---
title: 'FEGA25: Semi-Analytic Model of Galaxy Formation'
url: https://www.emergentmind.com/topics/fega25
type: topic
---

# FEGA25: Semi-Analytic Model of Galaxy Formation

Searching arXiv for FEGA25 and closely related papers to ground the article.
arXiv search query: FEGA25 Contini semi-analytic model AGN stellar halos black hole occupation baryonic content
FEGA25, short for “Formation and Evolution of GAlaxies” in its 2025 generation, is a semi-analytic model of galaxy formation and evolution developed within the FEGA series and implemented on merger trees extracted from dark-matter-only cosmological simulations. It extends FEGA24 by combining updated prescriptions for baryonic physics with a fully coupled active galactic nucleus feedback framework that includes negative, positive, and hot-gas-ejection modes, and it has been used to study halo baryon fractions, diffuse stellar components around bright central galaxies, and black hole occupation fractions across redshift [2502.19503, 2507.10673, 2511.01164, 2606.22758].

## 1. Lineage, numerical basis, and calibration

FEGA25 is described as a modern state-of-the-art semi-analytic model applied to halo merger trees from dark-matter-only simulations. Across its applications, it is run on three GADGET-4 simulations—YS50HR, YS200, and YS300—within a Planck 2018 cosmology with $\Omega_m=0.31$, $\Omega_\Lambda=0.69$, $n_s=0.97$, $\sigma_8=0.81$, and $h=0.68$. The merger trees are sampled with 100 snapshots from $z=20$ to $0$, evenly spaced in cosmic time [2507.10673, 2511.01164, 2606.22758].

| Simulation | Box size | DM particle mass |
|---|---:|---:|
| YS50HR | $50~{\rm Mpc}/h$ | $10^{7}\,M_\odot/h$ |
| YS200 | $200~{\rm Mpc}/h$ | $3.26\times10^{8}\,M_\odot/h$ |
| YS300 | $300~{\rm Mpc}/h$ | $2.2\times10^{9}\,M_\odot/h$ |

The three boxes are used complementarily: YS50HR supplies the highest mass resolution, YS200 samples groups and massive galaxies, and YS300 provides statistics for the most massive systems. Different FEGA25 applications impose different halo- and stellar-mass cuts. In the stellar-halo analysis, FEGA25 is run on central halos above study-specific thresholds, whereas the black-hole-occupation analysis imposes a stellar-mass cut $\log M_\star>7$ and minimum halo masses corresponding to roughly $10^4$ dark matter particles [2511.01164, 2606.22758].

Calibration is performed via MCMC against observed stellar mass functions from $z\sim3$ to $0$. The AGN-feedback papers emphasize a “single calibration target” philosophy in which stellar mass functions constrain the model, while quantities such as the star-forming main sequence, passive fractions, gas fractions, and baryon fractions are treated as predictions. Later FEGA25 applications inherit this calibrated framework when analyzing stellar halos, intracluster light, and black hole demographics [2502.19503, 2507.10673].

## 2. Coupled feedback architecture and baryon-cycle physics

A defining feature of FEGA25 is its coordinated treatment of supernova feedback and three AGN feedback channels. The radio-mode black-hole accretion rate is written as
$$
\dot{M}_{\rm BH}=\kappa_{\rm AGN}\left(\frac{f_{\rm hot}}{0.1}\right)\left(\frac{V_{200}}{200~{\rm km~s^{-1}}}\right)^3\left(\frac{M_{\rm BH}}{10^8~M_\odot h^{-1}}\right),
$$
with AGN mechanical power
$$
\dot{E}_{\rm radio}=\eta_{\rm rad}\,\dot{M}_{\rm BH}\,c^2,
$$
and a cooling rate reduced according to
$$
\dot{M}_{\rm cool,new}=\dot{M}_{\rm cool}-2\frac{\dot{E}_{\rm radio}}{V_{200}^2}.
$$
This is the negative mode: AGN heating offsets or fully suppresses cooling [2502.19503, 2507.10673].

FEGA25 also implements positive AGN feedback. When cooling is not entirely shut off, AGN activity can induce extra star formation through
$$
\dot{M}_\star=\left(\frac{\delta M_{\rm BH}}{M_{\rm BH}}\right)\left(\frac{\dot{M}_{\rm cool,new}}{10^8~M_\odot/h}\right).
$$
The key structural point is that negative and positive AGN feedback are tied to the same accretion efficiency parameter $\kappa_{\rm AGN}$ rather than being separately tuned. In the FEGA25 applications, this positive mode is strongest at high redshift, when cooling remains efficient, and fades at late times as negative feedback increasingly suppresses residual cooling [2502.19503, 2606.22758].

The third AGN mode ejects hot gas beyond the virial radius. FEGA25 contains two implementations. In AGNeject1,
$$
M_{\rm ejected}=\left(\frac{\delta M_{\rm BH}}{M_{\rm BH}}\right)\left(\frac{M_{\rm hot}}{10^8~M_\odot/h}\right)\left(1-\frac{V_{200}}{V_{\rm scale}}\right),
$$
whereas AGNeject2 activates only when AGN heating exceeds what is required to stop cooling, and then ejects
$$
M_{\rm ejected}=M_{\rm excess}\left(1-\frac{V_{200}}{V_{\rm scale}}\right).
$$
The ejected material is placed in an external reservoir and reincorporated at a rate
$$
\dot{M}_{\rm ej}=\gamma_2\,\frac{M_{\rm ej}}{t_{\rm reinc}}.
$$
For AGNeject1 the calibrated parameters are $\kappa_{\rm AGN}=2\times10^{-5}\,M_\odot\,{\rm yr^{-1}}$, $V_{\rm scale}=500\,{\rm km\,s^{-1}}$, and $\gamma_2=0.378$; for AGNeject2 they are $\kappa_{\rm AGN}=5\times10^{-5}\,M_\odot\,{\rm yr^{-1}}$, $V_{\rm scale}=415\,{\rm km\,s^{-1}}$, and $\gamma_2=0.367$ [2507.10673].

Supernova feedback remains essential. FEGA25 uses a redshift-dependent reheating and ejection scheme in which cold gas is reheated to the hot phase and, when the available SN energy is sufficient, hot gas is ejected beyond $R_{200}$. The SN parameters quoted for the baryon-fraction analysis are $\epsilon_{\rm reh}=0.7$, $V_{\rm reh}=70\,{\rm km\,s^{-1}}$, $\beta_{\rm reh}=3.5$, $\eta_{\rm ej}=0.15$, $V_{\rm ej}=70\,{\rm km\,s^{-1}}$, and $V_{\rm SN}=630\,{\rm km\,s^{-1}}$ [2507.10673].

Within FEGA25, the baryon cycle is therefore organized as inflow to a hot halo reservoir, cooling to a cold-gas component, star formation via the extended Kennicutt–Schmidt relation, SN-driven reheating and ejection, AGN-driven suppression of cooling, AGN-induced star formation when residual cooling persists, and AGN-driven hot-gas removal. A consistent implication of the FEGA25 studies is that hot-gas ejection by AGN is introduced specifically to reduce halo hot-gas fractions without significantly altering the stellar and cold-gas components [2502.19503, 2507.10673].

## 3. Halo baryon fractions and gas-phase partitioning

FEGA25 has been used explicitly to study the baryonic content of halos over a wide mass range. The baryon fraction inside $R_{200}$ is defined as
$$
f_b(M_{\rm halo})=\frac{M_{\rm baryons}(<R_{200})}{(\Omega_b/\Omega_m)M_{200}},
$$
and the model analyzes how this quantity responds to SN feedback and the two AGN hot-gas-ejection implementations [2507.10673].

For AGNeject1, the normalized baryon fraction at $z=0$ rises from approximately $0.4$ at $\log M_{\rm halo}\sim 11.2$ to approximately $0.75$ in massive clusters, with very weak redshift evolution and no pronounced cavity. AGNeject2 behaves similarly at $z\gtrsim1$, but at $z=0$ develops a strong U-shaped feature—a “cavity”—over $11.8\lesssim\log M_{\rm halo}\lesssim13.4$, where intermediate-mass halos have depressed baryon fractions relative to both lower- and higher-mass systems [2507.10673].

The hot-gas fraction within $R_{200}$ rises from about $0.3$ at $\log M_{\rm halo}\sim12$ to about $0.65$ in massive clusters at $z=0$. On cluster scales, this implies that approximately $92$–$93\%$ of the baryons within $R_{200}$ are in the hot phase, with the remainder in stars and negligible cold gas. The difference between AGNeject1 and AGNeject2 in the cavity mass range is entirely attributed to hot gas: AGNeject2 ejects substantially more hot gas at $z<1$ [2507.10673].

The ejection efficiencies reported for FEGA25 clarify the division of labor between SN and AGN feedback. In AGNeject1, AGN-driven ejection at $z=0$ reaches approximately $17\%$ near $\log M_{\rm halo}\approx13$ and decreases to approximately $5\%$ at cluster scales. In AGNeject2 it is negligible for $z>1$ but rises sharply below $z=1$, reaching approximately $23\%$ near $\log M_{\rm halo}\sim13$ at $z=0$. Supernova-driven ejection is identical in the two AGN models, decreases with halo mass, increases with redshift, reaches approximately $30\%$ in the smallest halos considered, and remains approximately $5\%$ even on cluster scales [2507.10673].

These results support a mass-dependent division: supernova feedback dominates gas ejection in halos with $\log M_{\rm halo}\lesssim12$, whereas AGN feedback becomes increasingly important at higher halo masses. FEGA25 further reports that both AGNeject1 and AGNeject2 preserve the stellar and cold-gas components and successfully reproduce the stellar-to-halo mass relation up to redshift $3$. The contrast between a smooth AGNeject1 evolution and the late-time AGNeject2 cavity places FEGA25 in direct dialogue with hydrodynamical simulation trends sometimes associated with EAGLE-like monotonic behavior and IllustrisTNG/SIMBA-like cavity behavior [2507.10673, 2502.19503].

## 4. Bright central galaxies, intracluster light, and stellar halos

In FEGA25’s stellar-halo application, the diffuse stellar distribution around bright central galaxies is formalized by treating stellar halos as the inner part of the intracluster light rather than as an independent formation channel. The central galaxy, identified with the BCG in that analysis, is the bound bulge-plus-disk component, with half-mass radius
$$
R_{\rm BCG}=\frac{R_{\rm bulge}\,M_{\rm bulge}^*+1.68\,R_{\rm sl}\,M_{\rm disk}^*}{M_{\rm BCG}^*}.
$$
The ICL is assigned an NFW-like profile more concentrated than the host dark matter halo, with
$$
c_{\rm ICL}=\gamma\,c_{200},
$$
where $\gamma$ is typically in the range $1\lesssim\gamma\lesssim3$, and the transition radius is defined as
$$
R_{\rm trans}=\frac{R_{200}}{c_{\rm ICL}}.
$$
All ICL stars within $R_{\rm trans}$ are tagged as the stellar halo, so that
$$
{\rm SH}\equiv{\rm ICL\ stars\ with}\ r<R_{\rm trans}.
$$
The model does not explicitly evaluate whether those stars are gravitationally bound to the BCG; it adopts the interpretation of the SH as a transition zone between the BCG and the extended diffuse light [2511.01164].

This construction is coupled to explicit ICL formation channels. FEGA25 builds the ICL from stellar stripping of satellites, mergers, and pre-processing in progenitor environments. Stellar stripping is identified as the dominant channel in previous FEGA work; mergers are secondary, though their relative role depends on how mergers are defined across simulations and SAMs. In the merger channel, a fixed fraction of the satellite’s stellar mass is deposited directly into the ICL, with
$$
\langle f_{\rm sat\rightarrow ICL}\rangle\approx20\%\quad{\rm with\ scatter}\ \pm5\%.
$$
The SH therefore traces ex-situ assembly by construction: FEGA25 does not form SH or ICL stars in situ [2511.01164].

The structural predictions are specific. The distribution of $R_{\rm trans}$ peaks at $30$–$40$ kpc at all redshifts, with about $90\%$ of systems below $100$ kpc. The upper envelope grows with time: the most massive halos reach approximately $150$ kpc at $z=2$, approximately $300$ kpc at $z=1$, and approximately $400$ kpc at $z=0$. At $z=0$, the SH mass correlates approximately linearly in log–log space with both BCG mass and ICL mass, but the SH–ICL relation has significantly smaller scatter. For the SH–BCG relation, FEGA25 reports
$$
\log M_{\rm SH}\simeq0.99\,\log M_{\rm BCG}-2.07.
$$
This implies a nearly constant fraction $f_{\rm SH-BCG}=M_{\rm SH}/M_{\rm BCG}$ across $\log M_{\rm BCG}=9.5$–$13$, with mean $f_{\rm SH-BCG}\approx0.007$, a roughly $\pm1\sigma$ range of approximately $0.003$–$0.016$, and a roughly $\pm2\sigma$ range of approximately $0.001$–$0.035$ [2511.01164].

Halo concentration and ICL formation efficiency are the principal FEGA25 control parameters for SH structure. Less concentrated halos have larger $R_{\rm trans}$ and hence larger SH masses, even though earlier FEGA work found higher ICL fractions in more concentrated halos; FEGA25 reconciles this by distinguishing masses from fractions. The baryonic counterpart of this structural dependence is the ICL formation efficiency
$$
\epsilon_{\rm ICL}\equiv\frac{M_{\rm ICL}(z=0)}{M_{\rm halo}},
$$
with higher $\epsilon_{\rm ICL}$ at fixed $M_{\rm BCG}$ yielding more massive stellar halos [2511.01164].

## 5. Colors, metallicities, and comparison with deep imaging surveys

FEGA25 predicts rest-frame optical colors and metallicities for BCGs, stellar halos, and the ICL at $z=0$, $0.5$, $1$, and $2$. The color results are uniform in one respect: all three components redden with time. At all epochs BCGs are slightly redder than SH and ICL, while SH and ICL have very similar color distributions. At $z=2$ the distributions are broader and the separation between BCGs and the diffuse components is clearer; by $z=0$, the distributions narrow and SH and ICL become effectively indistinguishable in both $g-r$ and $r-i$ [2511.01164].

The metallicity evolution is more diagnostic. FEGA25 reports that at high redshift, especially at $z=2$, BCGs are substantially more metal rich than SH and ICL. In the metallicity distributions, the BCG peak is around $\log Z\sim-1.7$, while SH and ICL peak around $\log Z\sim-2.15$, producing a gap of approximately $0.4$ dex. By $z=0$, the BCG peak remains near $\log Z\sim-1.7$, but SH and ICL have become more metal rich and the gap shrinks to approximately $0.1$ dex. The mean metallicities then overlap within the scatter, and SH and ICL remain almost identical chemically as well as photometrically [2511.01164].

These predictions are compared with a combined observational sample labeled “VEGAS,” consisting of VEGAS deep VST imaging, the Fornax Deep Survey, and spectroscopic metallicities from Fornax3D and M3G. The observational structural radius $R_2$, defined from three-component surface-brightness decompositions, is adopted as the observational analog of the model transition radius. Its distribution is broadly similar to FEGA25’s $R_{\rm trans}$ distribution, though somewhat skewed to smaller radii [2511.01164].

The color comparison is described as excellent: the mean predicted SH color overlaps the bulk of the observed data, and within $\pm2\sigma$ almost all observed values lie inside the FEGA25 distribution. The metallicity comparison is less favorable. Observed SH metallicities are systematically lower than FEGA25 SH metallicities at $z=0$, and the observed metallicity distribution peaks at lower metallicity than the model. The interpretation advanced in the FEGA25 stellar-halo study is that real stellar halos in the observed sample may have had a larger contribution from disrupted low-mass dwarfs than predicted for the model’s central-galaxy sample. The same study notes that many VEGAS galaxies are satellites, whereas FEGA25 in that analysis is run only on centrals, which matters particularly for metallicity comparisons [2511.01164].

A central observational conclusion follows directly from these results: broadband colors alone are insufficient to disentangle stellar halos from the surrounding ICL. FEGA25 therefore identifies chemo-dynamical tracers—metallicity, $\alpha$-enhancement, and kinematics—as the decisive tests. Upcoming wide-field surveys such as LSST, WEAVE, and 4MOST are singled out as crucial because they can jointly map structure, metallicity, and kinematics in large galaxy samples [2511.01164].

