Struct Distribution 

pub struct Distribution { /* private fields */ }

Owns a sorted snapshot of observed values and answers arbitrary quantile queries.

Construction sorts the input once; queries are O(1) index lookups. Input is f64 throughout — Duration callers convert to milliseconds at the call site.

Implementations

impl Distribution

pub fn from_unsorted(values: Vec<f64>) -> Self

Sorts values and takes ownership. O(n log n).

Use from_sorted when values are already sorted to skip the sort step.

pub fn from_sorted(values: Vec<f64>) -> Self

Constructs a Distribution from an already-sorted Vec<f64>.

Caller is responsible for ensuring the input is sorted in ascending order. No sort is performed; construction is O(1).

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

Returns the value at quantile p in [0.0, 1.0] using the floor-index formula.

idx = (n as f64 * p).floor() as usize, clamped to [0, n-1]. Returns 0.0 for an empty distribution.

This formula is equivalent to the integer formula n * p / 100 (for integer percentile values 0–100) used in the ASCII table renderer, ensuring both output paths report identical percentile values.

pub fn min(&self) -> f64

Returns the minimum value, or 0.0 for an empty distribution.

pub fn max(&self) -> f64

Returns the maximum value, or 0.0 for an empty distribution.

pub fn mean(&self) -> f64

Returns the arithmetic mean, or 0.0 for an empty distribution.

pub fn is_empty(&self) -> bool

Returns true if the distribution contains no values.

pub fn len(&self) -> usize

Returns the number of values in the distribution.

pub fn value_at(&self, idx: usize) -> f64

Returns the value at a pre-computed index into the sorted backing array.

Use this when the caller has already computed the index via the integer formula (n * p / 100).min(n - 1) and wants to avoid the redundant float round-trip through quantile(idx as f64 / n as f64).

Panics

Panics if idx >= self.len(). The caller is responsible for clamping the index to [0, n-1] before calling this method.