Research Translated — Digital is better (A study of DDS and quantum computing)

We’re back at you with another edition of “Research Translated” (RT for short), where we translate quantum computing research into terms you can understand. This is an experimental segment we are trying out. Love it? Hate it? Let us know.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

Today, we are going to be talking about an article originally published in Cornell University’s arXiv.org titled Direct digital synthesis of microwave waveforms for quantum computing

First, some background…

a technique called circuit quantum electrodynamics (circuit QED).

QED is a way for scientists to study the interactions between matter and light. Usually, a photon (a unit of light) is coupled with a quantum object. Circuit QED specifically uses artificial atoms for their quantum objects. So when utilizing circuit QED, you would not use a natural atom like one of hydrogen or silver. You would use an artificially created one.

Scientists have been able to use different types of qubits as their artificial atoms in circuit QED. Circuit QED has had enough success with qubits that the technique is currently a leading contender for creating a universal quantum computer.

There is a problem, though.

Circuit QED requires microwaves (the waveform, not the machine) to work. For their microwave resonator, Raftery et al (the researchers of this paper) used what they called an Arbitrary Waveform Generator (AWG).

As the name implies, it is a device that generates arbitrary waveforms.

Currently, AWGs generate waves called control pulses. Control pulses are low-frequency, basic waves.

These control pulses are then upconverted to microwave frequencies. Upconversion is a technique used by researchers and engineers to change a signal from a lower-frequency signal to a higher frequency signal. The control pulses created by the researchers are relatively low frequency, and microwave frequencies are relatively high, so upconversion makes sense in this case.

Upconversion has some problems though. Mainly, it requires a lot of equipment. One qubit requires two AWG for upconverting, plus another device called a microwave generator, plus a fourth device called a mixer.

Furthermore, many of the implementations of circuit QED in research have their readout pulses sent down the same input line. Readout pulses are signals sent out that assist scientists in collecting results. By sending out their readout pulses on the same line as the input, scientists require even more equipment (specifically a generator, mixer, and an additional power combiner) in order to carry out this experiment.

For those counting, that’s 7 additional large machines that must be present to carry out this experiment!

To make matters even worse, each of these pieces of equipment requires very fine tuning for accurate results. In fact, it’s pretty easy to get inaccurate results with this many moving pieces, especially when the moving pieces are sensitive machines. Mixers, in particular, are easy to mess up.

Scientists are all about accuracy, and engineers are all about efficiency. And this process is neither accurate nor efficient.

If we could make this whole process go digital, the experiment would be a lot easier to carry out. At the very least, a digital version of the experiment would require a lot less equipment.

As it turns out, many modern AWGs can operate with a very high sampling rate, close to 92 gigasamples per second. That means we can sample a signal 92 billion times each second.

That sampling rate is so high that modern AWGs actually have the capability of sending out microwave frequency control pulses directly without doing any upconverting.

This system is called digital direct synthesis (DDS)

DDS has lots of benefits. As systems become more and more complex, this will become more important. By eliminating upconverting alone, we’re saving a lot of money on hardware. This also allows us to eliminate mixers, which removes a lot of places that could introduce errors into the system.

Many are not sure how viable this method actually is. In particular, can DDS generate microwave signals that are as high quality as those of the current method?

To figure this out, the researchers decided to show that using DDS would cause low single-qubit error rates in circuit QED.

The researchers chose to use a process called randomized benchmarking to measure the error rates.

And what were the results?

As predicted, DDS was a lot simpler to implement than the existing techniques with traditional AWGs in circuit QED. In particular, no mixers were required, which reduced error significantly.

The researchers found that a little more noise was created in the signal when using a DDS AWG than when using a traditional AWG, but they did not expect it to increase error rates in quantum computers since there are plans for future experiments that will change circuit QED in such a way to reduce that type of noise.

The randomized benchmarking experiments results were really good as a matter of fact. The results showed that the qubit control pulses generated by the DDS system had error rates of less than 5e-4 (!).

Think about that, that is 5 errors out of every 10,000 qubits. That is incredibly low!

While this result is incredible, it is worth noting that the experiments caused the researchers to find a hardware limitation when using a fully DDS system for circuit DDS. However, the defect they did find did not actually affect the randomized benchmark experiment directly, so the result still holds.

So what does this mean?

This experiment touches on two important challenges that need to be surmounted before quantum computers can be used widely.

First, error correction. Everything that is sent electronically get translated to and from bits and bytes (a process called serialization and deserialization, respectively). Sometimes these transformations are lossless, meaning that there is no loss of data. But sometimes they are lossy, meaning that some data may be lost. And when they are lossy (which is sometimes unavoidable or considered a tradeoff for better performance), we require error correction.

If quantum computers are going to be out there completing millions of computations a second, then we need to be sure that the computations return as accurate results as possible, otherwise, there is no way a human or a traditional computer could check all of their computations.

The second challenge is cost. Quite simply, the lower the cost of a quantum computer, the more governments, companies, and eventually consumers will adopt quantum computers. The DDS technique significantly lowers the cost of creating and manipulating qubits via circuit QED.

If circuit QED moved on to be the leading technique in creating a universal quantum computer, then thanks to DDS quantum computers would become cheaper, faster.