Quantum Kolmogorov Complexity and the Quantum Turing Machine

The purpose of this thesis is to give a formal definition of quantum Kolmogorov complexity and rigorous mathematical proofs of its basic properties. Classical Kolmogorov complexity is a well-known and useful measure of randomness for binary strings. In recent years, several different quantum generalizations of Kolmogorov complexity have been proposed. The most natural generalization is due to Berthiaume et al. [5], defining the complexity of a quantum bit (qubit) string as the length of the shortest quantum input for a universal quantum computer that outputs the desired string. Except for slight modifications, it is this definition of quantum Kolmogorov complexity that we study in this thesis. We start by analyzing certain aspects of the underlying quantum Turing machine (QTM) model in a more detailed formal rigour than was done previously. Afterwards, we apply these results to quantum Kolmogorov complexity. Our first result, based on work by Bernstein and Vazirani [4], is a proof of the existence of a universal QTM which simulates every other QTM for an arbitrary number of time steps and than halts with probability one. In addition, we show that every input that makes a QTM almost halt can be modified to make the universal QTM halt entirely, by adding at most a constant number of qubits. It follows that quantum Kolmogorov complexity has the invariance property, i.e. it depends on the choice of the universal QTM only up to an additive constant. Moreover, the quantum complexity of classical strings agrees with classical complexity, again up to an additive constant. The proofs are based on several analytic estimates. Furthermore, we prove several incompressibility theorems for quantum Kolmogorov complexity. Finally, we show that for ergodic quantum information sources, complexity rate and entropy rate coincide with probability one. The thesis is finished with an outlook on a possible application of quantum Kolmogorov complexity in statistical mechanics.