Armory
will be upgrading to BIP32, but at the moment it uses it's own homegrown version of "Type 2" deterministic wallets. After all, it was the first application to implement them, so I couldn't really have followed any standard
But it's not homomorphic encryption. Homomorphic encryption is pretty neat and enables some pretty cool capabilities, in general, but I haven't thought about whether it could be useful for Bitcoin. I bet, if we upgraded the scripting environment, we could find something interesting, but I'm not sure how useful it would be. It would certainly be a fun discussion to have...
For reference, Armory's wallet chain together like this:
PrivKey[0] = Random(32)
Chaincode = Random(32)
PubKey[0] = Priv2Pub(PrivKey[0])
Then
PrivKey[i+1] = (hash256(PubKey[i]) XOR chaincode) * PrivKey[i]
PubKey[i+1] = (hash256(PubKey[i]) XOR chaincode) * PubKey[i]
Where the multiplication (*) is scalar-multiplication-mod-N in the first line, and it's elliptic-curve-point-mult-by-scalar on the second line. The magic of elliptic curve math is that if N is equal to the number of points on the elliptic curve, then you end up with matching private and public keychains on both sides.
I had actually been meaning to document this precisely somewhere. I guess this was my excuse.