Quantum Noise Characterization

Quantum Computing Needs Noise Characterisation.

Fight against mistakes is the core of robust, error-tolerant quantum computing. The high error rates of Noisy Intermediate-Scale Quantum (NISQ) devices limit their calculation complexity. Working with superconducting qubits is complicated due to the quantum logic gate. To improve gate fidelity, one must first identify, understand, and quantify noise sources. The noise characterisation process provides the knowledge needed to build more reliable quantum processors.

Identification of Error Sources

Other reasons of quantum gate failures in superconducting systems include qubit instability and microwave control system problems. Researchers have examined the sources of these errors to understand their effects.

Control System Instability, Qubit

A qubit's transition frequency that differs from the control pulse frequency might generate off-resonance faults. Frequency drifts in the control system and qubit may require periodic calibration.

Long-term stability assessments are done by scientists. Over 20-hour investigations showed that qubit frequencies can change by several kilohertz. The drift is dominated by “pink noise,” caused by tiny qubit manufacturing faults. Compared to the qubit's fluctuations, the control electronics' frequency drift is often less than one hertz. The qubit is the main cause of frequency instability.

In-Control Signal Background Noise

Also essential is the microwave pulse background noise utilised to operate qubits. Signal-to-noise ratio (SNR), which compares the intended signal to background noise, is often used to evaluate these pulses. Additive noise may diminish gate quality by randomly activating the qubit and diverging from its intended state. By intentionally adding noise to control pulses in studies, the gate error rate climbs exponentially as SNR lowers. SNR is crucial for high-fidelity quantum control, as shown in this study.

Issues with Material Defects

TLS—microscopic material defects and qubit resonant interactions—cause most noise fluctuations in superconducting processors. The qubit's relaxation time (T₁) may alter due to unexpected interactions. Due to oscillations, error correction noise models might lose accuracy and the device's performance stability and uniformity. Some interactions can lead a qubit's T₁ value to change by nearly 300% in 60 hours.

Noise Characterisation Techniques

Many models and experiments are used to quantify and understand noise sources.

Ramsey Experiments: Long-term frequency variations of a qubit imitate the noise spectrum and predict its effect on gate fidelity. Varying SNR: Researchers can immediately see how gate fidelity affects SNR by adding controlled noise to control pulses.

Randomised benchmarking: This method determines single-qubit gate error rates. Experimental data and simulations that simply account for qubit decoherence can help scientists distinguish the error contribution from the control system and other fundamental sources.

Noise Modelling: The sparse Pauli-Lindblad (SPL) model provides a scalable framework for estimating gate noise. These models help capture noise features, reducing errors.

Enhancing Performance with Characterisation

The ultimate goal of noise characterisation is to improve quantum computing accuracy. Understand the noise environment to develop efficient error mitigation solutions like ZNE and PEC. These strategies use noise knowledge to correct compute results, improving accuracy without ideal hardware.

New stabilisation methods have been developed because to greater understanding of noise instabilities, especially qubit-TLS interactions. With the "optimised noise strategy," qubit coherence is improved by continuously monitoring the TLS environment and adjusting control parameters to avoid strong interaction durations. In the "averaged noise strategy," oscillations are reduced by averaging over different noise environments, making the system more stable and predictable.