Struct GridSampler 

pub struct GridSampler { /* private fields */ }

Exhaustive search over a caller-provided discrete grid.

Unlike crate::RandomSampler, which can sample continuous distributions directly, GridSampler only makes sense for parameters with a finite, pre-specified set of values — you must register each parameter’s grid up front with GridSampler::add_grid (or the typed helpers). The trials are enumerated as a mixed-radix odometer over the registered grids: trial n maps deterministically to one point in the Cartesian product, in the order the grids were added.

Notes / limitations:

• Because this is define-by-run, a given trial may not request every registered parameter (conditional spaces). Enumeration still ranges over the full product; unused coordinates simply don’t get read that trial.
• If more trials are run than there are grid points, the odometer wraps (trial n uses combination n % total).
• A parameter that is suggested but was never registered has no grid to draw from, so it falls back to a random draw (seeded, for reproducibility) and is documented as such rather than silently returning a constant.

Implementations

impl GridSampler

pub fn new() -> Self

A grid sampler with no parameters registered yet.

pub fn add_grid(self, name: &str, values: Vec<Value>) -> Self

Register a parameter’s grid of explicit values. Order of registration defines the odometer axis order.

pub fn add_float_grid(self, name: &str, values: &[f64]) -> Self

Register a float-valued grid.

pub fn add_int_grid(self, name: &str, values: &[i64]) -> Self

Register an integer-valued grid.

pub fn add_categorical_grid(self, name: &str, values: &[&str]) -> Self

Register a categorical grid.

pub fn grid_size(&self) -> usize

Total number of grid points (the product of all registered grid sizes). Running this many trials covers the whole grid exactly once.

Trait Implementations

impl Default for GridSampler

fn default() -> Self

Returns the “default value” for a type.

impl Sampler for GridSampler

fn suggest( &mut self, _study_state: &StudyState, trial: &Trial, param_name: &str, distribution: &Distribution, ) -> Value