Dual Galaxy Systems: AGN Merger Insights

• Dual Galaxy Systems (DGS) are physically associated galaxy pairs in merger interactions, often hosting dual accreting supermassive black holes.
• Studies show that DGSs have low simultaneous visibility (around 1–3%) due to rapid variability, separation effects, and survey selection biases.
• Robust identification of DGSs requires multiwavelength, high-resolution imaging and spectroscopy to overcome projection effects and resolution challenges.

Dual Galaxy System (DGS) denotes a physically associated galaxy pair in an interaction or merger state; in the dual-AGN literature, the term is commonly specialized to a merging pair in which each galaxy hosts an actively accreting supermassive black hole (SMBH), so that the galaxy-scale system is the host manifestation of a dual AGN. In this usage, the relevant dynamical regime is kiloparsec-scale, prior to the parsec-scale bound SMBH binary that forms after galactic coalescence. Across simulations and observations, DGSs appear as short-lived, merger-driven configurations whose detectability depends strongly on separation, obscuration, temporal variability, and the selection function of the observing band (Rosas-Guevara et al., 2018).

1. Definition, scope, and physical regime

In the EAGLE analysis, a dual AGN is defined operationally as two active black holes with three-dimensional separation $r<30\ \mathrm{kpc}$, excluding $r<1\ \mathrm{kpc}$ where the pair is not resolved; a “visible” dual AGN further requires both nuclei to exceed a hard X-ray threshold of $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$ in the 2–10 keV band, with “active” black holes defined by Eddington ratio $\geq 0.01$ (Rosas-Guevara et al., 2018). Within that framework, a DGS is the galaxy-scale expression of the dual-AGN phase. By contrast, the “binary BH system” denotes the later parsec-scale, gravitationally bound SMBH pair that forms after the galaxies coalesce and is smaller-scale than the DGS phase (Rosas-Guevara et al., 2018).

A related but distinct observational construct is the double nuclei galaxy (DNG), defined as a system with two or more well-separated central light peaks within one host envelope. When at least two of those nuclei are spectroscopically confirmed as AGN, the system is classified as a dual AGN or multiple AGN. This usage emphasizes merger remnants or very close interactions embedded within a common stellar envelope, rather than wider galaxy pairs selected purely by nearest-neighbor criteria (Bhattacharya et al., 2020).

The literature represented here therefore places DGSs across several connected regimes: dwarf–dwarf interacting binaries with common gaseous envelopes, dumb-bell systems in common halos, kpc-scale dual AGN/quasars in ongoing mergers, and merger remnants with multiple nuclei. This suggests that “DGS” functions as a scale-dependent interaction category whose most restrictive meaning is the dual-AGN stage, but whose broader observational usage can extend to any dynamically verified binary galaxy system with strong morphological or kinematic evidence of coupling (Makarov et al., 2013).

2. Incidence, demographics, and merger-driven triggering

The most explicit cosmological estimate in the supplied literature comes from the EAGLE Ref-L100N1504 simulation, a $(100\ \mathrm{cMpc})^3$ volume with dark-matter and baryonic particle masses of $9.7\times10^6\ M_\odot$ and $1.81\times10^6\ M_\odot$, respectively, and a Plummer-equivalent softening of $2.66\ \mathrm{ckpc}$ capped at $0.70\ \mathrm{pkpc}$ (Rosas-Guevara et al., 2018). At $z\simeq 0.8{-}1$, only about $r<1\ \mathrm{kpc}$0 of AGN are members of a visible dual system when both nuclei satisfy the hard X-ray threshold, and allowing only one AGN above threshold raises the fraction to about $r<1\ \mathrm{kpc}$1 (Rosas-Guevara et al., 2018).

EAGLE further quantifies detectability with

$r<1\ \mathrm{kpc}$2

where $r<1\ \mathrm{kpc}$3 is the time spent at separations below $r<1\ \mathrm{kpc}$4 and $r<1\ \mathrm{kpc}$5 is the subset during which both AGN are simultaneously above the visibility threshold (Rosas-Guevara et al., 2018). The average simultaneously visible fraction is about $r<1\ \mathrm{kpc}$6; $r<1\ \mathrm{kpc}$7 of dual systems have $r<1\ \mathrm{kpc}$8, only about $r<1\ \mathrm{kpc}$9 have $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$0, and maximum values in the sample reach about $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$1 (Rosas-Guevara et al., 2018). This low coincidence rate is attributed to rapid hard X-ray variability on Myr time-scales, which strongly suppresses simultaneous observability even when both SMBHs are episodically fueled (Rosas-Guevara et al., 2018).

The separation dependence is also highly structured. The visible dual-AGN distribution peaks sharply at $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$2, interpreted as the post-first-passage regime in which tidal torques, shocks, and dynamical friction accelerate orbital decay and funnel gas toward both nuclei (Rosas-Guevara et al., 2018). Host-galaxy context is correspondingly merger-dominated: about $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$3 of visible dual AGN have hosts undergoing or having undergone a merger in the previous $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$4 with stellar mass ratio $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$5 (Rosas-Guevara et al., 2018). Visible duals preferentially occupy more massive, gas-rich hosts, with median stellar mass around $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$6, about $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$7 dex above the overall AGN-host median; visible systems at $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$8 have median black-hole mass $L_{\mathrm{hX}}\geq 10^{42}\ \mathrm{erg\,s^{-1}}$9, and about $\geq 0.01$0 exceed $\geq 0.01$1 (Rosas-Guevara et al., 2018).

The redshift dependence is monotonic in the simulation: the visible dual fraction rises from about $\geq 0.01$2 at $\geq 0.01$3 to about $\geq 0.01$4 by $\geq 0.01$5 (Rosas-Guevara et al., 2018). A plausible implication is that DGS identification becomes increasingly relevant to high-$\geq 0.01$6 SMBH growth, even though the intrinsically low simultaneous-visibility probability keeps observed fractions small.

3. Observational diagnostics, survey strategies, and selection pitfalls

Hard X-ray selection is treated as the most direct tracer of accreting SMBHs through obscuration and is the band that most closely matches the EAGLE visibility criterion (Rosas-Guevara et al., 2018). Optical and infrared methods can identify candidates through double-peaked narrow lines or spatially resolved AGN emission, but are affected by extinction, projection, and fiber-collision limits at small separations (Rosas-Guevara et al., 2018). Radio imaging provides an independent route when compact cores are detected in both galaxies, and multiwavelength confirmation remains the standard for robust classification (Benítez et al., 2022).

Two large-scale search strategies illustrate current methodology. The first is image-first detection of double nuclei. GOTHIC uses SDSS DR16 $\geq 0.01$7 cutouts of $\geq 0.01$8, applies normalization and smoothing, uses a Canny edge detector to delineate the host envelope, derives a threshold from a Sérsic-like light-profile fit, and then performs iterated hill climbing on the thresholded intensity field to locate local maxima; a cross-band $\geq 0.01$9 consistency rule distinguishes “pure doubles” from “impure doubles” (Bhattacharya et al., 2020). On a 94-object benchmark, GOTHIC achieved 47 true positives, 0 false negatives, 40 true negatives, and 7 false positives; in a blind run on 1,000,000 SDSS DR16 objects it produced 104,412 DNG candidates, which after staged filtering and manual verification yielded 681 confirmed nuclei pairs and 159 dual AGN, including 2 triple AGN (Bhattacharya et al., 2020).

The second is astrometric pre-selection of quasar pairs. In the Gaia-based series culminating in 16 new dual quasars, neighboring Gaia sources with proper motion and parallax consistent with zero are selected around known quasars, and DESI DR1 spectroscopy is then used to classify candidates (Deng et al., 3 Mar 2026). Dual quasars are defined by a line-of-sight velocity difference threshold $(100\ \mathrm{cMpc})^3$0, with

$(100\ \mathrm{cMpc})^3$1

Among 52 newly classified pairs, 16 were confirmed dual quasars and 36 were projected quasar pairs; the confirmed duals span $(100\ \mathrm{cMpc})^3$2 with median redshift about 1.46 (Deng et al., 3 Mar 2026).

A central controversy concerns the long-standing use of double-peaked narrow emission lines (DPNELs) as DGS indicators. In the DGS interpretation, the red and blue narrow components should arise from independent narrow-line regions in two different galaxies, so flux ratios from the two components should not be intrinsically correlated (XueGuang, 17 Jul 2025). The test proposed in 2025 defines

$(100\ \mathrm{cMpc})^3$3

For 1618 SDSS DPN galaxies, the study finds

$(100\ \mathrm{cMpc})^3$4

with 1RMS scatter $(100\ \mathrm{cMpc})^3$5, Spearman $(100\ \mathrm{cMpc})^3$6, and $(100\ \mathrm{cMpc})^3$7; by contrast, control samples of real SDSS galaxy pairs show no significant analogous correlations (XueGuang, 17 Jul 2025). In an oversimplified simulation, reproducing the observed correlation requires at least $(100\ \mathrm{cMpc})^3$8 of the DPN galaxies to have double peaks not related to expected DGSs, and in AGN-dominated subsamples the inferred non-DGS fraction rises to about $(100\ \mathrm{cMpc})^3$9 (XueGuang, 17 Jul 2025). The practical conclusion is that DPNELs alone are unreliable DGS indicators, and spatially resolved spectroscopy plus multiwavelength confirmation are required for robust identification (XueGuang, 17 Jul 2025).

4. Confirmed dual-AGN and dual-quasar systems

Representative confirmed systems span low-redshift Seyfert pairs, cosmic-noon dual quasars, and wider-separation quasar pairs selected from survey data.

System	Separation and redshift	Defining evidence
IRAS 05589+2828 + 2MASX J06021107+2828382	$9.7\times10^6\ M_\odot$0 projected at $9.7\times10^6\ M_\odot$1	Optical, X-ray, and radio confirmation; ionized bridge
SDSS J0749+2255	$9.7\times10^6\ M_\odot$2 at $9.7\times10^6\ M_\odot$3	Two distinct hosts, tidal features, dual-quasar spectra
Gaia/DESI dual-quasar sample	$9.7\times10^6\ M_\odot$4, $9.7\times10^6\ M_\odot$5	16 spectroscopically confirmed dual quasars

The nearby pair IRAS 05589+2828 and 2MASX J06021107+2828382 is a well-characterized Seyfert 1/Seyfert 2 DGS with projected separation $9.7\times10^6\ M_\odot$6 and redshifts $9.7\times10^6\ M_\odot$7 and $9.7\times10^6\ M_\odot$8 (Benítez et al., 2022). Optical spectroscopy places both nuclei in the AGN/Seyfert region of BPT space, while the inter-nuclear bridge falls in a LINER-like domain; Copernico/AFOSC spectra reveal faint extended H$9.7\times10^6\ M_\odot$9 emission over about $1.81\times10^6\ M_\odot$0, and OAN-SPM spectra show extended [O III] across the bridge, supporting an ongoing wet merger (Benítez et al., 2022). In X-rays, IRAS 05589+2828 behaves as a Type 1 AGN with variability, whereas the companion is hard and heavily obscured, with hardness ratio

$1.81\times10^6\ M_\odot$1

consistent with a Type 2 AGN (Benítez et al., 2022). VLA 22 GHz imaging detects both nuclei as compact radio sources with spectral indices $1.81\times10^6\ M_\odot$2 and $1.81\times10^6\ M_\odot$3, confirming dual AGN activity in the radio band (Benítez et al., 2022). Black-hole masses are reported as $1.81\times10^6\ M_\odot$4 and $1.81\times10^6\ M_\odot$5, with Eddington ratios around $1.81\times10^6\ M_\odot$6 (Benítez et al., 2022).

At much higher redshift, SDSS J0749+2255 is an unambiguous dual quasar in a disk–disk merger at cosmic noon, with two type 1 nuclei separated by $1.81\times10^6\ M_\odot$7 at $1.81\times10^6\ M_\odot$8 and a line-of-sight velocity offset of about $1.81\times10^6\ M_\odot$9 (Chen et al., 2022). HST and Keck AO imaging detect two extended hosts beneath the quasar PSFs and reveal low-surface-brightness tidal features, while strong-lens models fail and multi-band flux ratios are chromatic, disfavoring lensing (Chen et al., 2022). Keck AO structural fits support disk dominance for both hosts, with $2.66\ \mathrm{ckpc}$0 and $2.66\ \mathrm{ckpc}$1, and stellar masses $2.66\ \mathrm{ckpc}$2 and $2.66\ \mathrm{ckpc}$3 (Chen et al., 2022). Balmer-based SMBH masses are approximately $2.66\ \mathrm{ckpc}$4 and $2.66\ \mathrm{ckpc}$5, bolometric luminosities are about $2.66\ \mathrm{ckpc}$6 and $2.66\ \mathrm{ckpc}$7, and the estimated SMBH dynamical-friction timescale to a bound binary is $2.66\ \mathrm{ckpc}$8 (Chen et al., 2022).

The Gaia-selected DESI DR1 sample extends the confirmed dual-quasar population to wider kpc-scale separations. The 16 newly confirmed systems occupy projected separations of roughly $2.66\ \mathrm{ckpc}$9 across $0.70\ \mathrm{pkpc}$0 (Deng et al., 3 Mar 2026). One case, J0023+0417, has nearly identical spectra and a red source near the midpoint and is therefore treated as a high-confidence strong-lensing candidate rather than a secure physical DGS, illustrating the persistent need to distinguish true duals from lenses even after spectroscopic pairing (Deng et al., 3 Mar 2026).

5. Dwarf systems, dumb-bell galaxies, and radio-loud duals

Not all DGSs are defined by dual AGN. The Lynx system LSB J0911+4238 + SDSS J091108.40+423922.1 is a dwarf–dwarf DGS composed of two late-type dwarf irregular galaxies linked by both morphology and H I kinematics (Makarov et al., 2013). The projected angular separation is $0.70\ \mathrm{pkpc}$1, corresponding to $0.70\ \mathrm{pkpc}$2 at the adopted distance of about $0.70\ \mathrm{pkpc}$3; heliocentric H I velocities are $0.70\ \mathrm{pkpc}$4 and $0.70\ \mathrm{pkpc}$5, giving $0.70\ \mathrm{pkpc}$6 (Makarov et al., 2013). GMRT maps show a continuous H I bridge and a smooth velocity gradient connecting the pair, while optical imaging shows the fainter member as a low-surface-brightness tidal stream-like structure in the halo of the brighter dwarf (Makarov et al., 2013). The orbital mass is estimated from

$0.70\ \mathrm{pkpc}$7

to be about $0.70\ \mathrm{pkpc}$8, with orbital mass-to-light ratio about 49 and crossing time

$0.70\ \mathrm{pkpc}$9

indicating a dynamically bound pair on the verge of merging (Makarov et al., 2013).

A different extreme is the “Twin Radio Galaxy” TRG J104454+354055, a dumb-bell DGS at $z\simeq 0.8{-}1$0 in which each member of a gravitationally bound pair hosts extended bipolar radio jets (Gopal-Krishna et al., 2022). The two massive ellipticals, at redshifts 0.16173 and 0.16278, are separated by about $z\simeq 0.8{-}1$1 with line-of-sight velocity difference about $z\simeq 0.8{-}1$2 (Gopal-Krishna et al., 2022). Both are low-excitation radio galaxies, and the pair is only the third known case of such twin-radio morphology (Gopal-Krishna et al., 2022). uGMRT imaging shows FR I, edge-darkened, wiggling bipolar jets on scales of about 221 kpc and 143 kpc for the two hosts; the overall source extends to at least about $z\simeq 0.8{-}1$3 (Gopal-Krishna et al., 2022). Unlike 3C 75, the jets do not show large-scale C-shaped distortion from an external crosswind, making the source a comparatively clean laboratory for sideways jet–jet interaction, orbital-motion-induced wiggles, and jet stability within a common halo (Gopal-Krishna et al., 2022).

These non-AGN or radio-specialized cases broaden the DGS concept beyond the dual-X-ray-selected regime. They show that DGS identification can rest on common gaseous envelopes, continuous velocity fields, dumb-bell stellar morphologies, or paired radio jets, not only on simultaneous luminous accretion.

6. Role in SMBH–galaxy co-evolution and outstanding problems

Across the supplied literature, DGSs are repeatedly treated as transition states linking galaxy mergers, synchronized or quasi-synchronized SMBH fueling, and the later formation of bound SMBH binaries. Kpc-scale dual quasars are described as direct precursors of binary SMBHs and therefore as relevant waypoints toward low-frequency gravitational-wave sources targeted by pulsar timing arrays and anticipated by LISA (Deng et al., 3 Mar 2026). In SDSS J0749+2255, the observation that both SMBHs already lie broadly on local $z\simeq 0.8{-}1$4–$z\simeq 0.8{-}1$5 relations despite the apparent lack of classical bulges has been taken to suggest that at least some SMBHs may assemble substantial mass before host bulge assembly (Chen et al., 2022).

Host-galaxy demographics also point to merger-linked baryonic transformation. In the GOTHIC sample, DNGs lie preferentially in the redder part of the color–magnitude diagram, and dual/triple AGN are especially concentrated at the red end; the study interprets this as evidence that star formation is quenched as nuclei approach and AGN fraction increases (Bhattacharya et al., 2020). In the EAGLE picture, however, observability is limited less by the rarity of merger-driven fueling itself than by the small overlap time during which both AGN exceed a fixed luminosity threshold (Rosas-Guevara et al., 2018). Taken together, these results imply that incompleteness in observed DGS samples is driven by both astrophysical intermittency and survey architecture.

Several methodological limits remain explicit. SDSS $z\simeq 0.8{-}1$6 fibers strongly bias against spectroscopic separation of very close nuclei, and in the GOTHIC pipeline 46,061 of 47,521 spectroscopic candidates after initial filtering still had both nuclei within one fiber (Bhattacharya et al., 2020). In the DPNEL problem, aperture mixing and multi-Gaussian decomposition complicate line-ratio interpretation, even though the observed $z\simeq 0.8{-}1$7–$z\simeq 0.8{-}1$8 correlation appears too strong to be explained by fitting artifacts or detection bias alone (XueGuang, 17 Jul 2025). At high redshift, the principal ambiguity is often lensing versus a physical dual, requiring resolved host-galaxy imaging, chromatic flux-ratio tests, and accurate redshift comparison (Chen et al., 2022).

The convergent observational agenda is correspondingly specific: multi-epoch hard X-ray monitoring to mitigate Myr-scale variability bias; IFU spectroscopy to map velocity fields and ionization structure across both nuclei and any bridge; high-resolution radio or VLBI imaging to isolate compact cores and jets; and high-resolution optical/near-IR imaging to distinguish dual hosts from lensed images or unresolved merger remnants (Rosas-Guevara et al., 2018). This suggests that the most robust DGS census will remain inherently multiwavelength and spatially resolved, with large survey pre-selection followed by targeted confirmation.