3
$\begingroup$

Let $p:\mathbb{C}^{n+1} \setminus \{0\} \to \mathbb{P}^n$ be the natural projection. Prove that there is no holomorphic map $s: \mathbb{P}^n \to \mathbb{C}^{n+1} \setminus \{0\}$ with $p \circ s = \operatorname{id}$.

I'm seeking for intuition for the problem. I started to think about what the map $s$ would have to be in order for this to hold and for example if $s([z_0:\dots:z_n]) = (z_0,\dots,z_n)$, the the map would give the identity, but this $s$ isn't even well-defined let alone holomorphic since $[z_0:\dots:z_n]=[\lambda z_0:\dots:\lambda z_n]$ for any non-zero $\lambda$, but clearly $(z_0,\dots,z_n) \ne (\lambda z_0,\dots, \lambda z_n) $ unless $\lambda = 1$.

Now my understanding of the projective space is very limited so I don't have any intuition as to why such map wouldn't exist and I would appreciate if anyone could shed some light in to why this shouldn't be true?

$\endgroup$
9
  • $\begingroup$ Do you have knowledge of algebraic topology/homological algebra? If so, you could show that the existence of such a map (which would be an isomorphism between $\mathbb{C}\backslash\{0\}$ and $\mathbb{P}^n$ would have to induce an isomorphism on homology groups (or fundamental groups, based on your knowledge)), from which you could derive a contradiction. It is a quick way of proving this, though admittedly not the most elementary one $\endgroup$
    – Azur
    Commented Sep 25, 2023 at 11:32
  • $\begingroup$ I have enough to see that this would be quite a nice solution, but I would be happy to get more intuition for the projective space rather than use the machinery from algebraic topology. Thanks for the nice idea though! @Azur $\endgroup$ Commented Sep 25, 2023 at 11:36
  • 4
    $\begingroup$ Recall that the only holomorphic maps $\Bbb P^n\to \Bbb C$ are constants. $\endgroup$ Commented Sep 26, 2023 at 23:29
  • $\begingroup$ @TedShifrin So $p \circ z_j : \Bbb P^n\to \Bbb C$ is constant for each of the coordinate functions $z_j$. Will this automatically imply that $s$ is itself constant also? $\endgroup$ Commented Sep 27, 2023 at 6:43
  • $\begingroup$ You mean $z_j\circ s$? $\endgroup$ Commented Sep 27, 2023 at 14:42

1 Answer 1

1
$\begingroup$

Since $\mathbb{P}^n$ is compact, $s(\mathbb{P}^n)$ is compact and thus closed and bounded. But that means there is a $p \in \mathbb{P}^n$ with $|s(p)|$ maximized. Pick an unitary matrix $M$ that takes $s(p)$ to $(a, 0, \ldots 0)$ where $a = |s(p)|$. Then for $g = Ms$, $|g|$ is also maximized at $p$, but now $g(p) = (a, 0, .., 0)$. This is clearly not maximal if $g_1$ is non-constant since $g_1$ would then be an open mapping. But if $g_1$ is constant, then this $|g(p)|$ can only be maximal if all the $g_i$ are constant, but then $s$ would be constant, contradicting $p \circ s = \mbox{id}$.

$\endgroup$
1
  • $\begingroup$ Oh right, that's much simpler. $\endgroup$ Commented Sep 27, 2023 at 2:06

You must log in to answer this question.

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