Module concrete_commons::dispersion
source · [−]Expand description
Noise distribution
When dealing with noise, we tend to use different representation for the same value. In general, the noise is specified by the standard deviation of a gaussian distribution, which is of the form $\sigma = 2^p$, with $p$ a negative integer. Depending on the use case though, we rely on different representations for this quantity:
- $\sigma$ can be encoded in the
StandardDev
type. - $p$ can be encoded in the
LogStandardDev
type. - $\sigma^2$ can be encoded in the
Variance
type.
In any of those cases, the corresponding type implements the DispersionParameter
trait,
which makes if possible to use any of those representations generically when noise must be
defined.
Structs
A distribution parameter that uses the base-2 logarithm of the standard deviation as representation.
A distribution parameter that uses the standard deviation as representation.
A distribution parameter that uses the variance as representation
Traits
A trait for types representing distribution parameters, for a given unsigned integer type.