The universal family of punctured Riemann surfaces is Stein
Extracted source text · arXiv 2511.07963
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% F. FORSTNERIC
%
% The universal family of punctured Riemann surfaces is Stein
%
% AMS-LaTeX 1.2 file for journals, based on amsart.cls.
%
% http://arxiv.org/abs/2508.13660
%
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% https://www.jmilne.org/not/Mxymatrix.pdf
\documentclass[11pt]{amsart}
\usepackage{amsfonts,amssymb,amscd,amstext,mathrsfs,mathtools}
%\usepackage[utf8]{inputenc}
\usepackage{hyperref}
\usepackage{verbatim}
\usepackage{t1enc}
\usepackage{graphics}
%\usepackage{multimedia}
%\usepackage{natbib}
\usepackage{graphicx}
\usepackage{times}
\usepackage{enumerate}
%\usepackage{titlesec}
\usepackage[up,bf]{caption}
\usepackage{color}
\input xy
\xyoption{all}
%TO WRITE TO DO NOTES
%\usepackage[color=blue!20]{todonotes}
\textwidth = 157mm
\textheight = 239mm
\evensidemargin=0mm
\oddsidemargin=0mm
\hoffset=4mm
\voffset=-25mm
\parskip =1mm
\parindent = 6mm
\linespread{1.09}
\pagestyle{plain}
%% To do not cut the words
\pretolerance=2000
\tolerance=3000
\numberwithin{equation}{section}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newtheorem{theorem}{Theorem}[section]
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{claim}[theorem]{Claim}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{corollary}[theorem]{Corollary}
%\newtheorem{remark}[theorem]{Remark}
\theoremstyle{definition}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{question}[theorem]{Question}
\newtheorem{problem}[theorem]{Problem}
\newtheorem{example}[theorem]{Example}
\newtheorem{conjecture}[theorem]{Conjecture}
%
% BOLDFACE CAPITAL
%
\newcommand\bA{\mathbf{A}}
\newcommand\bB{\mathbf{B}}
\newcommand\bC{\mathbf{C}}
\newcommand\bD{\mathbf{D}}
\newcommand\bE{\mathbf{E}}
\newcommand\bH{\mathbf{H}}
\newcommand\bI{\mathbf{I}}
\newcommand\bK{\mathbf{K}}
\newcommand\bL{\mathbf{L}}
\newcommand\bM{\mathbf{M}}
\newcommand\bN{\mathbf{N}}
\newcommand\bO{\mathbf{O}}
\newcommand\bP{\mathbf{P}}
\newcommand\bR{\mathbf{R}}
\newcommand\bS{\mathbf{S}}
\newcommand\bT{\mathbf{T}}
\newcommand\bU{\mathbf{U}}
\newcommand\bV{\boldface{V}}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% CALIGRAPHIC CAPITAL
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newcommand\Acal{\mathcal{A}}
\newcommand\Bcal{\mathcal{B}}
\newcommand\Ccal{\mathcal{C}}
\newcommand\Dcal{\mathcal{D}}
\newcommand\Ecal{\mathcal{E}}
\newcommand\Fcal{\mathcal{F}}
\newcommand\Hcal{\mathcal{H}}
\newcommand\Ical{\mathcal{I}}
\newcommand\Kcal{\mathcal{K}}
\newcommand\Lcal{\mathcal{L}}
\newcommand\Mcal{\mathcal{M}}
\newcommand\Ncal{\mathcal{N}}
\newcommand\Ocal{\mathcal{O}}
\newcommand\Pcal{\mathcal{P}}
\newcommand\Rcal{\mathcal{R}}
\newcommand\Scal{\mathcal{S}}
\newcommand\Tcal{\mathcal{T}}
\newcommand\Ucal{\mathcal{U}}
\newcommand\Vcal{\mathcal{V}}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% SMALL BOLDFACE
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newcommand\ba{\mathbf{a}}
\newcommand\be{\mathbf{e}}
\newcommand\bu{\mathbf{u}}
\newcommand\bv{\mathbf{v}}
\newcommand\bx{\mathbf{x}}
\newcommand\by{\mathbf{y}}
\newcommand\bw{\mathbf{w}}
\newcommand\bz{\mathbf{z}}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% MATH SCRIPT
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newcommand\Ascr{\mathscr{A}}
\newcommand\Bscr{\mathscr{B}}
\newcommand\Cscr{\mathscr{C}}
\newcommand\Dscr{\mathscr{D}}
\newcommand\Fscr{\mathscr{F}}
\newcommand\Iscr{\mathscr{I}}
\newcommand\Jscr{\mathscr{J}}
\newcommand\Lscr{\mathscr{L}}
\newcommand\Oscr{\mathscr{O}}
\newcommand\Sscr{\mathscr{S}}
\newcommand\Uscr{\mathscr{U}}
\newcommand\Gscr{\mathscr{G}}
\newcommand\Yscr{\mathscr{Y}}
\newcommand\Zscr{\mathscr{Z}}
\newcommand\Pscr{\mathscr{P}}
\newcommand\Qscr{\mathscr{Q}}
\newcommand\Vscr{\mathscr{V}}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% MATH BLACKBOARD
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newcommand\A{\mathbb{A}}
\newcommand\B{\mathbb{B}}
\newcommand\C{\mathbb{C}}
\newcommand\D{\overline{\mathbb D}}
\newcommand\CP{\mathbb{CP}}
\renewcommand\D{\mathbb D}
\newcommand\G{\mathbb{G}}
\renewcommand\H{\mathbb{H}}
\newcommand\K{\mathbb{K}}
\newcommand\M{\mathbb{M}}
\newcommand\N{\mathbb{N}}
\renewcommand\P{\mathbb{P}}
\newcommand\R{\mathbb{R}}
\newcommand\RP{\mathbb{RP}}
\renewcommand\S{\mathbb{S}}
\newcommand\T{\mathbb{T}}
\newcommand\U{\mathbb{U}}
\newcommand\Z{\mathbb{Z}}
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% FRAKTUR
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\newcommand\ggot{\mathfrak{g}}
\newcommand\hgot{\mathfrak{h}}
\newcommand\igot{\mathfrak{i}}
\newcommand\jgot{\mathfrak{j}}
\newcommand\kgot{\mathfrak{k}}
\renewcommand\igot{\mathfrak{i}}
\newcommand\mgot{\mathfrak{m}}
\newcommand\pgot{\mathfrak{p}}
\newcommand\qgot{\mathfrak{q}}
\newcommand\fgot{\mathfrak{f}}
\newcommand\Agot{\mathfrak{A}}
\newcommand\Igot{\mathfrak{I}}
\newcommand\Lgot{\mathfrak{L}}
\newcommand\Mgot{\mathfrak{M}}
\newcommand\Ngot{\mathfrak{N}}
\newcommand\Pgot{\mathfrak{P}}
\newcommand\Ogot{\mathfrak{O}}
\newcommand\Ygot{\mathfrak{Y}}
\newcommand\Ggot{\mathfrak{G}}
\newcommand\Hgot{\mathfrak{H}}
\newcommand\Zgot{\mathfrak{Z}}
%
% typewriter
%
\newcommand\jtt{\mathtt{j}}
\newcommand\ktt{\mathtt{k}}
\newcommand\ttI{\mathtt{I}}
%
% e, i, zero - mathmode
%
\newcommand\E{\mathrm{e}}
\newcommand\I{\imath}
\renewcommand\imath{\igot}
\newcommand\zero{\mathbf{0}}
\newcommand\Psf{\mathsf{P}}
%
% arrows
%
\newcommand\hra{\hookrightarrow}
\newcommand\lra{\longrightarrow}
\newcommand\Lra{\Longrightarrow}
\newcommand\longhookrightarrow{\ensuremath{\lhook\joinrel\relbar\joinrel\rightarrow}}
%
% tilde, widehat, di, dibar
%
\newcommand\wt{\widetilde}
\newcommand\wh{\widehat}
\newcommand\di{\partial}
\newcommand\dibar{\overline\partial}
\newcommand{\nad}[2]{\genfrac{}{}{0pt}{}{#1}{#2}}
%
% abbreviations
%
\newcommand\dist{\mathrm{dist}}
%\newcommand\span{\mathrm{span}}
\newcommand\length{\mathrm{length}}
\newcommand\reg{\mathrm{reg}}
\newcommand\Fr{\mathrm{Fr}}
\newcommand\Id{\mathrm{Id}}
\newcommand\pr{\mathrm{pr}}
\def\Ell1{\mathrm{Ell_1}}
\def\CEll1{\mathrm{CEll_1}}
\def\CAP{\rm{CAP}}
\def\rmax{\mathrm{rmax}}
\def\cB{\overline{\B}}
\def\Jst{J_{\mathrm{st}}}
\newcommand\Flux{\mathrm{Flux}}
\newcommand\GCMI{\mathrm{GCMI}}
\newcommand\CMI{\mathrm{CMI}}
\newcommand\NC{\mathrm{NC}}
\newcommand\Psh{\mathrm{Psh}}
\newcommand\MPsh{\mathrm{\mathfrak{M}Psh}}
\newcommand\MSPsh{\mathfrak{MS}\mathrm{Psh}}
\newcommand\NPsh{\mathrm{\mathfrak{N}Psh}}
\newcommand\NSPsh{\mathfrak{NS}\mathrm{Psh}}
\newcommand\MS{\mathfrak{MS}}
\newcommand\Co{\mathrm{Co}}
\newcommand\Conv{\mathrm{Conv}}
\newcommand\supp{\mathrm{supp}}
\newcommand\Qcal{\mathcal{Q}}
\newcommand\tr{\mathrm{tr}}
\newcommand\Hess{\mathrm{Hess}}
\newcommand\Aut{\mathrm{Aut}}
\newcommand\Aff{\mathrm{Aff}}
\newcommand\Pic{\mathrm{Pic}}
\newcommand\Hom{\mathrm{Hom}}
\newcommand\Lin{\mathrm{Lin}}
\newcommand\Ocalc{\overline{\mathcal{O}}}
\newcommand\Ocalcl{\Ocalc_{\mathrm{loc}}}
\newcommand\Pcalc{\overline{\mathcal{P}}}
\newcommand\Rcalc{\overline{\mathcal{R}}}
\newcommand\Oscrc{\overline{\mathscr{O}}}
\newcommand\Oscrcl{\Oscrc_{\mathrm{loc}}}
\newcommand\boldA{\mathbf A}
%\renewcommand\i{\imath}
% The following makes 2020 to appear in the MSC
\makeatletter
\@namedef{subjclassname@2020}{\textup{2020} Mathematics Subject Classification}
\makeatother
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
\begin{document}
\title{The universal family of punctured Riemann surfaces is Stein}
\author{Franc Forstneri{\v c}}
\address{Faculty of Mathematics and Physics, University of Ljubljana, Jadranska 19, 1000 Ljubljana, Slovenia}
\address{Institute of Mathematics, Physics, and Mechanics, Jadranska 19, 1000 Ljubljana, Slovenia}
\email{franc.forstneric@fmf.uni-lj.si}
\subjclass[2020]{Primary 32G15; secondary 32Q28, 32Q56}
% 30F60 Teichmüller theory for Riemann surfaces
% 32E10 Stein spaces
% 32G05 Deformations of complex structures
% 32G15 Moduli of Riemann surfaces, Teichmüller theory (complex-analytic aspects in several variables)
% 32H02 Holomorphic mappings, (holomorphic) embeddings and related questions in several complex variables
% 32L20 Vanishing theorems
% 32Q28 Stein manifolds
% 32Q56 Oka principle and Oka manifolds
% 32V05(2000?now) CR structures, CR operators, and generalizations
% 32V20(2000?now) Analysis on CR manifolds
% 32W05 ?¯¯¯ and ?¯¯¯-Neumann operators
\date{15 December 2025}
\keywords{Riemann surface, Teichm\"uller space, universal family, Stein manifold, Oka manifold}
\begin{abstract}
We show that the universal Teichm\"uller family of $n$-punctured compact Riemann
surfaces of genus $g$ is a Stein manifold for any $n>0$. We describe
its basic function theoretic properties and pose several
challenging questions. We show in particular that
the space of fibrewise algebraic functions on the universal family
is dense in the space of holomorphic functions, and there is a
fibrewise algebraic map of the universal family in a Euclidean space
which is an embedding over any given relatively compact domain
in the Teichm\"uller space. We also obtain a relative
Oka principle for holomorphic fibrewise algebraic maps of the universal family
to any flexible algebraic manifold.
\end{abstract}
\maketitle
%
% SECTION:
%
\section{Introduction}\label{sec:intro}
The notion of a Teichm\"uller space originates in the papers
\cite{Teichmuller1940,Teichmuller1943,Teichmuller1944}
of Oswald Teichm\"uller, who defined a
complex manifold structure on the set of isomorphism classes of
marked closed Riemann surfaces of genus $g$.
Ahlfors \cite{Ahlfors1960} showed that this complex structure
can be defined by periods of holomorphic abelian differentials.
In \cite{Teichmuller1944}, Teichm\"uller also introduced the
universal Teichm\"uller curve -- a space $V$ over a Teichm\"uller
space $T$ whose fibre above $t\in T$ is a Riemann surface
$(M,J_t)$ representing that point, also called the universal family
of Riemann surfaces over $T$ -- and showed that it has the structure
of a complex manifold. Teichm\"uller's theory was developed
by Ahlfors and Bers \cite{AhlforsBers1960} and by
Grothendieck, who gave a series of lectures in Cartan's seminar 1960--1961;
see the discussion and references in \cite{ACampoJiPapadopoulos2016}.
Grothendieck asked whether every finite dimensional Teichm\"uller space
is a Stein manifold \cite[p.\ 14]{Grothendieck1962}.
An affirmative answer was given by
Bers and Ehrenpreis \cite[Theorem 2]{BersEhrenpreis1964} who
showed that any finite dimensional Teichm\"uller space embeds
as a domain of holomorphy in a complex Euclidean space, hence
is Stein. (Another proof was given by Wolpert \cite{Wolpert1987};
see also the surveys by Bers \cite{Bers1972} and Nag \cite{Nag1988}.)
The Teichm\"uller space $T(M)$ of a Riemann surface $M$ is
finite dimensional if and only if $M=\wh M\setminus \{p_1,\ldots,p_n\}$
is a compact Riemann surface $\wh M$ of some genus $g\ge 0$
with $n\ge 0$ punctures. Such $M$ is said to be of finite conformal
type and its Teichm\"uller space is denoted $T(g,n)$.
The universal family $\pi:\wh V(g,n)\to T(g,n)$ is a holomorphic
submersion whose fibre over any point $t\in T(g,n)$ is the
compact Riemann surface $(\wh M,J_t)$ with the
complex structure $J_t$ determined by $t$, and with
$n$ canonical holomorphic sections $s_1,\ldots,s_n:T(g,n)\to \wh V(g,n)$
with pairwise disjoint images representing the punctures.
(See Nag \cite[pp.\ 322--323]{Nag1988}.) The open subset
\begin{equation}\label{eq:Vgn}
V(g,n) = \wh V(g,n)\setminus \bigcup_{i=1}^n s_i(T(g,n))
\end{equation}
of $\wh V(g,n)$ is the universal family of $n$-punctured
compact Riemann surfaces of genus $g$.
If $2g+n\ge 3$ then the Teichm\"uller family $\pi:V(g,n)\to T(g,n)$
is the universal object in the complex analytic category of topologically marked holomorphically varying families of $n$-punctured genus $g$ Riemann surfaces (see \cite[Theorem 5.4.3]{Nag1988}).
In this paper, we establish several function
theoretic properties of the universal family $V(g,n)$.
Our first main result is the following.
%
% The universal family is stein
%
\begin{theorem}\label{th:main}
The Teichm\"uller family $V(g,n)$ for $n\ge 1$ is a Stein manifold.
