Logo image
Adaptive Reconstruction of Cluster Halos (ARCH): Integrating Shear and Flexion for Substructure Detection
Preprint   Open access

Adaptive Reconstruction of Cluster Halos (ARCH): Integrating Shear and Flexion for Substructure Detection

Jacob Shpiece and David M Goldberg
ArXiv.org
25 Sep 2025
url
https://doi.org/10.48550/arxiv.2509.21686View
Open

Abstract

Physics - Cosmology and Nongalactic Astrophysics
We present ARCH (Adaptive Reconstruction of Cluster Halos), a new gravitational lensing pipeline for cluster mass reconstruction that applies a joint shear-flexion analysis to JWST imaging. Previous approaches have explored joint shear+flexion reconstructions through forward modeling and Bayesian inference frameworks; in contrast, ARCH adopts a staged optimization strategy that incrementally filters and selects candidate halos rather than requiring a global likelihood model or strong priors. This design makes reconstruction computationally tractable and flexible, enabling systematic tests of multiple signal combinations within a unified framework. ARCH employs staged candidate generation, local optimization, filtering, forward selection, and global strength refinement, with a combined fit metric weighted by per-signal uncertainties. Applies to Abell 2744 and El Gordo, the pipeline recovers convergence maps and subcluster masses consistent with published weak+strong lensing results. In Abell 2744 the central core mass within 300 $h^{-1}$kpc is$2.1\times 10^{14} M_\odot h^{-1}$ , while in El Gordo the northwestern and southeastern clumps are recovered at$2.6\times 10^{14} M_\odot h^{-1}$and$2.3\times 10^{14} M_\odot h^{-1}$ . Jackknife resampling indicates typical 1 $σ$uncertainties of$10^{12}-10^{13} M_\odot h^{-1}$ , with the all signal and shear+ $\mathcal{F}$reconstructions providing the most stable results. These results demonstrate that flexion, when anchored by shear, enhances sensitivity to cluster substructure while maintaining stable cluster-scale mass recovery.

Metrics

5 Record Views

Details

Logo image