local Doppler factor and the magnetic-field orientation on the sky.
The synchrotron emissivities depend on the electron number density , magnetic field , and (dimensionless) electron temperature , which are specified by the disk model.
RIAF-like phenomenological model
Following the radiatively inefficient accretion flow (RIAF) framework [58, 59, 60], we adopt power-law radial profiles anchored at a fixed reference radius (rather than at , so that the emission model is identical for all metrics and differences between images are purely due to the spacetime),
with , , [60], , G, and a Gaussian thickness envelope of angular half-width (fiducial value rad; we also show and in Sec. IV E). The envelope keeps the emission concentrated around the equatorial region while remaining a genuinely thick, quasi-spherical flow; since the flow is optically thin, only sets the overall normalization. The emissivity is the thermal-synchrotron fit of Ref. [73],
with , , , and the angle between the photon and the magnetic field in the fluid frame. This is how the observing frequency enters: Eq. (14) is evaluated at the emitted frequency with GHz, where is the redshift factor. With the above parameters at , so that the 230 GHz emission is concentrated within , as in EHT-like models. The fluid four-velocity is taken to be the ballistic, marginally bound “conical” flow used in the analytic thick-disk models [61, 62]: , , , and
with specific angular momentum (sub-critical in Schwarzschild, where capture requires ). In the few regions where the radicand is negative (pockets with in strong coupling, or the core of horizonless configurations) we use a
purely rotating flow with the same angular velocity ratio, capped at . This velocity field remains well defined even where no circular orbits exist (relevant for the strong Block-2 regime, where the ISCO/photon sphere structure can be substantially altered).
Analytic Hou disk model
As a complementary, more first-principles thick-disk prescription we formulate the analytic (“Hou”) disk model [61, 62], in which the accreting fluid is confined to surfaces of constant polar angle, , so that the continuity equation reduces to
with fixed at a reference radius (typically ). Projecting energy-momentum conservation along gives the first law
where, defining the proton-to-electron temperature ratio , the internal energy satisfies
The fluid four-velocity is taken either as purely radial free fall or as a ballistic combination of radial infall and angular motion (the ballistic-approximation accretion flow, valid where gravity dominates the near-horizon dynamics) [61]; a purely circular flow is excluded because requires a nonzero radial velocity to avoid a divergent magnetic field. Equation (15) is evaluated using the Lee-Wick determinant (unchanged from Schwarzschild since the angular part of the metric (2) is unmodified) and the geodesic appropriate to the Lee-Wick lapse function (3), so that all Lee-Wick corrections enter through in the geodesic/radial-velocity profile rather than through the disk model itself. In the present work the ballistic velocity field of this model is used in the numerical images (Sec. IV E), but the thermodynamic profiles are taken from the RIAF-like prescription above; a full implementation of the Hou thermodynamic model and its self-consistent magnetic field. modynamics and magnetic field is left for future work.

Figure 12. The same strong-coupling sequence (, ) and Schwarzschild, after convolving each ray-traced image with a Gaussian beam of FWHM 20 (5.3 for M87*), mimicking the angular resolution currently achievable with horizon-scale very-long-baseline interferometry.

Figure 13. Effect of the observer inclination at fixed . Increasing leaves the ring size essentially unchanged but produces a pronounced Doppler-boosted crescent. The images are free of polar-axis artifacts because each geodesic is integrated in its own orbital plane, which avoids the coordinate singularity of at the poles (Sec. IV E).
Expected imaging signatures
Both disk models produce, at 230 GHz, a central brightness depression (the shadow) surrounded by a bright, asymmetric photon ring whose diameter tracks from § 1.2, with higher-order (“”) sub-images nested just inside the direct-emission ring, as is generic for RIAF-like and Hou-model thick disks around compact objects [54, 62]. Guided by the hydrodynamic shock-cone results of Ref. [29], we expect:
In the Block-1 regime, images nearly indistinguishable from Schwarzschild: a circular, only mildly asymmetric photon ring, smooth linear-polarization ticks following the near-axisymmetric magnetic field threading the narrow, symmetric shock cone.
In the Block-2 regime, a smaller (or, close to the horizon-formation threshold, larger) photon ring tracking of Figure 2, with brightness asymmetries and depolarization patches correlated with the broad, asymmetric bow-shock/turbulent structure found in the GRHD simulations; the time-variable shock-cone dynamics (Figs. 12,17 of Ref. [29]) should manifest as enhanced intra-night image variability, analogous to the flux and morphology variability reported by the EHT for Sgr A* [39].
These expectations are tested quantitatively in Sec. IV E.
Numerical ray-traced images
Code and geodesic integration. All images in this paper were produced with our own Python code, written for this work; no public GRRT package was used. Because the metric is spherically symmetric, every null geodesic lies in a plane through the origin. For each pixel of the observer’s screen, with , we define the in-plane basis (the direction to the observer, at inclination ) and , where are
@page 11/19 starts_mid=1 ends_mid=1

