Title: Noisy intermediate-scale quantum computation with a complete graph of superconducting qubits: Beyond the single-excitation subspace

Abstract: There is currently a tremendous interest in developing practical applications of NISQ processors without the overhead required by full error correction. Quantum information processing is especially challenging within the gate model, as algorithms quickly lose fidelity as the problem size and circuit depth grow. This has lead to a number of non-gate-model approaches such as analog quantum simulation and quantum annealing. These approaches come with specific hardware requirements that are typically different than that of a universal gate-based quantum computer. We have previously proposed a non-gate-model approach called the single-excitation subspace (SES) method, which requires a complete graph of superconducting qubits. Like any approach lacking error correction, the SES method is not scalable, but it often leads to algorithms with constant depth, allowing it to outperform the gate model in a wide variety of applications. A challenge of the SES method is that it requires a physical qubit for every basis state in the computer's Hilbert space. This imposes large resource costs for algorithms using registers of ancillary qubits, as each ancilla would double the required graph size. Here we show how to circumvent this doubling by leaving the SES and reintroducing a tensor product structure in the computational subspace. Specifically, we implement the tensor product of an SES register holding ``data" with one or more ancilla qubits. This enables a hybrid form of quantum computation where fast SES operations are performed on the data, traditional logic gates and measurements are performed on the ancillas, and controlled-unitaries act between. As an application we give an SES implementation of the quantum linear system solver of Harrow, Hassidim, and Lloyd.