foxhunt/ml/tests/dqn_hyperparameters_fields_test.rs

//! Test DQNHyperparameters struct has hold_penalty field
//!
//! This test verifies that the DQNHyperparameters struct includes the new fields
//! needed for action-aware reward system (Wave 2 preparation).

use ml::trainers::dqn::DQNHyperparameters;

#[test]
fn test_dqn_hyperparameters_has_hold_penalty_field() {
    // Create hyperparameters manually (conservative() removed in Wave B)
    let hyperparams = DQNHyperparameters {
        learning_rate: 0.0001,
        batch_size: 128,
        gamma: 0.99,
        epsilon_start: 1.0,
        epsilon_end: 0.01,
        epsilon_decay: 0.995,
        buffer_size: 100000,
        min_replay_size: 1000,
        epochs: 100,
        checkpoint_frequency: 10,
        early_stopping_enabled: true,
        q_value_floor: 0.5,
        min_loss_improvement_pct: 2.0,
        plateau_window: 30,
        min_epochs_before_stopping: 50,
        hold_penalty: -0.001,
    };
    
    // Field should exist and have the default value of 0.01
    assert_eq!(
        hyperparams.hold_penalty,
        -0.001,
        "hold_penalty should be -0.001"
    );
}

#[test]
fn test_dqn_hyperparameters_manual_construction_with_new_fields() {
    // Test that we can manually construct DQNHyperparameters with new fields
    let hyperparams = DQNHyperparameters {
        learning_rate: 0.0001,
        batch_size: 128,
        gamma: 0.99,
        epsilon_start: 1.0,
        epsilon_end: 0.01,
        epsilon_decay: 0.995,
        buffer_size: 100000,
        min_replay_size: 1000,
        epochs: 100,
        checkpoint_frequency: 10,
        early_stopping_enabled: true,
        q_value_floor: 0.5,
        min_loss_improvement_pct: 2.0,
        plateau_window: 30,
        min_epochs_before_stopping: 50,
        hold_penalty: -0.001,
    };

    assert_eq!(hyperparams.hold_penalty, -0.001);
}

#[test]
fn test_hold_penalty_range() {
    // Test various penalty values (negative values penalize HOLD action)
    let test_penalties = vec![-0.01, -0.001, 0.0, 0.001, 0.01];
    
    for penalty in test_penalties {
        let hyperparams = DQNHyperparameters {
            learning_rate: 0.0001,
            batch_size: 128,
            gamma: 0.99,
            epsilon_start: 1.0,
            epsilon_end: 0.01,
            epsilon_decay: 0.995,
            buffer_size: 100000,
            min_replay_size: 1000,
            epochs: 100,
            checkpoint_frequency: 10,
            early_stopping_enabled: true,
            q_value_floor: 0.5,
            min_loss_improvement_pct: 2.0,
            plateau_window: 30,
            min_epochs_before_stopping: 50,
            hold_penalty: penalty,
        };

        assert_eq!(hyperparams.hold_penalty, penalty);
    }
}