- Extremely Metal-poor Galaxies -

The physics of galaxy formation and evolution during the first billion years after the Big Bang critically shape galaxy properties across cosmic time. With the launch of the James Webb Space Telescope (JWST), astronomers are desperately searching for the first galaxies to shed light on the physics igniting and driving the formation and evolution of galaxies. However, given the intrinsically faint emission from the first galaxies, a thorough investigation on the chemical evolution of these galaxies and the environment they reside in remains challenging even with JWST. Extremely metal-poor galaxies (XMP) in the nearby Universe harbour near-pristine gas, analogous to some of the first galaxies that formed in the early Universe. These galaxies are therefore of critical importance to the studies of galaxy formation and evolution as well as Big Bang nucleosynthesis in cosmology studies. These galaxies are located in the nearby Universe and can be assessed in detail with high-resolution imaging and spectroscopic data as well as integral field spectroscopy.

Despite a great effort of searching for XMP in the local universe, there are only a few hundreds being found. I have recently developed a machine learning pipeline [composed of multiple convolutional neural networks (CNN)] to accurately classify XMP and predict their metallicity (probed by the value of [NII/Hα]). This pipeline has been validated by new spectroscopic observation by the Isaac Newton Telescope (INT; see Figure 1). The paper (Cheng & Cooke) is in prep.

Fig.1: The comparison of the logarithmic value of [NII/Hα] between the CNN predictions and the preliminary measurements from newly observed spectra using INT. The gray dots are values from samples in literatures. The squares show the new results -- the blue squares are samples with good detection of [NII] lines, and the orange squares show the 2σ values due to the lack of clear detections of [NII] lines.