Skip to content

The code for vector transport free LBFGS quasi-Newton's optimization on the Riemannian manifolds

License

Notifications You must be signed in to change notification settings

bghojogh/LBFGS-Vector-Transport-Free

Repository files navigation

Vector Transport Free LBFGS on Riemannian Manifolds

The code for vector transport free LBFGS quasi-Newton's optimization on the Symmetric Positive Definite Riemannian manifolds. This project reduces vector transport and Riemannian metric to identity and Euclidean inner product, respectively.

This is the code for the following paper:

  • Reza Godaz, Benyamin Ghojogh, Reshad Hosseini, Reza Monsefi, Fakhri Karray, Mark Crowley, "Vector Transport Free Riemannian LBFGS for Optimization on Symmetric Positive Definite Matrix Manifolds", Proceedings of The 13th Asian Conference on Machine Learning (ACML), PMLR, vol. 157, pp. 1-16, 2021.
  • Link to paper: https://proceedings.mlr.press/v157/godaz21a.html
  • Link to paper at arXiv: https://arxiv.org/abs/2108.11019

Additional notes (related codes)

The code of manopt toolbox and its cautious RLBFGS solver:

The code of Mixest toolbox (it includes its RLBFGS solver):

The code of another version of Mixest toolbox and its RLBFGS solver:

The code of ROPTLIB toolbox and its RBFGS and RLBFGS solvers: