Harnessing Quantum Systems for Sensing, Simulation, and Optimization

Abstract

Quantum information science offers a remarkable promise: by thinking practically about how quantum systems can be put to work to solve computational and information processing tasks, we gain novel insights into the foundations of quantum theory and computer science. Or, conversely, by (re)considering the fundamental physical building blocks of computers and sensors, we enable new technologies, with major impacts for computational and experimental physics. In this dissertation, we explore these ideas through the lens of three different types of quantum hardware, each with a particular application primarily in mind: (1) networks of quantum sensors for measuring global properties of local field(s); (2) analog quantum computers for solving combinatorial optimization problems; and (3) digital quantum computers for simulating lattice (gauge) theories. For the setting of quantum sensor networks, we derive the fundamental performance limits for the sensing task of measuring global properties of local field(s) in a variety of physical settings (qubit sensors, Mach-Zehnder interferometers, quadrature displacements) and present explicit protocols that achieve these limits. In the process, we reveal the geometric structure of the fundamental bounds and the associated algebraic structure of the corresponding protocols. We also find limits on the resources (e.g. entanglement or number of control operations) required by such protocols. For analog quantum computers, we focus on the possible origins of quantum advantage for solving combinatorial optimization problems with an emphasis on investigating the power of adiabatic quantum computation with so-called stoquastic Hamiltonians. Such Hamiltonians do not exhibit a sign problem when classically simulated via quantum Monte Carlo algorithms, suggesting deep connections between the sign problem, the locality of interactions, and the origins of quantum advantage. We explore these connections in detail. Finally, for digital quantum computers, we consider the optimization of two tasks relevant for simulating lattice (gauge) theories. First, we investigate how to map fermionic systems to qubit systems in a hardware-aware manner that consequently enables an improved parallelization of Trotter-based time evolution algorithms on the qubitized Hamiltonian. Second, we investigate how to take advantage of known symmetries in lattice gauge theories to construct more efficient randomized measurement protocols for extracting purities and entanglement entropies from simulated states. We demonstrate how these protocols can be used to detect a phase transition between a trivial and a topologically ordered phase in $Z_2$ lattice gauge theory. Detecting this transition via these randomized methods would not otherwise be possible without relearning all symmetries.