NASA, Google buy quantum computer from B.C. firm

0

A quantum computer has been purchased from Burnaby, B.C.-based D-Wave Systems by a NASA-led group that aims to solve problems requiring creativity — something that conventional computers aren't good at.


The D-Wave Two is currently being installed at the new Quantum Artificial Intelligence Lab at NASA Ames Research Center in Moffett Field, Calif., and should be up and running this fall, D-Wave announced Thursday.

The lab is a collaboration among NASA, Google and the Universities Space Research Association, who expect to use the quantum computer to tackle problems related to machine learning, web searching, speech recognition, planning and scheduling, searching for exoplanets, and supporting operations in mission control centres. The USRA will also make the system available to other U.S. academic institutions.

"We are extremely pleased to make this announcement," said Vern Brownell, CEO of D-Wave in a statement.

"Three world class organizations and their research teams will use the D-Wave Two to develop real world applications and to support research from leading academic institutions. This joint effort shows that quantum computing has expanded beyond the theoretical realm and into the worlds of business and technology."

Quantum computers store data in units called qubits, analogous to the bits used in conventional computers. But while each conventional bit stores information as either 1 or 0, qubits make use of quantum mechanics — laws of physics that apply only to very small particles such as atoms — to encode information as both 1 and 0 at the same time.

Written By: CBC News
continue to source article at cbc.ca

NO COMMENTS

    • In reply to #1 by joost:

      Computational creativity is not a hardware problem. It’s a software problem.

      It’s likely a combination of both. In the late 50′s when the perceptron model for artificial neural networks was devised, the computing power for larger networks to be applied to problems did not exist. Similarly, as hardware improves the size of these networks, thus the complexity of problems to which they can be applied also increases. There are no doubt still algorithmic breakthroughs to be made in software, though these often go hand-in-hand with technological leaps in hardware.

      • In reply to #7 by Mister T:

        In reply to #1 by joost:

        Computational creativity is not a hardware problem. It’s a software problem.

        It’s likely a combination of both. In the late 50′s when the perceptron model for artificial neural networks was devised, the computing power for larger networks to be applied to problems did not…

        You are right about the relationship between smarter software and faster hardware but the perceptron is very limited in capability. It is the same as the statistical technique Principal component analysis. Minsky and Papert proved their limitations in 1969. It was not until the invention of back propagation that artificial neural computation took off. The cute stuff are now the restricted Boltzmann machines.

        Fast hardware helps with all this AI stuff, though GPUs are the usually way to accelerate RBMs. I’ve no idea if a quantum computer would help.

      • In reply to #7 by Mister T:

        In reply to #1 by joost:

        Computational creativity is not a hardware problem. It’s a software problem.

        It’s likely a combination of both. In the late 50′s when the perceptron model for artificial neural networks was devised, the computing power for larger networks to be applied to problems did not…

        I disagree. All the computers that you have ever used and almost certainly ever will use are Turing Machines and Turing proved that from a computational standpoint all Turing Machines are equivalent. There is no program with better “creativity” that you can run on one Turing Machine but not another. Joost was absolutely right making computers “creative” is all about the software not the hardware. Neural nets for example — a very different computing paradigm from traditional programming — still get compiled into code that runs on a Turing Machine.

    • In reply to #1 by joost:

      Computational creativity is not a hardware problem. It’s a software problem.

      Yes, and no. It’s true that you can, in theory, simply expand a current form of hardware to accommodate other, more complex problems, but in practice some materials are just too impractical for this to work. A computer made from electronic chips, for instance, is more efficient and cost-effective at solving mathematical theorems than an equivalent made from waterworks and pressure valves for the same purpose.

  1. Quantum computers can solve “try all possibilities” problems by
    using Quantum Mechanics to explore all possibilities simultaneously.
    I suspect encryption, except one-time pads will have to be revamped or abandoned to stay ahead.
    This is going to create a massive social problem and open astounding possibilities.

    We might be on the verge of picking nearly every digital lock and counterfeiting nearly every digital signature. This has huge implications for the military and banking. Much of this depends on the difficulty of finding the prime factors of a large number. I suspect this computer could be programmed to make short work of cracking a 512 bit public key into two prime factors.

    XOR one-time pads for encryption will still be immune. I wrote such a system for a wealthy Saudi earlier this year that is designed for cyberpeasants to use that hides most of the book keeping. I have permission to release it into the public domain next year.

