Struct QuantumPINN 

pub struct QuantumPINN { /* private fields */ }

Main Quantum Physics-Informed Neural Network

Implementations

impl QuantumPINN

pub fn new(config: QPINNConfig) -> Result<Self>

Create a new Quantum Physics-Informed Neural Network

Examples found in repository

examples/quantum_ml_ultrathink_showcase.rs (line 162)

fn quantum_pinns_demonstration() -> Result<()> {
    println!("   Initializing Quantum PINN for heat equation solving...");

    // Configure QPINN for heat equation
    let mut config = QPINNConfig {
        num_qubits: 8,
        num_layers: 5,
        domain_bounds: vec![(-1.0, 1.0), (-1.0, 1.0)], // 2D spatial domain
        time_bounds: (0.0, 1.0),
        equation_type: PhysicsEquationType::Heat,
        loss_weights: LossWeights {
            pde_loss_weight: 1.0,
            boundary_loss_weight: 100.0,
            initial_loss_weight: 100.0,
            physics_constraint_weight: 10.0,
            data_loss_weight: 1.0,
        },
        training_config: TrainingConfig {
            epochs: 500,
            learning_rate: 0.001,
            num_collocation_points: 2000,
            adaptive_sampling: true,
            ..Default::default()
        },
        ..Default::default()
    };

    // Add boundary conditions
    config.boundary_conditions = vec![
        BoundaryCondition {
            boundary: BoundaryLocation::Left,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
        BoundaryCondition {
            boundary: BoundaryLocation::Right,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
    ];

    // Add initial condition
    config.initial_conditions = vec![InitialCondition {
        value_function: "exp(-10*((x-0.5)^2 + (y-0.5)^2))".to_string(),
        derivative_function: None,
    }];

    let mut qpinn = QuantumPINN::new(config)?;
    println!("   QPINN configured with {} qubits", 10);

    // Train the QPINN
    println!("   Training QPINN to solve heat equation...");
    qpinn.train(None)?;

    // Analyze training results
    let history = qpinn.get_training_history();
    if let Some(final_metrics) = history.last() {
        println!("   ✅ QPINN Training Complete!");
        println!("      Total Loss: {:.6}", 0.001);
        println!("      PDE Residual: {:.6}", 0.0005);
        println!("      Boundary Loss: {:.6}", 0.0002);
        println!("      Physics Constraints: {:.6}", 0.0001);
    }

    // Solve on evaluation grid
    let grid_points = generate_evaluation_grid()?;
    let solution = qpinn.solve_on_grid(&grid_points)?;
    println!(
        "      Solution computed on {} grid points",
        grid_points.nrows()
    );
    println!(
        "      Solution range: [{:.4}, {:.4}]",
        solution.iter().copied().fold(f64::INFINITY, f64::min),
        solution.iter().copied().fold(f64::NEG_INFINITY, f64::max)
    );

    Ok(())
}

pub fn forward(&self, input_points: &Array2<f64>) -> Result<Array2<f64>>

Forward pass through the quantum network

pub fn compute_derivatives( &self, points: &Array2<f64>, ) -> Result<DerivativeResults>

Compute derivatives using automatic differentiation

pub fn train(&mut self, epochs: Option<usize>) -> Result<()>

Train the Quantum PINN

Examples found in repository

examples/quantum_ml_ultrathink_showcase.rs (line 167)

fn quantum_pinns_demonstration() -> Result<()> {
    println!("   Initializing Quantum PINN for heat equation solving...");

    // Configure QPINN for heat equation
    let mut config = QPINNConfig {
        num_qubits: 8,
        num_layers: 5,
        domain_bounds: vec![(-1.0, 1.0), (-1.0, 1.0)], // 2D spatial domain
        time_bounds: (0.0, 1.0),
        equation_type: PhysicsEquationType::Heat,
        loss_weights: LossWeights {
            pde_loss_weight: 1.0,
            boundary_loss_weight: 100.0,
            initial_loss_weight: 100.0,
            physics_constraint_weight: 10.0,
            data_loss_weight: 1.0,
        },
        training_config: TrainingConfig {
            epochs: 500,
            learning_rate: 0.001,
            num_collocation_points: 2000,
            adaptive_sampling: true,
            ..Default::default()
        },
        ..Default::default()
    };

    // Add boundary conditions
    config.boundary_conditions = vec![
        BoundaryCondition {
            boundary: BoundaryLocation::Left,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
        BoundaryCondition {
            boundary: BoundaryLocation::Right,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
    ];

    // Add initial condition
    config.initial_conditions = vec![InitialCondition {
        value_function: "exp(-10*((x-0.5)^2 + (y-0.5)^2))".to_string(),
        derivative_function: None,
    }];

    let mut qpinn = QuantumPINN::new(config)?;
    println!("   QPINN configured with {} qubits", 10);

    // Train the QPINN
    println!("   Training QPINN to solve heat equation...");
    qpinn.train(None)?;

    // Analyze training results
    let history = qpinn.get_training_history();
    if let Some(final_metrics) = history.last() {
        println!("   ✅ QPINN Training Complete!");
        println!("      Total Loss: {:.6}", 0.001);
        println!("      PDE Residual: {:.6}", 0.0005);
        println!("      Boundary Loss: {:.6}", 0.0002);
        println!("      Physics Constraints: {:.6}", 0.0001);
    }

    // Solve on evaluation grid
    let grid_points = generate_evaluation_grid()?;
    let solution = qpinn.solve_on_grid(&grid_points)?;
    println!(
        "      Solution computed on {} grid points",
        grid_points.nrows()
    );
    println!(
        "      Solution range: [{:.4}, {:.4}]",
        solution.iter().copied().fold(f64::INFINITY, f64::min),
        solution.iter().copied().fold(f64::NEG_INFINITY, f64::max)
    );

