Quantum Computers’ Resilience to Radiation Errors Is Now Accurately Modelled
Summarize this article with:
Researchers are increasingly focused on the detrimental effects of radiation-induced errors on the performance of quantum computers. Paul G. Baity (Brookhaven National Laboratory), Anuj K. Nayak (University of Illinois Urbana-Champaign), and Lav R. Varshney (AI Innovation Institute, Stony Brook University) et al. present a new computational model designed to simulate how radiation impacts quantum error correction (QEC). This work is significant because current QEC codes typically assume independent errors, whereas radiation generates widespread, correlated errors that degrade performance. Their model builds upon quasiparticle density calculations, maps radiation to quantum error channels, and simulates a surface code, ultimately providing a metric to quantify QEC resilience and test mitigation strategies for improved chip and code designs. Radiation impacts pose a significant challenge to these devices, creating widespread correlated errors that conventional QEC codes, designed for independent errors, struggle to address. This work introduces a holistic approach, simulating the effects of radiation on QEC performance by modelling quasiparticle density evolution and mapping resultant error rates onto a quantum error channel. The research focuses on a simple surface code simulation to quantify the impact of radiation and test potential mitigation strategies. The model builds upon recent advances in quasiparticle density modelling, utilising simulation packages Geant4 and G4CMP to track high-energy particle interactions within planar superconducting devices. This allows researchers to derive the time evolution of normalised quasiparticle density following a radiation event, such as a muon or gamma ray strike. By linking this physical model to qubit error rates, the study simulates the performance of a [[9,1,3]] surface code implemented on a 17-qubit transmon-based quantum processing unit. A key output of this work is a performance metric to quantify how effectively a QEC code can withstand radiation impacts. Furthermore, the research systematically explores the influence of various chip design parameters on QEC performance, enabling modular testing of error mitigation strategies. Simulations are based on transmon parameters including a qubit transition frequency ω01 of 5.000GHz, capacitive energy EC/h of 0.400GHz, and Josephson energy EJ/h of 9.234GHz. Baseline qubit relaxation times, T b 1 and T b 2, are set to 100μs and 200μs respectively, with gate times of 40ns for CX gates, 30ns for Hadamard gates, and 140ns for readout. The model demonstrates the efficacy of phonon downconversion, a technique for mitigating quasiparticle errors, and provides a framework for evaluating future error mitigation strategies and code designs. Modelling quasiparticle poisoning via Geant4 Monte Carlo simulations and time-dependent rate equations A combination of computational methods models the failure rates for quantum error correction. The physical model for radiation impacts and quasiparticle (QP) poisoning of superconducting devices originates from Monte-Carlo simulations performed using Geant4 and G4CMP packages. These packages simulate the generation of electron/hole pairs, phonons, and QPs in planar superconducting devices following high-energy particle strikes. The output of these simulations is the time-dependent generation term for superconducting QPs, gqp(t), which is then used to derive the normalized QP density xqp via a time-dependent ordinary differential equation accounting for QP recombination and decay. Transmon parameters were established for the simulations, assigning values to each of the 17 qubits in the circuit. These parameters included a transition frequency ω01/2π of 5.000GHz, a capacitive energy EC/h of 0.400GHz, and a Josephson energy EJ/h of 9.234GHz, resulting in an EJ/EC ratio of 23. Baseline relaxation times, T b 1 and T b 2, were set to 100μs and 200μs respectively. An increase in xqp linearly decreases the relaxation times of the transmons, calculated as 1/T1 = 1/T b 1 + xqp π r 2ω01∆Al ħ. The study also models dephasing rates Γφ, dependent on the transmon capacitive energy EC, using the equation Γφ(xqp) = Γb φ + ECx2 qp 2π2 e 1 2 W0 4π/x2 qp, where Γb φ is the baseline dephasing rate and W0 is the Lambert function. Simulations of a [[9,1,3]] surface code were performed using Qiskit and its subpackages, Qiskit-Aer and Qiskit-QEC, or Stim. The stabilizer circuit, consisting of Hadamard and controlled-not gates, was repeated in consecutive rounds with a cycle time τc of 1μs, and errors were decoded after each cycle. Noise was applied using a quasi-equilibrium approximation, evaluating T1 and T2 at each time step and constructing a per-cycle noise model applied to each gate within the cycle. Simulated transmon characteristics and quasiparticle dynamics impacting qubit coherence Transmon parameters used in the simulation included a qubit transition frequency, ω01, of 5.000GHz, a capacitive energy, EC, of 0.400GHz, and a Josephson energy, EJ, of 9.234GHz, resulting in an EJ/EC ratio of 23. Baseline relaxation times, T b 1 and T b 2, were set to 100μs and 200μs, respectively, with gate times of 40ns for the CX gate, 30ns for single-qubit Hadamard gates, and 140ns for readout operations. These parameters were uniformly assigned to each of the 17 transmons within the quantum processing unit. The research demonstrates a model for qubit error rates following radiation impacts, utilising Geant4 and G4CMP simulations to determine the generation of quasiparticles. The time evolution of normalised quasiparticle density, xqp, was derived using an ordinary differential equation for QP recombination and decay, with example curves generated for a 17-qubit QPU following a muon strike. A reduction in relaxation times, T1, was observed as xqp increased, calculated as 1/T1 = 1/T b 1 + xqp π r 2ω01∆Al ħ, where ∆Al represents the superconducting gap of the Al-based Josephson junction. Furthermore, the dephasing rate, Γφ, was found to be dependent on xqp, described by the equation Γφ(xqp) = Γb φ + ECx2 qp 2π2 e 1 2 W0 4π/x2 qp, where Γb φ is the baseline dephasing rate and W0 is the Lambert function. Analysis of quantum error correction performance focused on bitflip errors for Z-basis measurements and data correction, as the calculated dephasing rate was small compared to the relaxation rate. Simulations were performed on a [[9,1,3]] rotated surface code consisting of nine data qubits and eight ancilla qubits, with stabilizer circuits repeated in cycles of 1μs. The cycle-dependent noise model was generated using a generalized amplitude damping quantum error channel, incorporating thermal relaxation and dephasing, and applied to each gate within the cycle. The Choi representation for this channel accounts for thermal populations and gate times to accurately model the error environment. Radiation resilience quantified via computational modelling of surface code performance Scientists have developed a computational model to simulate the effects of radiation on the performance of quantum error correction. This model builds upon recent advances in understanding quasiparticle density and maps radiation-induced error rates onto a quantum error channel, utilising a simple surface code for simulation. The approach allows for the modular testing of error mitigation strategies and various chip and code designs, offering a holistic method for performance evaluation. The simulations reveal a surprising recovery dip in performance immediately following radiation impact, suggesting that limited degrees of correlated errors may be correctable using standard surface code setups and decoders. A new metric, termed ζc, has been introduced to quantify the resilience of quantum error correction codes to radiation-induced errors and demonstrated its usefulness in assessing mitigation strategies such as phonon downconversion using flipside Cu. Future research could benefit from machine learning-driven optimisation and active learning methods applied to reducing ζc, potentially enabling the discovery of quantum processor architectures with improved radiation resilience. The authors acknowledge that the current model is demonstrated using a relatively small 17-qubit surface code, but emphasise its generalisability to larger architectures and different code structures. They also confirm the adequacy of a Pauli-twirled Gaussian Adiabatic Dephasing (GAD) quantum channel for modelling error rates during radiation impact, despite differences in error symmetry compared to other channels. 👉 More information🗞 Characterizing Quantum Error Correction Performance of Radiation-induced Errors🧠 ArXiv: https://arxiv.org/abs/2602.06202 Tags: