Additional event information

• Special Talks

  Speaker: Vital Fernandez (STScI)
  Title: ASPECT: An Algorithm for Blind Redshift and Line Measurements         
  Abstract: In this talk, we present ASPECT (Automatic SPEctra Components Tagging), an algorithm designed to identify astronomical spectroscopic features, such as emission and absorption lines, cosmic rays or continua and noise regions, using machine-learning models. We also demonstrate its application to the measurement of object redshifts and line fluxes (Fernández et al in preparation).

  The training and implementation of the algorithm are feasible with modest computational resources, and it has been successfully applied to the analysis of the JWST CEERS (Finkelstein et al 2024) and CAPERS (Dickinson et al 2025) surveys. Finally, we discuss how ASPECT can be integrated into the LiMe (LIne MEasuring) package (Fernández et al 2024) and the SpecSy (SPECtra Synthesis) online platform.

  Speaker: Doyeon Avery Kim (STScI)
  Title: Toward a Scalable Pipeline for UV Absorption Spectroscopy
  Abstract: Ultraviolet absorption spectroscopy is a powerful tool for probing the chemical and kinematic properties of diffuse gaseous media, yet current analysis workflows are often subjective and not fully optimized for reproducibility or scale. In this talk, I will present progress toward a unified Bayesian framework that combines state-of-the-art stellar atmosphere models with probabilistic inference to standardize continuum placement, component decomposition, and multi-ion fitting in UV spectra. I will illustrate how this pipeline can be used both to characterize feedback-driven outflows in the Milky Way and to measure chemical abundances and dust depletion in the diffuse interstellar medium across a range of metallicities, from the Milky Way to the Small Magellanic Cloud.

  Speaker: Benjamin Gibson (STScI)
  Title: Bridging the Divide: Studying the Resolved and Integrated Light of Nearby Galaxies in High Resolution
  Abstract: Our unique perspective from within the Milky Way gives us the ability to take resolved images and spectra of individual stars throughout most of its bar, bulge, and disk. However, taking spectra of individual stars in the centers of other galaxies is impracticable due to crowding. As such, we must use integrated light to characterize stellar populations, which often overlap spatially. Furthermore, spectral modelling techniques must account for the variety of stellar species, gas, dust, and dynamics that affect the shape of spectra. 

  I have developed a novel framework to analyze stellar populations in M31 using high resolution integrated light spectra from the APOGEE survey. By making some crucial, physically motivated assumptions about the bulk motions of stars in M31’s various components, I measured individual metallicity gradients in M31’s classical bulge and bar, and identified a chemodynamically distinct component in M31’s disk.

  I now study some galaxies that are slightly further away: M82, NGC4449, and NGC7793. These galaxies are actively star forming and have interesting dust properties, which I will probe via SED fitting to panchromatic photometry from Hubble and JWST. In the future I plan to combine these techniques with my novel spectroscopic framework to more fully characterize the overlapping stellar populations of nearby galaxies. Such work is vital for expanding our understanding of galactic structure, formation, and evolution in the local universe.

  Speaker: John Soltis (STScI)
  Title: Learning Gaia: Working Towards a Probabilistic Gaia Foundation Model
  Abstract: Even when trained to do relatively simple tasks, like predict the next word in a sentence, machine learning models trained on very large datasets are capable of performing a plethora of complex, previously unspecified tasks. The astronomy community has long sought to develop this type of model, often called a foundation model, using astronomical data. I will discuss ongoing work to develop such a model using Gaia Data Release 3 data, with the aim of creating a model that can accurately reproduce the distribution of Gaia-observed objects. Such a model would be useful for a variety of tasks in astronomy, including analyzing the structure of the Milky Way, identifying structures like streams and clusters, and classifying other objects of interest.