Large-scale Environmental Dependence of AGN Activity Revealed by JWST Pa$α$ Emitters

abstract · abstract

The role of the large-scale environment in regulating AGN activity remains an open question. We investigate the Mpc-scale environments of 70 AGNs at $z=1.1$-$1.5$ using a highly complete sample of 3,419 Pa$α$ emitters with accurate spectroscopic redshifts from JWST slitless spectroscopic survey. This dataset allows us to robustly trace the environments and quantify the environmental dependence of AGN properties. We measure the eighth-nearest-neighbor surface density of Pa$α$ emitters around each AGN, probing $\sim$1.5 pMpc scale environments. We find that this environmental density exhibits a weak positive correlation with radio luminosity. Radio-bright AGNs tend to reside in statistically denser environments than their radio-faint counterparts. The environmental density also has a positive correlation with the Eddington ratio and a weak negative correlation with the central black hole mass, implying more efficient black hole accretion in dense environments. We construct Pa$α$ luminosity functions (LFs) in AGN environments and find that they are systematically elevated relative to the field Pa$α$ LF. In addition, their faint-end slopes are shallower, suggesting accelerated galaxy growth and enhanced quenching of star formation. As Pa$α$ emitters trace gas-rich, star-forming galaxies, our results support a scenario in which the large-scale cold gas promotes AGN activity at cosmic noon.

S1: I Introduction / block 1 · html

Different studies report inconsistent results of the AGN large-scale environmental dependence on AGN and host-galaxy properties, from the nearby Universe to cosmic noon. The relation between radio power and environmental density is controversial. While some studies have reported no significant dependence [ 67 , 30 , e.g.,] , others have found a negative trend [ 43 , e.g.,] or a positive trend [ 11 , e.g.,] . The environmental dependence of M BH M_{\mathrm{BH}} and accretion rate shows no significant trends with substantial scatter [ 23 , 68 , e.g.,] . Such studies mostly rely on photometric data and ground-based spectroscopic surveys. The former suffer from photometric-redshift uncertainties, while the latter are restricted to bright galaxies and provide an incomplete census of the surrounding galaxy population.

S3.SS4: III.4 Luminosity Functions and Overdensities / block 1 · html

We perform a Markov Chain Monte Carlo (MCMC) analysis using the Python code emcee [ 13 ] to determine the best-fit parameters Φ ∗ \Phi_{*} , α \alpha , and L ∗ L_{*} . The resulting LFs are shown in Figure 5 , with the corresponding values presented in Appendix C .

S3.SS4: III.4 Luminosity Functions and Overdensities / block 1 · html

We find a flattening trend at the faint end of the environmental LFs, with the faint-end slope α \alpha becoming shallower from the field to the AGN environment, and further to the radio-bright AGN environment. This trend is consistent with literature results on H α \alpha LFs and galaxy stellar mass functions in overdense regions [ 63 , 14 , 36 , e.g.,] . The shallower faint-end slope may reflect accelerated galaxy growth and environmental processes. Simulations of galaxy evolution find early formation of massive galaxies and rapid merging and quenching activity in proto-cluster environments compared to field galaxies, driven by the high halo density and top-heavy halo mass functions [ 52 ] .

S4: IV Summary and Implication / block 1 · html

We have statistically investigated the large-scale environments of 70 AGNs using a highly complete Pa α \alpha emitter sample in the COSMOS-3D field at z = 1.1 z=1.1 – 1.5 1.5 . We constructed M stellar M_{\rm stellar} - and SFR-matched control samples and measured the eighth-nearest-neighbor surface density of Pa α \alpha emitters around both AGNs and control galaxies. This environmental density probes Mpc-scale environments. We confirm that the environmental density distribution of radio-bright AGNs is statistically higher than that of their control galaxies. We further find a weak positive correlation between environmental density and radio luminosity, together with a positive correlation with λ Edd \lambda_{\rm Edd} and a weak negative correlation with M BH M_{\rm BH} , particularly for the radio-bright AGN sample. We also construct Pa α \alpha LFs in AGN environments and find that they exhibit shallower faint-end slopes and systematically higher number densities than the field Pa α \alpha LF.

S4: IV Summary and Implication / block 1 · html

Since Pa α \alpha emitters trace actively star-forming galaxies, regions with higher Pa α \alpha emitter densities are likely associated with more abundant cold-gas reservoirs. The environmental dependence of λ Edd \lambda_{\rm Edd} and M BH M_{\rm BH} provides direct evidence for enhanced AGN accretion and rapid SMBH growth in gas-rich environments. The shallower faint-end slopes of the environmental LFs further support the accelerated galaxy evolution. The weak positive environmental dependence of radio luminosity suggests that radio emission is moderately enhanced in denser environments. Other processes related to the central engine, such as black hole spin and host-galaxy gas dynamics, are also expected to regulate the radio power of AGNs and introduce scatter to the environmental dependence. Moreover, the observed radio emission contains contributions from both AGN activity and host-galaxy star formation. In addition to producing radio emission, AGN activity may indirectly boost the total radio output by enhancing star formation through positive feedback [ 27 , 28 , e.g.,] .

S4: IV Summary and Implication / block 1 · html

Overall, our results support a scenario in which large-scale supply of cold gas could regulate both black-hole growth and galaxy evolution at cosmic noon. High-quality JWST WFSS surveys have opened a new era for AGN environmental studies. The upcoming Roman Space Telescope will provide environmental measurements over a larger cosmic volume. In the future, growing spectroscopic datasets will enable more comprehensive environmental investigations for diverse AGN populations at different redshifts.

S2.SS1: II.1 Highly Complete Pa α \alpha Emitter Sample / block 1 · html

We restricted the Pa α \alpha emitter sample to a high-completeness subsample, since completeness corrections become increasingly uncertain at the low-completeness end. We first limited the redshift range to 1.1 < z < 1.5 1.1<z<1.5 to avoid ineffective wavelength regions of the grism spectroscopy. We then applied a line-flux threshold of 0.035 mJy, corresponding to the 5 σ \sigma detection and 5.4 × 10 − 18 5.4\times 10^{-18} erg s -1 cm -2 at observed-frame 4.4 μ \mu m. With these selections, the average completeness in each simulated flux bin exceeds 60%. The resulting parent sample consists of 3,419 Pa α \alpha emitters, of which 74% have completeness values above 80%. In the subsequent environmental density measurement, the incompleteness of each Pa α \alpha emitter will be further corrected. We show the sky distribution of this Pa α \alpha emitter sample in the left panel of Figure 1 .

S2.SS3: II.3 Control Samples / block 1 · html

We constructed control samples to evaluate the AGN-related environmental effect without biases caused by the host galaxies. The M stellar M_{\mathrm{stellar}} and SFR measurements were adopted from [ 16 ] . Our Pa α \alpha AGNs are hosted by more massive galaxies with median log ⁡ ( M stellar / M ⊙ ) ∼ 10.84 \log(M_{\mathrm{stellar}}/M_{\odot})\sim 10.84 , compared to the overall Pa α \alpha emitters with median log ⁡ ( M stellar / M ⊙ ) ∼ 9.96 \log(M_{\mathrm{stellar}}/M_{\odot})\sim 9.96 . At z ≳ 1 z\gtrsim 1 , enhanced star formation is also observed in galaxies hosting radio AGNs [ 10 , 39 , 28 , e.g.,] . We generated control samples for the full AGN sample, the radio-bright AGN sample, and the radio-faint AGN sample, separately. For each AGN in a given sample, we randomly selected a Pa α \alpha emitter with M stellar M_{\mathrm{stellar}} within ± \pm 0.25 dex and SFR within ± \pm 0.5 dex of the AGN host galaxy. These matching tolerances generally represent the smallest 16th–84th percentile ranges of M stellar M_{\mathrm{stellar}} and SFR distributions of three AGN samples. We excluded 435 galaxies with radio detections in this control construction [ 26 ] . We repeated this procedure for 100 times and obtained 100 independent control samples with the same size of the corresponding AGN sample.

