Erratum for the short (PRL) version of "Physical limits of heat-bath
algorithmic cooling".
-----

There is a numerical error in an example given in the paper. The
following sentence appears in paragraph 4 of page 2:

"For example, perfect implementation of the PPA on a 5-qubit molecule
(4 computation qubits and one refrigerant) would yield a qubit of bias
16 epsilon, implying a 256-fold decrease in scan duration compared to
cooling without algorithmic amplification."

The sentence should read:

"For example, perfect implementation of the PPA on a 5-qubit molecule
(4 computation qubits and one refrigerant) would yield a qubit of bias
8 epsilon, implying a 64-fold decrease in scan duration compared to
cooling without algorithmic amplification."

The long (SICOMP) version of the paper carries the correct sentence.