🔭 astrophysics

Robust AGN and host-galaxy decomposition in optical spectral fitting

This paper introduces a robust optical spectral-fitting method combining pPXF and PyQSOFit to reliably disentangle AGN and host-galaxy components, demonstrating that stellar and black-hole mass estimates remain consistent across varying AGN activity phases when appropriate quality cuts are applied.

Original authors: C. Aydar, A. Merloni, G. Zeltyn, C. Andonie, B. Trakhtenbrot, S. Bernal, Q. Wu, J. Buchner, M. Salvato, T. Dwelly, S. F. Anderson, R. J. Assef, F. E. Bauer, W. N. Brandt, S. LaMassa, M. L. Martínez-Al

Published 2026-05-01

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: C. Aydar, A. Merloni, G. Zeltyn, C. Andonie, B. Trakhtenbrot, S. Bernal, Q. Wu, J. Buchner, M. Salvato, T. Dwelly, S. F. Anderson, R. J. Assef, F. E. Bauer, W. N. Brandt, S. LaMassa, M. L. Martínez-Aldama, A. L. Rankine, D. P. Schneider, Y. Shen, J. R. Brownstein, H. Javier Ibarra-Medel, A. M. Koekemoer, M. Krumpe, S. Morrison, K. Nandra, C. A. Negrete Peñaloza, S. F. Sanchez

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine looking at a busy city street at night from a distance. You see a dazzling, bright spotlight (the Active Galactic Nucleus, or AGN) shining right in the center, surrounded by the softer, steady glow of streetlights and building windows (the host galaxy's stars).

The problem for astronomers is that when they look at these "city streets" in the sky, the bright spotlight often washes out the details of the surrounding buildings. They want to know two things:

How big and heavy is the central spotlight (the supermassive black hole)?
How big and heavy is the city around it (the galaxy)?

If they can't separate the two, their measurements are a mess. This paper introduces a new, robust "digital filter" to cleanly separate the spotlight from the city lights in the light spectra (the rainbow of light) coming from these galaxies.

The Problem: A Blended Mess

When astronomers look at a galaxy with a black hole, the light is a jumbled mix.

The AGN (Spotlight): It's a super-bright, featureless beam of light that changes brightness over time.
The Galaxy (City): It's made of stars, which have specific "fingerprints" (absorption lines) in their light that tell us how fast the stars are moving and how heavy the galaxy is.

If the AGN is too bright, it hides the star fingerprints. If the AGN dims, the stars become visible again. The authors needed a way to mathematically "subtract" the AGN to see the stars clearly, and then "subtract" the stars to see the AGN clearly, all from the same piece of data.

The Solution: A Two-Step "Peeling" Process

The authors created a method that uses two specialized software tools, working like a team of expert editors:

Step 1: The "Star Counter" (pPXF): First, they use a tool designed for normal galaxies. It looks at the light and tries to match it to a library of "star templates." However, because the AGN is there, this tool might get confused. So, the authors tweaked it to say, "Okay, I see the stars, but I also see this weird extra light. Let's estimate how much of the total light is the AGN and how much is the stars." This gives them a rough estimate of the galaxy's mass and how fast its stars are moving.
Step 2: The "AGN Specialist" (PyQSOFit): Once they have a good estimate of the star light, they digitally subtract it from the image. What's left is the pure AGN light. Now, they hand this "cleaned" light to the second tool, which is an expert at analyzing quasars (bright AGNs). This tool measures the black hole's mass by looking at the speed of gas swirling around it.

The Analogy: Think of it like trying to hear a whisper in a noisy room.

Old way: You try to guess the whisper while the music is playing. You might get it wrong.
New way: You first record the music (the AGN), subtract it from the total sound, and then listen to what's left (the whisper/stars). Then, you take that whisper and analyze it on its own.

How They Tested It: The "Changing Look" Trick

To prove their method works, they didn't just use computer simulations (which can be faked). They used real data from the Sloan Digital Sky Survey (SDSS) in a clever way:

The "Changing-Look" AGNs: They found 32 galaxies where the AGN acts like a dimmer switch. Sometimes it's blindingly bright (Bright State), and sometimes it fades significantly (Dim State).
The Logic: The galaxy itself (the stars) and the black hole's mass do not change when the AGN dims. They are constant.
The Test: They applied their method to the "Bright" version and the "Dim" version of the same galaxy. If their method is good, the calculated mass of the galaxy and the black hole should be exactly the same in both cases, even though the light looks totally different.

The Result: It worked! When the AGN wasn't too overwhelmingly bright (specifically, when it contributed less than 80% of the total light), the measurements for the galaxy and black hole were consistent between the bright and dim states.

Key Findings in Plain English

The "80% Rule": If the AGN is providing more than 80% of the light, it's too bright to accurately measure the galaxy's properties. The "spotlight" is just too blinding. But if it's below 80%, the method is reliable.
Black Hole Mass: To get a reliable black hole mass, the gas swirling around it needs to be moving fast enough and be bright enough to be clearly distinguished from the background noise. The authors set strict rules (like requiring a signal 3 times stronger than the noise) to ensure they aren't guessing.
The "Breathing" Effect: They noticed something fascinating with the "Changing-Look" galaxies. The gas clouds around the black hole seem to "breathe." When the black hole gets brighter, the gas clouds expand and slow down slightly. When it dims, they shrink and speed up. This is expected physics, but they found some galaxies where the gas didn't behave this way, suggesting those specific systems might be chaotic or not fully settled.
Consistency: When they compared their results to other methods (like taking pictures of the galaxies and analyzing the light in 2D), their spectral method gave very similar answers for the galaxy's mass.

The Bottom Line

This paper presents a reliable, automated recipe for untangling the light of a galaxy from its central black hole. It allows astronomers to take a single snapshot of a galaxy, separate the "city" from the "spotlight," and accurately measure the weight of the stars and the size of the black hole. This is crucial for understanding how black holes and their host galaxies grow up together over billions of years.

The authors conclude that their method is ready for the future, where massive new surveys will capture millions of these spectra, helping us map the co-evolution of the universe's most massive objects.

1. Problem Statement

Active Galactic Nuclei (AGN) and their host galaxies are often observed as a single, blended spectrum in large-scale surveys (e.g., SDSS, DESI). Disentangling the emission from the supermassive black hole (SMBH) accretion disk and broad-line region (BLR) from the underlying stellar population of the host galaxy is critical for deriving accurate physical parameters.

The Challenge: Standard fitting methods often fail to simultaneously model the complex AGN continuum (power-law, Fe II, Balmer continuum) and the stellar population (Simple Stellar Populations - SSPs) without introducing degeneracies.
The Consequence: Inaccurate decomposition leads to biased measurements of stellar mass ( $M_*$ ), stellar velocity dispersion ( $\sigma_*$ ), and black hole mass ( $M_{BH}$ ). These biases undermine the reliability of scaling relations used to study the co-evolution of galaxies and SMBHs.
Specific Need: A robust, automated method is required to handle the diversity of AGN spectra (varying obscuration, luminosity, and activity phases) across millions of spectra from upcoming surveys.

2. Methodology

The authors propose a novel, two-step iterative spectral fitting approach that combines the strengths of two existing software packages: pPXF (Penalized Pixel-Fitting) and PyQSOFit.

Step 1: Host Galaxy Modeling (pPXF)
- Goal: Model and subtract the stellar population contribution while accounting for AGN contamination to avoid overestimating stellar mass.
- Implementation: The code fits the observed spectrum using a linear combination of:
  - SSPs: E-MILES templates with Padova isochrones and a Salpeter IMF.
  - AGN Components: A featureless power-law continuum, Fe II pseudo-continuum (optical and UV), Balmer continuum, and high-order Balmer emission.
  - Emission Lines: Gaussians for both narrow (NELs) and broad (BELs) lines.
- Output: The weight of the AGN components relative to the SSPs yields the AGN continuum fraction ( $f_{AGN}$ ). The kinematic moments provide the stellar velocity dispersion ( $\sigma_*$ ) and aperture stellar mass ( $M_*$ ). The SSP templates are then subtracted from the observed spectrum.
Step 2: AGN Modeling (PyQSOFit)
- Goal: Fit the residual AGN-dominated spectrum with high precision.
- Implementation: PyQSOFit fits the isolated AGN spectrum using:
  - Continuum: A power-law combined with a third-order polynomial (to account for intrinsic dust reddening) and flexible Balmer/Fe II components.
  - Emission Lines: 1–3 Gaussians per line to disentangle blended features (e.g., [N II] vs. H $\alpha$ ).
- Output: Precise measurements of broad-line flux, Full Width at Half Maximum (FWHM), and continuum luminosity, which are used to calculate single-epoch SMBH masses.
Validation Strategy:
- Internal Validation: Using 62 objects with multiple SDSS observations:
  - 32 Changing-Look AGN (CL-AGN) in bright and dim states.
  - 15 Quasars (QSO) and 15 Galaxies (GAL) with multiple single-epoch and stacked spectra.
- External Validation: Comparing results against:
  - Photometric image decomposition (Li et al. 2024, HSC).
  - SED fitting codes (GRAHSP, Buchner et al. 2024; CIGALE, Yu et al. 2023).
  - High-resolution spectroscopy (VLT/X-Shooter and Gemini/GMOS) for $\sigma_*$ .

3. Key Contributions

Hybrid Fitting Framework: The integration of pPXF (optimized for stellar kinematics) and PyQSOFit (optimized for AGN line profiles) creates a robust pipeline that minimizes the degeneracy between young stellar populations and AGN continuum.
Quality Criteria Definition: The authors establish specific thresholds for reliable measurements:
- Host Properties: Reliable only when $f_{AGN} < 0.8$ (AGN contributes less than 80% of the continuum).
- SMBH Mass: Reliable only when broad-line flux and FWHM have a signal-to-noise ratio (error) $> 3$ ( $3\sigma$ confidence).
- Velocity Dispersion: Reliable only when $\sigma_* > 70$ km s $^{-1}$ (SDSS instrumental resolution limit).
Quantified Uncertainties: The paper provides systematic uncertainty estimates for key parameters derived from this method, offering a benchmark for future large-scale surveys.

4. Key Results

A. Host Galaxy Properties

Stellar Mass ( $M_*$ ):
- Internal comparisons show an intrinsic scatter of 0.14 dex for spectra with $f_{AGN} < 0.8$ .
- External comparisons with photometric methods (HSC, GRAHSP, CIGALE) show systematic offsets of 0.05–0.19 dex and scatters of 0.31–0.41 dex. The authors' method is competitive with, and in some cases more precise than, these photometric techniques.
- Conclusion: The method yields consistent stellar masses when the AGN does not dominate the continuum.
Stellar Velocity Dispersion ( $\sigma_*$ ):
- Internal scatter is 35 km s $^{-1}$ (improving to 27.5 km s $^{-1}$ for S/N > 10).
- Measurements are unreliable below the SDSS resolution limit ( $\sigma_* < 70$ km s $^{-1}$ ).
- External validation against VLT/Gemini data confirms a scatter of 34 km s $^{-1}$ , validating the method's consistency despite different slit widths.

B. AGN Properties (Black Hole Mass)

Single-EPOCH Mass Estimates:
- Using H $\alpha$ and H $\beta$ virial estimators, the method produces consistent mass estimates across different epochs for stable systems (QSO and GAL samples).
- The scatter for "Clean" samples (reliable broad lines) is 0.10 dex (H $\alpha$ ) and 0.13 dex (H $\beta$ ), which is significantly lower than the typical 0.3–0.5 dex systematic scatter inherent to single-epoch methods.
- Changing-Look AGN (CL-AGN): These objects show larger scatter (0.24 dex) and deviations from virial expectations. Some exhibit "breathing" BLRs (correlated changes in luminosity and line width), while others deviate significantly, suggesting non-virialized systems during spectral transitions.

C. Scaling Relations

The derived $M_{BH}$ vs. $M_*$ and $M_{BH}$ vs. $\sigma_*$ relations align well with established low-redshift scaling relations (e.g., Kormendy & Ho 2013; Reines & Volonteri 2015), confirming the method's ability to extract physically consistent parameters for co-evolution studies.

5. Significance and Outlook

Survey Readiness: This methodology is specifically designed to handle the massive data volumes expected from next-generation surveys like DESI, 4MOST, and PFS, as well as the SDSS-V Black Hole Mapper (BHM) program.
Reliability: By defining strict quality cuts ( $f_{AGN} < 0.8$ , $3\sigma$ line detection), the authors provide a "safe" protocol for extracting host and AGN properties from blended spectra, reducing the risk of biased scientific conclusions.
Future Application: The authors plan to apply this method to the full eFEDS/SPIDERS sample (~60,000 spectra requiring decomposition). The code will be released as part of the SDSS Value Added Catalog, enabling the community to study the co-evolution of SMBHs and galaxies with unprecedented precision.
Scientific Insight: The analysis of CL-AGN reveals that while many follow the "breathing" BLR model, a subset exhibits non-virialized behavior, offering new clues about the dynamics of AGN spectral transitions.

In summary, this paper presents a validated, robust, and automated framework for decomposing AGN and host galaxy spectra, providing a critical tool for the next generation of extragalactic astronomy.