nlsq.stability.recovery.OptimizationRecovery¶

class nlsq.stability.recovery.OptimizationRecovery(max_retries=3, enable_diagnostics=True)[source]¶

Bases: object

Automatic recovery from optimization failures.

This class provides multiple recovery strategies for handling optimization failures including: - Parameter perturbation - Algorithm switching - Regularization adjustment - Problem reformulation - Multi-start optimization

max_retries¶

Maximum number of recovery attempts

Type:: int

strategies¶

List of recovery strategies to try

Type:: list

diagnostics¶

Diagnostics collector for monitoring

Type:: nlsq.diagnostics.OptimizationDiagnostics

stability_guard¶

Numerical stability checker

Type:: nlsq.stability.NumericalStabilityGuard

__init__(max_retries=3, enable_diagnostics=True)[source]¶

Initialize recovery system.

Parameters:

max_retries (int) – Maximum recovery attempts
enable_diagnostics (bool) – Enable diagnostic collection

recover_from_failure(failure_type, optimization_state, optimization_func, **kwargs)[source]¶

Attempt recovery from optimization failure.

Parameters:

failure_type (str) – Type of failure (‘convergence’, ‘numerical’, ‘memory’, etc.)
optimization_state (dict) – Current state of optimization
optimization_func (callable) – Function to retry optimization
**kwargs – Additional arguments for optimization function

Returns:

success (bool) – Whether recovery succeeded
result (dict) – Recovered optimization result or error info

Return type:

tuple[bool, dict]