LUT box size

FHE Libraries Zama TFHE-rs
1

In your function:

pub fn generate_programmable_bootstrap_glwe_lut<F, Scalar: UnsignedTorus + CastFrom>(

polynomial_size: PolynomialSize,

glwe_size: GlweSize,

message_modulus: usize,

ciphertext_modulus: CiphertextModulus,

delta: Scalar,

f: F,

) → GlweCiphertextOwned

you have a comment which outlines the box size for the LUT generation is N / ( p/2) however in the code it is implemented as box_size = N / p so I am wondering is the comment an error?

here:

// N/(p/2) = size of each block, to correct noise from the input we introduce the

// notion of box, which manages redundancy to yield a denoised value

// for several noisy values around a true input value.

let box_size = polynomial_size.0 / message_modulus;
2

hello @un5had0w

it looks like a typo, probable cause was that p was seen as the precision including the padding bit before (which means you would have useful_p = p / 2)

I’ll fix the comment, thanks

1 Like
3

Ahh, ok makes sense thanks for clearing that up.