---
title: 'Theta Partial Operator: Theory & Applications'
url: https://www.emergentmind.com/topics/theta-partial-operator
type: topic
---

# Theta Partial Operator: Theory & Applications

Searching arXiv for the cited papers to ground the article in current records.
The expression **Theta Partial Operator** has two closely related but distinct meanings in the literature. In the strict operator-theoretic sense, it denotes the operator \(\theta(y\mathbf D_{\lambda})\), introduced from the \(\lambda\)-derivative operator \(\mathbf D_\lambda\) and used to generate generalized Lambert series and multivariate analogues [2507.15160]. In the older and larger partial-theta literature, the central object is instead the **partial theta function**
\[
\theta(q,x)=\sum_{j=0}^{\infty}q^{j(j+1)/2}x^j,
\]
studied as an entire function of \(x\) for fixed \(|q|<1\); there, “spectrum” refers not to an operator spectrum in the linear-algebraic sense but to the exceptional parameter values \(q\) for which \(\theta(q,\cdot)\) acquires a multiple zero [1504.00885]. The topic therefore spans both a recent operator construction and a mature analytic theory of partial theta series, zeros, spectral values, modularity defects, and \(q\)-series transformations.

## 1. Terminology and defining objects

The classical partial theta function is the one-sided series
\[
\theta(q,x)=\sum_{j=0}^{\infty} q^{j(j+1)/2}x^j,
\]
called “partial” because it resembles the Jacobi theta function but retains only the nonnegative indices. For each fixed \(q\) with \(|q|<1\), \(\theta(q,\cdot)\) is an entire function of \(x\). In this setting, the **spectrum** is the set of those values of the parameter \(q\) for which \(\theta(q,\cdot)\) has a multiple zero; equivalently, it is the set of exceptional parameter values where the zero set becomes degenerate [1504.00885].

By contrast, the 2025 operator-theoretic usage defines the **Theta Partial operator**
\[
\theta(y\mathbf D_{\lambda}),
\]
from the \(\lambda\)-derivative operator \(\mathbf D_\lambda\). The paper introducing it states that the operator is built so that acting on monomials \(x^n\) produces the kernel
\[
\frac{x^{n+1}}{x-\lambda^n y},
\]
which is then used as the basic building block for generalized Lambert series [2507.15160].

This terminological bifurcation is important. Several papers explicitly note that earlier studies of partial theta functions are not about an operator in the usual sense, whereas the later operator paper introduces a genuine formal operator series. A plausible implication is that the phrase “Theta Partial Operator” should be interpreted contextually: in analytic-function theory it usually points back to the partial theta function itself, while in the recent Lambert-series literature it refers to \(\theta(y\mathbf D_{\lambda})\).

## 2. Classical partial theta function and the spectral problem

For \(q\in(0,1)\), the partial theta function has no positive real zeros and has infinitely many negative real zeros. A central functional identity is
\[
\theta(q,x)=1+qx\,\theta(q,qx). \tag{FE}
\]
When all zeros are simple and negative, the function admits the factorization
\[
\theta(q,x)=\prod_{j=1}^{\infty}\left(1-\frac{x}{\xi_j}\right). \tag{product}
\]
The real-zero geometry is organized by a sequence of **spectral values**
\[
0.3092493386\ldots=\tilde q_1<\tilde q_2<\cdots,\qquad \tilde q_j\to 1^-,
\]
which are exactly the values of \(q\) for which \(\theta(q,\cdot)\) has a multiple real zero [1504.05786].

At \(q=\tilde q_j\), the function \(\theta(\tilde q_j,\cdot)\) has a double zero \(y_j\), this double zero is negative, it is the rightmost real zero, and all the other real zeros are simple. For \(q\neq \tilde q_j\), there are no multiple real zeros. The asymptotic behavior of these spectral values and double zeros is
\[
\tilde q_j = 1-\frac{\pi}{2j}+\frac{\log j}{8j^2}+O\!\left(\frac{1}{j^2}\right),
\qquad
y_j=-e^{\pi}e^{-\frac{\log j}{4j}+O(1/j)}.
\]
The first spectral double zero is
\[
y_1=-7.5032559833\ldots
\]
[1504.05786].

Near \(q=0\), the spectrum is absent: for \(|q|\le 0.108\), the entire function \(\theta(q,\cdot)\) has no multiple zeros. This provides a spectral gap near the origin in parameter space and shows that the zero set is simple in that regime [1504.00885].

## 3. Zero geometry, confinement, and global structure

A major theme of the theory is the geometric localization of zeros. For every \(q\in(0,1)\), all zeros of \(\theta(q,\cdot)\) lie in the domain
\[
\{\operatorname{Re} z<0,\ |\operatorname{Im} z|\le 132\}\ \cup\ \{\operatorname{Re} z\ge 0,\ |z|\le 18\}.
\]
Equivalently, zeros in the right half-plane must lie in the closed disk \(|z|\le 18\), and zeros in the left half-plane satisfy \(|\operatorname{Im} z|\le 132\) [1710.01575].

Multiple zeros satisfy a much coarser but uniform global bound: for any \(q\in\mathbb D_1=\{q\in\mathbb C:|q|<1\}\), any multiple zero \(\zeta\) of \(\theta(q,\cdot)\) satisfies
\[
|\zeta|\le 8^{11}=8589934592.
\]
The same paper gives a smaller parameter range with no multiple zeros at all: if \(q\in\mathbb D_{c_0}\) with
\[
c_0=0.2078750206\ldots,
\]
then \(\theta(q,\cdot)\) has no multiple zeros [1602.08937].

For complex conjugate pairs of zeros, sharper explicit bounds are known. For \(q\in(0,1)\), all complex conjugate pairs belong to
\[
\{\,\Re x\in(-5792.7,0),\ |\Im x|<132\,\}\ \cup\ \{\,|x|<18\,\},
\]
and for \(q\in(-1,0)\) they belong to
\[
\{\,|\Re x|<364.2,\ |\Im x|<132\,\}.
\]
These bounds formalize the statement that nonreal zeros are confined to fixed explicit regions of the plane [1902.01726].

The zero set
\[
S:=\{(q,z)\in\mathbb D\times\mathbb C:\theta(q,z)=0\}
\]
is a connected analytic subset of \(\mathbb D\times\mathbb C\), and it is smooth at every point \((q^\flat,z^\flat)\) such that \(z^\flat\) is a simple or double zero of \(\theta(q^\flat,\cdot)\). The paper proving this uses the differential equation
\[
2q\frac{\partial\theta}{\partial q}=z\frac{\partial^2}{\partial z^2}(z\theta), \tag{5}
\]
together with monodromy of zero branches around spectral values [1911.08841].

These results fit together with the bifurcation picture: as \(q\) increases through a spectral value, the rightmost two real zeros coalesce into a double zero and then split into a complex conjugate pair. This is the basic mechanism by which nonreal zeros are born in the real-\(q\) theory.

## 4. Asymptotics of zeros and stabilized coefficient structure

For small \(q\), the zeros admit precise Laurent expansions. If \(-\xi_j\) denotes the \(j\)-th zero, then
\[
-\xi_j = -q^{-j} + (-1)^j q^{\,j(j-1)/2} + o\!\left(q^{j(j-1)/2}\right).
\]
Equivalently,
\[
-\xi_j = -q^{-j} + (-1)^j q^{j(j-1)/2} \left(1 + \sum_{k\ge 1} g_{j,k} q^k\right).
\]
A central stabilization theorem states that, for each fixed \(k\), the coefficient \(g_{j,k}\) becomes independent of \(j\) once \(j\) is large enough: there exists a formal power series
\[
1+\sum_{k\ge1} r_k q^k
\]
such that
\[
g_{j,k}=r_k,\qquad k=1,\dots,j,\quad j\ge2.
\]
The stabilized coefficients satisfy
\[
r_k=\sum_{\nu=1}^{\infty}(-1)^{\nu-1}(2\nu+1)\,r_{k-\nu(\nu+1)/2},
\qquad
r_0=1,\qquad r_k=0\quad\text{for }k<0,
\]
and are exactly the coefficients of
\[
\frac{1}{(q)_\infty^3} = \sum_{k\ge0} r_k q^k.
\]
The first values are
\[
r_0=1,\quad r_1=3,\quad r_2=9,\quad r_3=22,\quad r_4=51,\quad r_5=108,\quad r_6=221,\quad r_7=429,\ldots
\]
[1504.00883].

Uniform zero-counting results extend the small-\(|q|\) asymptotics to annuli away from \(0\) and \(1\). For any \(0<\delta_0<\delta<1\), there exists \(n_0\in\mathbb N\) such that for every \(q\) with
\[
\delta_0\le |q|\le \delta,
\]
and every \(n\ge n_0\), the function \(\theta(q,z)\) has exactly \(n\) zeros, counted with multiplicity, in the disk
\[
|z|<|q|^{-n-1/2}.
\]
This theorem is proved by comparison with the product
\[
u(q,z):=\prod_{v=1}^{\infty}(1+q^v z),
\]
whose zeros are exactly at \(-q^{-v}\) [1607.05453].

The asymptotic density of real zeros is also known near the singular limits \(q\to\pm1\). For \(q\to1^-\) and \(a\ge e^\pi\), there are
\[
o\!\left(\frac1{1-q}\right)
\]
real zeros in \([-e^\pi,0)\) and
\[
\frac{\ln(a/e^\pi)}{1-q}+o\!\left(\frac1{1-q}\right)
\]
real zeros in \([-a,-e^\pi]\). For \(q\to-1^+\) and \(a\ge e^{\pi/2}\), the analogous counts are
\[
o\!\left(\frac1{1+q}\right)
\]
in \([-e^{\pi/2},e^{\pi/2}]\) and
\[
\frac{\ln(a/e^{\pi/2})}{2(1+q)}+o\!\left(\frac1{1+q}\right)
\]
in each of the intervals \([-a,-e^{\pi/2}]\) and \([e^{\pi/2},a]\) [1911.08841].

## 5. The Theta Partial operator \(\theta(y\mathbf D_{\lambda})\)

The operator-theoretic construction starts from the \(\lambda\)-derivative
\[
(\mathbf{D}_{\lambda}f)(x)=
\begin{cases}
\dfrac{f(\lambda x)}{x},&\text{ if }x\neq0,\\[4pt]
\lim_{x\rightarrow0}\dfrac{f(\lambda x)}{x},&\text{ if }x=0,
\end{cases}
\]
whenever the limit exists. Among its stated properties are linearity, the action on monomials
\[
\mathbf{D}_{\lambda}\{x^{n}\}=\lambda^{n}x^{n-1},\qquad n\in\mathbb N,
\]
and the product rule
\[
\mathbf{D}_{\lambda}\{f(x)g(x)\}=f(\lambda x)\mathbf{D}_{\lambda}g(x)
=\mathbf{D}_{\lambda}f(x)\cdot g(\lambda x)
=x\mathbf{D}_{\lambda}f(x)\mathbf{D}_{\lambda}g(x).
\]
The operator itself is defined by the formal series
\[
\theta(yD_{\lambda})=\sum_{n=0}^{\infty}\lambda^{\binom{n}{2}}y^nD_{\lambda}^n
\]
[2507.15160].

Its core monomial identity is
\[
\theta(yD_{\lambda})\{x^n\}=\frac{x^{n+1}}{x-\lambda^n y}.
\]
The derivation uses
\[
D_\lambda^k\{x^n\}=\lambda^{nk-\binom{k}{2}}x^{n-k},
\]
so that
\[
\theta(yD_\lambda)\{x^n\}
=\sum_{k=0}^\infty y^k\lambda^{nk}x^{n-k}
= x^n\sum_{k=0}^\infty \left(\frac{\lambda^n y}{x}\right)^k
=\frac{x^{n+1}}{x-\lambda^n y},
\]
assuming \(|\lambda^n y/x|<1\) [2507.15160].

This identity is the mechanism by which the operator converts ordinary generating functions into Lambert-type series. A plausible implication is that \(\theta(y\mathbf D_\lambda)\) plays, for these generalized Lambert kernels, a role analogous to that played by exponential or shift operators in more classical generating-function calculus.

## 6. Generalized Lambert series and multivariate extensions

The basic generalized Lambert series introduced from the operator is
\[
\sum_{n=0}^{\infty}a_{n}\frac{x^{n+1}}{x-\lambda^ny}z^n.
\]
If
\[
f(x)=\sum_{n=0}^{\infty}a_n x^n,
\]
then under the convergence conditions
\[
|y|<|x|<R,\qquad |xz|<R,\qquad |\lambda|<1,
\]
one has the generating-function identity
\[
\sum_{n=0}^{\infty}a_{n}\frac{x^{n+1}}{x-\lambda^ny}z^n
=
\sum_{k=0}^{\infty}\left(\frac{y}{x}\right)^k f(\lambda^k xz).
\]
The paper also defines
\[
L(x,y,z;\lambda)=\sum_{n=0}^{\infty}\frac{x^{n+1}}{x-\lambda^ny}z^n,
\]
and notes that for \((x,y,z,\lambda)=(1,q,q,q)\) or \((q,q^2,1,q)\),
\[
L(q)=qL(1,q,q,q)=qL(q,q^2,1,q),
\]
where
\[
L(q)=\sum_{n=1}^{\infty}\frac{q^n}{1-q^n}
\]
is the classical Lambert function [2507.15160].

The same operator viewpoint yields several higher constructions. The **generalized Lambert-Mehler type series** is
\[
\sum_{n=0}^{\infty}a_{n}\frac{x^{n+1}}{x-\lambda^ny}\cdot\frac{z^{n+1}}{z-\lambda^nw}t^n,
\]
with the identity
\[
\sum_{n=0}^{\infty}a_{n}\frac{x^{n+1}}{x-\lambda^ny}\cdot\frac{z^{n+1}}{z-\lambda^nw}t^n
=
\sum_{k=0}^{\infty}\sum_{n=0}^{\infty}\left(\frac{w}{z}\right)^k\left(\frac{y}{x}\right)^n f(\lambda^{k+n}txz).
\]
The **generalized Lambert-Rogers type series** is
\[
\sum_{n=0}^{\infty}\sum_{m=0}^{\infty}a_{n}b_{m}\frac{x^{n+m+1}}{x-\lambda^{n+m}y}t^ns^m,
\]
and satisfies
\[
\sum_{n=0}^{\infty}\sum_{m=0}^{\infty}a_{n}b_{m}\frac{x^{n+m+1}}{x-\lambda^{n+m}y}t^ns^m
=
\sum_{k=0}^{\infty}\left(\frac{y}{x}\right)^k f(\lambda^k tx)g(\lambda^k sx).
\]
There are also double-sum bivariate generalized Lambert series and multivariate generalized Lambert series of the form
\[
\sum_{n=0}^{\infty}a_{n}\prod_{i=1}^{m}\frac{x_i^{n+1}}{x_i-\lambda_i^ny_i}\,z^n,
\]
with the multivariate identity
\[
\sum_{n=0}^{\infty}a_{n}\prod_{i=1}^{m}\frac{x_i^{n+1}}{x_i-\lambda_i^ny_i}\,z^n
=
\sum_{k_1,k_2,\ldots,k_m=0}^{\infty}
\prod_{i=1}^{m}\left(\frac{y_i}{x_i}\right)^{k_i}
f(\lambda_1^{k_1}\cdots\lambda_m^{k_m}x_1\cdots x_m z)
\]
[2507.15160].

A substantial part of the operator paper is a catalog of explicit specializations obtained from elementary functions:
\[
\frac{1}{1\mp x},\qquad \ln(1\pm x),\qquad e^x,\qquad \cos x,\qquad \sin x.
\]
The same framework is also applied to simplicial polytopic numbers and Lucas sequences. This gives the operator a clearly generative role: it is not merely a formal series, but a device for transporting ordinary generating functions into a structured family of Lambert kernels.

## 7. Related analytic, modular, and \(q\)-series frameworks

The operator viewpoint sits inside a broader ecosystem of partial-theta constructions. One important strand constructs a single analytic function \(F(z)\) that interpolates between a mock theta function in the upper half-plane and a partial theta function in the lower half-plane. In that framework, the contour-integral expression
\[
P_{d,k}(z)=\frac{1}{2\pi i}\int_{|s|=r} \frac{a_k(s)e^{2\pi i z s}}{1-\zeta_{2k}^d q\,e^{2\pi s}}\,ds
\]
is the mechanism that produces one expansion for \(|q|<1\) and another for \(|q|>1\), and the same analytic construction yields Ramanujan’s third-order mock theta function upstairs and a corresponding partial theta function downstairs [1111.1495].

A different strand is algebraic rather than operator-theoretic. The bivariate representation
\[
0(q,a)=L(a,b)+bL(aq,bq)
\]
encodes a one-variable partial theta function through a shifted bivariate object \(L(a,b)\) and unifies identities of Andrews, Warnaar, and Schilling–Warnaar. The same paper derives a general Warnaar-type identity and a \(q\)-series transformation associated with Bailey pairs, emphasizing that there is no formal “partial-theta operator” but that the resulting transformations and recurrences behave in an operator-like fashion [1709.07389].

The identity-theoretic literature extends this picture further. One paper establishes a three-term extension of the Andrews–Warnaar partial theta identity and a two-term companion identity that yields partial and false theta identities and links the theory to big \(q\)-Jacobi polynomials [1907.08353]. Another confirms Wang and Ma’s conjecture by proving that, for every integer \(m\ge2\),
\[
0(q,a) = \sum_{k=0}^{m-1} \frac{(q^{m-k};q)_k\,(bq^{k-m+1};q)_k}{(q;q)_k\,Q_m(bq^{1-m})}\, P_m(aq^k,bq^k),
\]
and also shows that each \(P_m(a,b)\) is expressible as a linear combination of shifted partial theta functions [1805.01268].

There is also a higher-rank modular and representation-theoretic generalization. For a positive definite lattice with Gram matrix \(A\), the higher-rank Jacobi partial theta function
\[
P(u,\tau):=\sum_{k\in (\mathbb Z_{\ge0})^n} q^{\frac12 (k,Ak)}\, e^{2\pi i (u,k)}
\]
is regularized by
\[
P_\varepsilon(u,\tau) :=\sum_{k\in (\mathbb Z_{\ge0})^n} q^{\frac12 (k,Ak)}\, e^{2\pi i (k,u)}\, e^{2\pi i (k,\varepsilon)},
\]
and its modular \(S\)-transformation is described by a Gaussian-contour term \(h_\varepsilon(u,\tau)\) plus explicit contour-integral corrections. In ADE type, this leads to regularized Kostant partial theta functions and to computations of regularized quantum dimensions of modules of narrow logarithmic \(W\)-algebras [1607.08563].

Taken together, these directions show that “Theta Partial Operator” is not an isolated phrase but part of a larger technical landscape. In one direction it names a concrete operator \(\theta(y\mathbf D_\lambda)\) generating generalized Lambert series; in another it designates a family of analytic and algebraic mechanisms—spectral analysis of zeros, contour regularization, Bailey-pair transformations, and higher-rank modular completions—through which partial theta phenomena are organized across \(q\)-series, mock modularity, and representation theory.

Source: https://www.emergentmind.com/topics/theta-partial-operator