S3.SS2: III.2 Distributions of Controlled Density / block 1 · html

We compare the environmental density distributions of the radio-bright, radio-faint, and full AGN samples with those of their corresponding control samples in Figure 2 . We find that radio-bright AGNs reside in significantly denser environments than the AGNs in the control sample, with a median density excess of 0.14 dex and a Kolmogorov–Smirnov (KS) test p p -value = 0.003. This result is consistent with previous studies showing that radio AGNs preferentially inhabit overdense environments [ 67 , 23 , 43 , e.g.,] . Meanwhile, the environmental density distributions of the radio-faint and full AGN samples are indistinguishable from those of their respective control samples. The elevated environmental densities of radio-bright AGNs suggest that dense large-scale environments may enhance AGN radio emission. We further explore the environmental dependence of AGN radio properties in Section III.3 .

S1: I Introduction / block 1 · html

In this paper, we present the first results on broad-line AGNs at z > 5 z>5 from COSMOS-3D. This first dataset, covering 127 arcmin 2 and representing just 10% of the full survey, already demonstrates the capabilities of COSMOS-3D in studying the nature of high-redshift AGNs. This work serves as a preview of the full potential of the survey. Detailed analyses of the complete COSMOS-3D sample will follow in future papers once the full dataset is available.

S1: I Introduction / block 1 · html

The paper is organized as follows. In § II , we introduce the datasets and methods to select broad-line AGNs. We present the analysis and results in § III , including the photometric and spectroscopic properties, variability, and luminosity function. We discuss the abundance of z > 7 z>7 AGNs in § IV . We adopt the AB magnitude system for all photometric measurements in this paper. Throughout this work, a flat Λ \Lambda CDM cosmology is assumed, with H 0 = 70 ​ km ​ s − 1 ​ Mpc − 1 \rm H_{0}=70~km~s^{-1}~Mpc^{-1} , Ω Λ , 0 = 0.7 \Omega_{\Lambda,0}=0.7 , and Ω m , 0 = 0.3 \Omega_{m,0}=0.3 .

S4.SS1: IV.1 Reduced Spectra / block 1 · html

Each slit-loss correction is visually validated to ensure consistency and to avoid biases from photometric contamination or aperture mis-centering. The resulting correction curve is then applied to the extracted 1D spectra. To verify the accuracy of this correction, we compare synthetic photometry generated from the corrected spectra with broadband photometric measurements. This procedure yields flux corrections with a typical uncertainty of ± 0.2 \pm 0.2 dex and a median offset of zero, confirming that the method effectively recovers the total galaxy light. The corrected spectra are suitable for line luminosity measurements; however, for line ratio analyses (e.g., BPT diagram) we recommend using the uncorrected spectra to avoid introducing wavelength-dependent systematics from the correction process.

S4.SS4: IV.4 SED Catalog / block 1 · html

To accelerate the fitting procedure, we use the ANN-based Parrot emulator [ 63 ] , which speeds up the SED fitting without introducing significant biases. The emulator produces results consistent with traditional spectral population synthesis models, with deviations of only 25-40% in stellar mass, metallicity, SFR, and stellar age. The fitting accuracy is validated through comparisons with independent SED models from Z. Ji et al. [45] , showing deviations of less than 0.15 dex, and with UV-based fits from J. Lyu et al. [59] , confirming consistency in the rest-frame UV spectral region.

S3.SS2: III.2 Testing the effect of modifying the specific bins / block 1 · html

It is important that we use luminosity complete bins. Otherwise, we mix lower luminosity sources at somewhat lower redshift with higher luminosity sources at somewhat higher redshift within the same redshift bin. This would confuse our results in the presence of any luminosity-dependent effects. This is why a large number of lower luminosity AGN are excluded from the lower- z z bins in Figures 4 - 6 . We note that for the X-ray sample the luminosity complete limits are nearly the same for both redshift bins, whereas for the IR samples we have about 0.5 dex difference. While this is not a large difference, it is conceivable that the redshift dependent effects we see for the IR AGN are partly luminosity-dependent. We attempted to explore this using multiple luminosity bins per redshift bin but found that we do not have the statistics for conclusive results.

