Module adaptive 

Adaptive exploration scenario functions.

These scenarios exercise the adaptive code path (dispatch_split -> adaptive_split_on_discovery) with coverage-yield-driven batching and 3-level energy budgets.

Each scenario provides a minimal xorshift64 RNG via thread-local storage, wired into moonpool-explorer through set_rng_hooks. Since scenarios use fork(), each must run in its own process (nextest default).

Enums

AdaptiveTestError

Errors from adaptive exploration test scenarios.

Constants

CALL_COUNT

Number of RNG calls since last reseed.

RNG_STATE

Current xorshift64 RNG state.

Functions

count

Return the current RNG call count.

next_random

Advance the xorshift64 RNG and return the next value.

random_below

Return a random integer in 0..divisor.

reseed

Reseed the xorshift64 RNG and reset the call counter.

run_adaptive_dungeon_floors

Progressive multi-floor exploration.

run_adaptive_energy_budget

Verify that the 3-level energy budget (global, per-mark, realloc pool) constrains adaptive forking. Uses low energy values to ensure bounds.

run_adaptive_maze_cascade

Cascading probability gates with dependent locks.