Super telescope will overload computers

Scientists admit they'll be forced to throw out valuable data because today's computers aren't powerful enough to process all the information that will be generated by a proposed new super telescope.

The planned $2.3 billion dollar Square Kilometre Array (SKA) will be the largest and most sophisticated radio telescope ever built.

It will consist of 3000 radio antenas spread across thousands of square kilometres, all linked by optical fibres and computers. Scientists claim it will be capable of seeing the first stars as they begin to shine, and help astronomers study mysterious forces like dark matter and dark energy.

Professor Bryan Gaensler from the University of Sydney, says the SKA will look deeper into the cosmos than anything that's been built before.

"It will answer many outstanding questions about the universe and most likely raise many new ones," he says.

According to Gaensler, each of the 3000 dishes will be collecting data continuously and when combined, the SKA will produce nine million signals at once, enough to fill five thousand 160-gigabyte mp3 players every minute.

This huge amount of data presents a new problem to astronomers.

"Until now we've been recording all the data and you can spend years gently going through it and taking your time. But that won't happen with the SKA," says Gaensler.

"You can't even write that amount of data to disk because there are no hard drives big enough to store that much information that quickly. So you have to decide in real time what you keep and what you don't.

"The data is coming in too fast for people to decide, so you need a computer program to do it for you.

"But if the program selects the wrong thing, you'll end up throwing away something you need, and you can't get it back again."
Putting computers in charge

To deal with the problem Gaensler and colleagues are working on new intelligent computer algorithms to process the torrent of data.

"We need computers that can do the job of humans, but make decisions on a timescale of microseconds. It would decide if something is interesting or should be thrown away," he says.

"Undoubtedly we'll occasionally be throwing out important data."

But Gaensler says the software will designed to recognise important astronomical events.

"There are certain types of things in astronomy which are always interesting like explosions or some flash that appears and then it's gone".

Astronomers also hope 'Moore's law' will assist them in processing the volumes of data. This 1965 prediction by Intel co-founder Gordon Moore suggests that computing powers doubles in capacity every two years, a prediction that's proven amazingly accurate.

"Moore's law means what's impossible now will be challenging in five years, and straight forward in 10," says Gaensler.

He says this and the nature of the SKA will allow astronomers to increase its capacity to collect data over time gradually.

"The beautiful thing about the SKA is that it's sort of organic, you don't have to have it all working at once," Gaensler.

"I imagine when you first turn it on you're just going to see the tip of the iceberg and as computers get more powerful it will become a richer instrument, growing in power as time goes on."

The SKA will be built in either Australia and New Zealand or Southern Africa. A decision will be made on its location in 2012.

  • Digg
  • Del.icio.us
  • StumbleUpon
  • Reddit
  • RSS

0 comments:

Post a Comment