6 ms·

I actually dont understand most of quantum stuff. When they say 16 qbits, are these simulated qbits or real? Aren't real qbits impossible to make currently bec

I actually dont understand most of quantum stuff. When they say 16 qbits, are these simulated qbits or real? Aren't real qbits impossible to make currently because of noise problem? When I was doing my masters, people were talking about replacing heavily doped channel of silicon transister with carbon nanotube. Turns out the hardest problem was making a decent connection between nanotube and metal. I dont think the problem has been solved yet. I am guessing that well see a functional carbon nanotubes in silicon transistor at least a decade before we see real qbits? After all fabrication processes must be similar at this scale.

lucubratory 3y ago
There are real qubits, that suffer from all the reliability issues implied by what you're talking about. There are also error correction algorithms to try and salvage from the mess the computational equivalent of error-free real qubits, through error correction on a bunch of real-world, error-prone qubits. The industry standard (such as it is) when reporting qubit counts is that you report your qubit logical equivalent, because that's what counts for computation. If your chip has 500 qubits with a ruinous error rate that requires a god-tier error correction algorithm to achieve a functional result equivalent to a 12 qubit computer, you should report that as "We made a 12 qubit computer".
fastneutron 3y ago
A qubit is any addressable 2-level quantum system. Think spins, oscillation modes, energy levels, polarization, etc. All of these have been implemented and demonstrated to varying degrees, with superconductors and trapped ions (this article) being the farthest along. Noise is indeed a problem, and right now these systems can’t execute more than a few hundred operations at most before decohering to the point of randomness. There is an error rate threshold (~1e-3 - 1e-4) at which error correction becomes possible, and the field has been hovering around that threshold for the past 3 years or so. I think we’ll know within the next 5 years whether or not it’s possible to engineer these error correction schemes in a scalable way.

lucubratory 3y ago

There are real qubits, that suffer from all the reliability issues implied by what you're talking about. There are also error correction algorithms to try and salvage from the mess the computational equivalent of error-free real qubits, through error correction on a bunch of real-world, error-prone qubits. The industry standard (such as it is) when reporting qubit counts is that you report your qubit logical equivalent, because that's what counts for computation. If your chip has 500 qubits with a ruinous error rate that requires a god-tier error correction algorithm to achieve a functional result equivalent to a 12 qubit computer, you should report that as "We made a 12 qubit computer".

fastneutron 3y ago

A qubit is any addressable 2-level quantum system. Think spins, oscillation modes, energy levels, polarization, etc. All of these have been implemented and demonstrated to varying degrees, with superconductors and trapped ions (this article) being the farthest along. Noise is indeed a problem, and right now these systems can’t execute more than a few hundred operations at most before decohering to the point of randomness. There is an error rate threshold (~1e-3 - 1e-4) at which error correction becomes possible, and the field has been hovering around that threshold for the past 3 years or so. I think we’ll know within the next 5 years whether or not it’s possible to engineer these error correction schemes in a scalable way.