5. Searching For Anomolous Galaxies in Science Archives with Machine Learning

     

    ESA supervisor: David O´Ryan
    Collaborator(s): Sandor Kruk, Pablo Gomez

    Site: ESAC

    Galaxies host an assortment of strange morphological features depending on their evolutionary history. These can range from the formation of tidal features in galaxy interaction, the formation of bars or spiral arms due to secular processes, or the complete removal of gas in dense environments. Thus, by finding galaxies with eccentric and unique morphologies, we can probe at the most extreme processes which drive galaxy evolution.

    An example of an anomalous galactic system with an extreme morphology is the jellyfish galaxy. These systems reside in the densest of galactic environments, where the intergalactic medium acts to ram pressure strip the gas out of the galaxy. As the gas is removed, long tendrils form giving the system the unique morphology that led to its name. The removal of gas has disastrous consequences for the galaxy, rapidly quenching it and leaving it as a 'red and dead' galaxy where no further stars form.

    Due to the unique and extreme environments of these anomalous systems, we have observed relatively few of them. This, in turn, means they are difficult to study with any statistical significance. We propose to use a machine learning algorithm, specifically developed to classify galaxy morphology, to discover more of these systems in the Euclid science archives. We will create image cutouts of many sources from Euclid and use our machine learning algorithm to classify their morphology. We will use the novel power of the ESA Datalabs platform to achieve this. This platform will provide us with direct access to Euclid's science archives and allow us to efficiently explore Terabytes of observational data and create millions of source cutouts for classification.

    Project duration: 6 months.

    Desirable expertise or programming language:

    • Experience with the Python programming language.
    • Experience with Anaconda (or equivalent) and JupyterLab.
    • Experience with using and deploying machine learning algorithms.
    • Experience with handling large astronomical datasets and imaging.
    • Knowledge of extragalactic astronomy is desired, but not essential.
    • Knowledge of CUDA and GPU programming is desired, but not essential.

     

    To apply for this project please fill in an online application form through the following link.