\end{theorem}
This is a special case of Theorems \ref{th:Stein} and \ref{th:Stein2}
given in the sequel. The result is obvious if $g=0$ and $n\in \{1,2,3\}$
since $T(0,n)$ is then a singleton and $V(0,n)$ is biholomorphic to
$\C$, $\C^*=\C\setminus \{0\}$, and $\C\setminus\{0,1\}$,
respectively. If $2g+n\ge 3$ then $T(g,n)$ is biholomorphic
to a bounded topologically contractible
% (see \cite{FischerTromba1987})
Stein domain in $\C^{3g-3+n}$ \cite[p.\ 161]{Nag1988},
but it cannot be holomorphically realised as a convex domain
in $\C^{3g-3+n}$ if $g\ge 2$ since the Kobayashi metric and
the Carath\'eodory metric on it differ
(see Markovi\'c \cite{Markovic2018}), while they agree on
a convex domain (see Lempert \cite{Lempert1981}).
Note also that $\wh V(g,n)$ is holomorphically convex and
all holomorphic functions on it come from the base $T(g,n)$.
%
% ON PROOF
%
We provide two proofs of Theorem \ref{th:main}. The first more elementary one
in Sect.\ \ref{sec:proof} shows that $V(g,n)$ admits a strongly
plurisubharmonic exhaustion function, hence is Stein
by a theorem of Grauert \cite{Grauert1958AM}. The second
one follows from Theorem \ref{th:A-embedding} which shows that
holomorphic fibrewise algebraic functions on $V(g,n)$ separate
points and provide holomorphic convexity. The proof
uses Grauert's theorem on coherence of direct images
of coherent analytic sheaves and other techniques
of complex geometry.
Theorem \ref{th:main} has several interesting consequences.
By classical results of Remmert, Bishop, and Narasiman
(see \cite[Theorem 2.4.1]{Forstneric2017E} and the references
therein), $V(g,n)$ for $n\ge 1$
admits a proper holomorphic embedding
in $\C^N$ with $N=2\dim V(g,n)+1$, which equals
$6g-6+2n+1$ if $2g+n\ge 3$. When $\dim V(g,n)\ge 2$,
it also embeds properly holomorphically in $\C^N$ with
$N=\big[\frac{3\dim V(g,n)}{2}\big]+1$
(see Eliashberg and Gromov \cite{EliashbergGromov1992}
and Sch{\"u}rmann \cite{Schurmann1997}, or the
exposition in \cite[Secs.\ 9.3--9.4]{Forstneric2017E}).
Since the Teichm\"uller space $T(g,n)$ is contractible,
the submersion $\pi:V(g,n)\to T(g,n)$ is
smoothly trivial and the inclusion of any fibre of $\pi$ in $V(g,n)$
is a homotopy equivalence. It follows that for any manifold $Y$,
the restriction of a continuous map $V(g,n)\to Y$
to any fibre of $V(g,n)$ lies in the same homotopy class.
The following is a corollary to this observation, Theorem \ref{th:main},
and the main result of Oka theory \cite[Theorem 5.4.4]{Forstneric2017E}.
%For the theory of Oka manifolds we refer to \cite[Chap.\ 5]{Forstneric2017E} and
(See also the surveys \cite{Forstneric2023Indag,Forstneric2025ICM,ForstnericLarusson2011}.)
%
% Maps to Oka manifolds
%
\begin{corollary}\label{cor:1}
Let $\pi:V(g,n)\to T(g,n)$ be as above, $n\ge 1$,
and let $Y$ be an Oka manifold. There is a holomorphic map
$V(g,n)\to Y$ in every homotopy class. Furthermore, a holomorphic map
$M_t\to Y$, $t\in T(g,n)$, from any fibre of $V(g,n)$ extends to a holomorphic
map $V(g,n)\to Y$. More generally, given a closed
complex subvariety $T'$ of $T(g,n)$, every continuous map
$F_0:V(g,n)\to Y$ which is holomorphic on $\pi^{-1}(T')$
is homotopic to a holomorphic map $F:V(g,n)\to Y$ by a
homotopy which is fixed on $\pi^{-1}(T')$.
\end{corollary}
If $n\ge 1$ and $(g,n)\ne (0,1)$ then $V(g,n)$ is not simply connected,
and its homotopy type is that of a finite bouquet of circles.
In this case, homotopically nontrivial maps $V(g,n) \to Y$ exist
whenever the manifold $Y$ is not simply connected.
This gives the following corollary. The last statement follows
by taking the Oka manifold $Y=\C^*$.
The result obviously holds for $g=0,\ n=1$ since $V(0,1)=\C$.
\begin{corollary}\label{cor:2}
If $n\ge 1$ and $Y$ is an Oka manifold which is not
simply connected, there is a holomorphic map $V(g,n)\to Y$
which is nonconstant on every fibre. In particular, $V(g,n)$ admits a nowhere
vanishing holomorphic function which is nonconstant on every fibre.
\end{corollary}
Another way to obtain fibrewise nonconstant holomorphic maps
from $V(g,n)$ to an Oka manifold is to inductively use the Oka property with
approximation on compact holomorphically convex subsets
of the Stein manifold $V(g,n)$; see \cite[Theorem 5.4.4]{Forstneric2017E}.
The Oka principle can be used to obtain many further properties
of holomorphic universal families $V(g,n)\to Y$ in any Oka manifold.
%
% FIBREWISE ALGEBRAIC FUNCTIONS
%
Since every fibre of $V(g,n)$ is an affine algebraic curve,
it is natural to expect that $V(g,n)$
admits holomorphic functions which are algebraic on every fibre.
%of the Teichm\"uller projection $V(g,n)\to T(g,n)$.
This is indeed the case. We show in Theorem \ref{th:OkaWeil}
that the algebra $\Ascr(V(g,n))$ of fibrewise algebraic functions
on $V(g,n)$ is dense in the algebra $\Oscr(V(g,n))$
of holomorphic functions in the compact-open topology.
Furthermore, functions in $\Ascr(V(g,n))$ provide a proper embedding
of $V(g,n)$ in a Euclidean space over any relatively compact domain
in $T(g,n)$ (see Theorem \ref{th:A-embedding}), and
fibrewise algebraic maps of $V(g,n)$ to
any flexible algebraic manifold satisfy a
relative Oka principle (see Theorem \ref{th:AOP}).
The following problem remains open.
\begin{problem}
Does $V(g,n)$ for $n\ge 1$ admit a proper holomorphic embedding in
some $\C^N$ which is algebraic on every fibre?
\end{problem}
%
% VECTOR BUNDLES ON V(g,n)
%
The fact that $V(g,n)$ for $n\ge 1$ is homotopy equivalent to a
bouquet of circles implies that every complex vector bundle on
$V(g,n)$ is topologically trivial. Since $V(g,n)$ is Stein,
the Oka--Grauert principle (see Grauert \cite{Grauert1958MA}
or \cite[Theorem 3.2.1]{Forstneric2017E}) implies the following.
\begin{proposition}\label{prop:VB}
Every holomorphic vector bundle on $V(g,n)$ for $n\ge 1$
is holomorphically trivial.
\end{proposition}
This fails in the algebraic
category, even on a single fibre. In fact, the tangent bundle
of a finitely punctured compact Riemann surface is not
algebraically trivial in general.
\begin{corollary}\label{cor:VFF}
Assume that $g\ge 0$ and $n\ge 1$.
\begin{enumerate}[\rm (a)]
\item There exists a nowhere vanishing holomorphic
vector field $\xi$ on $V(g,n)$ which is tangent to the fibres
of the projection $\pi:V(g,n)\to T(g,n)$, that is, $d\pi(\xi)=0$.
\item
With $\xi$ as in (a), there exists a holomorphic $1$-form $\theta$ on $V(g,n)$
satisfying $\langle \theta,\xi\rangle=1$. In particular, $\theta$
is nowhere vanishing on the tangent bundle to any fibre of $\pi$.
\end{enumerate}
\end{corollary}
\begin{proof}
Part (a) follows by applying Proposition \ref{prop:VB} to the
holomorphic line bundle $\ker d\pi \to V(g,n)$, the vertical tangent
bundle of the holomorphic submersion $\pi:V(g,n)\to T(g,n)$.
To see (b), consider the following short exact sequence of
vector bundles over $V(g,n)$:
\[
0\lra \ker d\pi \longhookrightarrow T V(g,n)
\stackrel{\alpha}{\lra} H:=TV(g,n)/\ker\pi \lra 0.
