Once again I’m at the SC17 Supercomputing conference with four amazing year 10 students from John Monash Science School. This time we’ve dragged a chem teacher along for the ride. As always I intend to blog and it will be largely a stream of consciousness set of notes. Not very polished, because otherwise they’ll never make it on here! But hopefully useful, or at least interesting. I also don’t vouch for the accuracy or correctness of what I write. What you see is what you get. 🙂
Yesterday we started Supercomputing with a bang at the DWave Quantum Computing Workshop. While some of the material was out of our reach, it was a fascinating introduction to the history of Quantum Computing, and a more realistic estimate of the state of the art than the popular media tends to give.
Although the media often describes Quantum Computing as just around the corner, it’s important to note that we do not yet have what they call “Quantum Advantage” – in other words although we have some small quantum computers, solving a problem on a quantum computer remains slower than solving it on a classical computer.
That’s now.
What will we have in 10, 20, or 30 years’ time?
There are some seriously head spinny aspects to Quantum Computing.
For starters, with quantum systems you can massively increase the size of your dataset without increasing computation time.
I can’t explain how that works (I suspect there really aren’t that many people who can!), but I know that it’s a game changer.
On a normal computer if, for example, you have to search for a piece of information in a dataset the size of a short book, it will take much less time than search for the same piece of information in a large dataset – think, War and Peace. And far longer again – to the point of impossibility – to search in a library sized dataset, or worse, the internet.
On a quantum computer, the size of the dataset doesn’t impact on the runtime.
That changes everything.
For example, much of the tech we use to encrypt things and protect your data relies on the fact that to try every possible password simply takes more compute time than anyone has reasonable access to. We have the tech to hack most encryption, but if it’s going to take a million years on the fastest possible computer, then realistically your data is safe. (Unless your password is “password”, or one of the other not terribly startling variants!)
Quantum computers might be able to hack your password immediately.
Boom. Move over Facebook, privacy really will be dead then.
One of the other interesting things about quantum computers is that they don’t necessarily give you a single answer. For example, if you are looking for the best route to the airport for the whole taxi fleet in a city, it’s possible to use DWave to give you a set of good answers. Not one best answer, but a range of answers that work. You can see how that might be better – because it’s faster, but also because the one best answer might be out of action suddenly, due to an accident.
If using classical computing to find the best answer will take hours, but a quantum computer can give you 1000 pretty good answers in moments, that’s a step forward.
There are some applications that currently run on quantum computers, but for the most part things are so experimental that they’re not producing solutions that work right now. But in exploring these systems, learning their limitations and their advantages, people are finding new problems that quantum computing might provide radical new solutions to, and finding new ways to work with these systems. So although the quantum computers we have right now are relatively small, and not yet faster than classical computers, the more we play with them, the readier we will be to jump on the bigger, more powerful computers as they come online.
One of the things I find really interesting about the quantum computing industry is that it is a deep investment in what has at least in the past been a very speculative area. It flies against the trend I see so much of in science and technology these days where the funding all too often comes with explicit requirements for immediate payoffs.
We heard from John Sarrao from Los Alamos that they are deliberately investing in a kind of “playtime” on their DWave machine in the certain knowledge that it won’t solve immediate problems, but that it is setting them up to get smart people working together and forming communities to think and explore and find a whole range of possible futures. More than all of the quantum computing technology we learnt about, this is the part I found really exciting. Back in the heyday of Xerox Parc this kind of investment in an as yet unspecified future was extraordinarily productive, and we need far more of it.
This is how you create innovation – by giving smart people the chance to play, and the opportunity to collaborate. Not by stamping your feet and demanding that people innovate!
That’s just a small slice of the ways my brain was exploded yesterday. Let there be more of it!
Compute It Simple 