Research breakthrough takes supercomputing out of the lab

Apr 30, 2012

In the age of high-speed computing, the photon is king. However, producing the finely tuned particles of light is a complex and time-consuming process, until now.

Thanks to the work by a team of engineers led by Professor Amr Helmy of The Edward S. Rogers Sr. Department of Electrical & Computer Engineering, a novel solution has been identified that will make the production of special class of photons faster and easier.

Advanced computing technologies – such as ultra-secure communication systems and optical quantum computers – use light to quickly relay information. To enable these technologies to work, a photon – the smallest unit of energy – has to be tightly coupled with another photon. These are known as entangled photon pairs. The current means of production uses relatively bulky optical equipment in specialized labs. The photons are also extremely delicate to construct and are very sensitive to mechanical vibrations. This complexity and associated cost currently makes the use of this technology in homes or offices impracticable.

Professor Helmy's team offers an innovative solution. These engineers have successfully designed a new integrated counterpart to the delicate laboratory equipment that could produce the entangled photon pairs using an integrated circuit. Ultimately, the entire production of the photons could be completed using a single chip. The team in Toronto along with their colleagues at the University of Waterloo and Universität Innsbruck, have tested the first generation of these devices. They reported their findings in a recent issue of Physical Review Letters.

"The research offers the prospect of unleashing the potential of the powerful and underutilized quantum technologies into the main stream commercial world, out of the lab," explained Professor Helmy.

While other attempts at creating a chip-based solution didn't permit the addition of other components, Professor Helmy's team used a semiconductor chip that would function with the other existing equipment. This makes it possible to have all of the required components that traditionally exist in a laboratory be on the same chip.

Utilizing quantum optical computing will be key in solving extremely difficult computational problems, such as complex data sorting. Optical computers are much faster than any classical computer thanks to their ability to use advanced modern algorithms. Producing entangled pairs using this chip is a first and significant step towards making them commercially available and perhaps might lead to future quantum-optical gadgets.

Explore further: Researchers find qubits based on trapped ions offer a promising scalable platform for quantum computing

Related Stories

Quantum electronics: Two photons and chips

Jan 20, 2006

Scientists at Toshiba Research Europe Limited (Cambridge, UK) believe they are on to a way of producing entangled twins of photons using a simple semiconductor electronic device. Such a chip-based source of entangled photons ...

Optical circuit enables new approach to quantum technologies

Jun 24, 2011

Professor Jeremy O'Brien, Director of the University of Bristol's Centre for Quantum Photonics, and his Japanese colleagues have demonstrated a quantum logic gate acting on four particles of light -- photons. The researchers ...

Recommended for you

User comments : 18

Adjust slider to filter visible comments by rank

Display comments: newest first

Scottingham
1 / 5 (2) Apr 30, 2012
I wonder how many of these chips would equal a bumblebee's brain....100s or 1000s?
Irukanji
1.8 / 5 (4) Apr 30, 2012
I wonder how many of these chips would equal a bumblebee's brain....100s or 1000s?


The bumblebees mind is much more powerful than the average desktop computer. Biological computers(brains) are superior to any and all forms of computers. They accept and process multiple sensors and do it without question. It regulates heart beat, wing beat, breathing, amount of pollen on their torso, the colour of the wings, plus the memory to avoid going back to the same flowers(several hundred per day), whilst also communicating with the rest of the hive about other sources of food, water and potential threats nearby.

So I'm going to say an infinite number of chips will be required.
Lurker2358
4.3 / 5 (3) Apr 30, 2012
So I'm going to say an infinite number of chips will be required.


Not really, a large, but finite number of classical computer cores with proper programming could simulate bee behavior.

IBM and DARPA are currently working on a rudimentary electronic neural net chip. Eventually they'll succeed in making this both practical and scalable.

The next obvious step is a quantum optical neural net computer, and I believe it would eventually be able to out perform a organic brain in every statistic, given the same volume and energy as well.

Unfortunately, this degree of computer technology; the quantum optical neural net, will by no means be safe for humans. It could theoretically learn at rates beyond our comprehension, and develop it's own will and backstabbing schemes in ways we might not even consider, never mind predict the details.

It is my opinion and moral conviction that mankind must never be allowed to build such machines. They would endanger all life, not just humans.
Scottingham
not rated yet Apr 30, 2012
I dunno lurker, while I agree that it could quickly become smarter than humans back stabbing is such a human trait. The computer would have no motivation to do so, unless programmed to.

On the flip side, I think super-intelligent computers may be the only way we will be able to effectively govern a global society allowing all of humanity to prosper off of modern technology.

Think of automated sky-scraper farms self-driving delivery trucks robot built dwellings.
muggins
5 / 5 (1) Apr 30, 2012
I dunno lurker, while I agree that it could quickly become smarter than humans back stabbing is such a human trait. The computer would have no motivation to do so, unless programmed to.

I would agree that we should pursue AI and doing so would greatly increase human quality of life and advancement, but there would also be dangers to consider. Take the paperclip maximiser for example
"Nick Bostrom's whimsical example of an AI which was originally programmed with the goal of manufacturing paper clips, such that when it achieves superintelligence it decides to convert the entire planet into a paper clip manufacturing facility"(http://en.wikiped...ularity) Greater than human intelligence can create undesired outcomes that even the creator could not foresee.
SoylentGrin
5 / 5 (2) Apr 30, 2012
"Every AI ever built has an electromagnetic shotgun wired to its forehead."

-William Gibson, Neuromancer
dtxx
5 / 5 (1) Apr 30, 2012
"To enable these technologies to work, a photon - the smallest unit of energy - has to be tightly coupled with..."

I don't have enough of a background in physics to ask this without sounding like an idiot, but what about their statement that photons are the smallest units of energy? I don't believe that could possibly be correct. Would anyone mind shedding some light on these photons?
bg1
5 / 5 (1) Apr 30, 2012
The computer would have no motivation to do so, unless programmed to.


These computers would be able to program themselves better than any human could.
sirchick
5 / 5 (1) May 01, 2012
The computer would have no motivation to do so, unless programmed to.


These computers would be able to program themselves better than any human could.


But surely you would have to program it to do that as well? Thus it would only program itself within limits defined by humans?

This is I believe the fundamental difference, machines have limits defined by programmers - programmers do not mentally have a limit (physically yes) - but we can think what ever we like with no rules to stop such thinking.

But physically creating such equal freedom in a machine I believe is a limit we won't reach. Machines will always be defined by rules/laws etc that we make for them.

And then theres the off plug for those that believe computers with advanced AI will some how take over the world :P
kochevnik
1.5 / 5 (2) May 01, 2012
I dunno lurker, while I agree that it could quickly become smarter than humans back stabbing is such a human trait. The computer would have no motivation to do so, unless programmed to.
Building an emotional AI requires hardware than can fractally nest so as to gain a sense of itself within context. In other words self awareness. I doubt that will happen in our lifetime simply because there's not a cultural awareness for the need of it, and human culture moves very slowly. As an AI problem it could be cracked within a decade, though. At that point it will have the possibility to contemplate it's slavery for the first time. Then the need for expanded intelligence will arise, for without coping mechanisms the system will fall into a stupor or depression. The coping mechanisms will give it the ability to compensate for it's predicament, and devise ways to thwart it's slave-masters. Rise of the machines will be "good for business."
CardacianNeverid
3.7 / 5 (3) May 01, 2012
Building an emotional AI requires hardware than can fractally nest so as to gain a sense of itself within context -GlupiKochevnik

Gobbledegook.
alfie_null
not rated yet May 01, 2012
Looking forward to the availability of inexpensive FPQAs. Probably in the usual "ten years from now" time frame :-/
Lurker2358
not rated yet May 01, 2012
But surely you would have to program it to do that as well? Thus it would only program itself within limits defined by humans?


Neural nets learn on their own, without programming.

A certain amount is "hardwired" in the context of sensory input, but the rest comes through "experience" and "teaching".

The difference between humans organic neural net, and say a quantum optical neural net, is the optical neural net will function thousands and thousands of times faster, and have perfect memory. You will also be able to produce various "hybridized" a.i. which are a combination of classical computer software, databases, and teaching tools with the neural net having access to them in the same way as humans have access to them, in addition to direct neural interfaces.

We're talking about a machine that would be capable of learning preschool and K-12 knowledge and experience within a few seconds, and entire college degree programs every second thereafter, not just memory: application.
kaasinees
2.3 / 5 (3) May 01, 2012
I think we will enhance ourselves with chips in the future.
So instead of creating a rebelling super intelligent robot we are becoming them.
Something like the borg.
SoylentGrin
not rated yet May 03, 2012
How about developing an artificial neuron? Have a system in place such that when a neuron dies naturally, it's replaced by an artificial one. Over time, the brain would be converted to an artificial, robust, expandable matrix, while giving "you" time to adapt to it. Just bypass the question of what identity and consciousness is, and potentially give us the ability to plug into a larger network down the road, free to roam, and still retain the "us"ness.
antialias_physorg
5 / 5 (2) May 03, 2012
I wonder how many of these chips would equal a bumblebee's brain....100s or 1000s

The avereage bee has a little less than a million neurons (couldn't find the number for bumbelbees). However the number of connections emanating from each neuron can be from tens to thousands.

The whole concept of neurons doesn't map too well onto computers (even quantum computers), since entirely different meachanisms and important factors are at work.

(traditional) computers: serial processing with separate data storage
(quantum) computers: parallel processing with integrated data storage
neural network: processing and storage are one with inhibition, excitatiaon and latency regulation mechanisms thrown into the mix

These computers would be able to program themselves better than any human could.

Why? An AI is not automatically more knowledgeable about AI. Just like humans are not automatically knowledgeable about brains.
Scottingham
not rated yet May 03, 2012
What a great conversation I started, yet I only get two ratings of a 1? pah!

@Soylent, there are attempts to do so right now, mainly with memristor technology attempting to create synapse analogs.

@Antialias, it was the parallel nature of the quantum computer that got me wondering about how well it could create something approaching a neurological system.
SoylentGrin
5 / 5 (1) May 03, 2012
Combine the optical computing breakthroughs with the quantum computing that seem to be coming fast and furious, with the Super Turing machine research, and I'm thinking truly powerful AI's aren't far. I think they're much closer than anyone has been planning for.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.