\]
Here, $TV(g,n)$ denote the tangent bundle of $V(g,n)$.
By Cartan's Theorem B the sequence splits, i.e. there is a
holomorphic vector bundle injection $\sigma:H\hookrightarrow TV(g,n)$
such that $\alpha\circ\sigma=\Id_H$. Hence,
$T V(g,n)= \ker d\pi \oplus \sigma(H)=\C\xi \oplus \sigma(H)$,
where $\xi$ is as in part (a).
The unique holomorphic $1$-form $\theta$ on $V(g,n)$
satisfying $\langle \theta,\xi\rangle =1$ and $\xi=0$ on $\sigma(H)$
clearly satisfies part (b).
%Note also that every holomorphic vector field on $T(g,n)$ admits a unique lift to a holomorphic vector field on $V(g,n)$ which is everywhere tangent to the subbundle $\sigma(H)$ of $V(g,n)$.
%Flows of such horizontal vector fields provide holomorphic families of biholomorphic maps from any relatively compact domain $\Omega$ in a fibre $M_{t_0}=\pi^{-1}(x_0)$ onto domains $\Omega_t\subset M_t$ in nearby fibres for as long as the flow exists. Clearly the flow does exist globally on $M_{t_0}$ since the fibres $M_t$ for $t$ neat $t_0$ are not all biholomorphic to $M_{t_0}$.
\end{proof}
By the Gunning--Narasimhan theorem \cite{GunningNarasimhan1967},
every open Riemann surface $M$ admits a holomorphic immersion
$f:M\to\C$. In view of Corollary \ref{cor:VFF} (b), the following
is a natural question.
\begin{problem}\label{prob:immersion}
Let $n\ge 1$.
Is there a holomorphic function $f:V(g,n)\to\C$ whose restriction
to every fibre of $V(g,n)$ is an immersion?
\end{problem}
By \cite[Theorem 1]{Forstneric2003AM} there exists a
holomorphic function $f:V(g,n)\to\C$ without critical points.
The problem is to find $f$ such that
$\ker df_z$ is transverse to $\ker d\pi_z$ at every point $z\in V(g,n)$.
Note that \cite[Corollary 8.3]{Forstneric2024Runge}
gives a smooth function $f:V(g,n)\to\C$ whose restriction to any
fibre is a holomorphic immersion.
Problem \ref{prob:immersion} is related to the question whether a
holomorphic 1-form $\theta$ in Corollary \ref{cor:VFF} (b)
can be made exact on every fibre of $\pi$ by multiplying it with
a suitably chosen nowhere vanishing holomorphic function on $V(g,n)$.
However, this is not the only problem. Since the Teichm\"uller submersion
$\pi:V(g,n)\to T(g,n)$ does not admit a holomorphic section when $g\ge 3$
(see Hubbard \cite{Hubbard1972,Hubbard1976} and
Earle and Kra \cite[p.\ 50]{EarleKra1976} for a precise description
of sections of $\wh V(g,n)$ and $V(g,n)$),
%
%who showed in that $V(g,0)\to T(g,0)$ has no holomorphic sections if $g\ge 3$ and six sections if $g=2$. Earle and Kra proved \cite[p.\ 50]{EarleKra1976} that for $n\ge 1$, $\pi:\wh V(g,n)\to T(g,n)$ has exactly $n$ sections (the canonical sections) if $g\ge 3$, and exactly $2n+6$ sections if $g=2$.
%
there is no natural way of choosing the initial point for computing
the fibrewise integrals of $\theta$, which would give a holomorphic family
of immersions on the fibres.
Another interesting question is whether the Riemann surfaces $M_t$
in the Teichm\"uller family $V(g,n)$, $n\ge 1$, admit a representation
as a family of conformal minimal surfaces in $\R^k$, $k\ge 3$,
whose $(1,0)$-derivatives depends holomorphically
on $t\in T(g,n)$. For background, % on minimal surfaces,
see \cite{Osserman1986} and \cite{AlarconForstnericLopez2021}.
By \cite[Corollary 8.6]{Forstneric2024Runge} the answer is affirmative
with continuous or smooth dependence on parameter.
It remains an open problem whether minimal surfaces in such families
can be chosen to be complete and with finite total curvature.
Each single surface in the family can be made such by
\cite{AlarconLopez2022APDE,AlarconLarusson2025Crelle}.
It is natural to wonder whether the results of this paper extend
to infinite dimensional Teichm\"uller families.
The analogous problem with continuous or smooth dependence
of the complex structures and the holomorphic maps on the parameter
was studied in \cite{Forstneric2024Runge}
for very general families of open Riemann surfaces $\{(M,J_t)\}_{t\in T}$,
where $M$ is a smooth open surface
and $J_t$ are complex structures of some local H\"older class on $M$
depending continuously or smoothly on the parameter $t$
in a topological space $T$. The Riemann surfaces in such families
need not belong to the same Teichm\"uller space.
For example, a punctured Riemann surface can be a member
of a family in which the punctures develop into boundary curves
and vice versa. Under mild assumptions
on the family $\{J_t\}_{t\in T}$,
$J_t$-holomorphic functions on $(M,J_t)$
satisfy the Runge approximation theorem with continuous
or smooth dependence on $t\in T$
\cite[Theorem 1.1]{Forstneric2024Runge}.
For a class of parameter spaces including finite CW complexes
we also have the Oka principle for continuous or smooth
families of holomorphic maps from $(M,J_t)$ to any Oka manifold
\cite[Theorem 1.6]{Forstneric2024Runge}.
These results were extended in \cite{ForstnericSigurdardottir2025}
to maps from tame families of Stein manifolds of arbitrary dimension
to Oka manifolds.
%
% PROOF OF THE MAIN THEOREM
%
\section{Proof of Theorem \ref{th:main}}\label{sec:proof}
In view of the description of the Teichm\"uller submersion
$\pi:V(g,n)\to T(g,n)$ \eqref{eq:Vgn}, Theorem \ref{th:main} is an
immediate consequence of the following result
with arbitrary Stein manifold as the base.
%See also Theorem \ref{th:Stein2} for a more general result.
%
% MAIN THEOREM
%
\begin{theorem}\label{th:Stein}
Assume that $X$ is a Stein manifold,
$Z$ is a complex manifold with $\dim Z=\dim X+1$,
$\pi:Z\to X$ is a surjective proper holomorphic
submersion with connected fibres,
and $s_1,\ldots,s_n:X\to Z$ $(n\ge 1)$ are holomorphic sections
with pairwise disjoint images. Then, the domain
$\Omega=Z\setminus \bigcup_{i=1}^n s_i(X)$ is Stein.
\end{theorem}
The assumption that the sections
$s_1,\ldots,s_n$ have pairwise disjoint images is inessential and
is imposed only for convenience of the proof; see Theorem \ref{th:Stein2}
for a more general result. The conclusion fails if the fibres
of $\pi$ have complex dimension $>1$, or if the sections
$s_i$ are not holomorphic. In such a case, the
domain $\Omega$ in the theorem fails to be
locally pseudoconvex at some boundary point $s_i(x)$, $x\in X$.
Note that Stein complements of smooth complex hypersurfaces
in compact K\"ahler manifolds have recently been
studied by H\"oring and Peternell \cite{HoringPeternell2024}
where the reader can find references to earlier works.
In our case, $Z$ is not compact unless $X$ is a point.
\begin{proof}[Proof of Theorem \ref{th:Stein}]
Let $\pi:Z\to X$ be as in the theorem.
Note that every fibre $Z_x=\pi^{-1}(x)$, $x\in X$, is a compact
connected Riemann surface, and the fibres are
diffeomorphic but not necessarily biholomorphic
to each other. Hence, $\{Z_x\}_{x\in X}$ is
a holomorphic family of compact Riemann surfaces and
$\Omega_x=Z_x\setminus \bigcup_{i=1}^n s_i(x)$ $(x\in X)$
is a holomorphic family of $n$-punctured Riemann surfaces.
