Joint Bayesian Source and Lens Reconstruction for Multi-messenger Binary Black Holes

This paper introduces *silmarel*, the first software package designed to jointly reconstruct gravitational wave sources and their lensing galaxies by integrating data from gravitational wave detectors with electromagnetic surveys like Euclid and Hubble, thereby enabling the identification of lensed binary black hole events in the era of multi-messenger astronomy.

The Gist

Imagine you are a detective trying to solve a cosmic mystery. You have two very different types of clues: one is a sound (a gravitational wave) and the other is a picture (light from a galaxy). Your goal? To prove that these two clues belong to the same event, even though they look and sound completely different.

This paper introduces a new digital tool called Silmaril (a name inspired by the "Silmarils" from fantasy literature, representing precious light) that helps astronomers connect these clues.

Here is the story of the paper, broken down into simple concepts:

1. The Cosmic "Funhouse Mirror"

Imagine looking at a distant star through a funhouse mirror made of a massive galaxy or a cluster of galaxies sitting between you and the star. This massive object bends space itself, acting like a lens.

• The Effect: Just like a funhouse mirror can stretch your reflection or show you multiple copies of yourself, this cosmic lens can bend light from a distant galaxy, creating multiple images or distorting its shape.
• The Sound: It does the same thing to "sounds" of the universe called Gravitational Waves (ripples in space caused by crashing black holes). If a black hole collision happens behind a lens, we might hear the same "chirp" multiple times, slightly delayed and louder or quieter each time.

2. The Problem: The "Ghost" Black Holes

For a long time, astronomers could only link these sound and picture clues if the event was a "loud" explosion involving stars (like two neutron stars crashing), which creates a bright flash of light.

• The Issue: Most black hole collisions are "silent" in light. They don't glow. They are like ghosts. We hear them (via gravitational waves), but we can't see them.
• The Clue: Even though the black holes are dark, they live inside galaxies. Those galaxies are bright. So, if we can find the lens that distorted the galaxy's picture, that same lens must have distorted the black hole's sound.

3. The Old Way vs. The New Way

The Old Way (Too Slow):
Previously, to prove a connection, scientists tried to simulate the entire universe from scratch. They would guess the shape of the lens, guess the black hole's details, and run supercomputers for weeks to see if the math matched. It was like trying to find a specific needle in a haystack by rebuilding the entire haystack, grain by grain, every time you looked.

The New Way (Silmaril):
The authors created Silmaril, a software package that acts like a smart shortcut.

• The Analogy: Imagine you already have a "fingerprint" of the black hole sound from previous observations (this is the data the LIGO/Virgo observatories already release). Instead of re-simulating the sound, Silmaril says, "Okay, we know what the sound should look like. Let's just check if the lens we see in the picture matches the changes in that sound."
• The Magic Trick: It separates the heavy lifting. It takes the "sound" data (which is already processed) and the "picture" data (the galaxy lens) and combines them mathematically without needing to run the massive simulations again.

4. Why This Matters

This tool is a game-changer for two reasons:

1. It finds the "Host": It can pinpoint exactly which galaxy the invisible black holes live in, turning a "ghost" event into a mapped location.
2. It's Super Fast: By using the "fingerprint" method (posterior distributions) instead of raw simulation, it turns a task that used to take weeks into something that can be done in hours or days.

5. The Future

The paper presents the "alpha version" (the first draft) of this tool. It's like a prototype car that drives well but needs a few more features. The team plans to make it even smarter, allowing it to work with different types of lens-mapping software and eventually helping us understand the nature of dark matter and the expansion of the universe.

In a nutshell:
Silmaril is a new translator that helps astronomers match the echoes of crashing black holes with the shadows of the galaxies they live in, using the cosmic lens as the connecting bridge. It turns a slow, impossible puzzle into a solvable mystery.

Technical Summary

Here is a detailed technical summary of the paper "Joint Bayesian Source and Lens Reconstruction for Multi-messenger Binary Black Holes" presented at the 44th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering.

1. Problem Statement

The paper addresses the challenge of identifying strongly lensed gravitational wave (GW) events, specifically Binary Black Hole (BBH) mergers, and associating them with their host galaxies and the lensing structures (galaxies or clusters) responsible for the lensing.

• The Core Difficulty: While GWs can be lensed, the BBHs themselves do not emit light. Unlike binary neutron star mergers (e.g., GW170817), which have bright electromagnetic (EM) counterparts, BBHs are "dark." Therefore, traditional multi-messenger association (matching light to GW) is impossible without identifying the host galaxy.
• The Hypothesis: If a GW event is lensed, its host galaxy is also lensed by the same structure. By identifying matching lenses in EM data (e.g., from Euclid, HST, JWST) and GW data, one can link the two.
• Computational Bottleneck: A rigorous joint Bayesian analysis requires evaluating a high-dimensional likelihood function that combines GW waveform generation (computationally expensive, taking hours to weeks) with EM lens reconstruction (forward modeling of galaxy morphology). Performing a full joint parameter estimation (JPE) for every potential lens-source pair is computationally prohibitive.
• Degeneracies: GW-only lens reconstruction suffers from "mass-sheet" and "similarity transformation" degeneracies because GWs lack the spatial resolution to uniquely constrain the lens model without EM priors. Conversely, EM data lacks the precise time-delay resolution provided by GWs.

2. Methodology: The silmarel Framework

The authors introduce silmarel (Statistical Inference for Lensing Multi-messenger Assocation, REconstruction, and Localisation), a Bayesian software package designed to bridge GW and EM data.

A. Theoretical Foundation

The method relies on the Lens Equation ($\vec{\beta} = \vec{\theta} - \vec{\alpha}(\vec{\theta})$) and the modification of GW waveforms by lensing. For a lensed GW image $j$, the waveform is:
$$\tilde{h}_{L,j}(f) = \sqrt{\mu_j} e^{-2\pi i (f \Delta t_j - \frac{n_j \pi}{2})} \tilde{h}_U(f)$$
Where $\mu_j$ is magnification, $\Delta t_j$ is time delay, and $n_j$ is the Morse factor (parity).

B. The "Fast Likelihood" Approximation

To overcome computational costs, silmarel reformulates the joint likelihood by replacing expensive full waveform generation with posterior-induced effective likelihoods. The authors make several key simplifications based on existing LVK data products:

1. Parameter Separation: Most BBH parameters (mass, spin) are unaffected by lensing. Only the effective luminosity distance and arrival time are modified by lensing parameters ($\mu_{rel}, \Delta t$).
2. Utilization of Existing Posteriors: Instead of re-running GW inference, the method uses pre-computed posterior distributions ($p(\vec{\phi}_{BBH}, \vec{\phi}_L | d_{GW})$) from LVK catalog releases.
3. Gaussian Approximation: The GW likelihood is approximated as a Gaussian function centered on the median time delay and magnification derived from the GW posteriors, rather than the full waveform.

The joint likelihood is expressed as:
$$p(d_{GW}, d_{EM}, d_{spect} | H_A) = \int p(\vec{d}_{GW} | \vec{\mu}_{rel}, \Delta \vec{t}) p(d_{EM}, d_{spect} | \vec{\theta}_L, \vec{\theta}_l, z_l, z_s) p(\vec{\theta}'') d\vec{\theta}''$$

C. Handling Uncertainties

A critical technical adjustment involves time delay uncertainty:

• GWs measure time delays with millisecond precision.
• EM imaging and lens models cannot reconstruct deflection potentials to this precision.
• Solution: The authors assign a larger, conservative uncertainty to the GW time delay term in the likelihood. This prevents the rejection of correct lenses simply because the EM model cannot match the extreme precision of the GW measurement, effectively bridging the resolution gap.

D. Implementation

The software integrates with lenstronomy (a standard EM lens modeling tool). It operates in stages:

1. EM Reconstruction: Reconstructs the lens and source galaxy morphology.
2. GW Localisation: Uses the reconstructed lens model to predict time delays and magnifications, comparing them against GW posteriors to calculate the association probability and refine the source location.

3. Key Contributions

• First Dedicated Software: silmarel is the first software package designed specifically to perform joint Bayesian analysis of lensed GW binaries and EM telescope observations.
• Computational Efficiency: By decoupling the full waveform generation from the lens reconstruction and using posterior-based likelihoods, the method reduces analysis time from weeks to a feasible timeframe, enabling the processing of real LVK data.
• Host Galaxy Identification: It provides a statistical framework to link "dark" BBH mergers to their host galaxies via the lensing structure, even in the absence of an optical counterpart to the merger itself.
• Precision Localisation: The method claims to localize GW sources within their host galaxies with precision orders of magnitude better than standard LVK sky localizations.

4. Results

The paper presents a demonstration using simulated data:

• Simulation: A mock observation of a lensed system was generated.
• Reconstruction: silmarel successfully reconstructed the lens mass profile and the source position.
• Visualization (Fig. 1): The results show the initial simulated observation, the final reconstructed lens, and the source plane localisation. The posterior contours (black) tightly enclose the true source position (pink star), demonstrating the method's ability to recover the source location and lens properties simultaneously.
• Performance: The "fast likelihood" approach successfully separated the EM reconstruction and GW localisation stages, validating the modular architecture.

5. Significance and Future Outlook

• Scientific Impact: This work enables the transition from searching for lensed GWs in isolation to a multi-messenger approach. It allows cosmologists to use lensed BBHs as standard sirens to probe the expansion of the universe, large-scale structure, and dark matter nature with unprecedented precision.
• Scalability: The framework is designed to scale with upcoming data from the LIGO-Virgo-KAGRA (LVK) network, Euclid, Hubble, JWST, and the Chinese Space Station Telescope (CSST).
• Future Directions:
  • Integrated Likelihoods: Moving from the current "fast estimator" (staged analysis) to a fully integrated likelihood that combines GW and EM reconstruction simultaneously.
  • Framework Expansion: Adapting silmarel to work with other EM reconstruction tools like herculens and PyAutoLens.
  • Machine Learning: Exploring ML approaches to further accelerate GW waveform generation and handle high-dimensional inference problems.

In conclusion, silmarel represents a pivotal step toward the practical realization of multi-messenger gravitational wave astronomy, transforming the theoretical potential of lensed BBHs into an analyzable reality.