---
title: Blaschke Winding Number Theory
url: https://www.emergentmind.com/topics/blaschke-winding-number
type: topic
---

# Blaschke Winding Number Theory

Searching arXiv for the cited papers to ground the article.
{"query":"2603.07258 Blaschke products and unwinding in higher dimensions", "max_results": 5}
{"query":"1808.00997 Non-integer valued winding numbers and a generalized Residue Theorem", "max_results": 5}
{"query":"1202.1669 The winding number of Pf+1 for polynomials P and meromorphic extendibility of F", "max_results": 5}
Blaschke winding number most naturally refers to the classical winding number carried by a Blaschke product on the unit circle: for a finite Blaschke product on the unit disk \(D\), the total change of argument along \(\partial D\) equals the number of zeros in \(D\), counted with multiplicity. In one complex variable this is simultaneously a topological degree, an argument-principle count, and the basic integer attached to inner factors. In several variables the situation is less direct: rational inner functions on the polydisk still admit a structured factorization and support convergence and unwinding theories, but the higher-dimensional framework does not define a global winding number on \(T^d\), relying instead on multi-degree data and one-dimensional slices [2603.07258]. Separately, a generalized non-integer winding number defined by Cauchy principal value extends contour index theory to points lying on the curve itself, though that work does not use the terminology “Blaschke winding number” [1808.00997].

## 1. Classical one-variable definition

In the unit disk \(D=\{z\in\mathbb C:|z|<1\}\), a finite Blaschke product has the form
\[
B(z)=z^k\prod_{j=1}^N e^{i\phi_j}\frac{z-a_j}{1-\overline{a_j}z},
\]
where \(k\ge 0\), \(|a_j|<1\), and the unimodular constants \(e^{i\phi_j}\) ensure \(|B(e^{it})|=1\) on \(\partial D\). In this setting, any rational inner function on \(D\) is a finite Blaschke product [2603.07258].

For such a product, with zeros \(a_j\) in \(D\) counted with multiplicity \(N\), the winding number around the origin is
\[
\operatorname{wind}(B;0)=\frac{1}{2\pi i}\oint_{|z|=1}\frac{B'(z)}{B(z)}\,dz=N,
\]
equivalently,
\[
\operatorname{wind}(B;0)=\frac{1}{2\pi}\int_0^{2\pi} d\,\arg B(e^{it})\,dt=N.
\]
These are the standard one-dimensional formulas given by the argument principle. Each elementary Blaschke factor
\[
\frac{z-a}{1-\overline a z}
\]
contributes one unit of winding.

This count is the core reason the term is associated with Blaschke products. The winding number is not merely a boundary invariant; it is exactly the interior zero count for inner rational functions on \(D\). In this sense, the Blaschke winding number is the degree of the boundary map \(\partial D\to T\), and its integrality reflects the discrete multiplicity structure of zeros in one complex variable.

## 2. Infinite products, unwinding, and Malmquist–Takenaka theory in \(D\)

An infinite Blaschke product is formed from infinitely many elementary Blaschke factors. Its convergence is governed by the classical Blaschke condition
\[
\sum_k (1-|a_k|)<\infty.
\]
Under this hypothesis, the product converges uniformly on compact subsets of \(D\) to an inner function; otherwise the product diverges to \(0\) in \(D\) [2603.07258].

The same one-variable framework supports unwinding expansions in Hardy space. The paper recalls expansions of the form
\[
f(z)=\sum_{n\ge 0} c_n \prod_{1\le j<n} B_j(z),
\]
where each \(B_j\) is a finite Blaschke product. In one dimension, this can be implemented by inner–outer factorization to remove zeros step by step. The relation to winding is immediate: in the classical disk theory, the successive extraction of Blaschke factors removes the zero structure that the winding number counts.

