International Symposium on Quantum Theory and Symmetries (QTS)
ALGORITHMIC COOLING AND SCALABLE QUANTUM COMPUTERS: WAYS TO IMPROVE THE SPACE-TIME REQUIREMENTS OF THE ALGORITHM
Tal Mor, Yossi Weinstein
Abstract
The "algorithmic cooling" of Boykin, Mor, Roychowdhury, Vatan and Vrijen (BMRVV) is a powerful cooling method for obtaining a large number of highly polarized spins in liquid nuclear-spin systems at finite temperature. The cooling is done by compressing the antropy of a string into a portion of it and then swaping that portion with rapidly thermal-relaxing bits, which transfers the antropy quickly to the environment. Algorithmic cooling could enable cooling spins to a very low temperature without cooling the environment and could resolve the scaling problem of NMR quantum computing devices. The BMRVV cooling algorithm is described in a simplified model in which all spins have identical bias (the minimum over all the different biases of the spins), and using blocks of m bits, hence the resulting cooling technique is non-optimal. We present here initial analysis of two improvements of the algorithm: working with variable-size blocks in the cooling process, and using better the fact that initial biases are not equal. As result, much less cooling steps are required in order to yield more spins which are highly polarized (with the desired bias).