Realistic Quantum Noise Simulation Validates Algorithms for Future Systems

A tunable noise model, calibrated using data from IBMQ systems and implemented on the Qaptiva simulator, accurately replicates hardware behaviour for quantum algorithms like GHZ state preparation and QAOA. A novel metric assesses noise model quality at scale, enabling extrapolation to future, partially fault-tolerant systems and informing algorithm development.

The fidelity of quantum computations is intrinsically linked to the accurate modelling of noise – imperfections inherent in quantum systems that degrade performance. As quantum processors scale, simulating these noisy environments becomes crucial for algorithm development and validation, often exceeding the capacity of available quantum hardware. Researchers from Eviden, alongside colleagues at the Technical University of Applied Sciences Regensburg and Siemens AG, address this challenge in a new study titled ‘Make Some Noise! Measuring Noise Model Quality in Real-World Quantum Software’. Stefan Raimund Maschek, Jürgen Schwitalla, Maja Franz, and Wolfgang Mauerer detail a scalable noise model, implemented on the Qaptiva simulator, calibrated using empirical data from IBM Quantum systems, and validated through simulations of Greenberger-Horne-Zeilinger state preparation and a Quantum Approximate Optimisation Algorithm applied to an industrial problem. Their work introduces a novel method for assessing noise model quality, enabling extrapolation to future, partially fault-tolerant systems and offering insights into the interplay between hardware modelling and algorithm design.

Enhanced Quantum Simulation Through Empirically Calibrated Noise Modelling

Accurate simulation is crucial for the development and evaluation of quantum algorithms, yet faithfully replicating the behaviour of current quantum hardware presents a significant challenge. Quantum systems are inherently susceptible to noise, which introduces errors and limits computational fidelity. Researchers have developed a new, scalable noise model implemented within the Qaptiva simulator, designed to address this limitation.

The model incorporates parameters derived from empirical noise data gathered from IBM Quantum systems. This calibration process ensures the simulation reflects the specific characteristics of real hardware, moving beyond simplified, theoretical noise assumptions. The noise considered includes both coherent and incoherent errors; coherent errors involve predictable phase and amplitude distortions, while incoherent errors are random and unpredictable, such as energy loss or unwanted interactions with the environment.

Validation of the model’s accuracy involved simulating the preparation of a Greenberger-Horne-Zeilinger (GHZ) state – a highly entangled quantum state sensitive to noise – and assessing the fidelity of the simulated output. Successful replication of the expected behaviour demonstrates the model’s ability to capture the effects of noise on entangled systems.

Furthermore, the model was applied to a practical problem using the Quantum Approximate Optimisation Algorithm (QAOA). QAOA is a heuristic algorithm designed to find approximate solutions to combinatorial optimisation problems. The researchers applied QAOA to an industrial use case, demonstrating the model’s utility in evaluating algorithm performance under realistic noise conditions.

The development of this empirically calibrated noise model within Qaptiva provides a more robust and reliable simulation environment. This advancement facilitates the development and testing of error mitigation strategies and allows researchers to more effectively evaluate the potential of quantum algorithms before deployment on physical hardware.

Note:

Qaptiva is a quantum computing simulation platform.
GHZ states are a specific type of quantum state exhibiting maximal entanglement between multiple qubits.
QAOA (Quantum Approximate Optimisation Algorithm) is a hybrid quantum-classical algorithm used for solving combinatorial optimisation problems.