Generalizations Of The Quantum Search Algorithm

Abstract

Quantum computation has attracted a great deal of attention from the scientific community in recent years. By using the quantum mechanical phenomena of superposition and entanglement, a quantum computer can solve certain problems much faster than classical computers. Several quantum algorithms have been developed to demonstrate this quantum speedup. Two important examples are Shor’s algorithm for the factorization problem, and Grover’s algorithm for the search problem. Significant efforts are on to build a large scale quantum computer for implementing these quantum algorithms. This thesis deals with Grover’s search algorithm, and presents its several generalizations that perform better in specific contexts. While writing the thesis, we have assumed the familiarity of readers with the basics of quantum mechanics and computer science. For a general introduction to the subject of quantum computation, see [1]. In Chapter 1, we formally define the search problem as well as present Grover’s search algorithm [2]. This algorithm, or more generally the quantum amplitude amplification algorithm [3, 4], drives a quantum system from a prepared initial state (s) to a desired target state (t). It uses O(α-1 = | (t−|s)| -1) iterations of the operator g = IsIt on |s), where { IsIt} are selective phase inversions selective phase inversions of the corresponding states. That is a quadratic speedup over the simple scheme of O(α−2) preparations of |s) and subsequent projective measurements. Several generalizations of Grover’s algorithm exist. In Chapter 2, we study further generalizations of Grover’s algorithm. We analyse the iteration of the search operator S = DsI t on |s) where Ds is a more general transformation than Is, and I t is a selective phase rotation of |t) by angle . We find sufficient conditions for S to produce a successful quantum search algorithm. In Chapter 3, we demonstrate that our general framework encapsulates several previous generalizations of Grover’s algorithm. For example, the phase-matching condition for the search operator requires the angles and and to be almost equal for a successful quantum search. In Kato’s algorithm, the search operator is where Ks consists of only single-qubit gates, which are easier to implement physically than multi-qubit gates. The spatial search algorithms consider the search operator where is a spatially local operator and provides implementation advantages over Is. The analysis of Chapter 2 provides a simpler understanding of all these special cases. In Chapter 4, we present schemes to improve our general quantum search algorithm, by controlling the operators through an ancilla qubit. For the case of two dimensional spatial search problem, these schemes yield an algorithm with time complexity . Earlier algorithms solved this problem in time steps, and it was an open question to design a faster algorithm. The schemes can also be used to find, for a given unitary operator, an eigenstate corresponding to a specified eigenvalue. In Chapter 5, we extend the analysis of Chapter 2 to general adiabatic quantum search. It starts with the ground state |s) of an initial Hamiltonian Hs and evolves adiabatically to the target state |t) that is the ground state of the final Hamiltonian The evolution uses a time dependent Hamiltonian HT that varies linearly with time . We show that the minimum excitation gap of HT is proportional to α. Also, the ground state of HT changes significantly only within a very narrow interval of width around the transition point, where the excitation gap has its minimum. This feature can be used to reach the target state (t) using adiabatic evolution for time In Chapter 6, we present a robust quantum search algorithm that iterates the operator on |s) to successfully reach |t), whereas Grover’s algorithm fails if as per the phase-matching condition. The robust algorithm also works when is generalized to multiple target states. Moreover, the algorithm provides a new search Hamiltonian that is robust against certain systematic perturbations. In Chapter 7, we look beyond the widely studied scenario of iterative quantum search algorithms, and present a recursive quantum search algorithm that succeeds with transformations {Vs,Vt} sufficiently close to {Is,It.} Grover’s algorithm generally fails if while the recursive algorithm is nearly optimal as long as , improving the error tolerance of the transformations. The algorithms of Chapters 6-7 have applications in quantum error-correction, when systematic errors affect the transformations The algorithms are robust as long as the errors are small, reproducible and reversible. This type of errors arise often from imperfections in apparatus setup, and so the algorithms increase the flexibility in physical implementation of quantum search. In Chapter 8, we present a fixed-point quantum search algorithm. Its state evolution monotonically converges towards |t), unlike Grover’s algorithm where the evolution passes through |t) under iterations of the operator . In q steps, our algorithm monotonically reduces the failure probability, i.e. the probability of not getting |t), from . That is asymptotically optimal for monotonic convergence. Though the fixed-point algorithm is of not much use for , it is useful when and each oracle query is highly expensive. In Chapter 9, we conclude the thesis and present an overall outlook.