Figure 12 (PDF p. 11). The same strong-coupling sequence (, ) and Schwarzschild, after convolving each ray-traced image with a Gaussian beam of FWHM 20 (5.3 for M87*), mimicking the angular resolution currently achievable with horizon-scale very-long-baseline interferometry.

Figure 13 (PDF p. 11). Effect of the observer inclination at fixed . Increasing leaves the ring size essentially unchanged but produces a pronounced Doppler-boosted crescent. The images are free of polar-axis artifacts because each geodesic is integrated in its own orbital plane, which avoids the coordinate singularity of at the poles (Sec. IV E).
modynamics and magnetic field is left for future work.
Expected imaging signatures
Both disk models produce, at 230 GHz, a central brightness depression (the shadow) surrounded by a bright, asymmetric photon ring whose diameter tracks from § 1.2, with higher-order (“”) sub-images nested just inside the direct-emission ring, as is generic for RIAF-like and Hou-model thick disks around compact objects [54, 62]. Guided by the hydrodynamic shock-cone results of Ref. [29], we expect:
In the Block-1 regime, images nearly indistinguishable from Schwarzschild: a circular, only mildly asymmetric photon ring, smooth linear-polarization ticks following the near-axisymmetric magnetic field threading the narrow, symmetric shock cone.
In the Block-2 regime, a smaller (or, close to the horizon-formation threshold, larger) photon ring tracking of Figure 2, with brightness asymmetries and depolarization patches correlated with the broad, asymmetric bow-shock/turbulent structure found in the GRHD simulations; the time-variable shock-cone dynamics (Figs. 12,17 of Ref. [29]) should manifest as enhanced intra-night image variability, analogous to the flux and morphology variability reported by the EHT for Sgr A* [39].
These expectations are tested quantitatively in Sec. IV E.
Numerical ray-traced images
Code and geodesic integration. All images in this paper were produced with our own Python code, written for this work; no public GRRT package was used. Because the metric is spherically symmetric, every null geodesic lies in a plane through the origin. For each pixel of the observer’s screen, with , we define the in-plane basis (the direction to the observer, at inclination ) and , where are the screen axes, and integrate
which follow from with . The position on the sky of the BH is , from which follow without any coordinate singularity; in particular, rays passing over the poles (which occur for at low inclination) are integrated without the pixel artifacts that a Hamiltonian integration produces there. Rays start at with fixed by the null condition and are integrated backward with a fourth-order Runge–Kutta scheme and step , , which resolves the winding near the photon sphere ( rad per step). Integration stops when (capture) or when the ray recedes beyond ; no ray reached the maximum of 9000 steps. and are tabulated on a logarithmic grid of points. The screen covers with pixels (0.1 per pixel). As a check, the critical curve of every image coincides with the circle of § 1.2 (cyan circles in Figure 8, Figure 11, 13), and the polarization transport was verified against the flat-space limit to machine precision.

Figure 8. Ray-traced 230 GHz images of the Lee-Wick black hole under thick-disk illumination for the weak-coupling sequence (with and observer inclination held fixed), compared with Schwarzschild (panel a). Images are computed with the code and emission model of Sec. IV E (thermal synchrotron, ballistic flow, toroidal field, , pixels). The cyan dotted circles mark the critical curves .

Figure 11. Ray-traced 230 GHz images of the Lee-Wick black hole under thick-disk illumination for the strong-coupling sequence (with and observer inclination held fixed), compared with Schwarzschild. Emission model and markings as in Figure 8. The bright ring visibly shrinks as increases, in direct correspondence with the shrinking shadow radius found analytically for this parameter sequence.
Radiative transfer. Along each ray we integrate the invariant optically thin transfer equation, , with computed exactly from the photon momentum and the ballistic four-velocity, and from Eq. (14) with the RIAF-like profiles of Sec. IV B. Emission is accumulated for . Absorption, Faraday rotation and conversion are neglected (optically thin limit), i.e. in Eq. (10). For the polarization, the electric vector in the fluid frame is (perpendicular to the photon direction and to the field ), with intrinsic fractional polarization . It is transported to the observer through the Walker-Penrose constants [71, 72]
which are conserved for any lapse in Eq. (2) ( follows from the Killing–Yano tensor and from the closed conformal Killing–Yano tensor ). On the screen the electric vector is , and each emission element contributes with EVPA measured east of north. We consider three field geometries in the fluid frame: toroidal (), mixed radial-toroidal (), and vertical (parallel to the spin axis); the field strength profile is the same in all cases.
Image-domain observables. To compare models quantitatively we compute, on images convolved with a 20 beam: the ring diameter (twice the median, over 180 position angles, of the radius of peak brightness), the brightness asymmetry along the ring, the net linear polarization , and the rotationally symmetric polarization mode [57]
over , with the position angle east of north. A purely radial EVPA pattern has and a purely azimuthal one .
Figure 8 shows the resulting images for the Block-1 sequence (, ): a dark shadow surrounded by a bright, asymmetric ring of higher-order images, with a pronounced Doppler-brightened crescent on the approaching side, essentially converged for as anticipated from Figure 2. The sharp-image ring diameters are for Schwarzschild and , respectively, i.e. identical within the pixel size.
Figure 9 shows the same sequence convolved with the 20 beam; the primary and higher-order images become blended. Figure 10 shows the corresponding intensity cuts along the - and -axes, which coincide with Schwarzschild in Block-1.

Figure 9. The same weak-coupling sequence (, ) and Schwarzschild, after convolving each ray-traced image with a circular Gaussian beam of FWHM , which corresponds to for M87* () and mimics the angular resolution currently achievable with horizon-scale very-long-baseline interferometry. The convolution blends the primary photon ring with the fainter higher-order sub-images.

Figure 10. Intensity cuts through the ray-traced images along the horizontal () and vertical () axes of the observer’s screen, for the weak-coupling sequence (; panels a,b) and the strong-coupling sequence (; panels c,d), at , compared with Schwarzschild (dashed). The peaks mark the photon ring; in Block-1 they coincide with Schwarzschild to within one pixel, while in Block-2 they are displaced outward by up to , following .
Figure 11 and its blurred counterpart Fig. 12 show the Block-2 sequence (, ): here the ring visibly shrinks as increases, in direct agreement with the trend of Figure 2(b) and Table I: the sharp-image diameters are , , and , i.e. , , and relative to Schwarzschild, and the beam-convolved ones , , versus .
Figure 13 and Figure 14 showcase the effect of the observer inclination at fixed : at the ring is nearly axisymmetric (), while at pronounced Doppler beaming produces a bright crescent on one side and a faint, thin arc on the other (), exactly the morphology anticipated qualitatively above and reported for Kerr and modified-gravity thick disks alike [62].

Figure 14. The same inclination sequence , , (fixed , ), after convolving each ray-traced image with a Gaussian beam of FWHM 20 as (5.3 for M87*), mimicking the angular resolution currently achievable with horizon-scale very-long-baseline interferometry. The Doppler-boosted crescent at high inclination remains clearly visible even after blurring.
Finally, Figure 15 shows the Stokes , , maps and the beam-convolved polarization fraction for the fiducial case , and Figure 16–Figure 17 show the polarization images (intensity with EVPA ticks colored by ) across the and sequences. The fractional polarization is suppressed in the bright crescent, where the EVPAs of the direct and lensed emission differ and are averaged within the beam. This depolarization is not imposed but emerges from the transport itself.

Figure 15. Stokes parameters , , and (normalized to the peak of ) and beam-convolved fractional linear polarization for the fiducial thick-disk image (, , toroidal field, ). Stokes is not shown: in the present optically thin model without Faraday conversion and with intrinsic circular emissivity neglected, .

