minimize(method=’trust-krylov’)

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

Minimization of scalar function of one or more variables using the GLTR Krylov subspace trust-region algorithm.

Warning

This minimizer is in early stages and has not been rigorously tested. It may change in the near future.

Parameters

  • fun (callable) – Scalar objective function to minimize.

  • x0 (Tensor) – Initialization point.

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

  • max_tr_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

Notes

This trust-region solver is based on the GLTR algorithm as described in 1 and 2.

References

1

F. Lenders, C. Kirches, and A. Potschka, “trlib: A vector-free implementation of the GLTR method for…”, arXiv:1611.04718.

2

N. Gould, S. Lucidi, M. Roma, P. Toint: “Solving the Trust-Region Subproblem using the Lanczos Method”, SIAM J. Optim., 9(2), 504–525, 1999.

3

J. Nocedal and S. Wright, “Numerical optimization”, Springer Science & Business Media. pp. 83-91, 2006.