How can we represent the space of matrices? E.g. A vector $z\in{}R^m$ in the column-space of matrix $A\in{}R^{m\times{}n}$ can be represented as $$ z=Ax $$ for some $x$.


In the following paper, in equation (7), the authors have noted that a particular null-space can be summarized in compact matrix notation as $$ Ce_ke^T_k+e_ke^T_kC$$

where $C$ is an arbitrary diagonal matrix (perhaps like $x$ in the example above) and $e_k$ is a vector of length $k$ of all ones. I did not understand how this expression can represent a matrix space.

Blaschko, Matthew B., Wojciech Zaremba, and Arthur Gretton. "Taxonomic Prediction with Tree-Structured Covariances." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases. 2013.

Link to the paper.

  • $\begingroup$ Can you consider changing the title and/or first sentence of your question to make it more clear what you are asking? $\endgroup$ – Stefan Smith Aug 30 '13 at 23:21

The expression can describe an space of matrices if you take $C$ to range over all diagonal matrices. As the space of diagonal $n×n$-matrices is naturally isomorphic to $ℝ^n$, you could write the mentioned space of matrices as $$ \{Ce_ke_k^T + e_ke_k^TC \mid C \in \mathrm{Diag}(ℝ, k)\} = \{x^T\!e_ke_k^T + e_ke_k^Tx \mid x \in ℝ^n\}$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.