Deep learning for slit-less Spectroscopic Redshift survey Simulator¶
Welcome to DISPERS!¶
Launched on July 1st, the Euclid mission heralds a new era in cosmological exploration. The goal of the DISPERS project is to take advantage of advanced Machine Learning (ML) approachs to be at the forefront of this revolution by exploring novel techniques in spectral calibration, Point Spread Function (PSF) and detector modeling, and instrument simulation.
Euclid/NISP¶
Euclid is a space mission led by the European Space Agency (ESA) with the primary goal of studying dark energy and dark matter, two mysterious components that make up most of the universe. By observing the accelerated expansion of the universe, Euclid aims to provide insights into the nature of these phenomena, enhancing our understanding of the cosmos.
The Near Infrared Spectrometer and Photometer (NISP) is one of the two instruments aboard the Euclid spacecraft, designed to capture near-infrared light from far galaxies. Its primary function is to measure the redshifts of galaxies accurately. Redshift is a crucial indicator of how much the universe has expanded since the light from a galaxy was emitted. By precisely measuring redshifts, NISP helps astronomers map the large-scale structure of the universe, providing essential data for Euclid's dark energy and dark matter studies.
Why DISPERS?¶
In the past few years, several data-driven approaches have been developed to generate synthetic images using machine learning (ML) approaches and have demonstrated their capability to generate realistic images (Lanusse et al. 2021, Guilloteau et al. 2019, Van Den Oord et al. 2016, Razavi et al. 2019, Smith et al. 2019). Currently, most of the literature focuses on photometric images simulations in the context of the studies of galaxy morphologies, which implies the modelling of both the instrument PSF and galaxy profiles. This project takes part in such cosmological simulation framework, aiming at developing a new and innovative method for simulating slitless spectroscopic images by relying on state-of-the-art ML algorithms.
What We Do¶
The project is organized around four aspects to evaluate and simulate the NISP instrument response. Those aspects are distributed under four work-packages. Three of them will be dedicated to the data analyses and simulation of specific properties of the NISP’s instrument response. The fourth work-package aims at compiling the different models to provide the end-to-end simulations that will allow fast on-sky simulations.
WP1: Spectral Calibration Models¶
WP1 focuses on developing precise calibration models essential for the NISP instrument. These models enable accurate adjustments to raw data, ensuring the reliability and integrity of the collected information. Calibration is vital for interpreting observations correctly and extracting meaningful scientific conclusions.
WP2: PSF Models¶
WP2 is dedicated to the creation of Point Spread Function (PSF) models. PSF describes how a point source appears after being detected by the instrument. Developing detailed PSF models is crucial for understanding the instrument's imaging capabilities. Accurate PSF models are foundational for various astronomical analyses, including flux measurements, source deblending and shape measurement.
WP3: Detector Models¶
WP3 focuses on constructing intricate models of the detectors used in the NISP instrument. These models account for detector responses to different wavelengths, temperatures, and illumination levels. Understanding detector behavior is essential for correcting observational data and ensuring the instrument's sensitivity across various conditions.
WP4: Simulation Models¶
WP4 is centered around the development of comprehensive simulation models. These models simulate the entire process of observation. Simulation models are invaluable for testing instrument performance, refining data analysis techniques, and predicting outcomes for different astronomical scenarios. They serve as indispensable tools for mission planning and scientific interpretation.