6
$\begingroup$

The complex orthogonal group $O(n+1, \mathbb{C})$ acts transitively on the complex quadric $$ Q_{n-1} := \{[z_0:z_1: \cdots :z_n] : z_0^2 + \cdots z_n^2 = 0 \} \subset \mathbb{CP}^n. $$

What is known about the geometry of curves, surfaces, invariant tensors, etc. in this geometry?

If I'm not mistaken, the case $n = 2$, where the quadric is a conic in the projective plane, is just conformal geometry.

$\endgroup$
2
  • 1
    $\begingroup$ Do you want to know about real curves, surfaces, etc. or do you want to know about complex curves and surfaces, etc? I'm assuming that you want to consider $Q_{n-1}$ as a homogeneous space of $\mathrm{O}(n{+}1,\mathbb{C})$, not as a Riemannian space. The case $n=3$ is not so interesting because $Q_2 = \mathbb{CP}^1\times\mathbb{CP}^1$. The case $n=4$ was treated classically, taking advantage of the classical isomorphism $B_2 = C_2$ and the case $n=5$ has a similar (but not as developed) story using $D_3 = A_3$. Above that, I think not much is known in detail. $\endgroup$ Commented Oct 30, 2016 at 1:02
  • $\begingroup$ Thanks @RobertBryant. Yes, I'm interested in the quadric as a homogeneous space of the orthogonal complex group and specially about the geometry of real and complex curves (the method I'm working on provides a way to study those, but I would like to compare to "classic" work). The cases n = 2, 3, 4, and 5 should suffice for me to get an idea of how the method works in this case. Do you have any references? $\endgroup$ Commented Nov 5, 2016 at 7:43

1 Answer 1

3
$\begingroup$

Consider a $(n+2)$-dimensional vector space $V$ over $\mathbb{K}=\mathbb{R}$ or $\mathbb{C}$. Let $V'$ be the set of all non-zero vectors. Then $V'$ is an open (real or complex) submanifold of $V$. The projective space $\mathbb{P}:=V'/\sim$ (where $\sim$ is the equivalence relation on $V'$ denied by $v\sim w$ iff there exists some $\lambda\in\mathbb{K}$ such that $w=\lambda v$), is a compact manifold of dimension $n+1$ (over $\mathbb{K}$), such that $V'\to\mathbb{P}$ is a submersion. In particular, $\mathbb{P}$ consists of all the lines $[v]=\{\lambda v :\lambda\in\mathbb{K}\}$ through $v\in V'$. Now, $T\mathbb{P}$ is identified with $TV'/\sim$ where we appropriate extend $\sim$ in $TV'\cong V'\times V$. In particular, if $E=\mathbb{P}\times V\to \mathbb{P}$ is the trivial bundle over $\mathbb{P}$ and $$F:=\{([v], x) : x\in[v]\}$$ denotes the canonical line bundle over $\mathbb{P}$, then one can show that there is a natural isomorphism of vector bundles $T\mathbb{P}\cong {\rm Hom}(F, E/F)$.

Let us pass now to the definition of the quadric. Assume $n\geq 1$ and consider the light cone in $V'$, i.e. the $(n+1)$-submanifold (hypersurface) $\cal{C}$ of $V'$ defined by $$ {\cal{C}}=\{v\in V' : g(v, v)=0\}, $$ where $g$ is a scalar product on $V$ satisfying specific properties (e.g. for $\mathbb{K}=\mathbb{R}$ the signature of $g$ must be $(p+1, q+1)$ with $p, q\geq 0$). Then, the quadric $Q$ is a $n$-dimensional submanifold of $\mathbb{P}$ defined by $$ Q=\{[v]\in\mathbb{P} : v\in\cal{C}\subset V'\}. $$ In particular, the map $V'\to\mathbb{P}$ restricts to a submersion ${\cal{C}}\to Q$. Also the vector bundle $E\to\mathbb{P}$ restricts to a vector bundle over $Q$ and similarly for the line bundle $F\to\mathbb{P}$.

Now, the Lie group ${\rm SO}(g)$ acts transitively on $Q$ with kernel $A=\{\rm Id \}$ if $n$ is odd, and $A=\{{\rm Id}, -{\rm Id}\}$ if $n$ is even. The quotient ${\rm SO}(g)/A$ is the (effective) Möbius group of transformations on $Q$.

Exercise: Show that $TQ\cong {\rm Hom}(F, F^{\perp}/F)$, where $F^{\perp}$ is the orthogonal subbundle of $F$ relative to the fibre metric $g$ on $E$.

Mention that although $g$ induces a scalar product in the fibres of the bundle $F^{\perp}/F\to Q$, the quadric $Q$ inherits only a weaker conformal structure, which is preserved by the Möbius group.

Example: The complex quadric, say of (complex) dimension n, $$ Q_{n}=\{[z]\in\mathbb{C}P^{n+1} : (z,z)=0\} $$ is diffeomorphic to the Grassmannian ${\rm Gr}_{+}(2, n):={\rm SO}(n+2)/({\rm SO}(n)\times{\rm SO}(2))$ of oriented two planes in $\mathbb{R}^{n+2}$.

The complex quadric admits a holomorphic conformal structure induced by the quadratic form $(z, z)$ on $\mathbb{C}^{n+2}$. It admits also a Kähler structure, induced by the hermitian form $(z, \bar{z})$, which can be thought of as induced by the embedding of $Q_{n}$ in $\mathbb{C}P^{n+1}$. A complex quadric doesn't admit any smooth complex-bilinear Riemannian metric, although it admits a Kähler metric as we said above. Further details on complex quadrics can be found in the books of Kobayashi and Nomizu.

$\endgroup$
3
  • $\begingroup$ Thanks for the answer. What do you mean by Möbius transformations on the quadric (when its complex dimension is not 1)? $\endgroup$ Commented Oct 26, 2016 at 8:07
  • 1
    $\begingroup$ Unfortunately Kobayashi and Nomizu don't treat the geometry of submanifolds in this geometry. $\endgroup$ Commented Oct 26, 2016 at 8:07
  • $\begingroup$ Indeed, although the complex quadric is described explicitly in Vol. 2, Example 10.6, page 278, of the book ok Kobayashi and Nomizu, nothing is said for submanifolds. If you are interested in totally geodesic submanifolds of $Q_{n}$, see the paper ''Totally geodesic submanifolds of the complex quadric'', of S. Klein and the references therein. $\endgroup$
    – 314159.
    Commented Oct 26, 2016 at 19:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .