Struct ExperimentStore 

pub struct ExperimentStore { /* private fields */ }

In-memory store for experiment tracking data.

Design

The store uses hash maps for O(1) lookups by ID, and stores metrics in a vector that can be filtered and sorted for time-series queries.

Time-Series Optimization

The get_metrics_for_run function returns metrics ordered by step, enabling efficient time-series visualization and analysis.

Implementations

impl ExperimentStore

pub fn new() -> Self

Create a new empty experiment store.

pub fn is_empty(&self) -> bool

Check if the store is empty (no experiments, runs, or metrics).

pub fn experiment_count(&self) -> usize

Get the number of experiments in the store.

pub fn run_count(&self) -> usize

Get the number of runs in the store.

pub fn metric_count(&self) -> usize

Get the number of metrics in the store.

pub fn add_experiment(&mut self, experiment: ExperimentRecord)

Add an experiment to the store.

pub fn get_experiment(&self, experiment_id: &str) -> Option<&ExperimentRecord>

Get an experiment by ID.

pub fn add_run(&mut self, run: RunRecord)

Add a run to the store.

pub fn get_run(&self, run_id: &str) -> Option<&RunRecord>

Get a run by ID.

pub fn get_runs_for_experiment(&self, experiment_id: &str) -> Vec<&RunRecord>

Get all runs for an experiment.

pub fn add_metric(&mut self, metric: MetricRecord)

Add a metric to the store.

pub fn get_metrics_for_run(&self, run_id: &str, key: &str) -> Vec<MetricRecord>

Get metrics for a specific run and key, ordered by step.

This is the primary query function for time-series metric data.

Arguments

• run_id - The ID of the run to query
• key - The metric key/name to filter by

Returns

A vector of metrics matching the run_id and key, sorted by step in ascending order.

Example

use trueno_db::experiment::{ExperimentStore, MetricRecord};

let mut store = ExperimentStore::new();

// Log some training metrics
for step in 0..100 {
    let loss = 1.0 / (step as f64 + 1.0);
    store.add_metric(MetricRecord::new("run-001", "loss", step, loss));
}

// Query the loss curve
let loss_metrics = store.get_metrics_for_run("run-001", "loss");
assert_eq!(loss_metrics.len(), 100);

Trait Implementations

impl Debug for ExperimentStore

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

Formats the value using the given formatter.

impl Default for ExperimentStore

fn default() -> ExperimentStore

Returns the “default value” for a type.