## 6. Black hole growth, occupation fractions, and major caveats

A distinct FEGA25 application addresses black hole occupation fractions. In that study, FEGA25 is described as not imposing a pre-existing BH seed population: black holes form naturally through gas accretion during quasar-mode events and then grow through merger-driven growth and radio-mode accretion. The quasar-mode accretion increment is parameterized as
$$
\Delta M_{\rm BH}=f_{\rm BH}\left(\frac{M_{\rm sat}}{M_{\rm cen}}\right)\frac{M_{\rm cold}}{1+\left(280~{\rm km~s^{-1}}/V_{200}\right)^2},
$$
with $f_{\rm BH}=0.066$. This no-explicit-seeding description is central to the occupation-fraction analysis, which then takes the reproduced black-hole mass function from at least $z=2$ to the present day as a prerequisite for studying occupation statistics [2606.22758].

The intrinsic occupation fraction is defined in stellar-mass bins as
$$
f_{\rm BH,occ}(>M_{\rm BH,thr}\mid M_\star)=
\frac{N_{\rm gal}(M_\star\ {\rm bin},\,M_{\rm BH}>M_{\rm BH,thr})}{N_{\rm gal}(M_\star\ {\rm bin})},
$$
with binomial uncertainty
$$
\sigma_f=\sqrt{\frac{f(1-f)}{N}}.
$$
FEGA25 focuses primarily on the thresholds $M_{\rm BH}>10^5\,M_\odot$ and $M_{\rm BH}>10^6\,M_\odot$. Across all boxes and redshifts, $f_{\rm BH,occ}$ rises monotonically with stellar mass, and massive galaxies are almost always occupied. The threshold choice is decisive: for $M_{\rm BH}>10^5\,M_\odot$, YS50 and YS200 can reach $f_{\rm BH,occ}\sim0.5$ near $M_\star\sim10^8$–$10^{8.5}\,M_\odot$, whereas for YS300 the same level is reached only near $M_\star\sim10^{10}\,M_\odot$ [2606.22758].

FEGA25 also finds that the central–satellite distinction matters. In YS50, satellites have slightly higher occupation fractions than centrals in the transition regime for the $10^5\,M_\odot$ threshold, while centrals and satellites are similar for the $10^6\,M_\odot$ threshold. In YS200, centrals rise more rapidly toward unit occupation, whereas satellites show a smoother and more delayed increase. The total occupation fraction is therefore interpreted as a population-weighted quantity whose shape depends on the mixture of centrals and satellites in a given simulation volume [2606.22758].

The redshift evolution is explicitly described as non-universal. YS50 and the hydrodynamical NewCluster zoom show higher occupation fractions at higher redshift for a given stellar mass, qualitatively similar to trends reported for Romulus and BRAHMA-like studies, whereas YS200 and YS300 show the opposite behavior. The FEGA25 black-hole analysis attributes this to a combination of numerical resolution, simulated volume, environment, sampled galaxy population, and progenitor bias at fixed stellar mass [2606.22758].

Several FEGA25 caveats recur across applications. The stellar-halo implementation defines SH stars geometrically from the ICL and does not evaluate binding energy, so kinematic separation between SH and ICL is not captured. The ICL profile is parametric, using a scaled NFW-like form with a single concentration factor $\gamma$. The SH mass is capped numerically so that $M_{\rm SH}\leq M_{\rm BCG}$, with any excess transferred to the BCG disk. No in situ star formation is permitted in SH or ICL. In the black-hole application, FEGA25 does not model Pop III versus direct-collapse seed channels, depends on dark-matter-only merger trees plus subgrid prescriptions, and predicts intrinsic occupation fractions rather than directly observable AGN fractions [2511.01164, 2606.22758].

Taken together, these studies define FEGA25 as a flexible semi-analytic framework in which galaxy growth, halo gas regulation, diffuse stellar structure, and black hole demographics are linked by a common set of merger trees and feedback prescriptions. The strongest recurring conclusion is methodological rather than singularly phenomenological: the inferred properties of halo gas, stellar halos, and black hole occupation fractions depend sensitively on how feedback, concentration, environment, and selection are encoded within the model [2507.10673, 2511.01164, 2606.22758].

Source: https://www.emergentmind.com/topics/fega25