Suppose we have a two-party protocol in the semi-honest model, which is based on an additive homomorphic cryptosystem like Paillier.(P2 will generate the public and private key and publish the public key to P1)
P2 will compute $R_1, ..., R_n$ using some operations and then send the vector $V=(R_1, ..., R_n)$ to P1. Since P1 has access to the private key, it will decrypt $R_1$ up to $R_n$.
In case where P1 is corrupted, to simulate the received message $V=(R_1, ..., R_n)$, is it suffice to just consider $n$ different dummy values and then encrypt them using the public key in order to make indistinguishable values? Or, when we want to prove the security in this case, we should necessarily show the operations P2 performed on its side? (I mean whether it is important to demonstrate how the simulator has reached to these dummy values($R'_1, ..., R'_n$)?