Saturday, March 7, 2015

Google claims qubit error correction

I am skeptical about whether quantum computers will ever be built, even tho Google, Microsoft, and Amazon are all spending millions of dollars on research. So I expect that the typical reader will assume that these companies employ 1000s of very smart people, and they do not waste their money on foolish dead-ends.

Here is the hype:
When scientists develop a full quantum computer, the world of computing will undergo a revolution of sophistication, speed and energy efficiency that will make even our beefiest conventional machines seem like Stone Age clunkers by comparison.

But, before that happens, quantum physicists like the ones in UC Santa Barbara’s physics professor John Martinis’ lab will have to create circuitry that takes advantage of the marvelous computing prowess promised by the quantum bit (“qubit”), while compensating for its high vulnerability to environmentally-induced error.

In what they are calling a major milestone, the researchers in the Martinis Lab have developed quantum circuitry that self-checks for errors and suppresses them, preserving the qubits’ state(s) and imbuing the system with the highly sought-after reliability that will prove foundational for the building of large-scale superconducting quantum computers.

It turns out keeping qubits error-free, or stable enough to reproduce the same result time and time again, is one of the major hurdles scientists on the forefront of quantum computing face.
Instead of "When scientists develop", it should say "In the unlikely event that scientists develop".

MIT Tech. Review reports:
A solution to one of the key problems holding back the development of quantum computers has been demonstrated by researchers at Google and the University of California, Santa Barbara. Many more problems remain to be solved, but experts in the field say it is an important step toward a fully functional quantum computer. Such a machine could perform calculations that would take a conventional computer millions of years to complete.

The Google and UCSB researchers showed they could program groups of qubits — devices that represent information using fragile quantum physics — to detect certain kinds of error, and to prevent those errors from ruining a calculation. The new advance comes from researchers led by John Martinis, a professor at the University of California, Santa Barbara, who last year joined Google to set up a quantum computing research lab ...

To make a quantum computer requires wiring together many qubits to work on information together. But the devices are error-prone because they represent bits of data—0s and 1s — using delicate quantum mechanical effects that are only detectable at super-cold temperatures and tiny scales. This allows qubits to achieve “superposition states” that are effectively both 1 and 0 at the same time, allowing quantum computers to take shortcuts through complex calculations. It also makes them vulnerable to heat and other disturbances that distort or destroy the quantum states used to encode information and perform calculations.

Much quantum computing research focuses on trying to get systems of qubits to detect and fix errors. Martinis’s group has demonstrated a piece of one of the most promising schemes for doing this, an approach known as surface codes. The researchers programmed a chip with nine qubits so that they monitored one another for errors called “bit flips,” where environmental noise causes a 1 to flip to a 0 or vice versa. The qubits could not correct bit flips, but they could take action to ensure that they did not contaminate later steps of an operation.
My prediction is that these companies will never see a dime of business value from this research.

Implementing the quantum error correction may well be a legitimate technical advance, but I suspect that this is just a disguised quantum experiment and does not give any scalable computing power.

Google is promising self-driving cars with 5 years or so. No promises are being made for quantum computers, as far as I know. If they were honest with their investors, what would they say? Those investors consider the self-driving cars a long-term project.

3 comments:

  1. Google: another completely useless entity staffed with con artists

    ReplyDelete
  2. If there was any talent at Google they would have leveraged their infrastructure and revolutionized quantum chemistry by now. But there is no talent at Google.

    ReplyDelete
  3. UCSB (more accurately, the ocean side estuary of Isla Vista) is the same circus clown car a great deal of the MWI lunacy arrived in, and every so often they have a virtual metaphysics nonsense con fab where they spout the latest mathematical happytalk on how they can prove 'x' based on imaginary black holes. As far a rackets go, it's good work with a beautiful view if you can afford to live in a town where a 1968 termite eaten tract home goes for 1.4 million dollars.

    ReplyDelete