S3.SS3: III.3 Testing the redshift-dependent role of environment / block 1 · html

Due to the low formal significance of our results in Figures 4 , 5 , and 6 , we employ bootstrapping to test the strength of the results, specifically the reversal in dependence seen in both IR samples around z ∼ 1.2 z\sim 1.2 . We randomly resampled the IR and IR-only AGN with replacement 1000 times and calculated the scores for each using equation 5 . The results are shown in Figure 7 . In each IR AGN sample we see a clear difference in the peaks of the scores of the bootstrapped samples. For AGN in the redshift range 0.1 < z < 1.2 0.1<z<1.2 the samples peak strongly at negative scores around ∼ − 0.5 \sim-0.5 with 95 % 95\% of the IR samples and 98 % 98\% of the IR-only bootstrapped samples having negative scores corresponding to negative dependencies between AGN occurrence and environmental density. For AGN in the redshift range 1.2 < z < 1.6 1.2<z<1.6 the samples peak with positive scores of ∼ 0.5 \sim 0.5 with 98 % 98\% of both the IR and IR-only bootstrapped samples having positive scores. These results highlight the result of Figures 5 , and 6 showing a reversal of dependence for both the IR AGN and the heavily obscured IR-only AGN samples around z ∼ 1.2 z\sim 1.2 .

article / block 1 · html

The spatial correlation between galaxies and the Ly α \alpha forest of the intergalactic medium (IGM) provides insights into how galaxies reionized the Universe. Here, we present initial results on the spatial cross-correlation between [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters and Ly α \alpha forest transmission at 5.4 < z < 6.5 5.4<z<6.5 from the JWST ASPIRE NIRCam/F356W Grism Spectroscopic Survey in z > 6.5 z>6.5 QSO fields. Using data from five QSO fields, we find 2 ​ σ 2\sigma evidence for excess Ly α \alpha forest transmission at ∼ 20 − 40 ​ cMpc \sim 20-40\,\rm cMpc around [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters at ⟨ z ⟩ ≃ 5.86 \langle z\rangle\simeq 5.86 , indicating that [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters reside within a highly ionized IGM. At smaller scales, the Ly α \alpha forest is preferentially absorbed, suggesting gas overdensities around [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters. Comparing with models, including THESAN cosmological radiation hydrodynamic simulations, we interpret the observed cross-correlation as evidence for significant large-scale fluctuations of the IGM and the late end of reionization at z < 6 z<6 , characterized by ionized bubbles over 50 ​ cMpc 50\rm\,cMpc around [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters. The required UV background necessitates an unseen population of faint galaxies around the [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters with average LyC leakage of log 10 ⁡ ⟨ f esc ​ ξ ion ⟩ / [ erg − 1 ​ Hz ] ≃ 24.5 \log_{10}\langle f_{\text{esc}}\xi_{\text{ion}}\rangle/[{\text{erg}^{-1}\text{Hz}}]\simeq 24.5 down to M UV = − 10 M_{\text{UV}}=-10 . Furthermore, we find that the number of observed [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters near individual transmission spikes is insufficient to sustain reionization in their surroundings, even assuming all [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters harbour AGN with 100 % 100\,\% LyC escape fractions. Despite broad agreement, a careful analysis of ASPIRE and THESAN, using the observed host halo mass from the clustering of [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters, suggests that the simulations underpred

S8: 8 Conclusions / block 1 · html

This completion of late reionization requires faint galaxies below our detection limit. The enhanced large-scale UV background for the excess Ly α \alpha forest transmission demands a collective population of fainter galaxies surrounding the observed [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters. The observed [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters fall short of providing the necessary ionizing budget, assuming reasonable values for the LyC escape fraction and ionizing photon production efficiency. Our analysis of the individual associations between [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters and Ly α \alpha forest transmission spikes further indicates that the LyC leakage from these emitters does not generate enough ionizing radiation to maintain the high ionization levels of the surrounding IGM, as evidenced by the presence of Ly α \alpha forest transmission spikes. This conclusion holds true even if we assume that all [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters host AGN activities and exhibit 100 % 100\,\% LyC escape fractions, suggesting that AGN alone are insufficient to drive reionization. Generally, an average LyC leakage of log 10 ⁡ ⟨ f esc ​ ξ ion ⟩ / [ erg − 1 ​ Hz ] ≈ 24.5 \log_{10}\langle f_{\rm esc}\xi_{\rm ion}\rangle/[{\rm erg^{-1}Hz}]\approx 24.5 down to galaxies with M UV ≈ − 10 M_{\rm UV}\approx-10 is required to establish a sufficient UV background. Thus, we conclude that an unseen population of fainter galaxies, or systems not selected as [ O ​ III ] {\rm[O\,{\scriptstyle III}]} emitters (or luminous populations residing outside the single NIRCam/WFSS field of view), is responsible for completing reionization.

