Some distances on subspaces

yifeng xue

Abstract


Let $H$ be an infinite separable complex Hilbert space. Let $V$ be a closed subspace of $H$ with $\dim V=n$ and let $MO_n(V)$ be the set of $n$--tuples $(e_1,\cdots,e_n)$ such that $e_1,\cdots,e_n$ are mutually orthonormal in $V$. In this short note, we first show that for an $n$--tuple of linearly independent unit vectors $(\alpha_1,\cdots,\alpha_n)$ in $H$, there is an $n$--tuple of mutually orthonormal vectors $(\gamma_1,\cdots,\gamma_n)=
(\alpha_1,\cdots,\alpha_n)A^{-1/2}$ such that
$$
\sum\limits^n_{j=1}\|\alpha_j-\gamma_j\|^2=\inf\big\{\sum\limits^n_{j=1}\|\alpha_j-e_j\|^2\vert\,(e_1,\cdots,e_n)\in MO_n(H)\big\},
$$
where $A$ is the Gram matrix of $(\alpha_1,\cdots,\alpha_n)$; then we define a new distance $d_1(M,N)$ for two subspaces $M$ and $N$ with $\dim M=\dim N<+\infty$ and give a formula to compute $d_2(M,N)$.


References


F. Deutsch, Best Approximation in Inner Product Spaces, CMS Books in Math. 7, Springer--Verlag, Now York, Berlin, Heidelberg, 2001.

S. Hu and Y. Xue, Completeness of $n$--tuple of projections in $C^*$--algebras, J. Operator Theory, {bf 72}(1) (2014), 49--70.

X. Sun and Q. Cheng, On subspace distance, in Lecture Notes in Computer Science, 4142, 81--89, Springer Berlin Heidelberg, 2006.

L. Wang, X. Wang and J. Feng, Subspace distance analysis with application to adaptive Bayesian algorithm for face recognition, Pattern Recognition {bf 39} (3) (2006), 456-464.

Y. Xue, Stable Perturbations of Operators and Related Topics, World Scientific, 2012.

G. Zuccon, L.A. Azzopardi and C.J. van Rijsbergen, Semantic Spaces: Measuring the distance between different subspaces, 225--236, P. Bruza, D. Sofge, W. Lawless, K. van Rijsbergen, M. Klusch (Eds), Quantum Interaction, Lecture Notes in Artificial Intelligence, 5494, Springer-Verlag, Berlin, 2009.


Refbacks

  • There are currently no refbacks.