You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
From analyzing the code for generating key-switching, it appears to follow the structure ( P \cdot s^2 - as + e ). However, based on the SEAL issue discussions, it seems to implement the HPS paper's method for key-switching as ( \tilde{q}_i \cdot q_i \cdot s^2 - as + e ) in RNS.
Is SEAL implementing the GHS variant of key-switching rather than the hybrid variant? Additionally, why does SEAL use only one ( P ), while the referenced papers describe multiple ( P_i )'s?
The text was updated successfully, but these errors were encountered:
kumarm17
changed the title
Maths behind switching level in CKKS
key switching in CKKS
Jan 20, 2025
From analyzing the code for generating key-switching, it appears to follow the structure ( P \cdot s^2 - as + e ). However, based on the SEAL issue discussions, it seems to implement the HPS paper's method for key-switching as ( \tilde{q}_i \cdot q_i \cdot s^2 - as + e ) in RNS.
Is SEAL implementing the GHS variant of key-switching rather than the hybrid variant? Additionally, why does SEAL use only one ( P ), while the referenced papers describe multiple ( P_i )'s?
The text was updated successfully, but these errors were encountered: