John Wu ISM*@ST Intro Talk (2021)

Transcript

1. John Wu ISM*@ST Intro talk
2. John F Wu Born in Columbia, MD Fun fact: my Chinese name, 吴凡, is pronounced Wú Fán, so I grew up thinking my middle name was Fan. My parents actually misspelled it Fong. Michael Pecirno - Minimal Maps Columbia, MD
3. 1991 - 1994: Ellicott City, MD Ellicott City, MD
4. 1994 - 1995: Lived in Lake Charles, Louisiana (actually Sulfur, LA) Lake Charles, LA
5. 1995 - 2000: Back to Maryland Ellicott City, MD
6. 1996 - 2000: Back to Maryland Ellicott City, MD 🦀 > 🦐
7. 2000 - 2009: Clinton, NJ Clinton, NJ
8. Clinton, NJ 2000 - 2009: Clinton, NJ
9. Clinton, NJ 2000 - 2009: Clinton, NJ
10. Clinton, NJ 2000 - 2009: Clinton, NJ
11. Clinton, NJ 2000 - 2009: Clinton, NJ
12. 2009 - 2013: Studied Physics at Carnegie Mellon University in Pittsburgh, PA Pittsburgh, PA
13. 2013 - 2019: PhD in Physics at Rutgers in Piscataway, NJ Piscataway, NJ Cape Town, South Africa New Jersey, USA Hạ Long Bay, Vietnam Iguazú Falls, Argentina Thesis defense
14. 2019 - now: Joined the ISM*@ST group at JHU + STScI in Baltimore, MD Baltimore, MD
15. 2019 - now: Joined the ISM*@ST group at JHU + STScI in Baltimore, MD Baltimore, MD Bethel Vera Wu (2020-11-07)
16. 2019 - now: Joined the ISM*@ST group at JHU + STScI in Baltimore, MD Baltimore, MD
17. science: galaxy formation and evolution, multi-phase gas methods: machine learning, multi-wavelength observations Research interests LBGs and analogs Massive clusters Multi-phase gas Dwarfs/satellites
18. science: galaxy formation and evolution, multi-phase gas methods: machine learning, multi-wavelength observations - Bright z ~ 9-10 LBG candidates (BoRG) - Hı emission out to z ~ 1.4 (LADUMA) - Dust, gas, and star formation in massive z ~ 1 clusters - ISM properties of z < 0.3 LBG analogs - Kinematic scaling laws for LBG analogs - Estimate gas metallicity from galaxy image cutouts - Connect Hı mass, morphology, and galaxy environment - Predict entire SDSS spectrum from a galaxy image - Select dwarf galaxy candidates for xSAGA and DESI LOWZ program
19. El Gordo Depressed sSFR and SFR / M gas in massive clusters For a sample of star-forming galaxies in four Sunyaev-Zel’dovich effect (SZE)-selected clusters, we find: - Increasing star-formation from 0.3 < z < 1.1 - Massive molecular gas and dust reservoirs - Depressed sSFR relative to field galaxies - No environmental trends in cluster cores Wu et al. 2018, ApJ, 853, 195 HST + ALMA CO
20. Local LBG analogs have extreme multiphase ISM properties From VLT/SINFONI observations of UV-bright LBG analogs at z < 0.3, we find: - Disordered kinematics - Mixed dust and ionizing stars - Hard ionization due to massive binaries? - Warm H 2 gas excited in PDRs - Merger-induced shocks(?) Wu et al. 2019, ApJ, 887, 251 F150LP F606W H 2 1-0 S(1)-S(7) Paα
21. Neural nets can estimate gas metallicity from galaxy images - Train a deep CNN to predict gas metallicity, 12 + log(O/H), to within 0.085 dex from gri SDSS imaging - Reconstruct the MZR purely from imaging with no additional scatter Wu & Boada 2019, MNRAS, 484, 4683 Metallicity Stellar mass Mass-metallicity relation (MZR) CNN Spectroscopy
22. ML connects Hı gas and environment with morphology - Estimate gas mass fraction purely from optical imaging - Measure how environmental overdensity covaries with the Hı-morphology relationship - Visualize and interpret CNN features corresponding to gas- rich and gas-poor predictions Wu 2020, ApJ, 900, 142
23. CNNs can predict entire SDSS spectrum from a galaxy image Wu & Peek 2020, NeurIPS workshop: ML4PS, 3 (arXiv:2009.12318) Holwerda, Wu, et al., 2021, ApJL submitted
24. CNNs can find satellite candidates across the sky extending the Satellites Around Galactic Analogs (xSAGA) Survey
25. and select dwarf galaxy candidates for DESI LOWZ survey Dark Energy Spectroscopic Instrument - LOWZ survey
26. science: galaxy formation and evolution, multi-phase gas methods: machine learning, multi-wavelength observations Research interests