After reading your paper carefully, as a symmetric encryption algorithm, the security is indeed higher as to resist linear attacks and differential attacks.
But if you want to strictly prove p!=np, you should rely on some kind of random number generation algorithm. or maybe i haven't fully understood it.