canspice.org

When I started out in astronomy I never meant to be an asteroid hunter. Now, five years after getting my BSc that’s what I’m turning into.

Okay, I’m not just hunting asteroids. I’m hunting things that pop, things that swell, and things that dance. Supernovae, variable stars, gamma ray bursts, these things will soon be my prey.

See, the telescope I work for (UKIRT) just took delivery of a wide field camera, the appropriately-named Wide Field Camera (or WFCAM for short). One of the problems with these large cameras is that they produce a tremendous amount of data, too much data to easily ship to astronomers doing research with the camera. In WFCAM’s case the peak data rate is about 460GB per 10-hour night. That’s about 100 DVDs full of data each night. Too much data for Joe Random Astronomer to reduce on their desktop. Thus the data is going to be shipped to a central clearing-house in the UK, reduced there, and the astronomer is going to be delivered a catalogue of detected objects. However, the data takes some time to ship to the UK from Hawaii, and once it’s there it’ll take some time to reduce and get into their database.

Because of this time-delay there’s the potential that time-sensitive detections will be discovered much too late to do anything about them. It’s no good to anybody if WFCAM observes a supernova and it only gets discovered three months later.

Enter the eSTAR/WFCAM Transient Object Detection Agent. That mouthful is what I presented in poster form at ADASS XIV in Pasadena last month. What it does is sit at the end of our data reduction pipeline at the telescope and detect transient objects. There’s a whole mess of things like cross-correlation and databases and THE GRID (oh there it is again). We’ll be implementing the whole thing early next year with the help of Alasdair Allan, who’ll hopefully be coming out to Hawaii for it.

And in the great tradition of presenting early results, I can gladly say that we got it working last week. With humans in place of all of the software, of course.

Last week I was looking at some of the results of a WFCAM run. It consisted of four sequences of the same field in four filters in the near infrared. I was blinking through the four results in an image viewer (and so much more!) called GAIA, and co-worker Vicki was marvelling at all the pixels when I scrolled over to a part of the frame and saw something moving. Vicki said it was a spaceship.

I looked it up the next day and found out it was discovered four years ago and has been named 2000 SB277. Want to see it? Here it is!

So our first detection of a transient object. Luckily we’re going to come up with software to find them because there’s no way I’m looking at every single result out of this instrument.

It’s a sad thing when an innocent life is dragged deep into despair and misery. Alcohol, drugs, sex, these things are toxic to the naive. I have a story to tell. I’ll call it Requiem For A Duck.

It started out innocently enough. I’d come home to find Duckie drinking beer.

He really seemed to get into it, but he still had his job at the carwash and wasn’t getting into any trouble.

I let him be.

Then he moved into illegal things.

He tried to explain it by making up a story about chronic pain.

I didn’t buy it, as his other cure for this “chronic pain” was to sit around all day and watch TV.

He’d lost his job at the carwash, but he somehow kept paying his share of the bills, so I let him be.

As a side-effect of his pot habit we always seemed to have a lot of chips around.

I guess he just got the munchies.

Occasionally I found him using my computer.

Invariably it was mother/daughter duck porn.

He always tried to cover it up, but I knew. Oh I knew.

This went on for a couple of months. I began to suspect he was getting into deeper and darker things, but he never brought any troubles home with him when he stumbled in at 2:30 each night.

One Saturday afternoon I got out of the shower to find him making a phone call.

I looked closer.

Duckie and I had an argument that night. He stormed out and didn’t come back for three days.

A couple of weeks later I came back to some strange noises from the bedroom

I pulled back the covers…

…and found him…

…in bed with some whore.

Another argument, broken dishes. Duckie was gone for a week.

He came back apologetic. I didn’t trust him, but what choice did I have? He was a friend, so I let him come back. Things were going well for a couple of months.

Then I came home and found this.

There was no denying it, he’d gone too far.

I ordered him out…

…and he went.

He turned back to say something…

…but I didn’t care. I wanted him out of my life, and he was.

I didn’t hear from him for six months. One day I came home to a jimmied lock. Not good. I went into my apartment and found a letter addressed to me.

I opened it.

“Good bye creul world…. Duckie never was a good spell… DUCKIE!”

I was too late.

I was too late.

I found the notebook I was using to jot down notes about ADASS XIV in Pasadena, so here’s a recap from day three.

It started off with a talk about the challenges of future robotic exploration from who was supposed to be Dr. Charles Elachi, the director of JPL, but he was busy dealing with present robotic exploration, so another guy from JPL gave the talk instead. I forgot to write down his name. It was Mars-heavy, which is fairly understandable considering JPL’s got the Spirit and Opportunity rovers on the surface of Mars at the moment.

Following that was a talk by Shep Doeleman about the Square Kilometer Array. The interesting bits I remember about it is that instead of having straight arms of antennas, like the Very Large Array in New Mexico has, a logarithmic spiral gives you a better point-spread function. They’re looking to get a square kilometer or so of dish space, so this involves a lot of dishes. The SKA group came up with an automated mapping program that would take roads and terrain into account when placing dishes. This is especially important for some versions of the SKA (they don’t know what form it’ll take just yet) because, taking the US SKA proposal as an example, if you put the core of the spiral in southwest USA (New Mexico, Arizona, somewhere around there) then the outer dishes won’t be a couple of miles away, they’ll be in Canada and Mexico. Thus you don’t have a small-ish site to place dishes in, you’ve got the entire continent, and you have to place them in places where you can get to them.

After this David Bohlender of the Canadian Astronomical Data Centre gave a talk about the Gemini Science Archive, and Ghaleb Abdulla gave a talk about middleware for data reduction pipelines. Since I support a data reduction pipeline you’d think this last talk would hold some kind of interest, but I can’t remember a thing about it and don’t have anything written about it.

Greg Schwarz gave an interesting presentation about linking data sets presented in astronomical journal articles (specifically the AAS journals) to virtual observatories and databases, allowing people to actually go look at the data talked about in an article. Of course, the opposite is possible, where if you get some data out of an archive you’ll be able to find out what papers were written using that data. It’s a good way for data providers (I think that’s what they’re calling telescopes in this virtual observatory age) to determine how useful their data is and what scientific impact they’re having, and it’s a good way to make your telescope more visible as well.

William Joy, one of the developers of SAOImage DS9 talked about, guess what, DS9. I enjoyed his talk, as it seemed to relate to ORAC-DR quite a bit. Both projects don’t have heaps of people working for them (1.5 for DS9, about 2 for ORAC-DR). Both projects try to keep adding functionality but also have to maintain the dykes against the flood of changing hardware and software that they rely on. Both projects rely heavily on outside contributions, DS9 for various libraries and ORAC-DR for Starlink’s data manipulation. Both projects realize that they need to spend more time improving documentation. And both projects get feedback and comments from their users. All in all it was a fine presentation. This presentation is available for download.

Emmanuel Bertin gave an invited talk about (and no snickering please) SExtractor, one of the most widely-used pieces of software available for the astronomical community. Unfortunately he tried to cram too much information about his object detection techniques into the available time, so he was rushed for the last half. All I have written about he talk was that SExtractor has been around for about 10 years and has been mainly static since 1998. Version 3 is supposed to be coming out around the end of 2005. Hopefully he’ll have the documentation improved by then!

Alex Gray talked about picking out quasars from a huge dataset, using various techniques to try to get to O(N) times for classification, and Alvaro Sato talked about detecting rare objects in massive datasets. Sato said their technique was able to find all of the fake rare objects they injected into a dataset, but also classified 10% of the other objects as being rare, a percentage that is much too high. They’re still working on their algorithms.

Then it was lunchtime, and after lunch I missed three and a half presentations because I went ice skating instead.

I made it back for the last bit of something about W projection (available for download), and luckily got to see one of the most interesting talks in the whole conference. Eric Sessoms’ talk was boringly titled “Hardware Acceleration for Astronomical Data Analysis”, but it was quite cool to see. He talked about using graphics cards for data processing, since they are able to push through about 6 times as many GFLOPS as high-end CPUs can. Of course, they’re a little more difficult to control, and you don’t get important programming techniques like looping or branching, but for doing things that are straight-foward like fast Fourier transforms, you can blaze through data like gangbusters. More information is supposed to be found at gpgpu.org but I can’t currently connect to it.

And that was ADASS XIV! Next one up is about 60 miles outside of Madrid in San Lorenzo de El Escorial.

Alright, so I ballsed-up my prediction that Kerry could win, and we’ve got another four years of a Bush presidency. Whoops.

Here are a few predictions for the coming four years:
1. The Democrats decide to try to confuse the voting public by going back to the name “Republican Party”.
2. Canada becomes a right-wing country as all the American “liberals” flee the US.
3. The 2004-2005 NHL season is cancelled. Bush is blamed.
4. Reports of spontaneous head explosions increase ten-fold. Victims span the political spectrum, as Republican heads explode in anger and Republican heads explode in ego.
5. George W. Bush burns down a church filled with disabled war veterans in Alabama and that state still votes Republican for president. Republicans (are you confused yet?) chastise themselves for forgetting to plant cute fluffy bunnies in the church. (Hello Amy!)
6. Americans discover that gasoline is cheaper than beer and start sniffing gas for recreational purposes. Anheuser-Busch merges with Shell.
7. The 2005-2006 NHL season is cancelled. Bush is blamed. Bob Goodenow is proclaimed Public Enemy #1 in Canada and Paul Martin declares a jihad against him.

It’s 7:55pm HST, or 12:55am EST, and CNN currently has Bush ahead of Kerry 246-195. I don’t know why they haven’t called Ohio or New Hampshire yet, since both look fairly safe (Bush for Ohio, Kerry for New Hampshire). If things progress as they did in the 2000 elections I think that means that Kerry will win. Nothing seems to suggest that something out of the ordinary will happen, so maybe tomorrow when I wake up it’ll be in a Kerry USA. That’d be kind of nice.