Cooling of Spins via Quantum Data Compression

To achieve scalability of NMR computers, one needs a large number of highly polarized spins in liquid nuclear-spin systems at finite temperature. In quantum computing terminology, such spin-half states are (almost) pure qubit states. Producing highly polarized spins (almost pure qubit states) out of non-polarized spins (non-pure qubit states) is sometimes called "purification ". From a thermodynamic point of view, purification can be viewed as cooling spins to a very low temperature and can be achieved via data compression algorithms on spins. In this work, we study how classical data compression codes can be used to design cooling algorithms. Specifically, we utilize fixed-rate block compression codes for long molecules and variable-rate lossless compression codes for short molecules and and provide a rudimentary analysis of cooling performance for both short and long molecules.