The paper's conclusion [1] does a much better job of summarizing what they did than the posted article. For example, they explicitly note limitations:
> Because the QMT representation uses frequencies to encode the state, the number of qubits would be limited by bandwidth. Using an electronic voltage as a physical representation, we find that frequencies in the practical range of 0.1 Hz to 100 GHz would allow for about n=40 entangled qubits to be realized. [...] Since the quantum state is explicitly encoded in the signal spectrum, security applications such as quantum key distribution would not seem to be amenable to this approach, although they may be emulated.
And they even note that their approach has not been demonstrated to be superior:
> It remains to be shown that this approach can be realized in a fault-tolerant manner, using existing quantum error correction techniques, and that it ultimately has utility and merit over current approaches to quantum computing—or standard digital computers, for that matter.
My takeaway is that the paper's intended goal is to explain a neat, perhaps practical, way to emulate some qubits. It doesn't do anything "magical", it encodes all of the amplitudes into the signal (in a real quantum computer, you'd have no direct access to the amplitudes) and explains how to emulate operations being applied.
This is something that has been noticed before mathematically, but hard to do in practice. The article is maybe a little confused. The limitation on size comes from the limit on frequency/amplitude resolution - for a given scale and resolution, the need to distinguish states grows combinatorially. It can be overcome by making the system smaller. Eventually you can rely on quantum effects to help, when you are no longer emulating a quantum system, you've basically built one.
You could, but it is a bit pointless. The idea is that mathematically a superposition of analog waves looks the same as the superposition of quantum states, so you can get exactly the same behaviour using the natural laws of physics. A digital system that modelled the waves would require an absurd amount of resolution since the interference effects need to be modelled accurately. In essence you are throwing away most of your information capacity by switching to digital on/off signals from superimposed waves even though you are using the same wires. The reason quantum computing is interesting in the first place is to use all that wasted information to produce a Massive gain in computing power in a much smaller system.
Digital circuits work fine for simulating, and emulating, quantum computers. They probably chose analog circuits because that's what they like to work with, or because the more-direct encoding allows for efficiencies.
Not quite true. The analog version is inherently parallel, the digital version is not (cf the MATLAB simulation discussed in the paper). The parallelism of the analog system increases exponentially with it's size, the digital system only linearly. So an analog computer built in the way they indicate might be able to create a useful and powerful practical system to test quantum algorithms on long before quantum computing becomes a reality, whereas the digital model cannot be more powerful than the computer it is running on and will struggle to model even moderately complicated quantum systems.
Just because you can't determine what it will do doesn't mean that it's nondeterministic. You just don't have enough information about the hundreds of billions of radio sources scattered all throughout the galaxy. You probably don't have any better information about what the radio station a few kilohertz away will broadcast next, either.
> Because the QMT representation uses frequencies to encode the state, the number of qubits would be limited by bandwidth. Using an electronic voltage as a physical representation, we find that frequencies in the practical range of 0.1 Hz to 100 GHz would allow for about n=40 entangled qubits to be realized. [...] Since the quantum state is explicitly encoded in the signal spectrum, security applications such as quantum key distribution would not seem to be amenable to this approach, although they may be emulated.
And they even note that their approach has not been demonstrated to be superior:
> It remains to be shown that this approach can be realized in a fault-tolerant manner, using existing quantum error correction techniques, and that it ultimately has utility and merit over current approaches to quantum computing—or standard digital computers, for that matter.
My takeaway is that the paper's intended goal is to explain a neat, perhaps practical, way to emulate some qubits. It doesn't do anything "magical", it encodes all of the amplitudes into the signal (in a real quantum computer, you'd have no direct access to the amplitudes) and explains how to emulate operations being applied.
1: http://iopscience.iop.org/1367-2630/17/5/053017/article#njp5...