Three Columbia Students in Internship to Streamline One of World's Largest Observatories

November 15, 2021

Three students from Columbia - one undergraduate and two graduate students – participated in an internship program sponsored by Universidad Adolfo Ibáñez's (UAI) School of Engineering and Sciences (FIC) where they researched "Application of Predictive and Prescriptive Analytics Tools in Observatory Operational Data," led by Rodrigo Carrasco (GSAS'13), FIC Professor of Operations Research and a member of the Santiago Center Advisory Board.

The work was to benefit the Atacama Large Millimeter/submillimeter Array (ALMA) observatory, an astronomical interferometer of 66 radio telescopes in northern Chile's Atacama Desert. The observatory results from an international partnership involving Canada, Chile, Europe, Japan, South Korea, Taiwan, and the US, and with a price tag at some US$1.4 billion, at the time of its launch in 2011, it was considered the most expensive ground-based telescope in operation.

The three students – María Teresa Tome, a computer science undergraduate at Columbia College; Zichan Liu, who is studying a Master of Science (MS) in Business Analytics at the School of Engineering and Applied Science (SEAS); and Mengyao He, SEAS MS student of Operations Research - took part in the program together with six students and professors from FIC.

The students worked on developing tools for the automation of operational processes in observatories, such as systems maintenance and data processing.

"We divided the students into three teams mixed between UAI and Columbia, and each group worked on different aspects of a particular Data Science problem, developing solutions for fault detection and diagnosis and maintenance prescription with real data. The idea is to later publish [findings] in scientific articles, and that they be implemented or in the process of being implemented in a telescope in Chile," Carrasco said.

He highlighted the work carried out by the research participants to enable better decisions in complex, data-intensive environments. One team developed a new tool to quickly identify certain patterns in large amounts of data, to be able to warn of possible failures in the systems. Another group developed a new system based on artificial intelligence to characterize a server's processing workloads to predict how long it would take and, together with this, develop tools that recommend how to order the workloads in a system to reduce their average waiting time.

"We worked together for two months and for the Columbia students it was a great experience to collaborate with UAI students, complementing their knowledge. For the latter it meant working with foreign students, in English and learning more about other universities and how they face research problems. The instance also showed that the problems we are solving in Chile are cutting-edge," Carrasco added.