In this Thesis, several theoretical and experimental issues on ensemble quantum computers are studied, with the goal of detailed understandings to build up a practical ensemble quantum computer.
One of the most fundamental issues on quantum computation is how to realize a computation of interest by a given quantum system. A Hamiltonian-based method is suggested to partially resolve this problem. It shows a way to understand computations in terms of evolution of Hamiltonians, and vice versa. Although it is not a general method, it can be applied to almost all meaningful operations such as controlled-gates. Also, a scheme to store and retrieve continuous quantum programs via quantum states is presented, which can be used to construct a probabilistic programmable quantum computer. It is pointed out that one of the important differences between classical and quantum programs is the continuity of variables to represent information about the programs. Handling the continuous information with finite resource is the key point of building a programmable quantum computer. This subject will be very important, because it is directly related to the quantum version of the von Neumann machine.
The standard models of quantum computers employ single quantum systems, while the successful experiments have used ensembles of single quantum systems. The differences between single-system and ensemble models are discussed in terms of the correlations among qubits. It is shown that meaningful results of computations can be obtained by an ensemble model so that an ensemble quantum computer is feasible. This is a really fruitful result because ensembles are easy to handle with current technologies and robust against noises. Thus it may move up the appearance of real quantum computers.
Quantum algorithms are implemented by a nuclear magnetic resonance quantum computer. Experimental features are studied, which will be useful because several NMR techniques can be employed in not only NMR bu...