Dear friends,
I am happy to report that our observations with the Green Bank Telescope (GBT) last month were smooth and successful. The students in the Spring 2019 edition of the UCLA SETI course were able to participate remotely from our computer lab on campus. We observed 16 solar-type stars near the plane of the Galaxy and collected over 4 terabytes of raw data. The signal that we sampled is a time-varying voltage that is proportional to what the telescopes "sees" (technically, it is proportional to the strength of the electric field at the focal point of the antenna). Whenever we conduct this type of observations, we sample this voltage 800 million times per second, which allows us to split the signal into at least 100 million individual channels for analysis. Because the data rate is so large, we use only 4 symbols to represent the voltage samples: 00, 01, 10, and 11. In computer parlance, we record 2 bits per voltage sample, where each bit is a binary unit of information, i.e., 0 or 1. One might think that recording only 2 bits per sample would yield a poor quality representation of the signal. It turns out that our detection efficiency (technically known as quantization efficiency) is 88% as good as that of a signal recorded with infinite precision. If we recorded 4 bits per sample, the efficiency would increase to almost 99%, but in that case we would have to halve the sampling rate and the number of channels available for analysis. We prefer to maximize the number of channels that we analyze at the expense of slightly reduced signal fidelity.
|
|
|
|
The UCLA Spring 2019 SETI class consists of 19 undergraduate students and 3 graduate students, most of whom are pictured here. Graduate student Paul Pinchuk volunteers as a teaching assistant for the class.
|
|
|
|
|
|
As usual, the students in the SETI course have different levels of programming expertise. Regardless of skill level, we ensure that all students in the class are capable of programming the essential elements of our data-processing pipeline on their own. For instance, they learn to compute power spectra with the Fast Fourier Transform algorithm. They learn to shift consecutive spectra to account for motion between the emitter (whatever or wherever it is) and the telescope. They all write a program that enables them to detect the signal from the Voyager 1 spacecraft, which is humanity's most distant ambassador as well as a perfect test case for recognizing artificial signals. At this point in the course, the students are examining the millions of candidate signals that we detected last month and quantifying the properties of the most interesting signals. They are also developing code to improve our data-processing pipeline. The range of projects is inspiring. The students are improving the robustness and usefulness of our database queries, the speed of our unpacking algorithms, our baseline correction algorithms, the pairing of signals observed in two individual scans, the identification of known interferers such as GPS satellites, etc. The graduate students are tackling advanced projects such as implementing our calculation of power spectra with graphical processing units (GPUs) and implementing the detection of anomalous signals with machine learning techniques.
On a related note, UCLA graduate student Paul Pinchuk passed his qualifying exam with flying colors last year. He has formed a doctoral committee composed of 4 faculty members in the Department of Physics and Astronomy (Carter, Fitzgerald, Margot, Saltzberg). This Tuesday, he will take the "advancement to candidacy" exam on his way to obtaining the PhD. Paul will present his research plan and the doctoral committee will evaluate whether Paul has acquired sufficient knowledge of the field and whether his proposed doctoral project is of reasonable scope. Paul will defend his dissertation after he completes his project, likely within the next two years. He will be one of only a handful of students to ever obtain a PhD in the field of SETI.
Warm regards,
Jean-Luc Margot
|
|
|
|
|
