Let $S=G/K$ and $S^*=G^*/K^*$ by symmetric spaces, with Riemannian metrics given by the Killing forms $B$ and $B^*$. We call $S$ and $S^*$ dual if

  • there exists a Lie algebra isomorphism $\phi\colon \mathfrak k \to \mathfrak k^*$ such that $$B^*(\phi V, \phi W) = B(V,W) \text{ for all }V,W\in \mathfrak k$$
  • there is a linear isometry $\psi\colon \mathfrak p \to \mathfrak p^*$ such that $$[\psi X,\psi Y]=-\phi[X,Y]\text{ for all } X,Y\in \mathfrak p$$

where $\mathfrak g=\mathfrak k \oplus \mathfrak p$ is the Cartan decomposition.

I have to show that $S$ and $S^*$ have opposite curvature. I already know that $R(X,Y)Z=[[X,Y],Z]$, so if I have two orthonormal vectors $X,Y\in \mathfrak p$, I get $\operatorname{sec}(X,Y) = \langle R(X,Y)X,Y\rangle=\langle [[X,Y],X],Y\rangle = \langle [X,Y],[X,Y]\rangle$. Therefore it remains to show $$-\langle [X,Y],[X,Y]\rangle_S = \langle [\psi X,\psi Y],[\psi X,\psi Y]\rangle_{S^*}$$

My problem seems to be that I'm not able to show that $S$ and $S^*$ are of different type. If I assume that, for example, $S$ is of non-compact type and $S^*$ is of compact type, then

$$\langle [\psi X,\psi Y],[\psi X,\psi Y]\rangle_{S^*} = -B^*([\psi X,\psi Y],[\psi X,\psi Y]) = -B^*(-\phi[X,Y],-\phi[X,Y])=-B([X,Y],[X,Y])=-\langle [X,Y],[X,Y]\rangle_S$$

But I have no idea how to show that, if $S$ is of non-compact type, then $S^*$ is of compact type. This however seems to be the main reason for the change of sign in the calculation. Any help?

EDIT: I'm not so sure anymore whether this problem is answerable with a positive answer. A counterexample would also be nice.


1 Answer 1


I would like to know where you take that definition of duality from. If you use the following notion of duality, it is very easy to prove that: if $S$ is of non-compact type, then $S^*$ is of compact type.

Let $\mathfrak{g}:=\mathfrak{p}\oplus[\mathfrak{p},\mathfrak{p}]$. Then, you can consider the complexification of $\mathfrak{g}$ and $G$. Let us call it $\mathfrak{g}^{\mathbb{C}}$ and $G^{\mathbb{C}}$, respectively. Hence, we can define $\mathfrak{g}^*=\mathfrak{k}\oplus i\mathfrak{p}$. It is easy to check that $\mathfrak{g}^*$ is a Lie subalgebra of $\mathfrak{g}^{\mathbb{C}}$. On the other hand, let us define $$S^*:=G^*/K,$$ where $G^*\subset G^{\mathbb{C}}$ is the connected subgroup of $G^{\mathbb{C}}$ with Lie algebra $\mathfrak{g}^*$. This is the dual symmetric space to $S$.

Notice that the Killing form of $\mathfrak{g}$ or $\mathfrak{g}^*$ is just the restriction of the Killing form of $\mathfrak{g}^\mathbb{C}$ to $\mathfrak{g}$ or to $\mathfrak{g}^*$, respectively. Then, the Killing form of $\mathfrak{g}$ is negative definite. Now for every non-zero $X\in\mathfrak{p}$, $$ B_{g^*}(i X, i X)=B_{\mathfrak{g}^{\mathbb{C}}}(i X, i X)=-B_{\mathfrak{g}^{\mathbb{C}}}(X,X)=-B_{\mathfrak{g}}(X,X).$$

Thus, if for instance $S$ is of non compact-type, then, $B_{g^*}(i X, i X)<0$ for every non-zero $X\in\mathfrak{p}$, which implies that $S^*$ is of compact type.

Observe that the reverse implication is also inmediate from the above argument.


You must log in to answer this question.

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