| Meno: | Rudolf |
|---|---|
| Priezvisko: | Vadovič |
| Názov: | Utilization of HOPSO optimizer within Variational Quantum Eigensolvers |
| Vedúci: | doc. RNDr. Martin Plesch, PhD. |
| Rok: | 2026 |
| Kµúčové slová: | Variational Quantum Eigensolver, Harmonic Oscilaltor based Particle Swarm Optimization, quantum error mitigation, zero-noise extrapolation, depolarizing noise |
| Abstrakt: | The Variational Quantum Eigensolver (VQE) is a hybrid quantum-classical algorithm widely investigated for near-term quantum hardware. Its performance depends criti- cally on two components: the classical optimizer used to navigate the parameter space, and the error mitigation techniques used to recover accurate energy estimates from noisy quantum evaluations. This thesis applies the Harmonic Oscillator based Particle Swarm Optimization (HOPSO) algorithm to VQE on the four-qubit hydrogen molecule benchmark, and conducts a systematic empirical study of HOPSO under multiple noise models and error mitigation strategies. Three depolarizing noise models spanning two orders of magnitude in error probability are studied, alongside shot noise at multiple measurement counts. Two error mitigation methods are evaluated: Zero-Noise Ex- trapolation with three extrapolation schemes and two noise scaling approaches, and Probabilistic Error Cancellation applied post-hoc. HOPSO is compared against three established optimizers (PSO, DE, COBYLA) under the studied noise configurations, and a representative IBM fake backend is used to validate the findings under realistic device noise. The principal findings are that HOPSO recovers high-quality parameter vectors under isolated depolarizing noise without any mitigation, that applying ZNE during the optimization loop on this benchmark degrades the actual quality of found parameters across all studied noise configurations, and that the relative ranking of opti- mizers under combined gate and shot noise places Differential Evolution slightly above HOPSO, with both above PSO. The conditions under which ZNE during optimization is counterproductive on this benchmark are characterized empirically and contrasted with the post-hoc application regime in which mitigation is typically evaluated in the literature. |
Súbory bakalárskej práce:
| Utilization of HOPSO optimizer within Variational Quantum Eigensolvers.pdf |
Súbory prezentácie na obhajobe: