Adiabatic quantum computation with spin ensembles

In the standard approach to adiabatic quantum computing (AQC), quantum information stored on qubits are adia- batically evolved to find the lowest energy state of a problem Hamiltonian. Here we investigate a variation of AQC where spin ensembles are used in place of qubits. The use of ensembles duplicates the quantum information, and allows errors to be suppressed during the adiabatic evolution. We show that there are two distinct types of problem Hamiltonians under this mapping, where the first excited state is a single particle perturbation on the ground state (Type I); or a macroscopically distinct fully polarized state (Type II). For Type I problems, we find that the minimum gap for large ensembles is well predicted by mean-field theory and the AQC performance improves with ensemble size, realizing error-suppression. For Type II problems, the performance of the AQC is mixed, and the gap can increase or decrease with ensemble size depending on the problem instance. The incidences of the Type II problems are greatly suppressed with ensemble size, and consequently for randomly chosen problem instances the success probability of the scheme improves with the ensemble size. Our approach shows that it is possible to perform AQC without the necessity of controlling individual qubits, which allows for an alternative route towards implementing AQC.