Hello there,
I am trying to study the BGV scheme as described in your implementation, in particular I tried to implement a toy modulus switching. Given a ciphertext with an error bound B^2, the idea is to scale it down by a factor B.
Nevertheless, while playing around, I was thinking about this: is there something preventing me to perform a mod switch on a fresh ciphertext? Theoretically is should scale the error B by a factor 1/B, almost removing the original error.
What am I missing? Thank you!!