Figure 16. Polarization images of the thick-disk emission for the weak-coupling sequence (, ) and Schwarzschild, for a toroidal field, showing the beam-convolved total intensity map with superimposed electric-vector position-angle (EVPA) ticks, colored by the local degree of linear polarization . Ticks are shown where .

Figure 17. Polarization images of the thick-disk emission for the strong-coupling sequence (, ) and Schwarzschild; same conventions as Figure 16. The degree of linear polarization is suppressed in the bright crescent, where differently oriented EVPAs from the direct image and the photon ring are averaged within the beam, and recovers away from it.
Quantitative comparison with Schwarzschild. Figure 18 and Table II compare, at fixed plasma model, the image observables of Lee-Wick and Schwarzschild black holes at . The main results are: (i) In Block-1 all observables agree with Schwarzschild to (, ) and in diameter, so that Block-1 is observationally indistinguishable at present. (ii) In Block-2 the beam-convolved ring diameter (toroidal field) changes by , , , and for , , , and (). (iii) The phase is remarkably robust: for all configurations with a single photon sphere it changes by for the toroidal and mixed fields and by for the vertical field, i.e. at low inclination traces the magnetic-field geometry rather than the metric. (iv) The amplitude is the most sensitive polarimetric discriminant: for the mixed field it increases from (Schwarzschild) to , , and ( to ) for , , at , and by for . We attribute this mainly to the larger Lyapunov exponent of Block-2 photon spheres (§ 1.2 D): the sub-images, whose EVPA pattern is rotated with respect to the direct image [74], are fainter, so that the beam-averaged polarization pattern is less scrambled. (v) In configurations with two unstable photon spheres and in horizonless configurations, rotates by and (mixed field), respectively, and decreases by and , because additional rings with differently oriented EVPAs contribute within the beam. (vi) At high inclination () the differences are larger in phase: for the mixed field changes from (Schwarzschild) to , , for , , (), and for the toroidal field increases from to –. These are the distinctive polarization signatures of the Lee-Wick geometry within the present model; they are differential signatures, comparable in size to the effects of changing the field geometry, so that a test requires a joint fit of plasma and metric parameters.

Figure 18. Quantitative comparison of beam-convolved (20 ) image observables between Schwarzschild and Lee-Wick black holes at the M87*-like inclination , for toroidal (blue), mixed radial-toroidal (red), and vertical (green) magnetic fields: (a) amplitude and (b) phase of the coefficient, (c) net linear polarization, and (d) ring diameter. The last two configurations, (1, 2.5) and (1, 1), have two unstable photon spheres and no horizon, respectively. Gold bands: M87* 2017 ranges of and [37, 38]; our model, which has no Faraday rotation or turbulent field component, overpredicts , so only the differences between metrics at fixed plasma model are meaningful.
Multiple photon spheres and horizonless configurations. Figure 19 illustrates the discussion of § 1.2 D. For , the dark central region has radius , while the brightest ring lies on the outer critical curve, , and the annulus between them is filled with lensed emission and a second, inner photon ring. After convolution, the image shows a ring with a strongly reduced central depression. For the horizonless configuration the image consists of the outer photon ring at and a set of bright inner rings formed by rays that cross the core; the beam-convolved image is centrally peaked. Both cases are therefore distinguishable from a Schwarzschild-like BH by morphology alone.
Dependence on the disk thickness. The fiducial thickness parameter is rad. Figure 20 compares , , and . The position of the sharp photon ring is unaffected ( and for Schwarzschild, – for ), as expected since it is fixed by the critical curve. In contrast, the total flux increases by a factor from to , the brightness asymmetry decreases from to –, and the beam-convolved ring diameter increases from to (Schwarzschild) and from to , i.e. by –. A change of disk thickness can therefore mimic in the blurred image a shift comparable to the Block-2 deviations, which is another reason why the shadow size should be inferred jointly with the flow geometry.

Figure 20. Dependence of the 230 GHz images on the disk thickness parameter (0.15, 0.3, 0.6 rad) for Schwarzschild (top) and , (bottom), at (toroidal field). Panels (d) and (h) show horizontal intensity cuts normalized to the maximum of each image.
Connection with shock-cone/bow-shock structures. To make the link with the GRHD morphology of Ref. [29] explicit, we superpose on the fiducial flow a parametrized density enhancement mimicking a shock cone (Figure 21). Its density contrast, , corresponds to a strong adiabatic shock, and the opening width distinguishes the narrow cone of Block-1 () from the broad bow shock of Block-2 (); the azimuth of the axis is varied relative to the observer. The results answer the three questions raised. (1) Size and geometry. The size of the shock structure does not set the scale of the ring: the sharp-image diameter varies by (– for the cone, – for the bow shock), because it is fixed by the critical curve. The beam-convolved diameter, however, depends on where the extra emission sits and varies from to (cone) and from to (bow shock). (2) Brightness. The shock modulates the flux strongly: relative to the shock-free flow the 230 GHz flux increases by factors – for the narrow cone and – for the broad bow shock, with the largest enhancement when the dense region lies on the far side of the BH (), where it is strongly lensed into the ring. (3) Viewing azimuth. Since the structure breaks axisymmetry, all image properties depend on the observer azimuth: the brightness asymmetry along the ring varies between and (cone

Figure 21. Beam-convolved 230 GHz images (, toroidal field) including a parametrized shock-density enhancement , with and the angle between the emission point and the shock axis lying in the equatorial plane at azimuth relative to the observer. Top: narrow downstream shock cone () on the Block-1 metric (4, 1). Bottom: broad bow-shock-like structure () on the Block-2 metric (1, 1.3). The first column has no shock; the other columns differ only in the azimuth of the shock axis.
| Schw. | 11.34 | 0.301 (0.499) | — | |
| (4,1) | 11.34 | 0.301 (0.499) | 0% | |
| (2,1) | 11.37 | 0.305 (0.500) | +1% | |
| (1.5,1.5) | 11.74 | 0.362 (0.505) | +20% | |
| (1.1,1) | 12.96 | 0.420 (0.522) | +40% | |
| (1,1.3) | 12.46 | 0.446 (0.516) | +48% | |
| (1,1.5) | 11.57 | 0.455 (0.504) | +51% | |
| (1,2.0) | 8.99 | 0.364 (0.471) | +21% | |
| (1,2.5) | 7.80 | 0.261 (0.445) | ||
| (1,1.0) | 3.94 | 0.210 (0.386) |
Table 2. TABLE II. Beam-convolved image observables at for the mixed radial-toroidal field (toroidal-field values of in parentheses). is the ring diameter in units of ; denotes the relative change with respect to Schwarzschild.
and between 0.74 and 0.91 (bow shock), and the position angle of the brightest spot shifts by up to 34 for the bow shock. The azimuthal dependence is therefore much stronger for the broad Block-2 structure than for the narrow Block-1 cone, which provides an imaging counterpart of the dynamical distinction found in Ref. [29]. For a nonstationary shock cone, the same mechanism converts its oscillation into flux and morphology variability.
Observational Strategy and Discussion
The combination of (i) the shadow-radius map of § 1.2, (ii) the polarized thick-disk imaging pipeline of Sec. IV, and (iii) the QPO frequency ratios already computed hydrodynamically in Ref. [29] constitutes a multi-messenger, horizon-scale test of Lee-Wick gravity: (a) EHT-type shadow measurements of M87* and Sgr A* directly bound (§ 1.2 E); (b) polarimetric EHT images [37, 38] constrain the near-horizon magnetic-field topology and, through the disk models of Sec. IV, indirectly constrain via the induced ring asymmetry and depolarization; and (c) X-ray timing missions such as eXTP [75] and STROBEX [76] can test the predicted LFQPO-to-HFQPO transition, providing an independent, accretion-scale handle on the same parameters. Because these three observables respond differently to (exponential damping) and (oscillatory modulation), their combination in principle breaks the degeneracy between the two Lee-Wick couplings that any single observable alone would leave partially unresolved.
Our results allow a more careful assessment of each channel. (a) With present EHT precision, the shadow size already excludes two strong-coupling bands (Figure 3), but the bound depends on the mass-distance priors and on a Kerr-calibrated ring-to-shadow factor which, as shown in Sec. IV E, changes by in Block-2, and by a similar amount with the disk thickness; a self-consistent analysis would require fitting the Lee-Wick images directly to the interferometric data. Block-1 () remains unconstrained. (b) The polarimetric amplitude changes by up to in Block-2, whereas at low inclination is set by the field geometry. Since the absolute of M87* (0.04–0.07 [38]) is strongly reduced by Faraday rotation and turbulence, which are absent from our model, a polarimetric test requires a joint fit of plasma and metric parameters; polarization is thus a complementary, not a standalone, probe. (c) Morphology rules out the horizonless corner, since a central brightness depression is observed for M87* and Sgr A*. (d) In the longer term, the photon-ring Lyapunov exponent (Table I), accessible to space-VLBI photon-ring interferometry [56], offers a nearly plasma-independent test, with exceeding by , and up to a factor , in Block-2.
Conclusions
