Eddington-Ratio Distribution

• Eddington‐ratio distribution is the probability function describing how supermassive black holes or AGN accrete at fractions of their Eddington limit, often modeled with Schechter, broken‐power-law, or log-normal forms.
• It is derived using rigorous observational and simulation methods that correct for selection biases and flux limits, revealing a rise toward lower accretion rates and a sharp cutoff at high values.
• The distribution underpins theoretical models of black hole and galaxy coevolution by informing AGN duty cycles, fueling modes, and feedback processes.

The Eddington-ratio distribution quantifies the probability that a supermassive black hole (SMBH) or active galactic nucleus (AGN) of given mass is accreting at a given fraction of its Eddington limit, λ ≡ L_bol / L_Edd. This distribution is a cornerstone for demography, fueling, and feedback studies of black hole–galaxy coevolution, serving as the statistical kernel that links accretion events to galaxy and black hole growth, observed luminosity functions, and the mapping between observed and relic black hole mass functions.

1. Formal Definition and Parametric Forms

The Eddington ratio, λ = L_bol / L_Edd, normalizes the bolometric luminosity to the Eddington limit, $L_{\rm Edd}=1.26\times10^{38}(M_{\rm BH}/M_\odot)$ erg s$^{-1}$.

The Eddington-ratio distribution function (ERDF), often denoted as ξ(λ) = dN/d log λ or Φ(λ), is typically parameterized in several analytic forms:

• Schechter Form:

$$\Phi(\lambda) = \Phi_*\left(\frac{\lambda}{\lambda_*}\right)^{\alpha}\exp\left(-\frac{\lambda}{\lambda_*}\right)$$

with normalization $\Phi_*$, characteristic Eddington ratio $\lambda_*$, and faint-end slope α. This form captures a power-law rise at low λ with exponential cutoff around $\lambda_*$ (Schulze et al., 2010, Jones et al., 2016, Blanton et al., 18 Nov 2025).

• Broken-Power-Law:

$$\xi(\lambda) = \xi^*\left[\left(\frac{\lambda}{\lambda^*}\right)^{\delta_1}+\left(\frac{\lambda}{\lambda^*}\right)^{\delta_2}\right]^{-1}$$

with low- and high-λ slopes δ₁, δ₂ and break λ* (Weigel et al., 2017, Bernhard et al., 2018, Ananna et al., 2022, Ananna et al., 2022).

• Log-normal:

$$\psi_{\rm Edd}(\lambda) = \frac{1}{\ln10\,\sqrt{2\pi}\,\sigma_\lambda}\,\exp\left[-\frac{(\log\lambda-\log\lambda^*)^2}{2\sigma_\lambda^2}\right]$$

with dispersion σλ (in dex) and mass- or redshift-dependent mean (DeGraf et al., 2012, He et al., 2023, Suh et al., 2015, Panda et al., 1 Oct 2025).

• Composite Forms:

Mixtures of Gaussians, Gaussians plus power-laws, or mass-dependent forms are used when reproducing broad features across populations or cosmic time (Shankar et al., 2011, Kelly et al., 2010).

2. Observational Inference and Correcting Systematics

Observed ERDF shapes depend sensitively on selection functions, sample definitions, and emission-line or photometric proxies. Optical, X-ray, and infrared studies demonstrate that:

• Without corrections, observed ERDFs often display an artificial turnover at $\lambda \sim 0.1$ due to flux limits (Schulze et al., 2010, Schulze et al., 2010, Schulze et al., 2010).
• Rigorous maximum-likelihood or forward-modeling methods, accounting for individual detection thresholds and full selection biases, recover an intrinsic ERDF that generally rises as a Schechter-like power law to low λ, with a sharp cutoff near λ ~ 0.1–1 (Schulze et al., 2010, Blanton et al., 18 Nov 2025, Schulze et al., 2010).
• Proper treatment of “outshining” by star formation and use of specific diagnostic diagrams (e.g., BPT or 3D line ratio spaces) is essential for optical/narrow-line samples (Jones et al., 2016, Blanton et al., 18 Nov 2025).

