Crosstalk Superconducting Qubits Mitigation

Short answer: Crosstalk in superconducting processors comes from unwanted couplings between qubits, above all the always-on ZZ coupling between transmons. Published work proposes three ways to reduce it: dynamical decoupling pulses, tunable couplers that switch the interaction off when idle, and theory tools for picking device parameters. The experiments cited ran on small cloud-accessible processors, and the coupler comparisons are numerical.

Where crosstalk comes from

A 2022 Physical Review Applied paper says currently available superconducting processors with interconnected transmon qubits are noisy and prone to errors from sources such as open quantum system effects and spurious interqubit couplings, called crosstalk. Static ZZ coupling between qubits in transmon architectures is always present and contributes to both coherent and incoherent crosstalk errors, so suppressing it is a key step towards higher fidelity. A 2024 arXiv preprint (2408.15402) adds that always-on ZZ coupling shifts energy levels conditional on qubit states and so affects gate fidelities, and calls unwanted coupling a major challenge for scaling up.

Dynamical decoupling

The 2022 paper proposes dynamical decoupling to suppress crosstalk, and reports experiments on several IBM quantum cloud processors. The authors report improvements in quantum memory and in the performance of single-qubit and two-qubit gate operations. Their open quantum system simulations of multiqubit processors agree well with the experiments. They also analyse the protocol with a simple analytical model and stress the importance of the qubit drive frequency in interpreting the results.

Tunable couplers

A recent Quantum Science and Technology paper uses the density-matrix renormalization group to study lattices of superconducting qubits with tunable couplers designed to decouple qubits when idle. It computes residual ZZ interactions and the inverse participation ratio in computational basis states. For transmon qubits outside the straddling regime, the authors report that tunable C-shunt flux couplers are significantly more efficient at suppressing ZZ interactions than tunable transmons. They also find that a proposed fluxonium architecture with tunable transmon couplers keeps its strong ZZ suppression in larger systems, with a higher inverse participation ratio than transmon lattices, which they take to suggest lower crosstalk when designed for similar performance. This is a numerical result.

Parameter regions

The 2024 preprint introduces a diagrammatic perturbation theory and a state-assignment algorithm to understand ZZ coupling. For two fixed-frequency transmons joined by a flux-tunable transmon coupler, it finds three types of parameter regions with zero or near-zero ZZ coupling, all accessible with current technology. It also finds nearby regions of strong ZZ coupling that could be used for adiabatic controlled-phase gates. The authors say their methods apply to many types of qubits.

Limits of these results

The experiments used small processors, and the coupler and parameter results come from simulation and theory, so they do not show how a large device would behave. The papers also address different crosstalk mechanisms and are not a like-for-like comparison.

FAQ

What is ZZ coupling?

An always-on interaction between neighbouring transmons that shifts energy levels depending on the qubit states.

Does dynamical decoupling help?

The 2022 paper reports improvements in quantum memory and in single-qubit and two-qubit gates on IBM cloud processors.

Which coupler design suppresses ZZ more?

For transmons outside the straddling regime, the numerical study reports tunable C-shunt flux couplers beat tunable transmons.

Are zero-ZZ settings reachable?

The 2024 preprint finds three types of near-zero regions for its two-transmon system, accessible with current technology.

Does this solve crosstalk at scale?

No. These are small-system experiments and calculations.