Galaxy Evolution with Manifold Learning

Area of Science:

• Cosmic Evolution
• Astrophysics
• Data Science

Background:

• Galaxies formed from early Universe density fluctuations.
• Understanding galaxy formation and evolution is complex due to vast astrophysical data.
• Conventional physics-based methods struggle with high-dimensional datasets.

Purpose of the Study:

• To elucidate the physics of galaxy evolution using advanced data science techniques.
• To analyze galaxy properties across cosmic time.
• To overcome limitations of traditional methods in handling large astrophysical datasets.

Main Methods:

• Applied manifold learning, a data science technique.
• Utilized a feature space defined by galaxy luminosities and cosmic time.
• Analyzed a dataset spanning the Universe's 13-billion-year history.

Main Results:

• Discovered a low-dimensional nonlinear structure termed the
• galaxy manifold.
• Found galaxy evolution is well-described by two parameters on this manifold: star formation and stellar mass evolution.
• Demonstrated that ultraviolet-optical-near-infrared luminosity space captures key evolutionary paths.

Conclusions:

• Manifold learning provides a powerful new approach to understanding galaxy evolution.
• The identified galaxy manifold simplifies the complex process of galaxy formation and change.
• Future work can connect manifold coordinates to specific physical quantities.