S2.SS3: 2.3 Depletion and replenishment scenarios / block 1 · html

In this study, we implement both approaches and compare the results. For the depletion scenario, as noted above, the galaxy counts decrease over time; therefore, the size of the initial distribution is set to 500,000 galaxies. In contrast, for the replenishment scenario, the initial distribution contains 50,000 galaxies, and we assume that the number of galaxies replenished equals the number of mergers (i.e., refill-ratio is 1), thereby the sample size remains constant. The replenished galaxies are generated in the same way as generating the initial distribution (section 2.1 ): we assume the same Schechter function (equation 3 ) for M ∗ M_{*} and generate M BH M_{\rm BH} following equation 4 with the same σ init \sigma_{\rm init} . We limit the log ⁡ ( M ∗ / M ⊙ ) \log\left(M_{*}/M_{\odot}\right) of replenished galaxies to the range [ 6 , 8 ] \left[6,8\right] , corresponding to the low-mass side of the initial distribution (figure 1 ), under the assumption that newly formed galaxies through structure formation are predominantly low-mass systems. In sections 3.1 and 3.2 , we present the results obtained with the depletion and replenishment scenarios, respectively.

S3.SS1.SSS1: 3.1.1 Evolution on the 𝑴 𝐁𝐇 − 𝑴 ∗ \bm{M_{\rm BH}\,\mathchar 45\,M_{*}} plane / block 1 · html

One possible cause of this higher intercept is that M BH M_{\rm BH} is summed directly (i.e., not the sum of their logarithms) during each merger event, which biases the resulting sum toward higher log ⁡ M BH \log M_{\rm BH} . A similar effect could occur in the M ∗ M_{*} direction. However, in this study, we limit the range of mass ratios ( μ \mu ), allowing μ > 1 / 40 \mu>1/40 at most. This restriction reduces the upward shift in log ⁡ M ∗ \log M_{*} relative to log ⁡ M BH \log M_{\rm BH} , leading to a tendency to shift toward the overmassive side. Because the resulting relation depends strongly on the assumptions for the initial distribution, which is highly uncertain at this stage, we do not focus on the absolute values of its slope and intercept. Instead, we concentrate solely on the evolution of the scatter.

S4.SS5: 4.5 Are HELM galaxies preferentially edge-on? / block 1 · html

Figure 10 shows the distribution of the axis ratio for the HELM samples as well as the control sample. As for the sizes, we show the results using the F277W filter, as dusty dwarf are generally undetected at shorter wavelengths. The median axis ratio of the control sample is b / a = 0.51 b/a=0.51 , with the HELM samples having similar median values, that is b / a = 0.47 b/a=0.47 for the HELM- 1 ​ σ 1\sigma , b / a = 0.51 b/a=0.51 for the HELM- 2 ​ σ 2\sigma , and b / a = 0.62 b/a=0.62 for the HELM- 3 ​ σ 3\sigma . We verified the statistical significance of these differences using the one-sided and the two-sided KS tests, but we can only exclude that the HELM- 1 ​ σ 1\sigma sample is extracted from a greater or similar distribution than the control sample with a confidence of 98%. The absence of a correlation between the dust attenuation and the axis ratio (Fig. 11 ) supports that the HELM galaxies are not preferentially edge-on galaxies.

