Widening the Bit: Extending QCLAB to Qudits and Generalizing State Preparation
SIMONE IANNICIELLO
Abstract
The research on qudits, which are the d-level generalization of qubits, as of now is in very early stages, in particular from the software and simulation point of view. This thesis actively extends QCLAB, the MATLAB toolbox for quantum circuits simulation, to qudits of arbitrary dimension d by introducing the necessary circuit representation and simulation as well as gates, building on top of the original functionalities. As a case study, we use the extended tool for qudit state preparation: building a circuit that loads a given vector into a quantum register. We implement the Bullock, O'Leary, and Brennen state preparation construction, which is optimal in gate count, and generalize the KP-Tree data structure and the associated state preparation method of Kerenidis and Prakash to qudits. The generalized method comes in two variants, built on Givens rotations and on Householder reflectors. Under the QRAM model, the Householder variant loads a state with fewer memory queries than the qubit KP-Tree. We validate the correctness of the constructions across several qudit dimensions.