💥 Understanding Dark Matter through Extragalactic Shocks

Summary

Junior Undergraduate Research project (CNPq/PIBIC-EM, Call 94/2022), advised by Prof. Ana Cecília Soja at IFF Bom Jesus do Itabapoana. I tested the accuracy of Dawson et al.’s (2013) code — which estimates the elapsed time since two galaxy clusters collided, via Monte Carlo — against ZuHone et al.’s (2018) dynamical simulations, as an indirect way to study dark matter’s behavior during extreme collisions.

Galaxy Cluster Shocks

🌌 The problem: how do you “see” dark matter?

Galaxy clusters are the largest gravitationally-bound structures in the Universe — and when two of them collide, the event is among the most energetic known. During a collision, a cluster’s three components (galaxies, intracluster gas, and dark matter) behave very differently: galaxies, made of ordinary matter but very sparse relative to each other, pass through each other with barely any interaction; gas, also ordinary matter, collides and is slowed by friction; and dark matter appears to follow the galaxies, but not exactly — indirect evidence that it interacts little (or not at all) through channels beyond gravity. The most famous example is the Bullet Cluster, whose gravitational-lensing maps show exactly this spatial separation between the three components.

Since dark matter can’t be observed directly, nor can a cluster collision be repeated in a lab, the strategy is indirect: compare dynamical simulations against statistical estimation methods for observational parameters (relative masses, redshift, projected separation), and check whether they agree.

🎯 Objective

Assess the accuracy of Dawson’s (2013) code — which uses a Monte Carlo method to estimate the elapsed time since a pair of clusters first collided, based on relatively simple-to-obtain observational parameters — by comparing its results against the known “ground truth” of ZuHone et al.’s (2018) high-resolution dynamical simulations.

🔧 Methodology

The work followed four stages:

1. Familiarization with fundamental Astronomy concepts (parallax, OBAFGKM spectral classification, stellar evolution, Hubble’s morphological galaxy classification — elliptical, spiral, irregular) and with the physics of colliding clusters.
2. Compiling and understanding Dawson’s code — first validated against the Bullet Cluster’s own reference case (masses $1.5\times 10^{14}$ and $1.5\times 10^{15}{M}_{\odot }$, projected separation of 720 kpc), reproducing the original result. The core function, MCEngine, takes both clusters’ masses, redshift, and projected distance, generating $10^4$ Monte Carlo samples (convergence already observed from $10^3$ iterations onward).
3. Obtaining ZuHone et al.’s (2018) data — the Galaxy Cluster Merger Catalog, a repository of hydrodynamical cluster-merger simulations organized by mass ratio (1:1, 1:3, 1:10) and impact parameter (0, 500, 1000 kpc). The work focused on the 3 simulations with zero impact parameter (collision in the plane of the sky).
4. Applying Dawson’s method to each ZuHone simulation, comparing the code’s estimated post-collision time against the simulation’s known ground truth, with uncertainty estimated via np.quantile over the $10^4$ Monte Carlo samples.

📊 Results

ZuHone’s simulation reveals an oscillatory pattern: clusters start at maximum separation, collide (black line, first collision), separate again to a new maximum — smaller than the first, due to energy lost in the collision — and so on.

Mass ratio	1st collision time (Gyr)
1:1	1.32
1:3	1.20
1:10	1.04

Comparing Dawson’s code against ZuHone’s data in the window between the first collision and the following maximum separation — the only window where Dawson’s method applies — the code’s results agree, within uncertainties, with the simulation for all three tested mass ratios (1:1, 1:3, and 1:10).

Systematic bias found

Despite the good overall agreement, Dawson’s code’s central estimates showed a systematic tendency to underestimate the simulation’s true time — a bias that needs deeper investigation in future work, though it doesn’t invalidate the method’s overall viability.

🧾 Conclusion

Dawson’s (2013) method proved reliable within uncertainties for estimating the elapsed time since a cluster collision, within the method’s own stated range of validity — but with a systematic underestimation bias deserving future investigation. The natural next step is extending the analysis to ZuHone’s non-zero impact-parameter scenarios (off-sky-plane collisions), not yet tested in this work.

🏆 Presentations and awards

This project was presented at FEBRACE 2023 and MOSTRATEC 2023 (Novo Hamburgo, Brazil) — see the MOSTRATEC coverage (Portuguese only).

🔗 References and related

• Dawson, W. A. (2013) — The Dynamics of Merging Clusters: A Monte Carlo Solution Applied to the Bullet and Musket Ball Clusters, ApJ 772, 131. MCMAC code.
• ZuHone, J. et al. (2018) — The Galaxy Cluster Merger Catalog: An Online Repository of Mock Observations from Simulated Galaxy Cluster Mergers, ApJS 234, 4.
• Clowe, D. et al. — the Bullet Cluster, the classic evidence for spatial separation between dark matter and gas.
• MOSTRATEC 2023 — coverage of this project’s presentation (Portuguese only)
• Anomaly Detection in Gaia Data — another Astronomy research project, also grounded in the dynamics/kinematics of gravitational systems
• Simulating the Impact of Satellites on Astronomical Observations — following project, also computationally focused on astronomical data
• CursoON — Lecture 05 — another context for non-luminous mass/dark matter within the Galaxy