[elhigh]

One wonders if you could do a sort of super-Atkinson cycle, like a 2:1 ratio instead of the 1.3:1 ratio of current modified Atkinson cycles. If you're injecting only a tiny amount of pure oxygen and a small amount of water to do the expanding, how much air you compress during the compression cycle wouldn't matter so much, would it?

Compression stroke = 40mm, expansion stroke = 89.4 (for instance).

If you're using water injection for expansion mass, that'll absorb a tremendous amount of heat. Could you do away with a radiator (and all its associated parasitic losses)?

Hmm, maybe not all. You'd probably want to reuse that water, or else have to carry around a large/heavy tank full of water. So a condenser is needed, aka a radiator. Dang it.