foxhunt/crates/ml/tests/ppo_reward_normalizer_tests.rs

#![allow(
    clippy::assertions_on_constants,
    clippy::assertions_on_result_states,
    clippy::clone_on_copy,
    clippy::decimal_literal_representation,
    clippy::doc_markdown,
    clippy::empty_line_after_doc_comments,
    clippy::field_reassign_with_default,
    clippy::get_unwrap,
    clippy::identity_op,
    clippy::inconsistent_digit_grouping,
    clippy::indexing_slicing,
    clippy::integer_division,
    clippy::len_zero,
    clippy::let_underscore_must_use,
    clippy::manual_div_ceil,
    clippy::manual_let_else,
    clippy::manual_range_contains,
    clippy::modulo_arithmetic,
    clippy::needless_range_loop,
    clippy::non_ascii_literal,
    clippy::redundant_clone,
    clippy::shadow_reuse,
    clippy::shadow_same,
    clippy::shadow_unrelated,
    clippy::single_match_else,
    clippy::str_to_string,
    clippy::string_slice,
    clippy::tests_outside_test_module,
    clippy::too_many_lines,
    clippy::unnecessary_wraps,
    clippy::unseparated_literal_suffix,
    clippy::use_debug,
    clippy::useless_vec,
    clippy::wildcard_enum_match_arm,
    clippy::else_if_without_else,
    clippy::expect_used,
    clippy::missing_const_for_fn,
    clippy::similar_names,
    clippy::type_complexity,
    clippy::collapsible_else_if,
    clippy::doc_lazy_continuation,
    clippy::items_after_test_module,
    clippy::map_clone,
    clippy::multiple_unsafe_ops_per_block,
    clippy::unwrap_or_default,
    clippy::assign_op_pattern,
    clippy::needless_borrow,
    clippy::println_empty_string,
    clippy::unnecessary_cast,
    clippy::used_underscore_binding,
    clippy::create_dir,
    clippy::implicit_saturating_sub,
    clippy::exit,
    clippy::expect_fun_call,
    clippy::too_many_arguments,
    clippy::unnecessary_map_or,
    clippy::unwrap_used,
    dead_code,
    unused_imports,
    unused_variables,
    clippy::cloned_ref_to_slice_refs,
    clippy::neg_multiply,
    clippy::while_let_loop,
    clippy::bool_assert_comparison,
    clippy::excessive_precision,
    clippy::trivially_copy_pass_by_ref,
    clippy::op_ref,
    clippy::redundant_closure,
    clippy::unnecessary_lazy_evaluations,
    clippy::if_then_some_else_none,
    clippy::unnecessary_to_owned,
    clippy::single_component_path_imports,
)]
//! Test suite for PPO RewardNormalizer
//!
//! TDD Red Phase: Tests written FIRST before implementation
//! These tests define the expected behavior of the RewardNormalizer module

// Import from PPO module (will fail until implementation exists)
use ml::ppo::reward_normalizer::RewardNormalizer;

#[test]
fn test_reward_normalizer_welford() {
    // Test Welford's algorithm correctness with known statistics
    let mut normalizer = RewardNormalizer::new();

    // Add values: [10, 20, 30, 40, 50]
    let values = vec![10.0, 20.0, 30.0, 40.0, 50.0];
    for &value in &values {
        normalizer.update(value);
    }

    // Expected mean: 30.0
    // Expected variance: 200.0 (population variance)
    // Expected std: 14.142
    let (mean, std) = normalizer.get_stats();
    assert!((mean - 30.0).abs() < 1e-6, "Mean should be 30.0, got {}", mean);
    assert!(
        (std - 14.142135623730951).abs() < 1e-6,
        "Std should be ~14.142, got {}",
        std
    );

    // Test normalization: (40 - 30) / 14.142 ≈ 0.707
    let normalized = normalizer.normalize(40.0);
    assert!(
        (normalized - 0.7071067811865475).abs() < 1e-6,
        "Normalized value should be ~0.707, got {}",
        normalized
    );

    // Verify count
    assert_eq!(normalizer.count(), 5, "Count should be 5");
}

#[test]
fn test_reward_normalizer_numerical_stability() {
    // Test with extreme values to ensure numerical stability
    let mut normalizer = RewardNormalizer::new();

    // Add extreme positive and negative values
    let extreme_values = vec![-1000.0, -500.0, 0.0, 500.0, 1000.0];
    for &value in &extreme_values {
        normalizer.update(value);
    }

    let (mean, std) = normalizer.get_stats();

    // Mean should be 0.0 (symmetric distribution)
    assert!(
        mean.abs() < 1e-6,
        "Mean should be ~0.0 for symmetric values, got {}",
        mean
    );

    // Std should be ~707.1 (for uniform extreme distribution)
    assert!(
        std > 700.0 && std < 710.0,
        "Std should be ~707.1, got {}",
        std
    );

    // Normalize a new extreme value
    let normalized_extreme = normalizer.normalize(1000.0);

    // Should be bounded: 1000 is 1 std above mean
    assert!(
        normalized_extreme > 1.0 && normalized_extreme < 2.0,
        "Normalized extreme should be ~1.58, got {}",
        normalized_extreme
    );

    // Test with very small epsilon (no division by zero)
    normalizer.update(0.0);
    let normalized_zero = normalizer.normalize(0.0);
    assert!(
        !normalized_zero.is_nan() && !normalized_zero.is_infinite(),
        "Normalized value should not be NaN or Inf"
    );
}