The High Altitude Observatory (HAO/NCAR) is hiring a post-doctoral fellow to work on the development of spectral line inversion tools based on machine-learning techniques to interpret spectropolarimetric observations of the solar chromosphere.
Spectropolarimetry, the science of measuring the intensity and polarization of light as a function of wavelength, is a very powerful tool for remote sensing of the thermodynamic and magnetic properties of astrophysical plasmas. The interpretation of the polarized spectrum of the Sun’s chromosphere is an area of research still under development. In particular, scattering polarization and the Hanle effect imprint signatures on the chromospheric spectra that are of high diagnostic value, yet difficult to interpret. The complexity of the forward problem alone poses a significant challenge to our current computing capabilities.
New facilities, most notably NSO’s Daniel K. Inouye Solar Telescope, will soon be producing large volumes of data that will need to be digested in automated ways in order to provide high-level data products to the scientific community. On the other hand, designing and running efficient and reliable inversion schemes to routinely interpret such large data volumes is practically unaffordable from a computational point of view. Combining forward modeling with machine learning techniques in order to develop quick-look inversion tools will greatly advance our ability to interpret chromospheric polarization spectra in an efficient manner, and open the path to exploring more efficient algorithms for full spectro-polarimetric inversions.
The successful candidate will work at HAO in collaboration with Drs. Roberto Casini, Rebecca Centeno, Natasha Flyer, and Ricky Egeland. She or he will use HAO’s HanleRT code (del Pino Alemán, Casini, & Manso Sainz 2016) to create training databases of synthetic polarization profiles for various solar atmospheric models and spectral lines. These will be used to explore different machine learning and pattern recognition techniques to parametrize the inverse mapping of the synthetic databases to solar atmospheric models. The final deliverable of this project will be to develop quick-look inversion methods for certain chromospheric spectral lines.
This is a two year appointment, renewable after the first year, subject to satisfactory performance. The successful candidate may start as soon as October 1, 2019.
Due to the restructuring of NCAR’s job application system, we will not be able to officially accept applications until September 16. However, please email Rebecca Centeno (rce@ucar.edu) if you are interested in applying.
Requirements: PhD in Physics, Astrophysics, or related sciences; working knowledge of machine learning techniques, experience with high performance computing and data visualization methods and tools (e.g., IDL, Python) will be necessary. A working knowledge of solar spectroscopy and spectropolarimetry is highly desirable.