Theoretical Methods and Experimental Implementation of Heat-Bath Algorithmic Cooling
by
Jose M. Fernandez
Ecole Polytechnique de Montreal
Coauthors: Experiment: Gilles Brassard (U of Montreal), Raymond Laflamme (Waterloo/IQC/PI), Tal Mor (Technion), Yossi Weinstein (Technion). Theory: Seth Lloyd (MIT), Tal Mor (Technion), Vwani Roychowdhury (UCLA).

Algorithmic Cooling is a generic term for techniques leading to the purification of qubit registers whose initial state is not pure. This term encompasses polarisation transfer techniques of NMR (pre-dating both QC and NMR-based QC) and also in-place quantum compression schemes such as those of Schulman and Vazirani. The entropy conservation law imposes, however, severe restrictions on the efficiency of such techniques, often referred to as the Shannon Bound. In addition, the necessity that all transformations applied be unitary (i.e. inner-product preserving), imposes an even tighter bound, referred to as the Sørensen Bound.

A new kind of algorithmic cooling which bypasses these bounds by pumping out excess entropy into the environment (i.e. heat bath) has been proposed in 1999 by Boykin, Mor, Roychowdhury, Vatan, and Vrijen and has been referred to as "non-adiabatic" or "heat-bath" algorithmic cooling. This talk presents an improved non-adiabatic cooling algorithm, more efficient in terms of the number of required qubits, which achieves an exponential increase in polarisation with a linear number of qubits. Furthermore, we report on the first proof-of-concept non-adiabatic cooling experiment ever, performed in April 2002 at the Université de Montréal on a 3-spin molecule.

Date received: March 11, 2004