Intrinsic representation of tangent vectors and vector transports on matrix manifolds

نویسندگان

  • Wen Huang
  • Pierre-Antoine Absil
  • Kyle A. Gallivan
چکیده

The quasi-Newton methods on Riemannian manifolds proposed thus far do not appear to lend themselves to satisfactory convergence analyses unless they resort to an isometric vector transport. This prompts us to propose a computationally tractable isometric vector transport on the Stiefel manifold of orthonormal p-frames in R. Specifically, it requires O(np) flops, which is considerably less expensive than existing alternatives in the frequently encountered case where n ≫ p. We then build on this result to also propose computationally tractable isometric vector transports on other manifolds, namely the Grassmann manifold, the fixed-rank manifold, and the positivesemidefinite fixed-rank manifold. In the process, we also propose a convenient way to represent tangent vectors to these manifolds as elements of R, where d is the dimension of the manifold. We call this an “intrinsic” representation, as opposed to “extrinsic” representations as elements of R, where w is the dimension of the embedding space. Finally, we demonstrate the performance of the proposed isometric vector transport in the context of a Riemannian quasi-Newton method applied to minimizing the Brockett cost function. This paper presents research results of the Belgian Network DYSCO (Dynamical Systems, Control, and Optimization), funded by the Interuniversity Attraction Poles Programme initiated by the Belgian Science Policy Office. This work was supported by grant FNRS PDR T.0173.13. Wen Huang Department of Mathematical Engineering, ICTEAM Institute, Université catholique de Louvain, B-1348 Louvain-la-Neuve, Belgium. Tel.: +32-10-478005 Fax: +32-10-472180 E-mail: [email protected] P.-A. Absil Department of Mathematical Engineering, ICTEAM Institute, Université catholique de Louvain, B-1348 Louvain-la-Neuve, Belgium. K. A. Gallivan Department of Mathematics, 208 Love Building, 1017 Academic Way, Florida State University, Tallahassee FL 32306-4510, USA.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A 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 کامل

Toolbox for Grassmann Manifold Computations

This a description and user guide for an object oriented toolbox written in matlab for computations defined on Grassmann manifolds and products of Grassmann manifolds. It implements basic operations as geodesic movement and parallel transport of tangent vectors. 1 Content of toolbox There are two classes: • @grass — Contains functions, methods and operations for objects on one Grssmannian. • @p...

متن کامل

Statistical cosymplectic manifolds and their submanifolds

    In ‎this ‎paper‎, we introduce statistical cosymplectic manifolds and investigate some properties of their tensors. We define invariant and anti-invariant submanifolds and study invariant submanifolds with normal and tangent structure vector fields. We prove that an invariant submanifold of a statistical cosymplectic manifold with tangent structure vector field is a cosymplectic and minimal...

متن کامل

On the k-nullity foliations in Finsler geometry

Here, a Finsler manifold $(M,F)$ is considered with corresponding curvature tensor, regarded as $2$-forms on the bundle of non-zero tangent vectors. Certain subspaces of the tangent spaces of $M$ determined by the curvature are introduced and called $k$-nullity foliations of the curvature operator. It is shown that if the dimension of foliation is constant, then the distribution is involutive...

متن کامل

Compression of Breast Cancer Images By Principal Component Analysis

The principle of dimensionality reduction with PCA is the representation of the dataset ‘X’in terms of eigenvectors ei ∈ RN  of its covariance matrix. The eigenvectors oriented in the direction with the maximum variance of X in RN carry the most      relevant information of X. These eigenvectors are called principal components [8]. Ass...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:
  • Numerische Mathematik

دوره 136  شماره 

صفحات  -

تاریخ انتشار 2017