Crate atune_showcase 

The live atune browser demo.

A seeded study tunes a real reinforcement learner in the page (the default cartpole problem, D113: Augmented Random Search training a CartPole-v1 policy, one full training run per trial) or minimizes a 2-D test function (Rastrigin, Ackley, Rosenbrock, Himmelblau, or a custom formula) at a chosen pace while the page draws the sampled points over the problem’s heatmap, a model map, a chart of the run and, for cart-pole, the best policy balancing the pole. Every catalog sampler runs under every scheduler from a seed and budget the visitor sets; the page states facts about each run and never ranks the samplers.

• controller owns the study and runs at most one ask/train/tell transaction per call. It has no window, clock, or browser dependency, so host tests drive the exact sequence the browser replays.
• problem, cartpole, training, heatmap, formula, landscape, and belief have no window, clock, or browser dependency.
• app paints a Session into any egui::Ui, with chart’s gap and parameter charts and atune_gui’s read-only history chart through StudyObservation.
• web (wasm only) wires the semantic DOM controls, the eframe canvas, visibility pausing, and the portfolio embed messages. It is started only by the atune_showcase binary.

Modules

app

The canvas side of the demo: frame pacing and egui painting.

belief

What a run has learned about the function, as the page draws it (D95).

cartpole

CartPole-v1 and the Augmented Random Search learner behind the cartpole problem (D113).

chart

The demo’s own charts beside atune_gui’s history chart (D95): each trial’s gap to the reference minimum on a log scale, and each parameter per trial.

controller

The demo’s run controller: one seeded study, advanced one trial at a time.

formula

The formula language of the demo’s custom landscape (D94).

heatmap

Pure geometry for the demo’s maps: sampled value grids, their color ramps and contour lines, and the mapping from parameter space to the unit square.

landscape

The custom objective (D94): a Formula on the fixed box [-5, 5]².

model_job

Deterministic model jobs shared by the sliced canvas and browser worker.

problem

The objectives the demo optimizes.

training

The simulated training run a scheduled trial reports from.