The Malmquist–Takenaka construction makes this precise at the level of orthogonal decomposition. For a single Möbius factor
\[
B_a(z)=\frac{z-a}{1-\overline a z},
\]
the space \(H^2(D)\ominus B_aH^2(D)\) is one-dimensional, with orthonormal vector
\[
\frac{\sqrt{1-|a|^2}}{1-\overline a z}.
\]
The projection of \(f\) onto this defect space is
\[
\frac{1-|a|^2}{1-\overline a z}\,f(a).
\]

This one-dimensionality is decisive. In the classical theory, one factor corresponds to one scalar degree of freedom and one unit of winding. That exact correspondence is what later fails in several variables.

## 3. Rational inner functions and Blaschke-type convergence on the polydisk

On the polydisk \(D^d\), the distinguished boundary \(\partial^\ast D^d\) is identified with \(T^d\), and the Hardy space is \(H^2(D^d)\), written \(H^d\). For a polynomial \(P\in\mathbb C[z_1,\dots,z_d]\) of multi-degree \(\operatorname{dgr}P=(\deg_1P,\dots,\deg_dP)\), with no zeros in \(D^d\cup\partial^\ast D^d\) and no common factor with \(P^\ast\), one defines
\[
P^\ast(z)=z^{\operatorname{dgr}P}\,\overline P(z_1^{-1},\dots,z_d^{-1}).
\]
The class of such polynomials is denoted \(\mathcal P_d\). Under these hypotheses, \(P^\ast/P\) is an inner function on \(D^d\). The general structure theorem states that any rational inner function on \(D^d\) is of the form
\[
a z^m P^\ast(z)P(z)^{-1},
\]
with \(a\in\mathbb C\) unimodular, \(m\in\mathbb N^d\), and \(P\in\mathcal P_d\) [2603.07258].

The higher-dimensional analogue of the Blaschke condition is expressed through
\[
\alpha(P)=\frac{P^\ast(0)}{P(0)},\qquad
\beta(P)=
\begin{cases}
\overline{\alpha(P)}/|\alpha(P)|,&\alpha(P)\neq 0,\\
1,&\alpha(P)=0.
\end{cases}
\]
For \(P\in\mathcal P_d\), the diagonal slice
\[
w\longmapsto \frac{P^\ast(w,\dots,w)}{P(w,\dots,w)}
\]
is inner on \(D\), has value \(\alpha(P)\) at \(0\), and therefore satisfies \(|\alpha(P)|<1\).

The main convergence theorem is a precise several-variable counterpart of the classical Blaschke criterion:
\[
\prod_{n\ge 1}\beta(P_n)\frac{P_n^\ast}{P_n}
\quad\text{converges uniformly on compact subsets of }D^d
\]
if and only if
\[
\sum_{n\ge 1}(1-|\alpha(P_n)|)<\infty.
\]
The complementary theorem states that if
\[
\sum_{n\ge 1}(1-|\alpha(P_n)|)=\infty,
\]
then
\[
\bigcap_n B_nH^d=\{0\},
\qquad
B_n=\prod_{1\le j<n}\frac{P_j^\ast}{P_j}.
\]

The proofs proceed by slicing: the several-variable problem is reduced to one-dimensional inner functions on suitable slices, including diagonal slices and fixed-boundary slices. This suggests that the higher-dimensional theory retains the analytic mechanism of Blaschke products without retaining a single global winding number comparable to the degree of a map \(\partial D\to T\).

## 4. Orthogonality, projections, and unwinding beyond one dimension

The polydisk theory develops several unwinding procedures. If \(P_n\in\mathcal P_d\) and
\[
\sum (1-|\alpha(P_n)|)=\infty,
\]
set \(B_n=P_n^\ast/P_n\). For \(f\in H^d\), define recursively
\[
f_0=f,\qquad
g_n=\operatorname{Proj}_{(B_{n+1}H^d)^\perp}(f_n),\qquad
f_{n+1}=\frac{f_n-g_n}{B_{n+1}}.
\]
Then
\[
f=\sum_{n\ge 0} g_n\prod_{1\le j\le n} B_j,
\]
with mutually orthogonal terms in \(H^d\). The divergent-product condition forces the tail to vanish in \(H^d\), so the expansion converges [2603.07258].

The same paper gives an adaptive unwinding based on greedy projection over a compact set \(\mathcal K\subset\mathcal P_d\), and a less greedy unwinding based on a Malmquist–Takenaka-type orthonormal system. The orthogonality mechanism is encoded by the statement: for \(P\in\mathcal P_d\), \(f/P\) is orthogonal to \((P^\ast/P)H^d\) if and only if \(\hat f(j)=0\) for all \(j\succeq \operatorname{dgr}P\). The associated orthonormal system is
\[
\left(\frac{\nu(P_n)}{P_n}\right)\prod_{1\le j<n}\frac{P_j^\ast}{P_j},
\qquad n\ge 1,
\]
where \(\nu(P)=\|P^{-1}\|^{-1}\).

A decisive difference from the disk appears here: when \(d>1\), the spaces
\[
B_nH^d\ominus B_{n+1}H^d
\]
are not finite-dimensional. Consequently, the several-variable Malmquist–Takenaka systems cannot be bases in the one-dimensional sense, and one factor no longer corresponds to one unit of winding.

The projection theory is explicit. If \(B\) is inner on \(D^d\), the orthogonal projection onto \(H^d\ominus BH^d\) has kernel
\[
K(z,\zeta)=\frac{1-B(z)\overline{B(\zeta)}}{\prod_{j=1}^d 2\pi(1-z_j\overline{\zeta_j})},
\]
so that
\[
\operatorname{Proj}_{(BH^d)^\perp}f(z)=\int_{T^d}K(z,e^{i\theta})f(e^{i\theta})\,d\theta.
\]
For tensor products of Möbius factors, the paper also gives explicit finite formulas for the corresponding projections.

The representation
\[
a z^m P^\ast P^{-1}
\]
isolates a monomial \(z^m\), and along a loop varying only \(z_j\) with the other variables fixed on \(\partial D\), this monomial contributes a winding \(m_j\) by the one-dimensional argument principle. The paper does not promote this to a general global index on \(T^d\). For completeness, one may consider a partial index in the \(j\)-th coordinate by integrating \(\partial_{\theta_j}\arg B\) with the other boundary variables fixed, but the paper does not adopt or develop this notion.

## 5. Generalized non-integer winding numbers and principal-value index theory

A distinct generalization of winding theory defines the index for piecewise \(C^1\) cycles even when the evaluation point lies on the curve. For a closed piecewise \(C^1\) curve avoiding \(a\), the classical winding number is
\[
\operatorname{Ind}_\gamma(a)=\frac{1}{2\pi i}\int_\gamma \frac{d\zeta}{\zeta-a}.
\]
If the point \(z_0\) may lie on the cycle \(C\), the generalized winding number is defined by Cauchy principal value:
\[
\operatorname{Ind}^{\mathrm{PV}}_C(z_0)
=
\frac{1}{2\pi i}\,\operatorname{PV}\!\int_C \frac{dz}{z-z_0}
=
\lim_{\varepsilon\to 0}\frac{1}{2\pi i}\int_{|z-z_0|>\varepsilon}\frac{dz}{z-z_0}.
\]
This produces non-integer values determined by the local geometry of the contact [1808.00997].

For a closed piecewise \(C^1\) immersion passing through \(z_0\), if the oriented angle between the incoming tangent at \(z_0\) and the negative of the outgoing tangent is \(\alpha\), then
\[
\operatorname{Ind}^{\mathrm{PV}}_\Gamma(z_0)=\frac{\alpha}{2\pi}.
\]
Thus a transverse crossing with \(\alpha=\pi\) contributes \(\pm 1/2\), while a tangential touch with \(\alpha=0\) contributes \(0\). With multiple contacts, the generalized index is the ordinary index of the part of the curve avoiding \(z_0\) plus the sum of the local contributions \(\alpha_j/(2\pi)\).

The paper also gives a real version with bounded integrand. For \(\Lambda(t)=x(t)+iy(t)\),
\[
\operatorname{Ind}^{\mathrm{PV}}_\Lambda(0)
=
\frac{1}{2\pi}\int_a^b
\frac{x(t)\dot y(t)-y(t)\dot x(t)}{x(t)^2+y(t)^2}\,dt.
\]
If \(\Lambda\) is a closed piecewise \(C^{1,1}\) immersion, this integrand is bounded. If \(\Lambda\) is \(C^2\) near a point \(\bar t\) with \(\Lambda(\bar t)=0\), its limit is \(K_\Lambda(\bar t)\|\dot\Lambda(\bar t)\|\), where \(K_\Lambda\) is the signed curvature.

This index underlies a generalized residue theorem. If \(f\) is holomorphic off a discrete set \(S\subset U\), and \(C\) is a null-homologous immersed piecewise \(C^1\) cycle containing only first-order poles of \(f\) on the cycle, then
\[
\operatorname{PV}\!\int_C f(z)\,dz
=
2\pi i\sum_{z\in S} n_z(C)\operatorname{Res}(f,z),
\]
where \(n_z(C)\) is the generalized winding number. The paper further describes when higher-order poles or essential singularities on the cycle are admissible, using flatness conditions and angle restrictions. Although this theory generalizes winding in a manner historically reminiscent of Blaschke-type index ideas, the paper explicitly does not use the terminology “Blaschke winding number” or “Blaschke’s index.”

## 6. Winding inequalities, Blaschke counting, and terminological scope

A third line of work studies winding numbers of boundary expressions of the form \(Pf+1\) on the unit circle. For a continuous \(g:bD\to \mathbb C\setminus\{0\}\), the winding number around \(0\) is the total change of argument divided by \(2\pi\), equivalently
\[
\operatorname{Wind}(g,0)=\frac{1}{2\pi i}\int_{|z|=1}\frac{g'(z)}{g(z)}\,dz
\]
when \(g\) is \(C^1\) on \(bD\). The central condition is imposed for every polynomial \(P\) such that \(Pf+1\) has no zero on \(bD\) [1202.1669].

The principal theorem states: if \(f\in C^\infty(bD)\) has at most finitely many zeros on \(bD\), each of finite order, and \(J\in\mathbb N\cup\{0\}\), then
\[
\operatorname{Wind}(Pf+1,0)\ge -J
\]
for each such polynomial \(P\) if and only if \(f\) extends meromorphically through \(D\) with at most \(J\) poles in \(D\), counted with multiplicity. In particular, for \(J=0\), the condition is equivalent to holomorphic extendibility in this class. The paper also proves the corresponding statement for real-analytic boundary data.

The connection with Blaschke products is explicit. For a finite Blaschke product
\[
B(z)=e^{i\phi}\prod_{k=1}^n\frac{z-a_k}{1-\overline{a_k}z},
\qquad |a_k|<1,
\]
the argument principle gives
\[
\operatorname{Wind}(B,0)=Z(B)=n.
\]
From the inner–outer viewpoint, the inner factor contributes exactly the zero count. The winding inequalities for \(Pf+1\) therefore emulate the same mechanism: they constrain the allowable interior zeros and poles of any candidate extension by the same counting principle that governs finite Blaschke factors.

The paper also identifies a limitation. If one replaces the constant \(1\) by a fixed factor \(R\) having zeros in \(D\), the analogous criterion can fail. The example
\[
f(z)=\frac{z}{z-1/2}\quad\text{on }bD,\qquad R(z)=z
\]
shows that one may have \(\operatorname{Wind}(f+zp,0)\ge 0\) for all relevant polynomials \(p\) even though \(f\) does not extend holomorphically through \(D\).

Taken together, these strands delimit the scope of the expression “Blaschke winding number.” In the strict classical sense it is the integer degree of a finite Blaschke product on \(\partial D\). In generalized contour theory, winding can become non-integer when the point lies on the curve. In several-variable unwinding on the polydisk, the role of winding is partially replaced by \(\operatorname{dgr}P\), the monomial factor \(z^m\), the parameters \(\alpha(P)\), and slice-wise one-dimensional reductions, rather than by a single global topological index.

Source: https://www.emergentmind.com/topics/blaschke-winding-number