S4.SS6: 4.6 Are HELM galaxies linked to galaxy interactions? / block 1 · html

Based on the results on the HELM − 1 ​ σ -1\sigma and HELM − 2 ​ σ -2\sigma samples, some HELM galaxies could be more clustered than their non-dusty counterparts. In this case, we need to understand if they are going through an interaction or if they are the product of interactions, like tidal dwarfs ( Zwicky, 1956 ; Duc and Mirabel, 1999 , e.g.) . Dynamical studies to verify the presence or absence of a dark matter halo and spectroscopic studies to derive gas abundances are beyond the possibilities of this work and JWST angular resolution limits any detailed morphological study. However, we can analyse if these galaxies are going through a burst of star-formation, possibly induced by galaxy interactions. Bisigello et al. (2023) have already shown, although using a selection based on colours and not on physical properties, that this scenario seems to be excluded. We confirm their finding in Fig. 13 . Indeed, the median sSFR of the HELM samples is smaller than the value of the non-dusty dwarfs, being log 10 ​ ( sSFR / yr − 1 ) = − 9.1 \rm log_{10}(sSFR/yr^{-1})=-9.1 for the non-dusty dwarfs and from log 10 ​ ( sSFR / yr − 1 ) = − 9.6 \rm log_{10}(sSFR/yr^{-1})=-9.6 to log 10 ​ ( sSFR / yr − 1 ) = − 9.4 \rm log_{10}(sSFR/yr^{-1})=-9.4 for the HELM samples. The two-sided KS test shows that we can exclude with a confidence larger than 99.9% that the HELM samples are derived from the same distribution as the non-dusty dwarfs. Therefore, we can exclude that the large dust attenuation of HELM galaxies is driven by a burst of star-formation. It is however necessary to take into account that the sSFR estimation is based on SED fitting with limited photometry, given that both the non-dusty dwarfs and the HELM galaxies are generally faint. In addition, considering the available data, the possibility that these objects are instead post-starburst galaxies is still totally open.

S4.SS1: 4.1 Science Drivers and Physical Probes for Cosmic-Web Studies with the SKA / block 1 · html

The quantitative scale of this programme is set by both SKA forecasts and SKA pathfinder results. Forecasts for an SKA1-Mid all-sky continuum survey such as SASS1 predict ∼ 5 × 10 8 \sim 5\times 10^{8} radio sources at ∼ 2 ′′ \sim 2^{\prime\prime} resolution and 2 ​ μ ​ Jy ​ beam − 1 2~\upmu\mathrm{Jy~beam^{-1}} rms ( Norris et al., 2015 ) . Recent EMU results already demonstrate that automated radio-galaxy cataloguing can operate at the required scale. In the 270 ​ deg 2 270~\mathrm{deg^{2}} EMU pilot survey, the RG-CAT pipeline produced 211,625 211{,}625 radio sources, including 10,414 10{,}414 extended radio sources ( Gupta et al., 2024 ) . The same work achieved IoU > 0.5 \mathrm{IoU}>0.5 for 99% of central radio-galaxy bounding boxes and placed 98% of predicted host positions within 3 ′′ 3^{\prime\prime} of the ground-truth CatWISE host. More recently, EMUSE applied morphology-based retrieval to first-year EMU main-survey data, covering ∼ 4,500 ​ deg 2 \sim 4{,}500~\mathrm{deg^{2}} , with ∼ 3 × 10 6 \sim 3\times 10^{6} detected radio sources and ∼ 1.7 × 10 5 \sim 1.7\times 10^{5} extended radio sources suitable for similarity search ( Gupta et al., 2025 ) . These precursor results show that a conservative SKA-era expectation of > 10 5 >10^{5} morphologically measurable radio galaxies is well grounded; first-year EMU-scale data already reach this order of magnitude for extended-source samples, and SKA depth, resolution, and sky coverage should expand the accessible population substantially. Citizen-science programmes could provide a complementary route for inspecting complex, rare, or ambiguous radio morphologies, while also generating validated training and test samples for increasingly automated machine-learning classification ( Hota et al., 2026 ) .