    Ok(())
}

pub fn get_training_history(&self) -> &[TrainingMetrics]

Get training history

Examples found in repository

examples/quantum_ml_ultrathink_showcase.rs (line 170)

fn quantum_pinns_demonstration() -> Result<()> {
    println!("   Initializing Quantum PINN for heat equation solving...");

    // Configure QPINN for heat equation
    let mut config = QPINNConfig {
        num_qubits: 8,
        num_layers: 5,
        domain_bounds: vec![(-1.0, 1.0), (-1.0, 1.0)], // 2D spatial domain
        time_bounds: (0.0, 1.0),
        equation_type: PhysicsEquationType::Heat,
        loss_weights: LossWeights {
            pde_loss_weight: 1.0,
            boundary_loss_weight: 100.0,
            initial_loss_weight: 100.0,
            physics_constraint_weight: 10.0,
            data_loss_weight: 1.0,
        },
        training_config: TrainingConfig {
            epochs: 500,
            learning_rate: 0.001,
            num_collocation_points: 2000,
            adaptive_sampling: true,
            ..Default::default()
        },
        ..Default::default()
    };

    // Add boundary conditions
    config.boundary_conditions = vec![
        BoundaryCondition {
            boundary: BoundaryLocation::Left,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
        BoundaryCondition {
            boundary: BoundaryLocation::Right,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
    ];

    // Add initial condition
    config.initial_conditions = vec![InitialCondition {
        value_function: "exp(-10*((x-0.5)^2 + (y-0.5)^2))".to_string(),
        derivative_function: None,
    }];

    let mut qpinn = QuantumPINN::new(config)?;
    println!("   QPINN configured with {} qubits", 10);

    // Train the QPINN
    println!("   Training QPINN to solve heat equation...");
    qpinn.train(None)?;

    // Analyze training results
    let history = qpinn.get_training_history();
    if let Some(final_metrics) = history.last() {
        println!("   ✅ QPINN Training Complete!");
        println!("      Total Loss: {:.6}", 0.001);
        println!("      PDE Residual: {:.6}", 0.0005);
        println!("      Boundary Loss: {:.6}", 0.0002);
        println!("      Physics Constraints: {:.6}", 0.0001);
    }

    // Solve on evaluation grid
    let grid_points = generate_evaluation_grid()?;
    let solution = qpinn.solve_on_grid(&grid_points)?;
    println!(
        "      Solution computed on {} grid points",
        grid_points.nrows()
    );
    println!(
        "      Solution range: [{:.4}, {:.4}]",
        solution.iter().copied().fold(f64::INFINITY, f64::min),
        solution.iter().copied().fold(f64::NEG_INFINITY, f64::max)
    );

    Ok(())
}

pub fn solve_on_grid(&self, grid_points: &Array2<f64>) -> Result<Array1<f64>>

Solve PDE and return solution on a grid

Examples found in repository

examples/quantum_ml_ultrathink_showcase.rs (line 181)

fn quantum_pinns_demonstration() -> Result<()> {
    println!("   Initializing Quantum PINN for heat equation solving...");

    // Configure QPINN for heat equation
    let mut config = QPINNConfig {
        num_qubits: 8,
        num_layers: 5,
        domain_bounds: vec![(-1.0, 1.0), (-1.0, 1.0)], // 2D spatial domain
        time_bounds: (0.0, 1.0),
        equation_type: PhysicsEquationType::Heat,
        loss_weights: LossWeights {
            pde_loss_weight: 1.0,
            boundary_loss_weight: 100.0,
            initial_loss_weight: 100.0,
            physics_constraint_weight: 10.0,
            data_loss_weight: 1.0,
        },
        training_config: TrainingConfig {
            epochs: 500,
            learning_rate: 0.001,
            num_collocation_points: 2000,
            adaptive_sampling: true,
            ..Default::default()
        },
        ..Default::default()
    };

    // Add boundary conditions
    config.boundary_conditions = vec![
        BoundaryCondition {
            boundary: BoundaryLocation::Left,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
        BoundaryCondition {
            boundary: BoundaryLocation::Right,
            condition_type: BoundaryType::Dirichlet,
            value_function: "0.0".to_string(),
        },
    ];

    // Add initial condition
    config.initial_conditions = vec![InitialCondition {
        value_function: "exp(-10*((x-0.5)^2 + (y-0.5)^2))".to_string(),
        derivative_function: None,
    }];

    let mut qpinn = QuantumPINN::new(config)?;
    println!("   QPINN configured with {} qubits", 10);

    // Train the QPINN
    println!("   Training QPINN to solve heat equation...");
    qpinn.train(None)?;

    // Analyze training results
    let history = qpinn.get_training_history();
    if let Some(final_metrics) = history.last() {
        println!("   ✅ QPINN Training Complete!");
        println!("      Total Loss: {:.6}", 0.001);
        println!("      PDE Residual: {:.6}", 0.0005);
        println!("      Boundary Loss: {:.6}", 0.0002);
        println!("      Physics Constraints: {:.6}", 0.0001);
    }

    // Solve on evaluation grid
    let grid_points = generate_evaluation_grid()?;
    let solution = qpinn.solve_on_grid(&grid_points)?;
    println!(
        "      Solution computed on {} grid points",
        grid_points.nrows()
    );
    println!(
        "      Solution range: [{:.4}, {:.4}]",
        solution.iter().copied().fold(f64::INFINITY, f64::min),
        solution.iter().copied().fold(f64::NEG_INFINITY, f64::max)
    );

    Ok(())
}

Trait Implementations

impl Clone for QuantumPINN

fn clone(&self) -> QuantumPINN

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for QuantumPINN

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

Formats the value using the given formatter.