Each $H_i=s_i(X)$ is a closed complex hypersurface in $Z$
whose ideal sheaf is principal, that is, it is locally near each
point of $H_i$ generated by a single holomorphic function.
Recall the following result (see Grauert and Remmert
\cite[Theorem 5, p.\ 129]{GrauertRemmert1979}):
If $Z$ is a Stein space and $H$ is a closed complex analytic hypersurface
in $Z$ (of pure codimension one)
whose ideal sheaf is a principal ideal sheaf, then $Z\setminus H$
is also Stein. If $Z$ is nonsingular
then the ideal sheaf of any closed complex hypersurface in $Z$ is a
principal ideal sheaf (see \cite[Chap.\ A.3.5]{GrauertRemmert1979}).
Hence, it suffices to prove the theorem in the case $n=1$,
that is, to show that the complement $Z\setminus s(X)$ of a
holomorphic section $s:X\to Z$ is a Stein manifold.
By a theorem of Siu \cite{Siu1976}, the Stein hypersurface $H=s(X)$
has a basis of open Stein neighbourhoods $U$ in $Z$.
Since $U\setminus H$ is a Stein manifold
%by the aforementioned theorem
\cite[Theorem 5, p.\ 129]{GrauertRemmert1979},
it admits a strongly plurisubharmonic exhaustion function
$\phi:U\setminus H\to \R_+$. To prove the theorem, we shall
construct a strongly plurisubharmonic exhaustion function
$Z\setminus H\to \R_+$; a theorem of Grauert \cite{Grauert1958AM}
will then imply that $Z\setminus H$ is Stein.
Fix a point $x_0\in X$ and set $z_0=s(x_0)\in H\subset Z$.
Since $\pi:Z\to X$ is a holomorphic submersion with compact
one-dimensional fibres, it is a smooth fibre bundle whose fibre
$M$ is a compact smooth surface.
In particular, there is a neighbourhood $X_0\subset X$ of $x_0$
such that the restricted bundle $Z|X_0=\pi^{-1}(X_0) \to X_0$ can
be smoothly identified with the trivial bundle
$X_0\times M\to X_0$. In this identification,
$z_0=(x_0,p_0)$ with $p_0\in M$. Since $\phi$ tends to $+\infty$
along $H$, there are small smoothly bounded open discs
$D\Subset D'\Subset M$ with $p_0\in D$ such that
\begin{equation}\label{eq:ineq1}
\inf_{p \in bD} \phi(x_0,p) >
\max_{p \in bD'} \phi(x_0,p).
\end{equation}
The set $O=M\setminus D$ is a compact bordered
Riemann surface with smooth boundary $bO=bD$, endowed
with the complex structure inherited by the identification
$M\cong Z_{x_0}=\pi^{-1}(x_0)$. Note that $bD'$ is contained
in the interior of $O$. It follows from \eqref{eq:ineq1} and standard
results that there is a smooth strongly subharmonic function
$u_0: O \to\R_+$ such that
\[
\text{$u_0 < \phi(x_0,\cdotp)$ on
$bO=bD$ and $u_0>\phi(x_0,\cdotp)$ on $bD'$.}
\]
Shrinking the neighbourhood $X_0\subset X$ of $x_0$ if necessary,
the following conditions hold for every $x\in X_0$, where we use
the smooth fibre bundle isomorphism $Z|X_0\cong X_0\times M$:
\begin{enumerate}[\rm (a)]
\item $s(x) \in D$,
\item the function $u(x,\cdotp)=u_0$ is strongly subharmonic on $O$
in the complex structure on $Z_x\cong M$,
\item $u(x,\cdotp) < \phi(x,\cdotp)$ on $bO=bD$, and
\item $u(x,\cdotp) > \phi(x,\cdotp)$ on $bD'$.
\end{enumerate}
Condition (b) holds since being strongly subharmonic on
a compact subset is a stable property under small smooth
deformations of the complex structure.
We define a function $\rho_{0}: (X_0\times M) \setminus H \to\R_+$
by taking for every $x\in X_0$:
\[
\rho_{0}(x,p)=
\begin{cases}
\phi(x,p), & p\in D\setminus \{s(x)\}; \\
\max\{\phi(x,p),u(x,p)\}, & p\in D'\setminus D; \\
u(x,p), & p \in M\setminus D'.
\end{cases}
\]
Note that $\rho_0$ is well defined, piecewise smooth, strongly subharmonic
on each fibre $Z_x\setminus\{s(x)\}$ $(x\in X_0)$, and it agrees with $\phi$
on $(X_0\times D)\setminus H$.
By using the regularised maximum in the definition of $\rho_{0}$
(see \cite[Eq. (3.1), p.\ 69]{Forstneric2017E}), we may assume that
$\rho_{0}$ is smooth and enjoys the other stated properties.
This construction gives an open locally finite cover $\{X_j\}_{j=1}^\infty$
of $X$ with smooth fibre bundle trivialisations
$Z|X_j\cong X_j\times M$, discs $D_j\subset M$ such that
$s(x)\in D_j$ for all $x\in X_j$, and smooth functions
$\rho_j:(Z|X_j)\setminus H \to \R_+$
such that $\rho_j$ is strongly subharmonic on each fibre
$Z_x\setminus \{s(x)\}$ $(x\in X_j)$
and it agrees with $\phi$ on $(X_j\times D_j)\setminus H$.
Let $\{\chi_j\}_j$ be a smooth partition of unity on $X$
with compact supports $\supp(\chi_j) \subset X_j$ for each $j$. Set
\[
\rho=\sum_{j=1}^\infty \chi_j \rho_j:Z\setminus H\to \R_+.
\]
By the construction, the restriction of $\rho$ to each fibre
$Z_x\setminus \{s(x)\}$ $(x\in X)$ is strongly subharmonic,
and there is an open neighbourhood $U_0\subset U\subset Z$ of $H$
such that $\rho=\phi$ holds on $U_0\setminus H$. In particular, $\rho$
is strongly plurisubharmonic on $U_0\setminus H$.
Note that for every compact set $K\subset X$,
the set $\pi^{-1}(K) \setminus U_0\subset Z\setminus H$ is compact.
Hence, by choosing a strongly plurisubharmonic exhaustion
function $\tau:X\to\R_+$ whose Levi form $dd^c \tau$ grows fast enough,
we can ensure that
\[
\rho+\tau\circ \pi: Z\setminus H\to\R_+
\]
is a strongly plurisubharmonic exhaustion function, thus proving
the theorem. Indeed, denoting
by $J$ the almost complex structure operator on $Z$ and
by $d^c$ the conjugate differential defined by
\[
(d^c \rho)(z,\xi)=-d\rho(z,J \xi)\ \
\text{for $z\in Z$ and $\xi\in T_z Z$},
\]
the function $\rho$ is strongly plurisubharmonic at $z\in Z$ if and only if
\[
(dd^c\rho)(z,\xi\wedge J\xi)>0\ \
\text{for every tangent vector $0\ne \xi\in T_z Z$}.
\]
(Up to a positive factor this equals the Laplace of $\rho$ on the
2-plane $\mathrm{span}(\xi,J\xi) \subset T_z Z$.)
Since $\rho$ is strongly subharmonic
on every fibre $Z_x \setminus \{s(x)\}$, $x\in X$, we have that
\[
(dd^c \rho)(z,\xi\wedge J\xi)>0 \ \
\text{if $z\in Z\setminus H$ and $0\ne \xi \in \ker d\pi_z$.}
\]
Hence, the eigenvectors of $(dd^c \rho)(z,\cdotp)$ associated
to non-positive eigenvalues lie in a closed cone $C_z\subset T_z Z$
which intersects $\ker d\pi_z$ only in the origin.
It follows that if $\tau:X\to \R$ is such that
$dd^c \tau>0$ is sufficiently big on $T_x X$ where $x=\pi(z)$,
then $dd^c\rho + dd^c (\tau\circ\pi) >0$ on $T_z Z$.
Furthermore, the estimates are uniform on the compact set
$\pi^{-1}(K) \setminus U_0$, where $U_0\subset Z$ is a neighbourhood
of $H$ such that $dd^c\rho>0$ on $U_0\setminus H$.
To see that $\tau$ can be chosen
such that $dd^c \tau$ grows as fast as desired, note that
if $h:\R\to\R$ is a $\Cscr^2$ function then for each point $x\in X$ and
vector $\xi\in T_xX $ we have that
\[
dd^c(h\circ \tau)(x,\xi\wedge J\xi) =
h'(\tau(x))\, (dd^c \tau)(x,\xi\wedge J\xi) +
h''(\tau(x))\, \big(|d\tau(x,\xi)|^2 + |d\tau(x,J\xi)|^2\big).
\]
Hence, if $\tau$ is a strongly plurisubharmonic exhaustion function on $X$
and the function $h:\R\to\R$ is chosen such that $h''\ge 0$ and
$h'$ grows sufficiently fast, then $dd^c (h\circ \tau)$ also grows
as fast as desired.
%This completes the proof of Theorem \ref{th:Stein}, and it also proves Theorem \ref{th:main}.
\end{proof}
A minor modification of the proof of Theorem \ref{th:Stein}
gives the following more general result.
%
% MAIN THEOREM 2
%
\begin{theorem}\label{th:Stein2}
Assume that $X$ is a connected Stein manifold,
$Z$ is a complex manifold,
$\pi:Z\to X$ is a surjective proper holomorphic
submersion with connected one-dimensional fibres,
and $H$ is a closed complex subvariety
of $Z$ of pure codimension one which does not contain
any fibre of $\pi$. Then, the domain $\Omega=Z\setminus H$ is Stein.
\end{theorem}
\begin{proof} %[Sketch of proof]
The hypotheses imply that $\pi|_H:H \to X$
is a finite holomorphic map (a branched holomorphic cover).
The proper mapping theorem of Remmert \cite{Remmert1957}
(see also \cite[p.\ 29]{Chirka1989})
implies that $\pi(H)$ is a closed complex subvariety of $X$
of pure dimension $\dim X$, hence $\pi(H)=X$ since $X$ is connected.
By \cite[Theorem 1 (d), p.\ 125]{GrauertRemmert1979},
the subvariety $H$ is Stein. A theorem of Siu \cite{Siu1976} implies
that $H$ has a basis of open Stein neighbourhoods $U\subset Z$.
By the argument in the proof of Theorem \ref{th:Stein},
the ideal sheaf of $H$ is a principal ideal sheaf, so
the manifold $U\setminus H$ is Stein for any Stein neighbourhood
$U\subset Z$ of $H$ \cite[Theorem 5, p.\ 129]{GrauertRemmert1979}.
The proof can now be completed by a similar argument as
in the proof of Theorem \ref{th:Stein},
and we leave the details to the reader.
\end{proof}
\begin{remark}
Theorem \ref{th:Stein2} still holds if the fibres of
the projection $\pi:Z\to X$ are not necessarily connected
provided that the subvariety
$H$ intersects every connected component of each
fibre $Z_x=\pi^{-1}(x),\ x\in X$, but it does not contain any such component.
The result also holds if $X$ is a Stein space, $Z$ is a complex space,
$H\subset Z$ is a complex hypersurface whose ideal sheaf is principal,
and the other conditions on the submersion $\pi:Z\to X$ and $H$
are satisfied.
\end{remark}
%
%
% SECTION: FIBREWISE ALGEBRAIC FUNCTIONS
%
%
\section{Fibrewise algebraic functions on $V(g,n)$} \label{sec:algebraic}
In this section, we show that the universal family $V(g,n)$
of $n$-punctured compact Riemann surfaces of genus $g$
for $n\ge 1$ has a large algebra of holomorphic functions which are
algebraic on every fibre of the Teichm\"uller projection
$\pi:V(g,n)\to T(g,n)$. Indeed, such functions are dense in the
algebra of holomorphic functions in the compact-open topology;
see Theorem \ref{th:OkaWeil}.
We shall consider the more general situation given
in Theorem \ref{th:Stein2}. Thus, we assume that $X$ is a connected
Stein manifold, $Z$ is a complex manifold,
$\pi:Z\to X$ is a surjective proper holomorphic submersion with
connected one-dimensional fibres, and $H$ is a closed complex subvariety
of $Z$ of pure codimension one which does not contain any fibre
of $\pi$. The domain $\Omega=Z\setminus H$ is Stein by
Theorem \ref{th:Stein2}, and every fibre $\Omega_x$ $(x\in X)$
is a finitely punctured compact Riemann surface, hence affine.
Let $\Ascr(\Omega)$ denote the $\C$-algebra of holomorphic functions
on $\Omega$ which are algebraic on every
fibre $\Omega_x$, $x\in X$. Clearly, $\Ascr(\Omega)$ is
a subalgebra of the algebra $\Oscr(\Omega)$ of holomorphic functions.
%
%
%
\begin{theorem}\label{th:A-embedding}
(Assumptions as above.) Given a relatively compact domain
$U\Subset X$, there exist finitely many functions
$f_1,\ldots,f_n\in \Ascr(\Omega)$ such that the map
$F:\Omega\to X \times \C^n$ given by
\begin{equation}\label{eq:F}
F(z) = (\pi(z), f_1(z),\ldots, f_n(z)),\quad z\in\Omega
\end{equation}
induces a proper embedding
$\Omega_U :=\Omega\cap \pi^{-1}(U) \to U\times \C^n$.
\end{theorem}
A map $F$ satisfying the conclusion of
theorem is called a {\em proper embedding over $U$}.
The same result holds if $X$ is a Stein space, $Z$ is a complex space,
and the other conditions on the submersion $\pi:Z\to X$ and the
hypersurface $H\subset Z$ hold; in particular, the ideal sheaf
of $H$ is principal. Theorem \ref{th:A-embedding} clearly implies that
holomorphic fibrewise algebraic functions on $\Omega$ separate
points and provide holomorphic convexity, thus providing another
proof that $\Omega$ (and in particular the universal
family $V(g,n)$) is a Stein manifold.
\begin{proof}
Let $L\to Z$ denote the holomorphic line bundle determined by the
divisor $H\subset X$, and let $\sigma_0:Z\to L$ be a holomorphic
section whose zero divisor equals $H$. (The existence
of such a section is tautological from the construction of $L$.)
Since the fibre $Z_x=\pi^{-1}(x)$ is a projective
curve for every $x\in X$, Serre's GAGA principle \cite{Serre1955AIF}
implies that the restricted line bundle $L_x:=L|Z_x \to Z_x$
is algebraic, and every holomorphic section of $L_x$ over $Z_x$
is algebraic. It follows that for any holomorphic section
$\sigma:Z\to L$, the quotient
$f=\sigma/\sigma_0$ is a well defined function in $\Ascr(\Omega)$.
The restriction of $f$ to $\Omega_x$ is algebraic and has an effective
pole at every end of $\Omega_x$ (that is, a point of
$H_x=H\cap Z_x$) at which $\sigma$ does not vanish.
Note that the line bundle $L_x\to Z_x$ is associated to the effective
divisor supported on $H_x$ in which the multiplicity of
a point $z\in H_x$ is the intersection number of $H$ with $Z_x$ at $z$.
In particular, $L_x$ is an ample line bundle on $Z_x$ for every $x\in X$.
Hence, some tensor power $L_x^{\otimes d} \to Z_x$ for $d=d(x)>0$
is very ample, so it admits finitely many holomorphic sections
$\sigma_{1,x},\cdots,\sigma_{n,x}:Z_x\to L_x$ such that the map
\begin{equation}\label{eq:Kodaira}
Z_x\ni z \longmapsto
\big[\sigma_{0,x}^{d}(z):\sigma_{1,x}(z):\cdots:\sigma_{n,x}(z)\big]
\in \CP^n
\end{equation}
is a holomorphic embedding.
Here, $\sigma_0^d$ denotes the $d$-th power of $\sigma_0$,
a section of the line bundle $L^{\otimes d}\to Z$.
Note that $\sigma_{i,x}/\sigma_{0,x}^{d}$ is a regular
algebraic function on $\Omega_x$ for every $i=1,\ldots,n$.
