Performance benchmarks of quantum computing applications
Abstract
The scope of the project is to develop a technical report that provides guidelines on how to benchmark quantum applications, i.e., to characterize the performance of quantum algorithms for classes of quantum applications, implemented on different quantum hardware. With specific well-defined benchmarks, quantum computing end-users will learn the performance of a specific quantum computing solution for specific use cases. At this application level, there is a need to benchmark the performance of methods/algorithms for classes of applications, implemented on different hardware. Comparing classical computing (especially HPC and its well-established libraries ) with the many different paradigms of quantum computing (digital, analog, etc.) is clearly a moving target. Hence, it is expected that this technical report will be regularly updated. The report should consider algorithms that can be implemented on Noisy Intermediate-Sc ale Quantum (NISQ) hardware (e.g ., variational hybrid methods), as well as algorithms that require Fault-Tolerant Quantum Computing (FTQC) capacities. While in principle the benchmarks for these two stages of quantum computing should not differ, their focus might be very different. In FTQC, the benchmark should focus largely on the application, rather than on how a computing device to solve a certain problem. For NISQ hardware, the application should again lie central. However, there can be more focus on benchmarking the quantum part specifically , as for NISQ devices the development of the quantum part is specifically of interest. Because of this, properties such as QPU time and fidelities can be relevant aspects of a NISQ application-oriented benchmark. Other aspects of application-oriented benchmarking to discussed in the report are: which properties are relevant to measure in such benchmarks (accuracy, speed, size, energy, implementation cost); the sort of applications to be considered by such benchmark; the scalability of quantum computing and when it will become advantageous over classical computing; implementation of a given quantum benchmark (specification, ease, degrees of freedom). The resulting technical report should give best practices in application-level quantum benchmarking and thereby form a reference frame for valid comparison of different (hybrid) quantum approaches to solving specific problems. The report will contribute to assess the potential of specific hardware for specific use cases and bring clarity on which type of quantum advantage quantum computers may provide (accuracy, speed, energy, implementation cost).
Begin
2024-03-06
WI
JT022005
Planned document number
DIN CEN/CLC/TR XXX-JT022005
Project number
04301104
Responsible national committee
NA 043-02-05 AA - Quantum Technologies
