A Pixon-based Method for Reverberation-mapping Analysis in Active Galactic Nuclei

Yan-Rong Li, Ming Xiao, Jianmin WangThe Astrophysical Journal 2021 · DOI

Reverberation mapping (RM) is a cornerstone technique in modern astrophysics, serving as our primary tool for "weighing" supermassive black holes and mapping the gaseous environments of Active Galactic Nuclei (AGNs). By observing the time delay between variations in the central continuum emission and the subsequent "echo" in broad emission lines, astronomers can infer the geometry and kinematics of the Broad-Line Region (BLR). However, the mathematical core of RM is an "ill-posed inverse problem"—reconstructing a coherent map from noisy, irregularly sampled time-series data is notoriously difficult.

This paper introduces a novel solution to this challenge by adapting the pixon algorithm, originally developed for image reconstruction, to the domain of reverberation mapping. This approach offers a more flexible and accurate alternative to traditional methods, providing a clearer view into the hearts of distant galaxies.

The Pixon Concept: Adaptive Resolution

The fundamental innovation of this work lies in the use of "pixons" rather than standard fixed pixels. In traditional reconstruction methods like the Maximum Entropy Method (MEM), a uniform smoothing constraint is often applied across the entire map. This can lead to over-smoothing of sharp features or the introduction of artifacts in low-signal regions.

The pixon algorithm, by contrast, utilizes a variable pixon basis. This allows the resolution of the reconstructed map to adjust locally based on the information content of the data. In regions where the signal is strong and features are sharp, the algorithm employs smaller pixons for high resolution. In noisier or featureless regions, it uses larger pixons to smooth out fluctuations, effectively "turning off" resolution where it isn't justified by the evidence.

Comparison of various pixon basis functions used to adjust local resolution.

As shown in the figure above, various kernel shapes—such as Gaussian, Lorentzian, or Top-hat functions—can serve as the basis for these pixons, allowing the model to adapt to different physical scenarios.

Methodology and Implementation

The authors developed a generic framework to implement the pixon algorithm for RM analysis. The process involves several key steps:

  1. Continuum Reconstruction: Since the driving continuum light curve is often gappy, it must be interpolated. The framework offers two paths: a pixon-based reconstruction and a Damped Random Walk (DRW) process, which is the standard statistical model for AGN variability.
  2. Transfer Function Recovery: The algorithm seeks the "simplest" model—the one with the minimum number of pixons—that provides an acceptable fit to the observed emission-line data.
  3. Optimization: A truncated Newton algorithm is used to maximize the posterior probability, iteratively refining the pixon sizes until the transfer function stabilizes.

Validation through Simulation

To test the performance of the pixon-based method, the authors used simulated datasets mimicking real AGN behavior. They tested the algorithm against various cadences (0.5 to 10 days) and signal-to-noise ratios (S/N 10 to 100).

Reconstruction of light curves and transfer functions using Pixon and DRW methods across different kernel types.

The results demonstrate high fidelity in recovering complex transfer functions. A key finding in the study is the method's ability to capture sharp, discontinuous features—like top-hat functions—that traditional methods like MEM typically round off. The pixon method consistently produced narrower, more accurate peaks and better-defined edges.

Key Insights and Future Directions

The pixon-based approach offers several advantages over the industry-standard MEM:

  • Local Adaptivity: It prevents noise in one part of the map from degrading resolution elsewhere.
  • Model Independence: The pixon-based continuum reconstruction is less dependent on specific physical models of AGN variability, making it robust for unusual sources.
  • Information Efficiency: It successfully ignores spurious features in low-information regions by increasing pixon size where data is sparse.

While this paper focuses on "velocity-unresolved" RM (1D time-delay maps), it lays the groundwork for more complex analyses. The authors note that an extension to "velocity-resolved" RM (2D maps of both velocity and delay) is forthcoming. Future work may also integrate Markov Chain Monte Carlo (MCMC) methods to provide formal uncertainty quantification, which is currently a limitation of the "best-fit" pixon approach.

Reference: Li, Y.-R., Xiao, M., & Wang, J.-M. (2021). A Pixon-based Method for Reverberation-mapping Analysis in Active Galactic Nuclei. The Astrophysical Journal, 921(2), 151. DOI: 10.3847/1538-4357/ac1c71

Summary updated

Abstract

Abstract We present an alternative method for reconstructing a velocity-delay map in reverberation mapping (RM) based on the pixon algorithm initially proposed for image reconstruction by Pina. The pixon algorithm allows for a variable pixon basis to adjust resolution of each image pixel according to the information content in that pixel, which therefore enables the algorithm to make the best possible use of measured data. The final optimal pixon basis functions would be those that minimize the number of pixons while still providing acceptable descriptions to data within the accuracy allowed by noises. We adapt the pixon algorithm to RM analysis and develop a generic framework to implement the algorithm. Simulation tests and comparisons with the widely used maximum entropy method demonstrate the feasibility and high performance of our pixon-based RM analysis. This paper serves as an introduction to the framework and the application to velocity-unresolved RM. An extension to velocity-resolved cases will be presented in a companion paper.
BibTeX
@article{li2021pixon,
  author = {Yan-Rong Li and Ming Xiao and Jianmin Wang},
  title = {{A Pixon-based Method for Reverberation-mapping Analysis in Active Galactic Nuclei}},
  journal = {The Astrophysical Journal},
  year = {2021},
  doi = {10.3847/1538-4357/ac1c71},
}
Figure