Skip to content
Quantum.world

The Signal

Researchers test Holstein-Primakoff boson simulation on noisy qubits

A Physical Review A paper reports simulations of bosonic dynamics on a cloud-based superconducting quantum processor. The study tests error mitigation and compares Trotter circuits with an optimized circuit for Jaynes-Cummings dynamics.

Researchers test Holstein-Primakoff boson simulation on noisy qubits

A bosonic mode encoded in qubits

A study published in Physical Review A on November 16 reports tests of a bosonic simulation method on a cloud-based superconducting quantum processor. Kelvin Yip, Alessandro Monteros, Sahel Ashhab and Lin Tian mapped bosonic modes onto an ensemble of superconducting qubits using the Holstein-Primakoff transformation.

Source journals.aps.org

The researchers simulated a driven harmonic oscillator, squeezed states and the Jaynes-Cummings model, which describes interactions between a two-level system and a bosonic mode. The paper reports two-excitation Jaynes-Cummings dynamics and squeezed-state generation involving qubit-qubit interactions, extending the demonstrations beyond the single-excitation regime.

In the mapping, collective states of the qubits represent the bosonic mode’s excitation states. The number of qubits sets a finite limit on that representation, while the simulations also approximate time evolution using Trotter steps. The study examined these algorithmic limitations alongside hardware errors, including entangling-gate infidelity, decoherence and readout errors.

Balancing mapping and hardware errors

For the driven oscillator, the team compared simulations using different numbers of qubits with the bosonic results. The paper reports that differences from the oscillator simulation became negligible when the drive was small, such as F = 0.3, or when the ensemble had 14 qubits. Increasing the ensemble size reduced mapping error, but the hardware measurements also accumulated more error as the number of qubits grew.

Because the oscillator circuit used single-qubit operations, readout was the main hardware error source in that test. Applying M3 measurement mitigation reduced the averaged difference between noisy hardware results and noiseless qubit results from 9.5 × 10−3 to 6.3 × 10−3, according to the paper.

The squeezing test used a five-qubit ensemble and required entangling gates. Each Trotter step included 35 CZ gates after the circuit was translated into the processor’s gate set. The study reports that error-mitigated measurements agreed better with noiseless qubit simulations, while the finite ensemble size caused larger discrepancies from the oscillator result at stronger squeezing.

Shallower circuits reduced measured error

For the Jaynes-Cummings simulations, the authors used the Suzuki-Trotter decomposition and a separately optimized, synthesized-unitary circuit. With two qubits representing the cavity mode, the total probability difference reached a minimum around four Trotter steps: more steps reduced approximation error but also increased hardware error.

Combining M3 measurement mitigation with zero-noise extrapolation reduced the averaged probability difference from 0.071 to 0.029 for the two-qubit encoding. In a three-qubit encoding, the reported difference fell from 0.131 to 0.042. The authors also report that entangling-gate errors were a dominant source of hardware error in these simulations.

For the synthesized-unitary comparison, a six-CZ circuit produced an average probability difference of 0.019. The corresponding reported figures were 0.30 for one Trotter step, which also used six CZ gates, and 0.06 for four Trotter steps, which used 18. The paper says the optimized approach can outperform Trotter decomposition for small systems, at the cost of additional classical computation for circuit optimization.

The authors describe the work as a demonstration of the feasibility and limitations of Holstein-Primakoff-based digital simulation on current processors. They say the framework could be extended to more complex spin-boson and multimode cavity models.

Sources

  1. American Physical Society
  2. arXiv

Related stories

Explore

More articles