To prove the theorem, we will show that for $d=d(x)>0$ big enough,
every holomorphic section $\sigma_x:Z_x\to L_x^{\otimes d}$
extends to a holomorphic section $\sigma:Z\to L^{\otimes d}$.
Assume for a moment that this holds true.
Extending all section $\sigma_{i,x}$ in \eqref{eq:Kodaira}
for $i=1,\ldots,n$ to $Z$, the map in \eqref{eq:Kodaira}
is an embedding for any base point in an open neighbourhood
$U_1\subset X$ of $x$. Hence, the map $F$ of the form \eqref{eq:F}
with the components $f_i=\sigma_i/\sigma_0^d\in \Ascr(\Omega)$
is a proper embedding over $U_1$.
Indeed, for any pair of points $x\in U_1$ and $z\in H_x$, at least one
of the sections $\sigma_{i,x}$ $(i=1,\ldots,n)$ does not vanish
at $z$. Since $\sigma_0(z)=0$, it follows
that $f_i=\sigma_i/\sigma_0^d$ has an effective
pole at $z$, which implies properness.
Repeating the same construction at other points
$x\in \overline U$ (for different values of $d=d(x)\in\N$)
and assembling the resulting functions as components of
a map \eqref{eq:F} yields a proper embedding over $U$
as in the theorem.
It remains to explain why every
holomorphic section $\sigma_x:Z_x\to L_x^{\otimes d}$
for $d>0$ big enough extends to a holomorphic section
$\sigma:Z\to L^{\otimes d}$. Let $\Lscr\to Z$ denote the
sheaf of holomorphic sections of the line bundle $L\to Z$;
note that $\Lscr$ is a coherent $\Oscr_Z$-analytic sheaf.
Since the holomorphic projection $\pi:Z\to X$ is
surjective and proper, the direct image sheaf $\pi_*\Lscr$
is a coherent analytic sheaf on $X$ by Grauert's direct image theorem
\cite[p.\ 207]{GrauertRemmert1984}. A section
of $\pi_*\Lscr$ over an open subset $V\subset X$ is the same thing
as a section of $\Lscr$ over $Z_V:=\pi^{-1}(V)$.
The argument given (in the algebraic case) in
\cite[\href{https://stacks.math.columbia.edu/tag/0D2M}{Tag 0D2M}]{stacks}
shows that for $d_0=d_0(x)>0$ big enough, the evaluation map
\begin{equation}\label{eq:eval}
E_x: \pi_*(\Lscr^{\otimes d})_x \to H^0(Z_x,\Lscr_x^{\otimes d})
\end{equation}
is surjective for $d\ge d_0$.
This means that every holomorphic section
$\sigma_x:Z_x\to L_x^{\otimes d}$ of the line bundle
$L_x^{\otimes d}\to Z_x$ extends to a holomorphic section $\sigma$
of $L^{\otimes d}|{Z_V}\to Z_V$ for an open neighbourhood
$V$ of $x$. Denote by $[\sigma]_x \in \pi_*(\Lscr^{\otimes d})_x$
the associated element of the stalk $\pi_*(\Lscr^{\otimes d})_x$,
so $E_x([\sigma]_x)=\sigma_x$. Since the sheaf $\pi_*\Lscr^{\otimes d}$
is $\Oscr_X$-coherent, Cartan's Theorem A
\cite[p.\ 124]{GrauertRemmert1979} shows that
$\pi_*(\Lscr^{\otimes d})_x$ is generated as an $\Oscr_{X,x}$-module
by global sections of $\pi_*(\Lscr^{\otimes d})$. Hence, there are sections
$\xi_1,\ldots,\xi_m \in \Gamma(X,\pi_*(\Lscr^{\otimes d}))$
and germs of holomorphic functions $g_1,\ldots,g_m\in \Oscr_{X,x}$
such that
\begin{equation}\label{eq:sigmax}
[\sigma]_x = \sum_{j=1}^m g_j [\xi_j]_x \in \pi_*(\Lscr^{\otimes d})_x.
\end{equation}
Since $X$ is Stein, there are functions $\tilde g_j\in \Oscr(X)$ such that
$\tilde g_j(x)=g_j(x)$ for $j=1,\ldots,m$. Then,
$
\tilde \sigma := \sum_{j=1}^m \tilde g_j \xi_j \in
\Gamma(X,\pi_*(\Lscr^{\otimes d}))
$
is a section whose value at $x$ equals
\[
E_x(\tilde \sigma) = \sum_{j=1}^m \tilde g_j(x) E_x (\xi_j) =
\sum_{j=1}^m g_j(x) E_x (\xi_j) = E_x([\sigma]_x) = \sigma_x
\]
(see \eqref{eq:sigmax}). This completes the proof.
\end{proof}
%
% REMARK
%
\begin{remark}\label{rem:A-embedding}
The integer $d_0=d_0(x)>0$ in the above argument may depend on $x$.
For this reason, we do not know whether there is a proper embedding
$F:\Omega\to X \times \C^n$ of class $\Ascr(\Omega)$
and of the form \eqref{eq:F} for some $n\in\N$.
However, such an embedding exists and can be chosen
fully algebraic if the base $X$ is affine algebraic
due to finiteness of the above procedure, using the Zariski
topology as opposed to the Hausdorff topology.
The following is a special case of a lemma due to
R.\ van Dobben de Bruyn \cite{deBruyn2018}.
This result will not be used in the paper and is stated only
for the sake of comparison.
\begin{lemma}
Assume that $X$ is an affine algebraic variety, $Z$ is an algebraic
variety, $\pi:Z\to X$ is a proper algebraic submersion with
one dimensional fibres, and $s_1,\cdots,s_m$ for $m\ge 1$
are algebraic sections. If all fibres
$Z_x\setminus \bigcup_{i=1}^m s_i(x)$ $(x\in X)$ are affine
then there exists a proper algebraic morphism
$F:\Omega=Z\setminus \bigcup_{i=1}^m s_i(X)\to X\times \C^n$
of the form \eqref{eq:F} for some $n\in\N$.
\end{lemma}
\end{remark}
Recall (see e.g.\ \cite{Hormander1990})
that a compact set $K$ in a complex manifold $\Omega$
is said to be $\Oscr(\Omega)$-convex if it equals its
$\Oscr(\Omega)$-convex hull
\[
\wh K_{\Oscr(\Omega)} =
\{z\in \Omega: |f(z)|\le \sup_K|f|\ \text{for all}\ f\in \Oscr(\Omega)\}.
\]
The following Oka--Weil type approximation theorem for the
Stein domain $\Omega=Z\setminus H$ in Theorem \ref{th:A-embedding}
shows that the algebra $\Ascr(\Omega)$ is dense in $\Oscr(\Omega)$
in the compact-open topology.
%
% OKA-WEIL THEOREM FOR A(\Omega)
%
\begin{theorem}\label{th:OkaWeil}
Let $\pi:Z\to X$, $H\subset Z$, and
$\Omega = Z\setminus H$ be as in Theorem \ref{th:A-embedding}.
Given a compact $\Oscr(X)$-convex set
$K\subset \Omega$, every holomorphic function
on a neighbourhood of $K$ is a uniform limit on $K$
of functions in $\Ascr(\Omega)$.
This holds in particular for the Teichm\"uller
family $\Omega=V(g,n)$ \eqref{eq:Vgn} for any $g\ge 0$ and $n\ge 1$.
\end{theorem}
\begin{proof}
Choose a relatively compact Runge domain $U\Subset X$
such that $\pi(K)\subset U$. Let
$F:\Omega\to X \times \C^n$ be a map of class
$\Ascr(\Omega)$, furnished by Theorem \ref{th:A-embedding}
(see \eqref{eq:F}), such that $F:\Omega_U \to U\times \C^n$
is a proper embedding. Its image $F(\Omega_U)$ is then
a closed complex submanifold of the Stein manifold $U\times \C^n$.
It follows that the set $F(K)$ is holomorphically
convex in $U\times \C^n$, and hence in $X\times \C^n$ since
$U$ is Runge in $X$. Let $f$ be a holomorphic function on a
Stein neighbourhood $V\subset \Omega$ of $K$. We may assume that
$\pi(V)\subset U$. Then, $\tilde f=f\circ F^{-1}:F(V)\to\C$ is a
holomorphic function on the open subset $F(V)$
in the closed complex submanifold $F(\Omega_U)$ of $U\times \C^n$.
