Class MultiFidelityTrainingHistory<T>

Namespace

AiDotNet.PhysicsInformed

Assembly

AiDotNet.dll

Training history for multi-fidelity PINN training.

public class MultiFidelityTrainingHistory<T> : TrainingHistory<T>, IMultiFidelityTrainingHistory<T>

Type Parameters

T

The numeric type.

Inheritance

object

TrainingHistory<T>

MultiFidelityTrainingHistory<T>

Implements

IMultiFidelityTrainingHistory<T>

Inherited Members

TrainingHistory<T>.Losses

TrainingHistory<T>.AddEpoch(T)

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

For Beginners: This class tracks the training progress of a multi-fidelity PINN. It extends the base TrainingHistory with additional metrics specific to multi-fidelity learning:

• LowFidelityLosses: Error on cheap/approximate data
• HighFidelityLosses: Error on expensive/accurate data
• CorrelationLosses: How well the model captures the relationship between fidelity levels
• PhysicsLosses: PDE residual errors

Typical Training Dynamics:

1. Early training: Low-fidelity loss dominates (most data)
2. Mid training: High-fidelity becomes important (precision matters)
3. Late training: Physics loss should be low (PDE satisfied)

Properties

CorrelationLosses

Gets the correlation losses per epoch.

public List<T> CorrelationLosses { get; }

Property Value

List<T>

HighFidelityLosses

Gets the high-fidelity data losses per epoch.

public List<T> HighFidelityLosses { get; }

Property Value

List<T>

LowFidelityLosses

Gets the low-fidelity data losses per epoch.

public List<T> LowFidelityLosses { get; }

Property Value

List<T>

PhysicsLosses

Gets the PDE residual losses per epoch.

public List<T> PhysicsLosses { get; }

Property Value

List<T>

Methods

AddEpoch(T, T, T, T, T)

Records metrics for a training epoch.

public void AddEpoch(T totalLoss, T lowFidelityLoss, T highFidelityLoss, T correlationLoss, T physicsLoss)

Parameters

totalLoss T

Combined loss from all components.

lowFidelityLoss T

Loss from low-fidelity data fitting.

highFidelityLoss T

Loss from high-fidelity data fitting.

correlationLoss T

Loss measuring fidelity correlation.

physicsLoss T

PDE residual loss.