Choosing the Appropriate Diagnostic Tool
In this tutorial, we give an overview of some of the options for characterizing your device using True-Q™’s protocols. As discussed in the Introduction to Quantum Gates and Circuits example, True-Q™ makes it straightforward to generate circuit collections for device characterization, but sometimes you may not know where to start or which protocol fits your use-case. This page will list some of the protocols True-Q™ offers for device characterization and suggest several that could be used as a starting point to better understand the noise profile of your device.
Protocols to Run First
There are three major types of error that we focus on in this tutorial:
State preparation and measurement errors are introduced when the initial state used by the device is not in the expected state or when measurements are imperfect.
Local gate errors are a type of gate error typically introduced by non-perfect gate calibration. These errors act on the same qudits that a gate is being applied to.
Crosstalk errors are another type of error introduced during the application of gates on a system, and are distinct from local gate errors in that they act on qudits that are not being acted upon by the gate being applied.
Characterizing Local Gate Errors
Gate errors are a ubiquitous source of error in quantum computing devices. In this section we focus on characterizing errors arising from calibration; in the following section we look at evaluating crosstalk errors, which are another type of error introduced when applying gates.
Streamlined Randomized Benchmarking (SRB) is the canonical protocol for characterizing noisy gatesets. We recommend this protocol as a first step for device characterization because it does not require a large number of circuits (~60 circuits are generally sufficient) and returns a figure of merit that most users of quantum computers are familiar with: the average process infidelity of the gateset. A tutorial on the various figures of merit quantified by True-Q™'s protocols is coming soon.
Extended Randomized Benchmarking (XRB) quantifies the stochastic error introduced by gates. After running SRB and XRB, we can compare the fitted parameters they return to see how much of the average gateset error is stochastic and how much is coherent. See Comparing Infidelities with SRB and XRB for an example showing how to retrieve this information. XRB requires ~180 circuits for local gates or ~720 circuits for 2-qubit gatesets, so it is a bit more expensive than SRB, however, it really is a must to see how much of the noise is coherent.
K-body Noise Reconstruction (KNR) is the go-to protocol if we want to retrieve information about the probabilities of Pauli errors for a given cycle. This protocol allows us to retrieve a much more complete picture of what form the noise takes, and is correspondingly more costly to run, requiring ~540 circuits by default if an entangling gate is involved.
Characterizing Crosstalk Errors
Crosstalk is a very common source of error in quantum computing devices, and one that is often not accounted for. True-Q™’s Crosstalk Diagnostics (CTD) provides a pre-packaged call to Streamlined Randomized Benchmarking (SRB) and Extended Randomized Benchmarking (XRB) on different label configurations that allows us to characterize crosstalk errors in a system. This set of protocols is a bit more expensive to run, but it is a must because crosstalk errors are ubiquitous in quantum systems. However, if you’re still getting the hang of True-Q™, feel free to leave this one for later.
Characterizing State Preparation and Measurement Errors
State preparation and measurement (SPAM) errors are another very common source of noise in quantum computing devices. Fortunately, True-Q™ has a Readout Calibration (RCAL) protocol built-in to characterize SPAM errors, and the capability to compensate for them when running other protocols.
Total running time of the script: ( 0 minutes 0.000 seconds)