    I trust the US government will keep a tight rein on whom is allowed to buy or access these computers.

    I have yet to learn what the programming languages for these beasts looks like. For all I know it might require building aux hardware.

    For more info see dwavesys.com

    There are a dozen or so related Youtube videos. Here are a few to get you started:

    youtube video1

    youtube video2

    youtube video3

    youtube video4

    I sent the sales people a query on cost. They require supercooling and a dedicated room, so I suspect they are not cheap.

    • In reply to #2 by Roedy:

      Quantum computers can solve “try all possibilities” problems by
      using Quantum Mechanics to explore all possibilities simultaneously.

      No that’s wrong although its a common misconception. There is no infinite improbability drive in a quantum computer that allows you to try infinite number of possibilities all at once. In fact a quantum computer isn’t really all that revolutionary in many ways from a computer science perspective. Its still a Turing Machine which means that in theory it can’t solve any problem that any other computer can’t solve. The difference is that for certain kinds of problems it can be massively parallel but the important thing is it still can’t solve NP complete problems. NP complete problems are problems that end up with exponentially large search spaces. For example playing chess by enumerating all the possible moves and coutner moves.

      I actually thought this article was terrible. I don’t know what in the world they are talking about with creativity and how quantum computers enhance it. Quantum computers — if they ever exist — will just be a lot faster on certain kinds of brute force search problems. From what I’ve seen so far there is no implied major breakthrough in general purpose computing any time in the forseeable future and certainly nothing related to “creativity”.

      Here is an article from Scientific American. Unfortunately its behind a firewall now but I got it when it was free if anyone wants a copy: http://www.scientificamerican.com/article.cfm?id=the-limits-of-quantum-computers

  2. One of the videos cleared something up. These are not programmable beasts. It is special-purpose quantum hardware designed to implement one algorithm, in this case a discrete optimisation. These are not capable of running Schor’s algorithm, discovered in 1994 that would allow you to find the prime factors of large numbers, useful in code cracking. It would be useful in classifying and searching images for example.

    So perhaps Google Image could separate out images of lambs the animal from Charles Lamb the poet and young Elvis from fat Elvis. Perhaps it could show images in roughly chronological order and watermark images with their dates.

    Apparently implementing Schor’s algorithm is harder and has less commercial potential. So the banks can rest easy for a while longer.

  3. What is the point of a computer that works out 2+2 and gets -191?

    Ok I don’t understand quantum computing in the least, but from from sketchy understanding of quantum mechanics, quantum computing is an oxymoron.

    • In reply to #5 by Krasny:

      What is the point of a computer that works out 2+2 and gets -191?

      Ok I don’t understand quantum computing in the least, but from from sketchy understanding of quantum mechanics, quantum computing is an oxymoron.

      “2 + 2 = 4(-ish, for quite a long time)”

    • In reply to #5 by Krasny:

      What is the point of a computer that works out 2+2 and gets -191?

      Ok I don’t understand quantum computing in the least, but from from sketchy understanding of quantum mechanics, quantum computing is an oxymoron.

      As I understand it from when I read about these (10 years ago) the advantage of quantum computers is the ability to store any number no matter how vast on a single qubit. A normal bit is obviously binary only, so to store large numbers in memory requires stringing together lots of bits.
      Being able to process with a smaller amount of units means processing can occur much much faster on number crunching problems.
      However even with this extra speed I am sceptical that means they can crack third party encryption algorithms quickly because there are a limit to how many qubits you can get to play together. Last I read it was seven.

      • In reply to #10 by mr_DNA:

        the advantage of quantum computers is the ability to store any number no matter how vast on a single qubit. A normal bit is obviously binary only, so to store large numbers in memory requires stringing together lots of bits. Being able to process with a smaller amount of units means processing can occur much much faster on number crunching problems.

        The qubit doesn’t give you much extra memory. As I understand it a qubit has four states as opposed to the normal 2 of a bit. So with 100 bits you have 200 possible states. With 100 qubits you have 400. Its more but that’s not where the power comes from. The power is (and Roedy had this right its just not infinite) is that you can explore an enormous number of parallel qubits all at once using the state of a few hundred particles and quantum entanglement. How you do that I have no idea but that is where the power comes from you can search through a state space in parallel all with one instruction but there is nothing really special about qubits as a data structure. Its unusual to use four states rather than two but I think its a natural design decision because of the way they use the particles.

Leave a Reply