A Fast and Scalable Approach to Controlling 100 Qubits and More - Q&A

September 2, 2020 by Tobias Thiele

On August 20, we talked about our fast and scalable approach to controlling 100 qubits and more. The webinar was recorded and is available on demand here.

In the first part of the webinar, we covered our fast approach and explained how to get a handle on the complexity of pulse sequences and measurements on a small number of qubits. At the example of our HDAWG Arbitrary Waveform Generator and UHFQA Quantum Analyzer, we argued why we can comfortably claim demonstration of Rabi oscillations in a day. In specific, we discussed the workflow of the UHFQA to do parallel single-shot readout of multiple qubites, and our gate-level sequencing approach to control single and two-qubit gates. As a highlight, we programmed a simple Ramsey sequence in 2 minutes.

In the second part, we discussed our scalable approach, i.e. our avenue to controlling 100 qubits and more. We explained how our Quantum Computing Control System (QCCS) is more than the sum of its parts, by demonstrating a feedback experiment. Furthermore, we highlighted our approach to reduce system complexity by integration of additional instrumentation into new instruments. Importantly - and for the very first time - we presented our latest innovation: the new SHFQA 8.5 GHz Quantum Analyzer.

In the remainder of this blog post, we answer all questions that were asked by the audience during our webinar and the Q&A session. The questions are sorted according to the different instruments that they relate to best.

Enjoy browsing them and - as always - if you have any questions, please get in touch with us.

HDAWG

Does pulse envelope shaping affect the decoherence time and qubit control?

There are several detrimental effects that can happen if one does not shape its control pulse. For example, a short pulse with sharp edges has a very wide frequency spectrum and can excite unwanted states. This reduces the fidelity of qubit gates and, ultimately, the fidelity of your quantum algorithm. Common approaches to mitigate these effects are to use DRAG or WAHWAH pulses.

How does one define a DRAG pulse on HDAWG without reducing instrument sampling rate?

Starting with LabOne 20.01, the signal routing and modulation functionality has been extended such that dual-phase digital modulation at full sampling rate is possible, e.g. for the generation of DRAG pulses. Please refer to Sec. 4.4 of the HDAWG User Manual for more details here.

You talked about fast single qubit gates by IBM that have been measured with your AWG, are there also measurements on fast 2 qubits gates?

Yes, for example - very recently - Leo DiCarlo from TU Delft achieved high-fidelity (>99.9%) controlled-Z two-qubit gates that operated at the speed limit (~43 ns total time). You can find the arXiv version here.

What's the latency of the real-time precompensation part in AWG?

This depends on which and how many filters are enabled. For example, the high-pass compensation uses about 40 ns, and the bounce-compensation only about 13 ns. The exact latency of the active configuration is displayed in the user interface and accessible from the API. More information can be found in the HDAWG User Manual here.

How much current and voltage can the AWG provide maximally?

In the amplified mode, the HDAWG can provide up to 2.5 V into 50 Ohm - bipolar. This corresponds to a current of up to 125 mA.

What is the frequency range of the internal oscillator used for wave modulation?

300 MHz with an anti-aliasing filter, 750 MHz in the direct mode. However, higher frequencies exceeding 1 GHz can be generated at the penalty of a reduced amplitude due to filtering.

UHFQA

Why does qubit state "1" have two red blobs in the IQ plane?

This was an exaggeration of qubit cross-coupling and meant to be purely illustrative.

How do you do fast averaging?

There are two aspects to that: on the signal generation and acquisition side. For generation, our sequencer allows to define common loop commands, such as a for, while, or repeat loops. These loops are compiled and then run in real time on the instrument. On the acquisition side, our result logger that runs on the FPGA ensures that the results get averaged in real time - and only at the end of an experiment communicates the results to a computer. This allows for different averaging modes such as point-wise or TV-mode averaging.

How many qubits can be analysed with this instrument?

The UHFQA can analyze up to 10 qubits in parallel using its 10 matched filter integration weights.

You mentioned that FPGA programming is not required to operate your system. But some users may be interested in customizing low-level operations. Are you going to provide this possibility? In what form?

With the exception of the PQSC (see below), we are convinced that our instruments are tailored to provide the functionality that you need and FPGA-programming is not needed by the user. We are continuously talking and collaborating with researchers and scientist all over the world to ensure that we provide the best solutions possible. Indeed, we continuously provide firm- and software updates of our instruments to the community free of charge. If you want to do an experiment and are missing a feature - we are more than happy to work with you and find a solution.

SHFQA

How can the SHFQA interface again with the QCCS?

The SHFQA features a 32-bit DIO connector and the Zurich Instruments ZSync link. Both of them can be used to transmit and receive Data and trigger signals.

What do you mean with "no mixer calibration" in SHFQA?

For optimal performance of some frequency conversion protocols, e.g. when using an IQ mixer, the system needs to be continuously recalibrated. This not only interrupts the flow of the experiment, but is tedious and error-prone. The SHFQA uses a double super-heterodyne frequency conversion scheme that does not rely on interference, but on filtering. No mixer-calibration is needed in this case - and the output and analysis bandwidth is clean and stable from the first turn-on of the instrument.

Is your SHFQA based on direct digital synthesis or analog up/down conversion? Will the next generation be based on DDS? On what time scale are you planning that shift to happen?

It relies on a double super-heterodyne upconversion approach, i.e. an analog frequency conversion. We are very convinced in the solution as it is wide-band, stable and does not require mixer calibrations. Direct RF, e.g. synthesis of a waveform at lower frequencies and then exploiting higher order Nyquist zones are an attractive solution, but current technology does not provide the optimal mix in performance and low-latency addressing needed for high-performance quantum computing applications.

How many qubits can be analysed with this instrument?

The SHFQA can have up to 16 matched filter integration weights for a single channel. This means a single channel can optimally measure up to 16 qubits, 8 qutrits or 5 ququads in parallel. For the 4-channel version of the instrument, this corresponds to 64 qubits, 32 qutrits, or 20 ququads.

PQSC

Can you tell something about the Programmable Quantum System Controller?

The PQSC is the central controller of our Quantum Computing Control System and interfaces to up to 18 instruments. When all are connected to a HDAWGs, this corresponds to 144 channels that are synchronized to sub-ns using the PQSC. It also allows to implement error correction and feedback protocols. Please contact us - and we are happy to discuss more.

What does the acronym PQSC stand for?

Programmable Quantum System Controller.

Do you have any error code decoders implemented in the PQSC?

Starting with LabOne 20.07, a decoder based on a configurable look-up-table is available on the PQSC. Please refer to the PQSC User Manual for more details here.

How programmable is the PQSC? Can we upload our own FPGA firmware? What does this mean in practice? Which software/equipment is needed? And how is it this feature supported?

Yes, you can. The PQSC will feature FPGA access that allows the user to implement its own decoders. We are still defining the best and most user-friendly way to open the FPGA to our customers. In the mean time - and in addition - we will be continuously implementing decoders and deploy them to the community.

What is the approximate cost of the PQSC?

Given that the PQSC is a vital part our Qantum Computing Control System, and requires the use of our ZSync and other instruments, the cost of the instrument is small compared to other offers. That way we actively support scaling ambitions, and feedback and error correction aspirations of even small systems. Please contact us - and we are happy to discuss the value of our PQSC that suits your needs best.

QCCS

What do you use for a clock?

All devices are normally connected either to a 10/100 MHz frequency reference clock provided by the lab to which all instruments are locked - in specific also third-party instruments such as microwave synthesizers.

What is the typical cost of the system?

This depends very much on the required configurations of the instruments and their numbers. Please contact us - and we are happy to discuss the value of our Quantum Computing Control System that suits your needs best.