D. A. Lidar

We show how to realize, by means of non-abelian quantum holonomies, a set of universal quantum gates acting on decoherence-free subspaces and subsystems. In this manner we bring together the quantum coherence stabilization virtues of decoherence-free subspaces and the fault-tolerance of all-geometric holonomic control. We discuss the implementation of this scheme in the context of quantum information processing using trapped ions and quantum dots.

A recently developed theory for eliminating decoherence and design constraints in quantum computers, ``encoded recoupling and decoupling'', is shown to be fully compatible with a promising proposal for an architecture enabling scalable ion-trap quantum computation [D. Kielpinski et al., Nature 417, 709 (2002)]. Logical qubits are encoded into pairs of ions. Logic gates are implemented using the Sorensen-Molmer (SM) scheme applied to pairs of ions at a time. The encoding offers continuous protection against collective dephasing. Decoupling pulses, that are also implemented using the SM scheme directly to the encoded qubits, are capable of further reducing various other sources of qubit decoherence, such as due to differential dephasing and due to decohered vibrational modes. The feasibility of using the relatively slow SM pulses in a decoupling scheme quenching the latter source of decoherence follows from the observed 1/f spectrum of the vibrational bath.

Motivated by the recently demonstrated ability to attach quantum dots to polymers at well defined locations, we propose a condensed phase analog of the ion trap quantum computer: a scheme for quantum computation using chemically assembled semiconductor nanocrystals attached to a linear support. The linear support is either a molecular string (e.g., DNA) or a nanoscale rod. The phonon modes of the linear support are used as a quantum information bus between the dots. Our scheme offers greater flexibiliy in optimizing material parameters than the ion trap method but has additional complications. We discuss the relevant physical parameters, provide a detailed feasibility study, and suggest materials for which quantum computation may be possible with this approach. We find that Si is a potentially promising quantum dot material, already allowing 5-10 qubits quantum computer to operate with an error threshold of 10^-3.