Struct Analytic 

pub struct Analytic {
    pub id: u64,
    pub family: Family,
    pub domain: Domain,
    pub pieces: Pieces,
    pub int: bool,
    pub form: Option<Arc<Linear>>,
    pub weight: Option<Arc<Likelihood>>,
}

Fields

id: u64

family: Family

domain: Domain

pieces: Pieces

int: bool

Whether it’s an int, as floor(x) is: then every piece is a whole number.

form: Option<Arc<Linear>>

For a function of several normal latents, its axis: a linear form of them, whose distribution is family. id is then 0.

weight: Option<Arc<Likelihood>>

The latent’s likelihood, with @integrate: a weight on family.

Implementations

impl Analytic

pub fn new(id: u64, family: Family) -> Self

pub fn same_axis(&self, other: &Analytic) -> bool

Whether it’s a function of the same axis as other.

pub fn linear(&self) -> Option<Linear>

The linear form of independent normal latents it is, if it’s one: affine in a normal latent or a form, and unrestricted.

pub fn renamed(&self, map: &BTreeMap<u64, u64>) -> Self

The same outcome of other latents: each as map renames it.

pub fn at(&self, c: f64, value: f64) -> Value

Its value where its axis is value, at the CDF coordinate c.

pub fn moved(&self, to: &Family) -> Self

The same function of the same values of the draw, in the coordinates of to, a distribution of the draw with the same support or less.

pub fn affine(&self) -> Option<Affine>

The function, when it’s affine everywhere.

pub fn piecewise_affine(&self) -> bool

Whether it’s affine on each piece: then it has a density where it isn’t constant.

pub fn key(&self) -> impl Ord + Hash + use<>

pub fn value(self) -> OpResult<Value>

pub fn is_discrete(&self) -> bool

Whether each of its values has positive probability: it’s a constant on each part of the domain, like floor(x).

pub fn atoms(&self) -> Vec<(Value, Domain, f64)>

Each value of a discrete outcome, with where the draw is when it’s that value, and that part’s share of the domain’s probability.

pub fn segments(&self) -> Vec<(f64, f64, Fun)>

Each part of the domain, with the piece of the function there, as (lo, hi, f).

pub fn has_atoms(&self) -> bool

Whether a single value has positive probability: a constant piece.

pub fn measure(&self, d: &Domain) -> f64

The measure of part of its coordinates: its length, or with a likelihood, the likelihood’s integral there.

pub fn share(&self, part: &Domain) -> f64

The probability of part of its coordinates, given its domain.

pub fn cdf(&self, x: f64) -> f64

pub fn pdf(&self, x: f64) -> f64

The density at x, of a piecewise affine outcome: the caller rules out atoms and other functions.

pub fn quantile(&self, q: f64) -> f64

pub fn integrated_moments(&self, tolerance: f64) -> Option<(f64, f64)>

The mean and variance by numerical integration, to tolerance, for @integrate where there’s no formula: None if an integral doesn’t converge.

pub fn moments(&self) -> Option<(f64, f64)>

The mean and variance, from formulas: None where a piece’s function has none for the draw’s family.

pub fn sd(&self) -> Option<f64>

Trait Implementations

impl Clone for Analytic

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Analytic

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

Formats the value using the given formatter.