Hands-on Superconducting Qubit Characterization

November 25, 2022 by Bruno Küng

This blog post accompanies the webinar "Superconducting Qubit Characterization", which introduced the basic physics of transmon qubits as well as the tools and methods required to tune them up. The webinar included a live Q&A session, but due to time constraints, not all of the viewers' questions could be answered within the duration of the session. Below, we address the technical questions asked during the webinar.

Please note that for a small handful of questions we would have needed further information to be able to answer them, hence we did not include them here. If you don't find the answer to your question or are interested to engage in a conversation, please contact us!

Watch the webinar recording

Hands-on Superconducting Qubit Characterization | Zurich Instruments Webinar

LabOne Q Code Download

The webinar featured 3 hands-on segments demonstrating measurements on real qubits at the ETHZ-PSI Quantum Computing Hub at the Paul Scherrer Institute:

  1. qubit and readout resonator spectroscopy
  2. pulsed qubit control and Rabi oscillations
  3. single-shot readout

All measurements were carried out with LabOne Q, the Zurich Instruments software framework for quantum computer control. The code used during the webinar is available on our github repository and can be executed on the SHFQC 8.5 GHz Qubit Controller.

To dive further into the capabilities and look-and-feel of LabOne Q, you can request a demo or browse through our documentation to find further examples. 

Q&A

Does the resonator measurement destroy/reset the qubit state?

The state of the qubit is not destroyed by the process. The measurement is what we call "projective". This means that if the qubit is in one of the computational states |g> (|0>) or |e> (|1>), it remains unchanged and the measurement merely reveals information about which state we have. However, if the qubit is in a superposition state, the measurement makes it collapse into one of the computational states at random.

What is the advantage of using transmons for your specific 17-qubit experiment?

Let me answer first more generally why we like to use superconducting qubits and why we think they are a useful platform for quantum computing. The main reasons are that they are solid state systems which are constructed relatively easily with basic microfabrication techniques and we have great freedom in choosing their parameters. This is in contrast with some naturally occurring qubits such as trapped ions whose fundamental parameters are basically fixed. Now as to why we use transmon qubits specifically - the reason does not have anything to do with error correction per se but it's rather because transmons are a mature type of qubit, well understood and conceptually simple, with a number of favorable properties when it comes to coherence times, susceptibility to noise, etc. That being said, there are of course other types of qubits, some of them also looking quite promising and recently attracting increasing attention - for example capacitively shunted flux qubits or fluxoniums.

What happens at the compilation step, and what at the execution step, of an experiment?

A quantum information processing experiment with LabOne Q is performed in two steps: In a first step, a Compiler generates the instrument settings, waveform data, and sequencer code for the AWGs. This step can be performed fully offline, allowing for more flexibility for scheduling experiments. Subsequently, the Controller accesses the pre-generated code and waveform data, uploads them to the devices and starts the online execution of the experiment.

How free are you in choosing the form of the controlling impulse?

LabOne Q offers two ways for defining the pulse shapes for your control signals: For commonly used control pulses, one can make use of the built-in pulse library. This library offers standard pulse shapes such as Gaussian or DRAG pulses, which can be parametrized with a small set of parameters, such as pulse amplitude or pulse width. In addition, LabOne Q allows you to specify arbitrary pulse shapes sample-by-sample, which can be generated via user code for example in the form of a numpy array.

Do you offer openQASM3 compatibility?

An OpenQASM 3.0 parser is currently under development. This parser will make it easier to interface LabOne Q with higher levels of your software stack. Please contact us to learn more.

What is the maximum number of qubits that can be read out with the quantum analyzer channel?

The quantum analyzer channel of the SHFQC contains up to 16 integration weight units. This is enough to read out 16 qubits simultaneously, while in practice it may be useful to assign one signal line to a smaller number of qubits, but use some of the integration weight units to read out higher excited states of a given qubit. For example, the 16 units allow for readout of 8 qutrits, or 5 ququads. The different readout signals need to be distributed over the bandwidth of the instrument, which is 1 GHz, or even more when accepting a reduced signal amplitude.

How many qubit can we control with your system? What is the parameter that limits the synchronization of many SHFQC?

Up to 18 SHFQC can be synchronized with a PQSC in a star-like topology, and this number is limited by the number of synchronization ports of the PQSC. The resulting system provides 108 microwave control channels and 18 readout channels. In case also fast flux channels are required, some of the synchronization ports of the PQSC are used to synchronize HDAWG. If your system requires more channels, please contact us. 

What is the latency of analog-to-analog feedback? What is the latency when a conditional pulse is the output of a different box?

The SHFQC can provide internal feedback, e.g. for active qubit reset in less than 350 ns. This time is measured as a last-sample-in to first-sample-out, that means, from the time the last sample of a readout signal is acquired on the signal input, to the first sample of a conditional control pulse appearing on the analog signal output. When passing to a different instrument via the PQSC, the latency is 550 ns.

What is the difference between superconducting and spin qubit control?

This question is best addressed by looking at the 3 types of signal interfaces separately: microwave control, flux control, and readout. For microwave control lines, there is a large overlap of the corresponding requirements. The control pulses have similar timescales and the used frequencies partially overlap, but spin qubit control sometimes requires higher frequencies, e.g. up to 20 GHz. Flux control signals in superconducting qubits have a close analogy in spin qubits: fast gate voltage signals. Both require unmodulated pulses with well-controlled timing and high vertical resolution. The Zurich Instruments HDAWG is a suitable instrument to generate both fast flux and voltage bias signals. Finally, qubit readout has the largest differences between superconductors and spins. Depending on the readout method, spin qubits often requires fast lock-in measurements at frequencies below 1 GHz, as offered by the UHFLI Lock-in Amplifier. The demodulated signals are processed using threshold detection or gated integration.

Do you allow users to use your platform on the web so that they can test them on a real qubit/s?

Our instruments are for example used on the quantum computers of Quantum Inspire which is publicly accessible on a high interface level. For a more direct experience of our instruments, the best is to contact us to arrange a demo, or even a test measurement in your lab.

Can I change the threshold for state discrimination as a result of the previous measurement (so in a few hundreds of nanoseconds)?

This specific real-time adjustment is not possible. However, it may be possible to address this use case differently by acting in real time on the drive pulse amplitude or phase.

Why do we have a sigal generator for readout, when we have already send RF pule by the main signal generator?

Qubit spectroscopy is a two-tone measurement, where one tone is used to change the qubit state, and one to measure the state. The two signals are guided to the sample on different signal lines.

Where does the requirement for the complex conjugate come from? (in the formula for weighted integration)

This is a matter of convention in the implementation of the weighted integration. In the SHFQC, the implementation is chosen to be consistent with the definition of a Fourier transform.

Is the Jupyter demo script used in the webinar integrated within LabOne Q or a separate python file repository?

The demo script is a separate file. The file is based on the standard LabOne Q examples, but was modified for the purpose of the webinar, in particular in the section for single-shot readout. The file is available for download on our github repository.

What amplifiers are used in your ADC? In my application I need amplifiers at 8 GHz for the development of a gamma radiation detector.

The amplifiers at the input stage cover a range between -50 dBm to +10 dBm. The inputs offer a particularly low noise in order to faithfully process signals from qubit readout. Please find noise specifications in the user manual (section Signal Inputs).

Do you have floating point support?

There is no general support for real-time floating point arithmetic on the SHFQC, but the AWG sequencer of the Signal Generator channel supports integer arithmetic. Furthermore, certain specialized operations (real-time phase, amplitude, and frequency increments) are supported with floating-point precision.

What is the goal of flux lines?

Flux lines are mostly used to control the interaction between qubits, for example to realize 2-qubit gate operations. Flux lines often allow to tune or modulate the resonance frequency of a sub-circuit such as a qubit, or a tunable coupling circuit. The interaction between qubits is then controlled by bringing circuits into and out of resonance, or by generating sideband transitions.