On Riemannian and non-Riemannian Optimisation, and Optimisation Geometry
نویسندگان
چکیده
Abstract Optimisation algorithms such as the Newton method were first generalised to manifolds by generalising components of algorithm directly: gradients replaced Riemannian gradients, straight lines geodesics, and so forth. This meant having endow manifold with a metric. Traditionally then, attention focused on geometry underlying manifold. However, we argue is not right focus because it does take cost function into consideration. For online optimisation problems requiring minimisation many different functions, most relevance family functions whole: if fit together in “nice” way, fast can be developed even individual are difficult optimise. In particular, non-convex necessarily problems. paper presents Riemannian-based homotopy for solving Geometry briefly explains how non-Riemannian (e.g., coordinate-adapted) algorithm.
منابع مشابه
Probing non-Riemannian spacetime geometry
The equations of motion for matter in non-Riemannian spacetimes are derived via a multipole method. It is found that only test bodies with microstructure couple to the non-Riemannian spacetime geometry. Consequently it is impossible to detect spacetime torsion with the satellite experiment Gravity Probe B, contrary to some recent claims in the literature.
متن کاملOn non-Riemannian geometry of superfluids
The Gross-Pitaevski (GP) equation describing helium superfluids is extended to non-Riemannian spacetime background where torsion is shown to induce the splitting in the potential energy of the flow. A cylindrically symmetric solution for Minkowski background with constant torsion is obtained which shows that torsion induces a damping on the superfluid flow velocity. The Sagnac phase shift is co...
متن کاملA Geometry Preserving Kernel over Riemannian Manifolds
Abstract- Kernel trick and projection to tangent spaces are two choices for linearizing the data points lying on Riemannian manifolds. These approaches are used to provide the prerequisites for applying standard machine learning methods on Riemannian manifolds. Classical kernels implicitly project data to high dimensional feature space without considering the intrinsic geometry of data points. ...
متن کاملRiemannian Geometry on Quantum Spaces
An algebraic formulation of Riemannian geometry on quantum spaces is presented, where Riemannian metric, distance, Laplacian, connection, and curvature have their counterparts. This description is also extended ∗email address: [email protected] to complex manifolds. Examples include the quantum sphere, the complex quantum projective spaces and the two-sheeted space.
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IFAC-PapersOnLine
سال: 2021
ISSN: ['2405-8963', '2405-8971']
DOI: https://doi.org/10.1016/j.ifacol.2021.06.119