Struct opendp::measures::SmoothedMaxDivergence

pub struct SmoothedMaxDivergence<Q>(PhantomData<Q>);

$\epsilon(\delta)$-approximate differential privacy.

The greatest divergence between any randomly selected subset of the support, with an additive tolerance for error.

The distance $d$ is of type SMDCurve, so it can be invoked with a $\delta$ to retrieve the tightest corresponding $\epsilon$.

Proof Definition

d-closeness

For any two vectors $u, v \in \texttt{D}$ and any choice of $\epsilon, \delta$ such that $\epsilon \ge d(\delta)$, we say that $M(u), M(v)$ are $d$-close under the smoothed max divergence measure (abbreviated as $D_{S\infty}$) whenever

D_{S\infty}(M(u) \| M(v)) = \max_{S \subseteq \textrm{Supp}(Y)} \Big[\ln \dfrac{\Pr[M(u) \in S] + \delta}{\Pr[M(v) \in S]} \Big] \leq \epsilon.

Tuple Fields

0: PhantomData<Q>

Trait Implementations

impl<Q> Clone for SmoothedMaxDivergence<Q>

fn clone(&self) -> Self

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<Q> Debug for SmoothedMaxDivergence<Q>

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<Q> Default for SmoothedMaxDivergence<Q>

fn default() -> Self

Returns the “default value” for a type.

impl<Q: Clone> FixDeltaMeasure for SmoothedMaxDivergence<Q>

type Atom = Q

type FixedMeasure = FixedSmoothedMaxDivergence<Q>

fn new_fixed_measure(&self) -> Fallible<Self::FixedMeasure>

fn fix_delta(&self, curve: &Self::Distance, delta: &Q) -> Fallible<(Q, Q)>

impl<Q> Measure for SmoothedMaxDivergence<Q>

type Distance = SMDCurve<Q>

Proof Definition

impl<Q> PartialEq<SmoothedMaxDivergence<Q>> for SmoothedMaxDivergence<Q>

fn eq(&self, _other: &Self) -> bool

This method tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

This method tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.