First Experimental Validation of the "Lagrange" On-Premise IQM Spark Superconducting Quantum Processor

This thesis presents an experimental benchmarking study of the IQM Spark, an on-premise five-qubit superconducting quantum device recently installed at Politecnico di Torino. The main objective is to provide a first validation of the quantum computer through practical experiments, in order to understand its capabilities and limitations. In order to better evaluate the results obtained from the quantum machine, the defined set of experiments has also been performed using classical emulation, both noiseless and noisy. In particular, this stage made it possible to validate the workflow and to prepare the circuits in advance. Then, all experiments have been repeated on the actual IQM Spark quantum computer, and the results have been compared with those obtained on a fake backend. The comparison has provided a measure of consistency between the emulator and the physical quantum processor. With the aim of providing an overview as wide as possible, both fault-tolerant and non fault-tolerant algorithms have been identified and implemented. More in detail, the study focuses on three applications. The first addresses a fault-tolerant version of the Deutsch-Jozsa algorithm, implemented using the error detection code [[4,2,2]]. This case shows how a simple quantum algorithm can be protected against noise by encoding logical qubits into a small code, and it highlights the trade-off between error detection and hardware resources. The second application investigates the Single-Impurity Anderson Model (SIAM), studied with a Variational Quantum Eigensolver (VQE) and, again, leveraging the [[4,2,2]] code to provide partial detection. For both the Deutsch-Jozsa algorithm and the SIAM problem, the experiments quantify the reduction of errors achieved by encoded implementations compared to their unencoded counterparts. The third application explores Quantum Reservoir Computing (QRC), a framework that leverages quantum systems for tasks such as time-series processing, classification and control. In this work, QRC is specifically applied to temporal signal prediction, with a model trained and tested on the NARMA10 sequence, a widely used nonlinear benchmark in time-series prediction, consisting of a tenth-order Nonlinear Autoregressive Moving Average system, which probes the ability of the model to capture complex temporal dependencies. Although the algorithms presented in this thesis have already been investigated on other quantum platforms based on different technologies, such as trapped ions and neutral-atom systems, they have never before been implemented on the IQM Spark quantum processor. The study required careful adaptation of the implementations to the hardware architecture and constraints, and it delivers a benchmark that highlights both the capabilities and the limitations of the device at the algorithmic level. In conclusion, this thesis shows that the IQM Spark quantum computer can already serve as a platform for preliminary implementations of error detection, variational algorithms and quantum machine learning models. Although the results are limited in scale, they establish the first experimental reference point for future research on the IQM Spark processor.