Skip to content

input.md

Latest commit

 

History

History
54 lines (30 loc) · 3.09 KB

input.md

File metadata and controls

54 lines (30 loc) · 3.09 KB
Raw

Section: Zeros of polynomial over the complex field and applications

1. Fundamental Theorem of Algebra: Every non-constant polynomial with complex coefficients has at least one complex root in $\mathbb{C}$.

Definitions and Concepts

  1. Complex Numbers ($\mathbb{C}$): The field of complex numbers, algebraically closed and the natural setting for the complete root theory of polynomials.

  2. Polynomial Ring $\mathbb{C}[z]$: The ring of polynomials in $z$ with complex coefficients. A non-constant polynomial has degree $n \ge 1$.

  3. Root (or Zero): $z_0 \in \mathbb{C}$ is a root of $P(z) \in \mathbb{C}[z]$ if $P(z_0)=0$. By the Factor Theorem, $(z-z_0)$ divides $P(z)$.

  4. Algebraically Closed Field: A field $F$ where every non-constant polynomial in $F[x]$ has a root in $F$. The FTA states: $\mathbb{C}$ is algebraically closed.

Lemmas

  • Lemma 1 (Liouville): Every bounded entire function on $\mathbb{C}$ is constant. If $P(z)$ has no roots, $1/P(z)$ is entire and bounded, hence constant — contradiction unless $P$ is constant.

  • Lemma 2 (Rouché / Winding Number): For large $|z|=R$, $P(z)$ is dominated by its leading term. The image winds around the origin $n$ times. No roots would force winding number $0$ — contradiction.

  • Lemma 3 (Galois–Sylow): The FTA requires topological completeness of $\mathbb{R}$. Using that every odd-degree real polynomial has a real root and every complex quadratic has a root, Sylow theory on Galois groups shows $\mathbb{C}$ has no nontrivial finite algebraic extensions.

Generalization and Intuition

  • Complete Factorization: Inductively, $P(z) = c(z-r_1)\cdots(z-r_n)$ over $\mathbb{C}$.

  • Unity of Disciplines: The FTA shows that an algebraic question (existence of roots) depends on topological/analytic properties of $\mathbb{R}$ and $\mathbb{C}$.

Related Open Problems

  • Smale’s 9th Problem: Existence of a deterministic algorithm approximating roots of polynomial systems in polynomial time on average over $\mathbb{C}$.

  • Random Polynomials (Kac): Statistical distribution of roots of random coefficient polynomials as degree $\to \infty$; connections to random matrix theory and quantum chaos.

Bridge Toward Neural Network Functions

For polynomials, zero-counting is naturally studied over $\mathbb{C}$.
We consider analytically extendable activations. A natural candidate is the sigmoid function

[ \sigma(z) = \frac{1}{1+e^{-z}}, ]

which extends meromorphically to $\mathbb{C}$ with simple poles at $z = i\pi(2k+1),,k\in\mathbb{Z}$.

We study functions $f: \mathbb{C} \to \mathbb{C}$ constructed from finite compositions and affine combinations of the complex sigmoid $\sigma(z)$, i.e., functions belonging to the algebra generated by $\sigma$ and polynomials, or more general sigmoid-based expressions.

Task: Generate Novel Mathematical Problems by Analogy with Polynomials

Core task:
Exploit the problem about functions arising from complex sigmoid networks exhibit a structural property analogous to the Fundamental Theorem of Algebra.