@dartin
There is a simple thermodynamic argument which says that to achieve COP = infinity all you need is passive temperature-dependent cooling!
Surpised that Rossi has never experimented with this... it give you a higher safety margin against runaway than you can get from controlled additive heating.