The Grassmannian $\text{Gr}(k,n)$ can be described as the quotient $$\text{Gr}(k,n) \cong \text{GL}(k,k) \ \backslash \ \text{Mat}^*_{\mathbb{R}}(k,n) $$

where $\text{Mat}^*_{\mathbb{R}}(k,n)$ is the set of real $k \times n$ matrices of rank $k$. Every point of $\text{Gr}(k,n)$ can be represented by a $k \times n$-matrix $M$. The Plücker coordinates on $\text{Gr}(k,n)$ are all the $k \times k$-minors of $M$.

I don't really understand this construction. I have been reading about how to construst $\text{Gr}(k,n)$ as a projective variety via embedding into $\mathbb{P}\mathbb{R}$ of dimension $n \choose k$ and expressing the $n \choose k$-minors by vanishing of homogeneous polynomials, but I am not sure how to relate that description to the one above.

I also came across a construction of $\text{Gr}(k,n)$ as quotient of $\text{GL}(k,k)$ by some stabilizer, but this seems to be related to its structure as a manifold, not as a variety, as far as I understood. My goal here is to understand another space, which is a subvariety in $\text{Gr}(k,n)$, but I don't really follow this definition of $\text{Gr}(k,n)$ to begin with.

Any help/explanation would be really great!


That should be $\text{Mat}^*_{\mathbb{R}}(k,n)/Gl(k,k)$. You might find it helpful to instead think of this as the quotient $\text{Mat}^*_{\mathbb{R}}(k,n)/\sim$, where $\sim$ is the relation by which $A \sim B$ if $A,B$ have the same row-space.

| cite | improve this answer | |
  • $\begingroup$ Hmmm, weird that there's a typo, because it's mentioned as $\text{Gr}(k,n) \cong \text{GL}(k,k) \ / \ \text{Mat}^*_{\mathbb{R}}(k,n)$ twice in the paper I am reading..Thanks for your answer either way! $\endgroup$ – mike Jun 30 at 20:14
  • $\begingroup$ @mike Is it instead written as $GL(k,k) \setminus \text{Mat}_{\Bbb R}^*(k,n)$? $\endgroup$ – Ben Grossmann Jun 30 at 20:17
  • $\begingroup$ Oh yes, i am def blind. feeling really silly now :D $\endgroup$ – mike Jun 30 at 20:22
  • $\begingroup$ @mike The fact that it's from the left is to emphasize that the action is multiplication from the left by elements of $GL(k,k)$ $\endgroup$ – Ben Grossmann Jun 30 at 20:23
  • $\begingroup$ Hmm Yes, so it's really just re-stating the same construction like with Plücker embedding? I mean, we have a map from $(k,n)$-matrices of rank $k$ to projective space of dimension $n \choose k$, where we consider the projective space in order to remain invariant under action of $\text{GL}(k,k)$, correct? $\endgroup$ – mike Jun 30 at 20:28

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.