# Euclid preparation. XXV. The Euclid Morphology Challenge -- Towards model-fitting photometry for billions of galaxies

**Authors:** S. Davini, N. Mauri, L. Patrizii, G. Sirri, Y. Wang, A. A. Nucita, O. Ilbert, M. Meneghetti,  Euclid Collaboration, G. Desprez, S. Paltani, J. Coupon, M. Brescia, S. Cavuoti, W. G. Hartley, A. Tramacere, M. Castellano, F. Dubath, E. Merlin, S. Andreon  
**arXiv:** [2209.12906](https://arxiv.org/abs/2209.12906)  
**Date:** 2022-09-26  
**Summary page:** https://www.summarizepaper.com/en/paper/2209.12906v1/

## Key points

- The Euclid mission aims to provide high-quality imaging for approximately 1.5 billion galaxies.
- The Science Ground Segment of the Euclid Consortium is developing a software pipeline that includes a model-fitting algorithm to process and analyze such vast amounts of data in real-time.
- The Euclid Morphology Challenge is an investigation comparing the performance of five model-fitting software packages on simulated Euclid data to identify the most suitable algorithm for implementation in the pipeline.
- Diagnostic plots were created to analyze results, and ad-hoc metrics were defined, including the global metric S used to evaluate each software package's performance on three different realizations: SS, DS (IE-only), and RM (multi-band).
- Each code had strengths and weaknesses; therefore, no rigorous evaluation or ranking was possible based solely on these metrics. However, all software packages provided acceptable to good results in at least some cases.
- Simultaneous fitting of multi-band data sets proved important as using information from deep high-resolution IE band helped obtain more accurate measurements on shallower NIR bands as well as ground-based LSST bands.
- While the study did not investigate the performance of model-fitting algorithms in the context of the Euclid Deep Survey, it is expected that results should be similar to those obtained for the Wide Survey at equivalent signal-to-noise ratio (S/N).
- The results set a baseline for deciding which algorithm to implement in the Euclid pipeline, with one possible approach being to use a combination of chosen algorithms. Ultimately, these findings will benefit both core science goals and legacy use within the community.

## Summary

The Euclid mission, led by the European Space Agency (ESA), aims to provide high-quality imaging for approximately 1.5 billion galaxies. To process and analyze such a vast amount of data in real-time, the Science Ground Segment of the Euclid Consortium is developing a software pipeline that includes a model-fitting algorithm. This algorithm will provide photometric and morphological estimates crucial to achieving the core science goals of the mission and for legacy science. The Euclid Morphology Challenge is an investigation comparing the performance of five model-fitting software packages on simulated Euclid data to identify the most suitable algorithm for implementation in the pipeline. The simulated dataset used in this study included five fields of view, each with three realizations of galaxy profiles: single and double S\'ersic, and 'realistic' profiles obtained with a neural network. For one field in the double S\'ersic realization, images were also simulated for three near-infrared bands ($Y_E$, $J_E$, $H_E$) and five Rubin/LSST optical complementary bands ($u$, $g$, $r$, $i$, and $z$). Diagnostic plots were created to analyze results, and ad-hoc metrics were defined. The global metric S was used to evaluate each software package's performance on three different realizations: SS, DS (IE-only), and RM (multi-band). Each code had strengths and weaknesses; therefore, no rigorous evaluation or ranking was possible based solely on these metrics. However, all software packages provided acceptable to good results in at least some cases. Additionally, it was observed that all codes tended to underestimate uncertainties at the bright end but provided reasonable estimates for intermediate- and faint-sources statistically. Simultaneous fitting of multi-band data sets proved important as using information from deep high-resolution IE band helped obtain more accurate measurements on shallower NIR bands as well as ground-based LSST bands. While the study did not investigate the performance of model-fitting algorithms in the context of the Euclid Deep Survey, it is expected that results should be similar to those obtained for the Wide Survey at equivalent signal to noise ratio (S/N). However, increased contamination and blending might worsen performance. The results of this study set a baseline for deciding which algorithm to implement in the Euclid pipeline, with one possible approach being to use a combination of chosen algorithms. Ultimately, these findings will benefit both core science goals and legacy use within the community.