minimize(method=’dogleg’)

torchmin.trustregion._minimize_dogleg(fun, x0, **trust_region_options)[source]

Minimization of scalar function of one or more variables using the dog-leg trust-region algorithm.

Warning

The Hessian is required to be positive definite at all times; otherwise this algorithm will fail.

Parameters

  • fun (callable) – Scalar objective function to minimize

  • x0 (Tensor) – Initialization point

  • initial_trust_radius (float) – Initial trust-region radius.

  • max_trust_radius (float) – Maximum value of the trust-region radius. No steps that are longer than this value will be proposed.

  • eta (float) – Trust region related acceptance stringency for proposed steps.

  • gtol (float) – Gradient norm must be less than gtol before successful termination.

Returns

result – Result of the optimization routine.

Return type

OptimizeResult

References

1

Jorge Nocedal and Stephen Wright, Numerical Optimization, second edition, Springer-Verlag, 2006, page 73.