The next generation of computers is a few years off, but it’s pretty damn cool.
The machine is the D-Wave 2X, and the only working model outside the company is in the Quantum Artificial Intelligence Lab. A joint project between Google, NASA, and the Universities Space Research Association, the lab will test-drive the 2X on some sticky problems in high-powered computing.
The 2X is a type of quantum computer, which means it uses devices that exploit quantum physics to replace transistors and other components of ordinary computers. The quantum nature of the inner workings in theory should make the computer solve problems much faster than anything else available, making it useful for a wide range of applications. While there are no fully quantum computers out yet, the 2X is the closest yet—assuming it works as advertised.
All ordinary computers—laptops, desktops, tablets, phones, e-readers, smartwatches, or whatever—are based on semiconductors, materials that conduct electricity reluctantly. That reluctance makes it easy to control the flow of power using devices like transistors, so that current is either flowing or not: represented by the numbers “1” for “on” or “0” for “off.” Combining the current through different parts of circuits allows computers to perform simple mathematical operations using just those two numbers. The power of a computer lies in doing lots and lots and lots of computations, faster than we perceive. (Note to experts: this is a simplified explanation. Don’t try this at home, kids!)
Quantum computers also use just two numbers, but instead of manipulating electric current, they manipulate “quantum states.” A quantum state contains a kind of list of all the possible configurations a particle (or other microscopic system) can have: its position, speed, energy, spin, and so forth. When two quantum systems interact with each other, or we perform a measurement in the lab, the quantum state describes how likely the outcome of that interaction or measurement was.
Until the measurement, though, the state is undetermined. For a quantum computer, a “quantum bit” or “qubit” could be either 0 or 1, but we don’t know until the computer reads it out in some way. One qubit, just like one bit in a normal computer, is pretty useless. However, if you have lots of qubits, you can perform many calculations simultaneously. Theoretically, a quantum computer could solve a given problem every possible way, including finding all the wrong solutions, in the amount of time it would take a normal computer to find a single solution to the same problem. That makes quantum computers useful for stuff like decryption and finding the optimal approach to performing searches.
But “theoretically” is the key word. Nobody has yet built a true quantum computer, and the D-Wave 2X is no exception. (More about what it is shortly.) One difficulty is that quantum states are delicate things: interactions between particles behave exactly the same way a measurement does, altering the state and screwing up whatever calculation we were trying to do. A larger problem is that all qubits in the computer need to be entangled with each other, meaning that their quantum states are linked up: a measurement on one qubit restricts the possible outcome of similar measurements on all the others. The more qubits, the harder the entanglement becomes.
Matthew R. Francis 09.30.151:00 AM ET