Programmable quantum simulation by dynamic Hamiltonian engineering

Quantum simulation is a promising near term application for mesoscale quantum information processors, with the potential to solve computationally intractable problems at the scale of just a few dozen interacting quantum systems. Recent experiments in a range of technical platforms have demonstrated the basic functionality of quantum simulation applied to quantum magnetism, quantum phase transitions, and relativistic quantum mechanics. In all cases, the underlying hardware platforms restrict the achievable inter-particle interaction, forming a serious constraint on the ability to realize a versatile, programmable quantum simulator. In this work, we address this problem by developing novel sequences of unitary operations that engineer desired effective Hamiltonians in the time-domain. The result is a hybrid programmable analog simulator permitting a broad class of interacting spin-lattice models to be generated starting only with an arbitrary long-range native inter-particle interaction and single-qubit addressing. Specifically, our approach permits the generation of all symmetrically coupled translation-invariant two-body Hamiltonians with homogeneous on-site terms, a class which includes all spin-1/2 XYZ chains, but generalized to include long-range couplings. Building on previous work proving that universal simulation is possible using both entangling gates and single-qubit unitaries, we show that determining the "program" of unitary pulses to implement an arbitrary spin Hamiltonian can be formulated as a linear program that runs in polynomial time and scales efficiently in hardware resources. Our analysis extends from circuit model quantum information to adiabatic quantum evolutions, where our approach allows for the creation of non-native ground state solutions to a computation.