Experimental Implementation of Randomized Compilation

Coherent errors in gate operations account for an increasing percentage of computational error budgets as the gate depths of quantum algorithms increase. Randomized compiling is an efficient method for converting these coherent errors into stochastic Pauli errors, whose aggregate imperfection scales more favorably with gate count. This conversion, known as Randomized Compiling, is achieved via substitution of portions of the algorithm of interest with easily pre-computed equivalent operations. The procedure allows for an increase in the total fidelity of an algorithm with minimal classical overhead. We demonstrate randomized compiling with an implementation of the HHL algorithm on 4 transmon qubits in the presence of artificial coherent noise, and compare performance to that of the non-randomized version of the algorithm.