banner

Projects

Extreme Precision Spectroscopy

Detecting Earth-like Planets in the Era of Extreme Precision Spectroscopy

PI: Andrew Howard (Division of Physics, Mathematics and Astronomy)

Since the discovery of that first exoplanet using Doppler measurements, we have developed long-term, high-precision spectroscopy, sensitive to radial velocity (RV) amplitudes at precisions of ∼1 meters-per-second for bright, ‘quiet’ Sun-like stars. Now, driven by the urge to detect Earth-like planets, the field of radial velocity (RV) exoplanet detection is attempting to move towards extreme precisions (∼10 centimeters-per-second). This effort requires extreme technical artistry and demands unprecedented performance from both hardware and software. At Caltech, the research group of Andrew Howard is leading the development of a next-generation extreme precision spectrograph for the twin 10 m Keck telescopes, called the Keck Planet Finder (KPF). KPF will achieve 30 centimeters-per-second precision on sky, opening new eyes to the population of habitable zone planets around smaller stars, or super-Earths around Sun-like stars. Perhaps most importantly, it will also attempt to achieve 10 centimeters-per-second precision to find true Earths – or at least reveal more fundamental technical and astrophysical limitations and pave the way for future instruments.

Groundbreaking and believable exoplanet discovery relies heavily on a data reduction pipeline (DRP) for RV analysis. To this end, the Schmidt Academy collaborated with the instrument team on developing a sophisticated and fully automated Data Reduction Pipeline (DRP), which will be delivered along with the  KPF spectrograph. The KPF DRP is expected to leverage the extensive heritage of field-tested processing algorithms, while still conforming to the standards set forth by the Keck Observatory. The Schmidt scholar was able to take on the existing code repository of the DRP and design an architecture that standardizes the diverse algorithms such that they can be run, logged, and tested within the DRP. The DRP is highly customizable and supports pipeline modification through “recipes” that outlines the series of steps the pipeline should takes. While versatile, the KPF DRP also sets forth a unified standard on how data reduction algorithms should be implemented. Specifically, the existing code was refactored into pipeline primitives in an object-oriented manner, and extensive unit tests and integration tests were written to maintain the algorithms’ integrity. Furthermore, data objects are designed within the pipeline to provide a unified data format that the KPF pipeline primitives can recognize. With these constructs in place, the KPF pipeline provide the basis for automated data reduction as well as an environment for validating novel algorithms.

Stellar Spectrum

Fig1: A raw science image of a stellar spectrum, which is expected to be the input to the DRP.

The KPF Pipeline is as robust as it is complex and requires serious effort in testing to prevent regression. To this end, the Schmidt scholar collaborated with the instrument team on the pipeline’s infrastructure for sustainability, such as development strategy, test automation, and documentation. The Schmidt scholar was able to introduce the git-flow branching strategy within the development team and set up automated testing environment via Jenkins that verifies the integrity of each git push and pull request. Additionally, the scholar set up automated documentation generation via sphinx, and hosted by readthedoc.com. Significant effort was put into user and developer guides to set examples of proper documentation for future developers. SASE handed the project to the instrument team with the necessary infrastructure along with ample examples to guide future algorithmic development.

Stellar Spectrum

Fig2: Output of the pipeline running the Template-Fit algorithm on extracted spectrums of Barnard’s Star (left) and Tau Ceti (Right). The Template-Fit algorithm is one of many implemented modules of the KPF Pipeline

https://exoplanets.caltech.edu/kpf/