We have derived the photon sphere condition and shadow radius of the non-rotating quantum Lee-Wick black hole and shown that the shadow size converges to the Schwarzschild value in the weak-coupling (Block-1) regime but becomes strongly, non-monotonically modified in the strong-coupling (Block-2) regime, in a manner consistent with the shock-cone/bow-shock morphology found in earlier BHL-accretion hydrodynamic simulations. We have mapped the horizon and photon-sphere structure of the full plane, showing that horizonless configurations are regular ultracompact objects with a critical curve but no dark shadow, and that in configurations with two unstable photon spheres the shadow is bounded by the inner critical curve while the outer one produces an additional bright ring. We have further developed the polarized radiative-transfer formalism, using both a phenomenological RIAF-like model and the analytic Hou disk model, needed to image a geometrically thick accretion flow around this spacetime, deliberately bypassing the thin-disk approximation in favor of a description appropriate to horizon-scale, EHT-like targets. Building on this formalism, we developed a dedicated ray-tracing and polarized-transfer code (planar geodesic integration, thermal synchrotron emission at 230 GHz, exact redshift for a ballistic flow, Walker-Penrose polarization transport) and produced shadow, blurred, intensity-profile, and polarization images across the Block-1 and Block-2 sequences and as a function of observer inclination. The images confirm quantitatively the trends of § 1.2, show that Gaussian-beam convolution at the current EHT resolution blends the primary and higher-order images, and demonstrate that the observer inclination alone generates a strong Doppler crescent even in the absence of spin.
Our main quantitative results are: (i) Block-1 images are indistinguishable from Schwarzschild to in all observables considered; (ii) in Block-2 the ring diameter deviates by up to , the photon-ring Lyapunov exponent increases from to –, and the amplitude increases by up to for a mixed field, while at low inclination is insensitive to ; (iii) the EHT shadow bounds for M87* and Sgr A* exclude two diagonal strong-coupling bands of the plane, while is allowed for all ; (iv) shock-cone and bow-shock structures modulate the flux by factors up to and make the brightness distribution strongly dependent on the viewing azimuth, while leaving the critical-curve size unchanged; and (v) the disk thickness affects the beam-convolved diameter by –17% and must be fitted jointly with the metric. A dedicated, full covariant polarized-transfer campaign including absorption, Faraday effects and GRMHD-based plasma models, combined with the GRHD-derived shock-cone morphology, is the natural next step toward a complete, quantitative multi-wavelength test of Lee-Wick quantum gravity at the horizon scale.
References
- [1]T. D. Lee and G. C. Wick, Nucl. Phys. B 9, 209 (1969).
- [2]T. D. Lee and G. C. Wick, Phys. Rev. D 2, 1033 (1970).
- [3]R. E. Cutkowsky, P. V. Landshoff, D. I. Olive, and J. C. Polkinghorne, Nucl. Phys. B 12, 281 (1969).
- [4]L. Modesto, Nucl. Phys. B 909, 584 (2016), arXiv:1602.02421.DOI
- [5]L. Modesto and I. L. Shapiro, Phys. Lett. B 755, 279 (2016), arXiv:1512.07600.DOI
- [6]M. Asorey, J. L. López, and I. L. Shapiro, Int. J. Mod. Phys. A 12, 5711 (1997), arXiv:hep-th/9610006.DOI
- [7]D. Anselmi and M. Piva, Phys. Rev. D 96, 045009 (2017), arXiv:1703.05563.DOI
- [8]J. Liu, L. Modesto, and G. Calcagni, JHEP 02, 140.
- [9]A. Accioly, B. L. Giacchini, and I. L. Shapiro, Phys. Rev. D 96, 104004 (2017), arXiv:1610.05260.DOI
- [10]B. L. Giacchini, Phys. Lett. B 766, 306 (2017), arXiv:1609.05432.DOI
- [11]I. L. Shapiro, in Handbook of Quantum Gravity, edited by C. Bambi, L. Modesto, and I. Shapiro (Springer Nature Singapore, 2023).
- [12]C. Bambi, L. Modesto, and Y. Wang, Phys. Lett. B 764, 306 (2017), arXiv:1611.03650.DOI
- [13]B. L. Giacchini and T. de Paula Netto, Eur. Phys. J. C 79, 217 (2019), arXiv:1806.05664.DOI
- [14]N. Burzillà, B. L. Giacchini, T. de Paula Netto, and L. Modesto, JCAP 11, 067, arXiv:2308.12810.
- [15]B. L. Giacchini and T. de Paula Netto, in Regular Black Holes: Towards a New Paradigm of Gravitational Collapse, edited by C. Bambi (Springer Nature Singapore, 2023) pp. 89–130.
- [16]T. de Paula Netto, B. L. Giacchini, N. Burzillà, and L. Modesto, Nucl. Phys. B 1007, 116674 (2024).
- [17]P. Nicolini, A. Smailagic, and E. Spallucci, Phys. Lett. B 632, 547 (2006), arXiv:gr-qc/0510112.DOI
- [18]P. Nicolini and E. Spallucci, Class. Quantum Grav. 27, 015010 (2010).
- [19]S.-S. Zhao and Y. Xie, Phys. Lett. B 774, 357 (2017).
- [20]L. Buoninfante and B. L. Giacchini, Phys. Rev. D 102, 024020 (2020), arXiv:2005.05355.DOI
- [21]H.-Y. Lin and X.-M. Deng, Eur. Phys. J. Plus 137, 176 (2022).
- [22]D. V. Singh, S. Upadhyay, and M. S. Ali, Int. J. Mod. Phys. A 37, 2250049 (2022), arXiv:2206.05057.DOI
- [23]E. T. Newman and A. I. Janis, J. Math. Phys. 6, 915 (1965).
- [24]M. Azreg-Aïnou, Phys. Rev. D 90, 064041 (2014).
