Although they have generated significant interest, near-term quantum algorithms face a major roadblock: the measurement problem. Briefly, standard methods for the estimation subroutine used in many of these algorithms to measure quantities are too slow for use on industrial problems. Example near-term quantum algorithms are the variational quantum eigensolver (VQE), the quantum approximate optimization algorithm (QAOA), quantum circuit Born machines (QCBM), and many others. Industrial applications of these include simulating molecules, optimizing vehicle routes, and generating images. Solving the measurement problem is therefore a critical goal for the field of applied quantum computing.
This work explores a promising solution to the measurement problem called robust amplitude estimation (RAE). We introduced this technique in previous work. In this new paper we investigate: what value does RAE provide when used on today’s devices? We compare the performance of RAE to the standard method of estimation. Using IBM’s noisy quantum devices, we find that RAE improves both accuracy and precision of Pauli expectation value estimates by a factor of three while using the same total runtime. Moreover, unlike the standard estimation method, with RAE, better quality devices will yield higher-precision estimates. We find that estimating expectation values with RAE
- can be used to speed up quantum computations on today’s devices
- gives a more promising route to achieving quantum advantage
The counter-intuitive message of this work is that we should use deeper and thus noisier quantum circuits in order to reduce runtime of estimating expectation values. With this finding, we hope quantum computing practitioners will “jump into the deep end” and start achieving high-precision estimates with shorter runtime.