6
$\begingroup$

Put $$A=\{(a_{0},a_{1},\ldots,a_{n}) \in \mathbb{C}^{n+1}\mid p(z)=a_{0}+a_{1}z+\ldots a_{n}z^{n} \;\;\text{is a one-to one function on the unit disc} \{z\in \mathbb{C} \mid |z|\leq 1\}$$

Is $\{(|a_{0}|,|a_{1}|,\ldots,|a_{n}|)\mid (a_{i})\in A\}$ a semi algebraic set in $\mathbb{R}^{n+1}$?

Of course this is true for $n=2$

$\endgroup$

1 Answer 1

16
$\begingroup$

It's fairly immediate from the Tarski-Seidenberg theorem that it's semi-algebraic. More exactly, if you can define your set in the language of real closed fields using first-order logic, then the Tarski-Seidenberg theorem guarantees that the set is semi-algebraic.

As a first step, the set $A$ is semi-algebraic, as it is defined by a formula $\forall_{z, w} [z \bar{z} \leq 1 \wedge w \bar{w} \leq 1 \wedge a_0 + a_1 z + \ldots + a_n z^n = a_0 + a_1 w +\ldots + a_n w^n] \Rightarrow z = w$. Abbreviating this as $\forall_{z, w} [|z| \leq 1 \wedge |w| \leq 1 \wedge p_a(z) = p_a(w)] \Rightarrow z = w$, this is equivalent to

$$\neg \exists_{z, w} (|z| \leq 1 \wedge |w| \leq 1 \wedge p_a(z) = p_a(w)) \wedge \neg (z = w)$$

where the subset $\{(z, w, a) \in \mathbb{C}^{n+3}: |z| \leq 1 \wedge |w| \leq 1 \wedge p_a(z) = p_a(w) \wedge z \neq w\}$ (before quantification) is clearly semi-algebraic. Existentially quantifying over $z, w$ corresponds to taking the image of that subset under the projection $(z, w, a) \mapsto a$; this image is semi-algebraic by Tarski-Seidenberg, and then the complement of that is also semi-algebraic.

Finally, your set is

$$\{(r_0, \ldots, r_n): \exists_{a \in \mathbb{C}^{n+1}} a \in A \wedge r_0^2 = a_0 \bar{a_0} \wedge \ldots \wedge r_n^2 = a_n \bar{a_n} \wedge r_0 \geq 0 \wedge \ldots \wedge r_n \geq 0\}$$

where, just as before (taking an image of a semi-algebraic set, defined in abbreviated form by $a \in A \wedge r^2 = a\bar{a} \wedge r \geq 0$, under the projection $(a, r) \mapsto r$), we easily conclude this is semi-algebraic.

$\endgroup$
3
  • $\begingroup$ Thank you very much for your answer.May you elaborate on your véry elegant answer in particular on the logical part? $\endgroup$ Jun 8, 2016 at 19:52
  • 1
    $\begingroup$ Let me know if you're asking about something else, but essentially I was converting a formula of the form $\forall_{z, w} A(z, w) \Rightarrow B(z, w)$ into $\neg \exists_{z, w} \neg (A(z, w) \Rightarrow B(z, w))$ (i.e., using $\forall = \neg \exists \neg$), and then converting $\neg (A \Rightarrow B)$ into $\neg (\neg A \vee B)$, and that into $A \wedge \neg B$. The reason for those manipulations is that Tarski-Seidenberg refers to taking images, directly tied to existential quantification since (for $\pi: (x, y) \mapsto y$) the image $\pi(A)$ is exactly $\{y: \exists_x (x, y) \in A\}$. $\endgroup$
    – Todd Trimble
    Jun 8, 2016 at 20:19
  • 2
    $\begingroup$ Quantification is all we have to worry about, as semi-algebraic sets are closed under the other logical operations $\neg, \wedge, \vee$, practically by definition. By the way, I first learned of all this by reading the book Tame Topology and O-minimal Structures by van den Dries, which I recommend as it is written for mathematicians with varying backgrounds (not including a background in logic particularly, besides what everyone knows). $\endgroup$
    – Todd Trimble
    Jun 8, 2016 at 20:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.