Component performance testing shows differences between cloud-base quantum computer systems

Characterizing a quantum computer’s performance is capturing increasing attention. While use of quantum computing remains confined to limited applications, companies have made cloud-based access to some platforms publicly available. Questions remain, however, over platform reliability.

Blinov et al. conducted the first test comparing the component performance of four of the top cloud-based quantum computing available to consumers. The researchers examined the state preparation and measurement (SPAM) fidelity, the quantum gate quality and aspects of circuit scaling performance in the IBMQ-16-Melbourne, IBMQ-Vigo, Rigetti Aspen-8 and IonQ platforms.

“One new and innovative aspect of our paper is the running and direct comparison between publicly available quantum computers that anybody can access on the internet,” said author Sergey Blinov. “Each circuit was designed to highlight particular aspects of the machines, such as gate fidelity and qubit connectivity.”

Quantum computing organizes data using the the qubit, which can exist in superposition state and has some attributes reminiscent of analog computing. Moreover, quantum operations are more prone to error, because qubits cannot be easily stabilized through measurement and feedback.

The group found superconducting systems, such as the IBM and Rigetti platforms, had higher error rates for SPAM. All systems experienced increased noise with increasing circuit depth.

“One unanticipated challenge in our work was dealing with the various hidden system transpilers or software preprograms that alter our algorithms to ostensibly make them more efficient,” said author Brandon Wu. “This makes it difficult to extract component performance metrics.”

The group looks to continue testing quantum systems through further testing of components to one day compare systems on high-level applications, such as fully blown algorithms.

Source: “Comparison of cloud-based ion trap and superconducting quantum computer architectures,” by S. Blinov, B. Wu, and C. Monroe, AVS Quantum Science (2021). The article can be accessed at https://doi.org/10.1116/5.0058187 .