![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
I feel very inspired by the Square Kilometre Array and have decided to educate myself further on the subject. I am required to collect professional development points every year to maintain my ACS certification and this will take care of a goodly chunk o' that. The current working title of what will probably be a Powerpoint presentation to myself is:
The Digital Logistics of Astrophysics: A guide fordummies those who lack the good taste to have studied the relevant subjects at graduate level.
Catchy, eh? And possibly a billion parsecs from where it's going to end up. I could be horribly abusing the whole school of astrophysics, for which I apologise in advance. I start from a position of humble ignorance.
I'm aware (and will no doubt take advantage of the fact) that I have friends in the physics, mathematics, computer science, broadcast technology and astronomy fields; I ask your indulgence while I work through the basics. Because you see I don't know what the data in a star catalogue looks like. Lassie? Maybe a lot of dates, times, elevations, frequencies, amplitudes, and pictures; very big numbers, lots of overlap, and a plethora of correlations, comparisons and extrapolations. Maybe lies, guesses, and approximations.
I know that noisy data doesn't compress well, and you can get into lots of trouble by trying. There's a start.
In the past I've learned about the relationships between physical and financial data in mining and manufacturing; I see this excercise as a variation on the theme with a new glossary and hopelessly large numbers. That said, I couldn't tell you what a million ounces of gold looks like, and I've created reports with figures like that. I'm not keen on raw numbers: I'd like to be able to think in pictures.
There's any amount of educational material around, and part of my challenge will be wading through it and defining my audience (apart from me). I'm thinking of pitching it as basic training material for new programmers and DBAs who might be joining in one the SKA projects. It's a not unreasonable goal to support the anticipated demands on the local IT industry. I may become the most employable person in Western Australia! And I'm a chick!
The trick will be identifying the core body of knowledge for that level without getting into the trippy space-is-curved bits. Or maybe the trippy bits are necessary after all to hold a sensible requirements discussion with the experts*.
Part of the process will be learning the vocabulary of supercomputering. I haven't a clue what a petraflop is; I don't even know if I just spelled it correctly.
I'm looking forward to finding out.
Thoughts anyone?
(*) Otherwise known as "faking it".
The Digital Logistics of Astrophysics: A guide for
Catchy, eh? And possibly a billion parsecs from where it's going to end up. I could be horribly abusing the whole school of astrophysics, for which I apologise in advance. I start from a position of humble ignorance.
I'm aware (and will no doubt take advantage of the fact) that I have friends in the physics, mathematics, computer science, broadcast technology and astronomy fields; I ask your indulgence while I work through the basics. Because you see I don't know what the data in a star catalogue looks like. Lassie? Maybe a lot of dates, times, elevations, frequencies, amplitudes, and pictures; very big numbers, lots of overlap, and a plethora of correlations, comparisons and extrapolations. Maybe lies, guesses, and approximations.
I know that noisy data doesn't compress well, and you can get into lots of trouble by trying. There's a start.
In the past I've learned about the relationships between physical and financial data in mining and manufacturing; I see this excercise as a variation on the theme with a new glossary and hopelessly large numbers. That said, I couldn't tell you what a million ounces of gold looks like, and I've created reports with figures like that. I'm not keen on raw numbers: I'd like to be able to think in pictures.
There's any amount of educational material around, and part of my challenge will be wading through it and defining my audience (apart from me). I'm thinking of pitching it as basic training material for new programmers and DBAs who might be joining in one the SKA projects. It's a not unreasonable goal to support the anticipated demands on the local IT industry. I may become the most employable person in Western Australia! And I'm a chick!
The trick will be identifying the core body of knowledge for that level without getting into the trippy space-is-curved bits. Or maybe the trippy bits are necessary after all to hold a sensible requirements discussion with the experts*.
Part of the process will be learning the vocabulary of supercomputering. I haven't a clue what a petraflop is; I don't even know if I just spelled it correctly.
I'm looking forward to finding out.
Thoughts anyone?
(*) Otherwise known as "faking it".
no subject
Date: 2010-02-17 04:08 pm (UTC)Although they didn't let the certain university clubs have access to the supercomputer so they're not as cool as they could be.
Sometimes it seems the important thing about having a supercomputer is being able to say you have one.
no subject
Date: 2010-02-17 04:20 pm (UTC)I went to a talk last year by Ron Ekers where he described 1950's radio astronomers spending literally _month's_ calculating FT's (not FFT's, they hadn't been invented yet) by hand, to produce a 16x16 pixel image...
Happily, optical astronomers never need to go near FFT's unless they're doing something really wierd.
no subject
Date: 2010-02-18 01:56 am (UTC)Have some Prefixes for biiiig numbers, incl. "peta-"
In computing, a "FLOPS" is a measure of processor speed - "floating-point operations per second". This doesn't specify which particular operations, nor does it consider various aspects of computer architecture (eg. bus speed, pipelining operations, memory-access speed, caching....). Not bad for a ballpark figure though.
I would combine these into a working definition of "petaflops" but i'm watching the snowboarding, and anyway the word is reminding of rabbits' ears for some reason.
no subject
Date: 2010-02-18 01:58 am (UTC)