I am trying to solve a problem that is asking for what conditions that make $<A>$ unique. It does not explain what <> represents, but I think it means the column space. A is a matrix where its columns are a subset of the k largest eigenvectors of B.

I do not understand what it means for the column space to be unique. But if I had to guess, I would take it to mean that A is full rank, i.e., the subset of eigenvectors that form A are linearly independent.

Does anyone have a different interpretation?

Edit: Actually I just confirmed that $<A>$ is the column space of A.


The column space is unique as it is defined as the smallest (wrt inclusion) subspace that contains all the columns of $A$ as elements. In this case it also is the set of all linear combinations of columns of $A$, which is minimal (as we can check it is a subspace and any subspace containing the columns must contain their linear combinations).

  • $\begingroup$ Thanks. So if $A$ is full rank, then the column space of A is unique? $\endgroup$ – Iamanon Nov 20 '18 at 21:44
  • $\begingroup$ @Iamanon It's always unique, but with full rank it is equal to the maximal subspace (the whole vector space the columns lie in), so $\mathbb{R}^n$ in the case of a real-valued $n \times m$ matrix. $\endgroup$ – Henno Brandsma Nov 20 '18 at 21:47
  • $\begingroup$ Oh I see. Then it seems the question does not make sense. The question is asking for me to give a condition that makes the subspace <A> unique, where <A> is colspan(A). Since <A> is defined as colspan(A), then by definition <A> is unique? $\endgroup$ – Iamanon Nov 20 '18 at 21:51
  • $\begingroup$ @Iamanon yes that's a nonsensical question. If we assume all the columns are linearly independent we at least can say that every element can be written in a unique way as a linear combination of columns. But that's not the same as unicity of the subspace, $\endgroup$ – Henno Brandsma Nov 20 '18 at 21:53
  • $\begingroup$ Actually, $<A>$ is defined as the column SPAN (not space) of A. The column space is unique., but I think the column span need not be composed of the span of linearly independent columns. For example, if A is rank 3. Then we can still say that the column span is the span of the 3 columns of A. But this would not be the column space of A. The column space of A would be the span of the two linearly independent columns. $\endgroup$ – Iamanon Nov 20 '18 at 21:55

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.