Function opendp::traits::samplers::discretize::sample_discrete_gaussian_Z2k

source · 
pub fn sample_discrete_gaussian_Z2k<T>(
    shift: T,
    scale: T,
    k: i32
) -> Fallible<T>
where
    T: CastInternalRational,
Expand description

Sample from the discrete gaussian distribution on $\mathbb{Z} \cdot 2^k$.

Implemented for floating-point types f32 and f64.

k can be chosen to be very negative, to get an arbitrarily fine approximation to continuous gaussian noise.

§Proof Definition

For any setting of the input arguments, return either Err(e) if there is insufficient system entropy, or Ok(sample), where sample is distributed according to a modified discrete_gaussian(shift, scale).

The modifications to the discrete gaussian are as follows:

  • the shift is rounded to the nearest multiple of $2^k$
  • the sample is rounded to the nearest value of type T.
  • the noise granularity is in increments of $2^k$.