Fourteen thousand feet atop Mauna Kea in Hawaii, the Gemini North telescope silently probes the universe. The newly completed 8.1-meter Gemini telescopes (the second Gemini telescope is located in the Chilean desert) are the flagship facilities for astronomers in seven countries. The Gemini telescopes are designed to provide the sharpest images of any telescope on our planet. Each clear night the telescopes feed an enormous volume of data back to scientists based at various points around the globe. Understanding the data obtained from a single night’s observation can take weeks of time for the astronomers working back at their home institutions.

Thanks to a new data reduction application written with Cocoa on a Power Mac G4 running Mac OS X, the time taken to analyze and interpret the data from Gemini has been shortened dramatically. Astronomers are using this new capability in an ambitious project whose aim is to determine just when the various galaxies in the universe were formed.

The Gemini telescopes are operated jointly by a partnership of national institutions in the United States, Canada, the United Kingdom, Chile, Australia, Argentina, and Brazil. A single night of observations on both telescopes costs roughly $100,000 and is allotted on a highly competitive basis, so time on the telescopes is precious.

Studying Far Away Galaxies
Recently, the University of Toronto’s Roberto Abraham, a professor of Astronomy and coprincipal investigator of the Gemini Deep Deep Survey (GDDS), joined together with scientists from Johns Hopkins University, Carnegie Institution of Washington, Oxford University, Massachusetts Institute of Technology (MIT), and the Gemini Observatory to undertake a ground-breaking survey to obtain spectra of the faintest galaxies ever studied.

In order to obtain spectra for these objects (about 100 million times fainter than the faintest galaxies detectable with the naked eye), modifications had to be made to Gemini’s control software to allow the telescope to use a newly developed technique called “nod and shuffle.” This moves the telescope back and forth rapidly during an exposure, subtracting out the interference of the atmosphere. As a result, data has been streaming back to the astronomers at an unprecedented pace.

Outpacing Existing Software
“The technical advances made possible by adding a new mode of operation to the telescope meant that we’d moved faster than the ability of the existing software to keep up. Even IRAF [Image Reduction and Analysis Facility], the ‘gold standard’ for astronomical data reduction software, couldn’t easily handle the kind of data we were taking in and it was hard to use,” Abraham notes.

“We’d spent all this money, time, and energy getting the hardware right, and were getting fantastic data, so we had to try to do the best job possible with the software too — even if this meant writing an entirely new bit of analysis software from scratch,” Abraham continues. “It was pretty tricky. The data was already flowing, so there really wasn’t time to get a professional programmer up to speed. We really needed something put together quickly and it was pretty clear I would have to do it myself.”

Cocoa Power and Speed
Abraham investigated several software development environments. But he says Cocoa on a Power Mac G4 with Mac OS X was the only one powerful enough to meet his needs, while adhering to the project’s demanding schedule.

“This was the biggest allocation of time on the telescope to any Canadian astronomer so we were under the gun from day one to deliver,” he says. “I looked around to see what the best tools available were for a working scientist who wanted to get a nontrivial GUI-based application put together really quickly. Cocoa was very obviously the best designed framework for this kind of thing, and Objective-C looked easy to learn. It was pretty clear that doing it on a Mac would be the fastest way to go.”

Two Weeks, a Mac, and a Manual
The astronomer built the application himself — quickly. Abraham recounts, “I sat down with a reference book and a laptop knowing nothing about Objective-C or Cocoa programming. Two weeks later I had a solid working prototype of our application that was already way better than anything the standard tools could provide. I’d never written a GUI-based app in my life, but Cocoa made it ridiculously easy.”