Search Search

SURF: Announcements of Opportunity

Below are Announcements of Opportunity posted by Caltech faculty and JPL technical staff for the SURF program. Additional AOs for the Amgen Scholars program can be found here.

Specific GROWTH projects being offerred for summer 2018 can be found here.

Each AO indicates whether or not it is open to non-Caltech students. If an AO is NOT open to non-Caltech students, please DO NOT contact the mentor.

Announcements of Opportunity are posted as they are received. Please check back regularly for new AO submissions! Remember: This is just one way that you can go about identifying a suitable project and/or mentor.

Announcements for external summer programs are listed here.

Students pursuing opportunities at JPL must be
U.S. citizens or U.S. permanent residents.

  << Prev    Record 66 of 151    Next >>           Back To List


Project:  Data-Driven Methods of Measuring Stellar Compositions from Keck Spectroscopy
Disciplines:  Astronomy, Astrophysics
Mentor:  Evan Kirby, assistant professor, (PMA), enk@astro.caltech.edu, Phone: 626-395-4200
Mentor URL:  http://astro.caltech.edu/~enk  (opens in new window)
Background:  Stars incorporate the material created in previous nucleosynthetic events. As a result, the compositions of stars contain a wealth of information about how supernovae explode, how galaxies evolve, and how stars change during their lifetimes. For example, the amount of manganese in a star is a sensitive indicator of the mass of the white dwarfs that exploded prior to the star's birth. The ratio of oxygen to iron traces the vigor of a galaxy's star formation. And the amounts of carbon and lithium visible in a star's atmosphere reflect the intensity of gas mixing within the star.

The compositions of stars are measured via spectroscopy. The traditional method involves isolating specific absorption lines of specific elements. When an electron in an atom or ion of a certain element absorbs a photon, an absorption line is created at a specific wavelength. Thus, the element can be identified by the line's wavelength, and the strength of the line is proportional to the amount of that element.

Unfortunately, the conversion between line strength and abundance is not always straightforward. The procedure depends on atomic data (energy levels and oscillator strengths) that are often poorly known. Furthermore, some approximations, like the assumption of local thermodynamic equilibrium, are inappropriate for some absorption lines in some stars.

There now exist more modern techniques to determine the abundances of stars. For example, machine learning can circumvent some of the deficiencies in the traditional techniques. "Data-driven" computer codes can learn the relationship between absorption line strength and abundance without ever knowing the underlying physics. In this method, the user trains the code on spectra of well-characterized stars with known elemental abundances. Then, the code is run on spectra where the abundances are unknown. By applying the learned relationship between spectrum and abundance, the abundances of the new stars can be determined.
Description:  Professor Kirby and his group at Caltech have measured the abundances of thousands of stars with traditional techniques. The source of the spectra is the DEIMOS spectrograph on the Keck II telescope. Our approach is to construct a physically motivated model of a spectrum and compare it to the observed Keck spectrum.

We are now interested in exploring data-driven methods for measuring these abundances. The role of the SURF student will be to learn a data-driven code, like The Cannon, and apply it to DEIMOS data. The student will construct a training set based on Keck/DEIMOS spectra and the abundances previously measured from our traditional techniques. Then, the student will apply the trained algorithm to a set of spectra where the previously known abundances have been censored. In this way, the data-driven results can be compared to the traditional results in a blind fashion.

It may even be the case that the data-driven method outperforms the traditional method. We can envision a competition to see which code does the best at measuring abundances. The Kirby group will make friendly wagers on the traditional technique vs. the data-driven technique. Note that the SURF student may even choose to bet against the data-driven model, depending on his or her faith in the physically motivated code vs. the data-driven code! The winners will be able to name the parameters (party theme, cake flavor, etc.) for a party at the end of the summer!
References:  Keck/DEIMOS
https://www2.keck.hawaii.edu/inst/deimos/

Catalog of Stellar Abundances from DEIMOS Spectra
https://arxiv.org/abs/1011.4516

A Tutorial for the "Traditional" Abundance Measurement Technique
http://www.astro.caltech.edu/~enk/Ay219/pages/Ay219_Stellar_Abundance_Measurements.pdf

The Cannon - A Data-Driven Method for Measuring Stellar Abundances
https://github.com/annayqho/TheCannon
Student Requirements:  Linux/Unix (required)
Python (required)
Ay20 (suggested)
Programs:  This AO can be done under the following programs:

  Program    Available To
       SURF    Caltech students only 

Click on a program name for program info and application requirements.


  << Prev    Record 66 of 151    Next >>           Back To List