- [25]S. P. Drake and P. Szekeres, Gen. Relativ. Gravit. 32, 445 (2000).
- [26]D. Batic, D. Dutykh, and B. L. Giacchini, Phys. Rev. D 110, 084032 (2024), arXiv:2410.13245.DOI
- [27]B. L. Giacchini and I. Kolář, arXiv:2406.00997 [gr-qc] (2024), arXiv:2406.00997.DOI
- [28]B. Holdom, Phys. Rev. D 66, 084010 (2002), arXiv:hep-th/0206219.DOI
- [29]O. Dommnez, G. Mustafa, M. Yousaf, F. Javed, I. Saidov, and F. Atamuratov, Chin. Phys. C 50, 085102 (2026), arXiv:2512.17358.
- [30]W. Kluzniak and M. A. Abramowicz, Astrophys. Space Sci. 300, 143 (2005).
- [31]R. A. Remillard and J. E. McClintock, Annu. Rev. Astron. Astrophys. 44, 49 (2006).
- [32]D. R. Pasham et al., Science 363, 531 (2019).
- [33]P. A. Charles and M. J. Coe, arXiv:astro-ph/0308020 (2003), arXiv:astro-ph/0308020.DOI
- [34]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 875, L1 (2019).
- [35]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 875, L5 (2019).
- [36]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 875, L6 (2019).
- [37]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 910, L12 (2021).
- [38]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 910, L13 (2021), arXiv:2105.01173.
- [39]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 930, L12 (2022).
- [40]Event Horizon Telescope Collaboration, Astrophys. J. Lett. 930, L17 (2022).
- [41]D. Psaltis et al., Phys. Rev. Lett. 125, 141104 (2020).
- [42]S. Vagnozzi et al., Class. Quantum Grav. 40, 165007 (2023), arXiv:2205.07787.DOI
- [43]V. Perlick and O. Y. Tsupko, Phys. Rep. 947, 1 (2022).
- [44]P. V. P. Cunha and C. A. R. Herdeiro, Gen. Relativ. Gravit. 50, 42 (2018).
- [45]J. L. Synge, Mon. Not. R. Astron. Soc. 131, 463 (1966).
- [46]J. M. Bardeen, in Black Holes (Les Houches), edited by C. DeWitt and B. S. DeWitt (Gordon and Breach, New York, 1973).
- [47]J.-P. Luminet, Astron. Astrophys. 75, 228 (1979).
- [48]H. Falcke, F. Melia, and E. Agol, Astrophys. J. 528, L13 (2000), arXiv:astro-ph/9912263.DOI
- [49]C. Bambi and K. Freese, Phys. Rev. D 79, 043002 (2009), arXiv:0812.1328.DOI
- [50]C. Bambi and K. Maeda, Phys. Rev. D 80, 024042 (2009), arXiv:0904.3575.
- [51]R. Kumar and S. G. Ghosh, Astrophys. J. 892, 78 (2020).
- [52]F. Atamuratov, B. Ahmedov, and A. Abdujabbarov, Phys. Rev. D 92, 084005 (2015).
- [53]A. Abdujabbarov, F. Atamuratov, Y. Kucukakca, B. Ahmedov, and U. Camci, Astrophys. Space Sci. 344, 429 (2013).
- [54]S. E. Gralla, D. E. Holz, and R. M. Wald, Phys. Rev. D 100, 024018 (2019).
- [55]S. E. Gralla and A. Lupsasca, Phys. Rev. D 101, 044031 (2020).
- [56]M. D. Johnson et al., Sci. Adv. 6, eaaz1310 (2020).
- [57]D. C. M. Palumbo, G. N. Wong, and B. S. Prather, Astrophys. J. 894, 156 (2020), arXiv:2004.01751.DOI
- [58]R. Narayan and I. Yi, Astrophys. J. 428, L13 (1994).
- [59]F. Yuan and R. Narayan, Annu. Rev. Astron. Astrophys. 52, 529 (2014).
- [60]A. E. Broderick, V. L. Fish, S. S. Doeleman, and A. Loeb, Astrophys. J. 735, 110 (2011).
- [61]Y. Hou, Z. Zhang, M. Guo, and B. Chen, arXiv:2309.13304 [gr-qc] (2023), arXiv:2309.13304.arxiv.org/abs/2309.13304
- [62]Z. Zhang, Y. Hou, M. Guo, and B. Chen, arXiv:2401.14794 [astro-ph.HE] (2024), arXiv:2401.14794.arxiv.org/abs/2401.14794
- [63]F. H. Vincent, T. Paumard, E. Gourgoulhon, and G. Perrin, Class. Quantum Grav. 28, 225011 (2011).
- [64]M. Mościbrodzka and C. F. Gammie, Mon. Not. R. Astron. Soc. 475, 43 (2018).
- [65]P. Dexter, Mon. Not. R. Astron. Soc. 462, 115 (2016)
- [66]H.-Y. Pu, K. Y. Yun, Z. Younsi, and S.-J. Yoon, Astrophys. J. 820, 105 (2016).
- [67]P. V. P. Cunha, E. Berti, and C. A. R. Herdeiro, Phys. Rev. Lett. 119, 251102 (2017), arXiv:1708.04211.DOI
- [68]V. Cardoso, A. S. Miranda, E. Berti, H. Witek, and V. T. Zanchin, Phys. Rev. D 79, 064016 (2009), arXiv:0812.1806.DOI
- [69]P. Kocherlakota et al. (Event Horizon Telescope), Phys. Rev. D 103, 104047 (2021), arXiv:2105.09343.DOI
- [70]Q. Gan, P. Wang, H. Wu, and H. Yang, Phys. Rev. D 104, 044010 (2021), arXiv:2105.11770.DOI
- [71]M. Walker and R. Penrose, Commun. Math. Phys. 18, 265 (1970).
- [72]P. A. Connors, T. Piran, and R. F. Stark, Astrophys. J. 235, 224 (1980).
- [73]P. K. Leung, C. F. Gammie, and S. C. Noble, Astrophys. J. 737, 21 (2011).
- [74]E. Himwich, M. D. Johnson, A. Lupsasca, and A. Strominger, Phys. Rev. D 101, 084020 (2020), arXiv:2001.08750.DOI
- [75]S. N. Zhang, A. Santangelo, Y. Xu, et al., Sci. China Phys. Mech. Astron. 68, 19502 (2025).
- [76]C. S. Froning et al., in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series (2024).