3. Physical Drivers, Galaxy Properties, and Universality

The functional form and parameters of the ERDF are physically motivated by models of AGN fueling, feedback, and galaxy environment:

• Self-Regulated Growth: Feedback expels fuel post-peak and enforces a power-law decay in accretion, predicting a power-law ERDF (Cao, 2010, Kelly et al., 2010).
• Accretion Modes: Radiatively efficient (thin-disk; X-ray/optical; blue/green hosts) and inefficient (ADAF/radio-mode; red quiescent hosts) populations are distinguished by broken-power-law ERDFs, often nearly mass-independent (Weigel et al., 2017, Ananna et al., 2022).
• Mass and sSFR Dependence: While some studies find nearly mass-independent ERDFs in star-forming systems (F_AGN~const above λ>10^{-3}), quiescent galaxies show a declining active fraction with increasing mass (Blanton et al., 18 Nov 2025). At high z, log-normal or Schechter ERDFs show negative mass dependence—massive SMBHs accreting at lower λ (He et al., 2023, Jones et al., 2016).
• Redshift Evolution: Characteristic λ and the high-λ ERDF tail both shift positively with redshift, generating higher fractions of AGN at high λ in the early Universe (“downsizing” or anti-hierarchical growth) (Suh et al., 2015, He et al., 2023, Shirakata et al., 2019, Shankar et al., 2011, Bernhard et al., 2018).
• Population Bimodality: Evidence for clear bi-modal ERDFs (e.g., in changing-look AGN) is limited to specific populations and generally absent in global samples, which are typically well modeled by unimodal forms (Panda et al., 1 Oct 2025).

4. Impact on Black Hole and Galaxy Evolution Frameworks

The ERDF is mathematically coupled to the evolution of the black hole mass function (BHMF) and AGN duty cycle:

• Continuity Equation: Black hole growth via accretion evolves under

$$\frac{\partial n(M_{\rm bh},t)}{\partial t}+\frac{\partial}{\partial M_{\rm bh}}[n(M_{\rm bh},t)\langle\dot M(M_{\rm bh},t)\rangle]=0$$

with

$$\langle\dot M(M_{\rm bh},z)\rangle = \int_{\lambda_{\min}(z)}^\infty \zeta(\lambda)\frac{(1-\eta_{\rm rad})\,\lambda\,L_{\rm Edd}(M_{\rm bh})}{\eta_{\rm rad}c^2}d\log\lambda$$

where $\zeta(\lambda)$ is the ERDF (Cao, 2010).

• AGN Duty Cycle: The fraction of SMBHs above given λ, $$P(\lambda>\lambda_0) = \int_{\lambda_0}^{\lambda_{\max}} \xi(\lambda)\,d\,\log\lambda\,/\,\int_{M_{\rm BH,\min}}^{M_{\rm BH,\max}} \Phi_{\rm tot}(M_{\rm BH})\,d\,\log M_{\rm BH}$$, is a key diagnostic of the SMBH activity timescale and fueling duty (Ananna et al., 2022, Ananna et al., 2022, Blanton et al., 18 Nov 2025).
• Host Coevolution: Mapping stellar mass functions onto X-ray/AGN luminosity functions using a mass-independent or mass-dependent ERDF—convolved with empirical $M_{\rm BH}−M_*$ relations—successfully reproduces observed AGN XLFs, and constraints on the normalization, shape, and redshift evolution of the ERDF critically inform theoretical models of SMBH/galaxy growth (Weigel et al., 2017, Delvecchio et al., 2020, Suh et al., 2015, Bernhard et al., 2018).

5. Empirical Results across Population and Cosmic Time

A synthesis of key empirical findings is shown in the following table, summarizing recent best-fit ERDF parameters in diverse environments and redshifts. (All λ are in units of $L_{\rm bol}/L_{\rm Edd}$.)

