Comparing Galaxy Formation Simulations and Observations with Machine Learning

Joel Primack
UC Santa Cruz
Tuesday, May 28, 2019
4:00 pm
NS2 1201
I will describe using high-resolution zoom-in hydrodynamic galaxy simulations to understand Hubble Space Telescope observations of galaxies during the period of most vigorous star formation (redshifts 1 to 3, “Cosmic High Noon”) in CANDELS, the largest-ever HST survey. HST observations show that most star-forming galaxies at z > 1 have elongated (prolate) stellar distributions [1,2] rather than being disks or spheroids, and our simulations explain why: stars initially form along the prolate central regions of dark matter halos [3,4]. These “pickle galaxies” [5] trace cosmic filaments in mock catalogs [6]. A large fraction of star-forming galaxies at redshifts 1 < z < 3 are observed to have massive stellar clumps [7] that appear to evolve [8] as they move toward the galaxy center, as in our simulations [9]. Gaseous disk instabilities, often triggered by major or minor mergers [10], lead to clump formation and to rapid gas inflows and central starbursts that reduce the half-light radius (a phenomenon that we call “wet compaction”) and that help to create compact central star-forming spheroids (“blue nuggets”), many of which have X-ray detected AGN [11]. Our simulations rather accurately reproduce observed galaxies in radius [12] and their time evolution [13]. Comparison of our zoom-in galaxy simulations with CANDELS observations using “deep learning” methods suggests that galaxies do typically evolve through prolateto-nugget-to-disk phases at the same stellar masses as in our simulations [14]. We are also trying to understand how disk galaxies decrease their ratio of gas velocity dispersion to rotation velocity and settle into thin disks by the present epoch [15], and what halo parameters affect the stellar radius of galaxies [16-20].

[1] van der Wel et al. 2014, Geometry of Star-forming Galaxies from SDSS, 3D-HST, and CANDELS,
[2] Zhang, Primack, et al. 2018, The Evolution of Galaxy Shapes in CANDELS: From Prolate to Oblate,
[3] Ceverino, Primack, Dekel 2015, Formation of Elongated Galaxies with Low Masses at High Redshift,
[4] Tomassetti et al. 2016, Evolution of Galaxy Shapes from Prolate to Oblate through Compaction Events,
[5] Primack, Why Do Galaxies Start Out As Cosmic Pickles, American Scientist, Sept-Oct 2018,
[6] Pandya, Primack, et al. 2019, Constraints on intrinsic alignments of elongated low-mass galaxies out to z 2.5 from CANDELS observations and mock lightcones, MNRAS submitted,
[7] Guo et al. 2015, Clumpy Galaxies in CANDELS I. The Definition of UV-Bright Clumps and the Fraction of Clumpy Galaxies at 0.5 ≤ z < 3,
[8] Guo et al. 2018, Clumpy Galaxies in CANDELS II. Physical Properties of UV-Bright Clumps at 0.5 ≤ z < 3,
[9] Mandelker et al. 2017, Giant Clumps in Simulated High-z Galaxies: Properties, Evolution and Dependence on Feedback,
[10] Inoue, Dekel et al. 2016, Non-linear Violent Disc Instability with High Toomre's Q in High-Redshift Clumpy Disc Galaxies,
[11] Observations: Barro et al. 2017, Structural and Star-forming Relations since z 3: Connecting Compact Star-forming and Quiescent Galaxies,;
Kocevski et al. 2017, CANDELS: Elevated Black Hole Growth in the Progenitors of Compact Quiescent Galaxies at z ˜ 2,; and references therein.
Simulations: Zolotov et al. 2015, Compaction and Quenching of High-z Galaxies: Blue and Red Nuggets in Cosmological Simulations,
[12] Tacchella et al. 2016, Evolution of Density Profiles in High-z Galaxies: Compaction and Quenching, Inside-Out,
[13] Tacchella et al. 2016, The Confinement of Star-Forming Galaxies into a Main Sequence through Episodes of Gas Compaction, Depletion, and Replenishment,
[14] Huertas-Company et al. 2018, Deep Learning Identifies High-z Galaxies in a Central Blue Nugget Phase in a Characteristic Mass Range, — “Face recognition for galaxies” press release:
[15] Ceverino et al. 2017, Formation and Settling of a Disc Galaxy During the Last 8 Billion Years in a Cosmological Simulation,
[16] Somerville et al. 2018, The Relationship Between Galaxy and Dark Matter Halo Size from z 3 to the Present,
[17] Jiang et al. 2018, Is the Dark-Matter Halo Spin a Predictor of Galaxy Spin and Size? MNRAS submitted,
[18] Lee, Primack, et al. 2017, Properties of dark matter haloes as a function of local environment density,; Lee, Primack, et al. 2018, Tidal stripping
and post-merger relaxation of dark matter haloes: causes and consequences of mass-loss,
[19] Dragomir et al. 2018, Does the Galaxy-Halo Connection Vary with Environment,
[20] Goh, Primack, et al. 2018, Dark Matter Halo Properties vs. Local Density and Cosmic Web Location,
James Bullock