What are explicit expressions for the operators $I,J$ and $K$ on the hyperkähler manifold $TS^2$ (or $TRP^2$), say, in the coordinates $(x_1,x_2, \alpha_1, \alpha_2)$ where a line $m\in TS^2$ is given parametrically by $(x_1+\alpha_1t,x_2+\alpha_2t,t)$? Are they known? I did not find them in the book "The geometry and dynamics of magnetic monopoles" by Atiyah and Hitchin, and could not extract them from it.
-
$\begingroup$ I think you could use the fact that the total space of the unit sphere bundle of $TS^2$ is $S^3$. I think. $\endgroup$– მამუკა ჯიბლაძეCommented Aug 17, 2016 at 19:40
-
2$\begingroup$ Explicit formulae for the operators in coordinates could be somewhat ungainly, especially since there are no global coordinates. Usually, the most useful way of specifying $I$, $J$, and $K$ on a hyperKähler manifold is to specify the three corresponding parallel $2$-forms in some manner. Would that be useful for your purposes? They aren't hard to describe in this particular case, and, once you know them, if you really want the operators themselves, it's just multivariable calculus to write them down in coordinates, given the $2$-forms. If the $2$-forms would be useful to you, let me know. $\endgroup$– Robert BryantCommented Aug 19, 2016 at 8:51
-
$\begingroup$ @RobertBryant: Thank you very much, this could be useful for my purposes. I have also another related question: I use the (real) coordinates $(x_1,x_2,α_1,α_2)$ since they are usual for Radon's transform. However, one often uses the (complex) Donaldson parameters $a_0,a_1,b_0$ verifying $a_0^2+b_0a_1^2=1.$ How are they related? $\endgroup$– sgvlCommented Aug 20, 2016 at 14:09
-
$\begingroup$ "on the hyperkähler manifold $TS^2$ (or $TRP^2$)" - You are not suggesting that you think the real projective plane is diffeomorphic to $S^2$, are you? $\endgroup$– QfwfqCommented Aug 21, 2016 at 22:34
1 Answer
I will explain how you can get the formulae for the closed $2$-forms that define the hyperKähler structure without introducing coordinates. Once you know the $2$-forms, you can find the $I$, $J$, and $K$ by well-known formulae.
Start with the fact that we can think of $TS^2$ as the space of pairs $(a,b)$ where $a$ and $b$ are vectors in $\mathbb{R}^3$ satisfying $a\cdot a = 1$ and $a\cdot b = 0$. (This embeds $TS^2$ as a submanifold of $\mathbb{R}^6 = \mathbb{R^3}\times\mathbb{R}^3$.) Let $u,v:TS^2\to\mathbb{R}^3$ be the $\mathbb{R}^3$-valued functions that satisfy $u(a,b) = a$ and $v(a,b) = b$. Evidently, they are smooth functions on $TS^2$ and satisfy $u\cdot u = 1 $ and $u\cdot v = 0$.
I'm going to be expressing things in terms of $1$-forms and $2$-forms, so it will be useful to write down some explicit $1$-forms in terms of $u$ and $v$ and their differentials. Define $\mathbb{C}^3$-valued $1$-forms on $TS^2$ by $$ \alpha = \mathrm{d}u - i\,(u\times \mathrm{d}u) $$ and $$ \beta = u\times\mathrm{d}v + i\,\bigl(u\times(u\times\mathrm{d}v)\bigr). $$ When you introduce your favorite coordinates on $TS^2$, you'll have $u$ and $v$ expressed in terms of those coordinates, so the above formulae will give you $\alpha$ and $\beta$ in terms of those coordinates as well. Since everything I write below will be expressed in terms of these forms, you should have no trouble expanding these quantities in terms of your favorite coordinates, though I admit that the answers might be tedious and unenlightening.
One can now compute that $\Upsilon = \mathrm{d}v\,\,{\hat\cdot}\,\,\alpha$ is a nowhere-vanishing $\mathbb{C}$-valued $2$-form on $TS^2$ that is closed and satisfies $\Upsilon\wedge\Upsilon = 0$. Thus, it is the holomorphic volume form for a unique complex structure on $TS^2$. [Notation: If $\phi$ and $\psi$ are $\mathbb{C}^3$-valued $1$-forms on a manifold $M$, then, the $2$-form $\phi\,\,{\hat\cdot}\,\,\psi$ is defined by the equation $$ (\phi\,\,{\hat\cdot}\,\,\psi)(x,y) = \phi(x)\cdot \psi(y) - \phi(y)\cdot\psi(x). $$ for all $x,y\in T_pM$.]
Now, let $r = v\cdot v\ge 0$, and consider the (real-valued) $2$-form $$ \omega = \frac{i}{\sqrt{1+4r}}\,\,\beta\,\,{\hat\cdot}\,\,\bar\beta + \frac{i\sqrt{1+4r}}{4}\,\,\alpha\,\,{\hat\cdot}\,\,\bar\alpha\,. $$
It is now not difficult to show that $\omega$ is closed, of type $(1,1)$, and positive with respect to the complex structure defined by $\Upsilon$. Moreover, $\Upsilon$ has (constant) unit volume with respect to the Kähler metric $g$ defined by $\omega$. In fact, $\omega^2 = 2\,\Upsilon\wedge\overline{\Upsilon}$. Thus, $\Upsilon$ is parallel with respect to $g$, which implies that the metric $g$ is hyperKähler.
-
$\begingroup$ Many thanks! Is it true that ${\mathbb C}^3$ on the line +9 should be replaced by ${\mathbb C}$? $\endgroup$– sgvlCommented Aug 22, 2016 at 11:09
-
$\begingroup$ @sgvl: No. $\alpha$ and $\beta$ are vector-valued $1$-forms because $u$ and $v$ are vector-valued functions. (Otherwise, what could $u\times \mathrm{d}u$ mean?) $\endgroup$ Commented Aug 22, 2016 at 11:24
-