S5: 5 Summary / block 1 · html

In this chapter, we have reviewed how radio galaxies trace, respond to, and influence the large-scale structure of the Universe, and outlined how the SKA, together with complementary multi-wavelength facilities, can turn these connections into quantitative and testable probes of cosmic-web evolution.

article / block 1 · html

A highly star-forming galaxy at z ∼ 2 z\sim 2 hosting an obscured, luminous active galactic nucleus (AGN) and a relativistic radio jet sets the stage for a cosmic crime scene. The victim is star formation, the suspect is AGN feedback. These systems offer a rare opportunity to catch this process in the act. We propose SHARP@ELT integral-field spectroscopic observations of heavily obscured, luminous AGN with resolved radio emission to witness the onset of feedback and its impact on the host interstellar medium (ISM). Such objects trace a short-lived ( < 10 5 <10^{5} yr) evolutionary phase in which a recent starburst, newly triggered radio jets, and a deeply embedded AGN coexist. In this phase, feedback is expected to suppress star formation while clearing the dusty nuclear regions. Using SHARP/VESPER, we will derive spatially resolved maps of stellar ages, star formation rate density, gas density, and ionization state, alongside the kinematics of stars and ionized gas. By directly comparing these properties with the radio structures, we will quantify the effects of both jet-driven and radiative feedback on the ISM. Our sample consists of eleven luminous, obscured AGN at 1.5 < z < 2.5 1.5<z<2.5 with resolved radio emission on scales of ∼ \sim 1–-15 kpc. Exploiting the VESPER multi-Integral Field Selector capability, we will obtain resolved continuum and emission-line maps for at least 110 galaxies at cosmic noon, enabling a comprehensive characterization of their environments, multiphase ISM, and nuclear activity within 55 hours of integration time. SHARP will thus reveal AGN feedback at the epoch when it is most effective, providing a decisive step toward understanding its role in galaxy evolution.

S2.SS1: 2.1 Selected targets: the radio DOGs / block 1 · html

These radio DOGs were selected as extremely red and bright mid-infrared (MIR) sources using the WISE all sky survey and as unresolved radio sources in the NVSS (45 ′′ beam) and FIRST ( ∼ \sim 5 ′′ beam) radio surveys ( Lonsdale et al., 2015 ) . The initial sample contains 167 radio DOGs, all are obscured AGN at z = 0.5 − 3 z=0.5-3 . Follow-up multi-frequency (150 MHz-–10 GHz) radio imaging with the JVLA, and the VLBA at various spatial resolutions (0.065 ′′ –2 ′′ ) revealed sub-galactic jets with intermediate power and peaked spectra for 72% of the sample ( Patil et al., 2020 ; Patil et al., 2022 ) , consistent with compact radio jets ( O’Dea and Saikia, 2021 ) (see some examples in Fig. 3 , and 4 ). Their high radio luminosity and morphologies (see Figs. 6 , and 3 ) imply the presence of a jet that has been recently launched. Many jets show bends and complex morphologies as seen in simulations of moderate power jets driving into a dense ISM. Visible and IR imaging, sub-mm/mm data and spectroscopic observations are available for 49 radio DOGs providing, through spectral energy distribution (SED) modeling ( Boquien et al., 2019 ; Efstathiou et al., 2000 , carried out with both CIGALE and radiative transfer models;) , stellar masses, SFRs and AGN fractions ( Lonsdale et al., 2015 ) . About half of these sources are well above ( > 0.5 >0.5 dex) the star forming main sequence ( Popesso et al., 2023 , MS;) and are thus considered starbursting. Many of these systems also exhibit high-excitation visible emission lines ( Kim et al., 2013 ; Lonsdale et al., 2015 ; Ferris et al., 2021 , Fig. 5 ; ) , implying that they are also radiatively efficient AGN. In at least one case, J1500 − - 06, there is evidence that either a wind or the jet is driving the ionised gas outward and generating shocks. This interpretation is supported by spectroscopic observations of J1500 − - 06 ( Ferris et al., 2021 , see Fig. 5 ; ) , which reveal a blueshifted component in several visible emission lines ([O II ] λ \lambda 3727, H α \alpha , [S II ] λ ​ λ \lambda\lambda 6716,6731, and [O III ] λ \lambda 5007) as well as unusual visible line flux ratios (e.g., high [OII]/[OIII]).