Study & Population / Redshift	ERDF Parametric Form	Key Parameters, Trends, and Results
(Schulze et al., 2010) HES BLAGN, z<0.3	Schechter: $$\Phi_*\left(\frac{\lambda}{\lambda_*}\right)^\alpha\exp(-\lambda/\lambda_*)$$	$\alpha=-1.95$, $\lambda_*=0.28$; steady power-law rise, exponential cutoff
(Blanton et al., 18 Nov 2025) MaNGA Seyferts	Schechter (Netzer 2019, Kormendy–Ho)	$\alpha = -0.67^{+0.15}_{-0.17}$; $F_{\rm AGN}(>\!10^{-3})=0.078$ mass-indep. in SF, declining in quenched
(Weigel et al., 2017) Swift-BAT X-ray, z~0.1	Broken power law: $$\xi=\xi^*[(\lambda/\lambda^*)^{\delta_1}+(\lambda/\lambda^*)^{\delta_2}]^{-1}$$	X-ray: $\log\lambda^*=-1.84$, $\delta_1=0.47$, $\delta_2=2.53$; mass-independent, different for radio/X-ray modes
(Bernhard et al., 2018) Host mass-dependent	Broken power law, 3 stellar mass bins	Low-mass SF: suppressed at $\lambda<0.1$; high-mass: broader ERDF; necessary for SFR–$L_X$ flatness
(He et al., 2023) HSC+SDSS $z\sim4$ BLAGN	Mass-dependent log-normal/Schechter	$\log\lambda^*=0.40$, $\sigma=0.32$, $k_\lambda=-0.19$ (peak shifts lower at high $M_{\rm BH}$)
(Suh et al., 2015) X-ray BLAGN, 1<z<2.2	Log-normal	$\langle\log\lambda\rangle=-0.6$, $\sigma=0.8$ dex; peak at $\lambda\sim0.25$
(DeGraf et al., 2012) Hydrodynamical sim., $z>4.75$	Log-normal, $\sigma_m\approx0.39$	$\langle\lambda\rangle\propto (1+z)^3$, peaks at $M_{\rm BH}\sim5\times10^{7}M_\odot$
(Ananna et al., 2022) BASS DR2 obscured/unobsc.	Broken power law	Unobsc.: $\delta_1=0.16$, $\delta_2=0.22$, $\lambda^*=1.1$; Obsc.: $\delta_1=0.27$, $\delta_2=0.32$, $\lambda^*=10^{-1.75}$
(Cao, 2010) Theoretical, z ≲ 5	Power-law time-weighted: $$\zeta(\lambda)\sim\left(\lambda/\lambda_{\rm peak}\right)^{-\beta_l}\exp(-\lambda/\lambda_{\rm peak})$$	$\beta_l=0.3$, $\lambda_{\rm peak}=2.5$; $\eta_{\rm rad,0}=0.11$; $\tau_Q\gtrsim0.5$ Gyr
(Jones et al., 2016) SDSS AGN (optical/X-ray)	Schechter: $\lambda^{-0.4}\exp(-\lambda/1)$	Consistent with X-ray and obscured samples; log-normal artifact in SF galaxies due to selection

Characteristic results are that ERDFs for AGN are generally broad (0.3–0.8 dex in log λ), strongly rising toward lower λ, and display redshift and mass-dependence mainly in their normalization and turnover location. Intrinsic mass-independence of ERDFs for radiatively efficient and inefficient modes is supported in the local Universe (Weigel et al., 2017, Ananna et al., 2022).

6. Theoretical and Simulation Perspectives

• Hydrodynamical simulations confirm log-normal forms for $P(\lambda|M_{\rm BH},z)$, with evolution of the mean and width tied to cosmological gas fraction and feedback processes (DeGraf et al., 2012, Shirakata et al., 2019).
• Continuity-equation and semi-analytic models systematically link the ERDF to black hole mass/bulge buildup, radiative efficiency, and duty cycles (Cao, 2010, Shankar et al., 2011).
• Eddington ratio–driven feedback models predict power-law or broken-power-law ERDFs reflecting the self-regulated shut-off of fueling (Cao, 2010, Kelly et al., 2010).
• Host and fueling mode dichotomy: Physical bimodality in ERDFs—reflecting hot/cold accretion modes, jet/radiative feedback, or merger/secular fueling—naturally emerges in population synthesis (Weigel et al., 2017, Ananna et al., 2022).

7. Astrophysical Implications and Open Directions

• Obscuration and AGN Geometry: Sharp transitions in ERDF shape between Type 1 and Type 2 AGN, with much steeper high-λ cutoffs for obscured systems, are predicted and observed, supporting radiation-regulated unification scenarios (Ananna et al., 2022, Ananna et al., 2022).
• AGN Duty Cycle and Cosmic Growth: The fraction of time/mass SMBHs spend above fixed λ is a direct probe of cosmic AGN “on” fraction and is sensitive to underlying ERDF parameters and their evolution (Blanton et al., 18 Nov 2025, Shirakata et al., 2019).
• Downsizing and Cosmic History: The CSFR/ERDF link, population-wide suppression of high-λ activity at late times, and the changing locus of black hole growth are all set by the underlying ERDF (Shankar et al., 2011, He et al., 2023, Delvecchio et al., 2020).
• Selection Effects and Completeness: Accurate inference of the intrinsic ERDF, especially in the low-λ regime, is fundamentally limited by survey depth and sample biases; next-generation IR/X-ray and deep emission-line surveys are critical for comprehensive mapping (Jones et al., 2016, Blanton et al., 18 Nov 2025, Shirakata et al., 2019).

The Eddington-ratio distribution thus encodes fundamental constraints on the stochasticity, regulation, and evolution of SMBH accretion and underpins both phenomenological and theoretical models of black hole–galaxy coevolution.