By the Cartan--Oka extension theorem, $\tilde f$ extends to a
holomorphic function on a neighbourhood of $F(K)$ in $U\times \C^n$.
(This can also be seen by precomposing $\tilde f$ by a holomorphic
retraction from a neighbourhood of $F(\Omega_U)$ in $U\times \C^n$
onto $F(\Omega_U)$, given by Docquier and Grauert
\cite{DocquierGrauert1960};
see also \cite[Theorem 3.3.3, p.\ 74]{Forstneric2017E}.)
By the Oka--Weil theorem, $\tilde f$ can be approximated
uniformly on $F(K)$ by functions $g\in \Oscr(X\times \C^n)$.
Every such function is of the form
$
g(x,w)=\sum_{\alpha\in \Z_+^n} g_\alpha(x) w^\alpha,
$
where $w=(w_1,\ldots,w_n)$ are coordinates on $\C^n$ and
$g_\alpha\in \Oscr(X)$ for every $\alpha\in \Z_+^n$. By cutting
off the power series at a finite level gives functions
$\Ascr(X\times \C^n)$ (polynomials in $w$
depending holomorphically on $x\in X$)
approximating $g$ on $F(K)$. Precomposing
such functions with $F$ gives the desired approximation
of $f$ by functions in $\Ascr(\Omega)$.
\end{proof}
The following is an immediate corollary to Theorem \ref{th:OkaWeil}.
\begin{corollary}
(Assumptions as in Theorem \ref{th:OkaWeil}.)
A compact set $K$ in $\Omega$ is $\Oscr(\Omega)$-convex
if and only if it is $\Ascr(\Omega)$-convex.
\end{corollary}
It follows that the functions in $\Ascr(\Omega)$ satisfy the
axioms of a Stein manifold, i.e., they separate points in $\Omega$
and provide holomorphic convexity.
%
%
%
\section{A relative Oka principle for fibrewise algebraic maps}
\label{sec:AOP}
In this section, we describe a relative Oka principle
(see Theorem \ref{th:AOP}) for holomorphic
fibrewise algebraic maps from the
Teichm\"uller family $V(g,n)$ $(n\ge 1)$, or more generally from any
manifold $\Omega=Z\setminus H$ as in Theorem \ref{th:A-embedding},
to a class of complex algebraic manifolds which includes
all flexible manifolds in the sense of Arzhantsev et al.\
\cite{ArzhantsevFlennerKalimanKutzschebauchZaidenberg2013DMJ}.
In order to state this result, we must
recall the following notions; see
Gromov \cite[0.5, p.\ 8.5.5]{Gromov1989},
\cite[Definition 5.6.13]{Forstneric2017E}, or
\cite[Definition 3.1]{Forstneric2023Indag}.
A {\em dominating holomorphic spray} on a complex manifold $Y$
is a holomorphic map $s:E\to Y$ from the total space of a
holomorphic vector bundle $E\to Y$ such that
\begin{equation}\label{eq:sprayonY}
s(0_y)=y\ \ \text{and}\ \ ds_{0_y}(E_y) = T_y Y
\ \ \text{for every $y\in Y$.}
\end{equation}
Here, $0_y$ is the origin of the fibre $E_y$ of $E$ over the point $y\in Y$,
and $E_y$ is considered as a $\C$-linear subspace of the tangent space
$T_{0_y}E$. (Identifying $Y$ with the zero section of $E$,
we have $TE|Y \cong TY\oplus E$.)
The spray is algebraic if the vector bundle $E\to Y$
and the map $s:E\to Y$ are algebraic.
A complex manifold $Y$ is {\em elliptic} if it admits a dominating
holomorphic spray, and is {\em special elliptic} if such a spray exists
on a trivial bundle $E=Y\times \C^m$ for some $m\in \N$.
An algebraic manifold $Y$ is {\em algebraically (special) elliptic}
if these conditions hold with algebraic sprays.
An algebraic manifold $Y$ is said to be {\em flexible}
\cite{ArzhantsevFlennerKalimanKutzschebauchZaidenberg2013DMJ}
if its tangent bundle $TY$ is pointwise spanned by finitely many LND's,
that is, algebraic vector fields $V_j$ with
complete algebraic flows $\phi_t^j$ $(t\in\C,\ j=1,\ldots,m)$.
The map $s:Y\times \C^m\to Y$ defined by
\[
s(y,t_1,\ldots,t_m)=\phi^1_{t_1}\circ \cdots \phi^m_{t_m}(y)
\]
is a dominating algebraic spray on $Y$, so a flexible manifold
is algebraically special elliptic. References to examples
of flexible manifolds can be found in
\cite[p.\ 394]{Forstneric2023Indag}. % (just above Example 6.3).
The following result is an analogue of the relative Oka principle
for algebraic maps from affine algebraic varieties to algebraically
elliptic manifolds; see
\cite[Theorem 3.1]{Forstneric2006AJM},
\cite[Theorem 6.15.1]{Forstneric2017E}, and
\cite[Theorem 6.4]{Forstneric2023Indag}.
%
%
% THEOREM ON ALGEBRAIC APPROXIMATION
%
\begin{theorem}\label{th:AOP}
Let $\pi:Z\to X$, $H\subset Z$, and $\Omega = Z\setminus H$
be as in Theorem \ref{th:A-embedding}, and let
$Y$ be an algebraically special elliptic manifold.
Given a map $f:Z\to Y$ of class $\Ascr(Z,Y)$ (i.e.,
$f$ is holomorphic and fibrewise algebraic),
a compact $\Oscr(\Omega)$-convex set $K\subset\Omega$, and
a homotopy of holomorphic maps $f_t:U\to Y$ $(t\in[0,1)$
on an open neighbourhood $U$ of $K$ with $f_0=f|_U$,
there are holomorphic maps $F:\Omega \times \C \to Y$
% of class $\Ascr(\Omega \times \C,Y)$
such that $F(\cdotp,0)=f$ and $F(\cdot,t)\in \Ascr(\Omega,Y)$
approximates $f_t$ as closely as desired uniformly on $K$
and uniformly in $t\in[0,1]$.
In particular, a holomorphic map $\Omega \to Y$ that is homotopic
to a map in $\Ascr(\Omega,Y)$ is a limit of maps in
$\Ascr(\Omega,Y)$ uniformly on compacts.
This holds in particular for the Teichm\"uller family
$\Omega=V(g,n)$ for any $g\ge 0$ and $n\ge 1$.
\end{theorem}
\begin{proof}[Proof]
It suffice to inspect the proof of \cite[Theorem 3.1]{Forstneric2006AJM}
and apply Theorem \ref{th:OkaWeil}.
Let $p:E\to Y$ be an algebraic vector bundle and $s:E\to Y$ be a dominating
algebraic spray. Consider the commuting diagram
\[
\xymatrixcolsep{4pc}
\xymatrix{
f^*E \ \ar[d]_{\tilde p} \ar[dr]^{\tilde s} \ar@{^{(}->}[r]^{\iota}
& E \ar[d]_{p} \ar@/^1pc/[d]^{s} \\
\Omega \ar[r]^{f} & Y
}
\]
where the map $\tilde p$ in the first column is the $f$-pullback of the
vector bundle $p:E\to Y$, the map
$\iota$ is the top row is the natural inclusion,
and $\tilde s=s\circ \iota$ is a dominating holomorphic spray over $f$
(the pullback of $s$ by $f$).
Note that $\tilde s=f$ holds on the zero section of $f^*E$.
Since $f$ is fibrewise algebraic and $s$ is algebraic,
the maps $\iota$ and $\tilde s$ are algebraic on
$f^*E|\Omega_x$ for every $x\in X$.
Replacing the spray bundle $(E,p,s)$ with a sufficiently high
iterate of itself (see \cite[Definition 6.3.5]{Forstneric2017E} for this notion)
and shrinking $U$ around $K$ if necessary,
\cite[Proposition 6.5.1]{Forstneric2017E} give a lift of
the homotopy of holomorphic maps $f_t:U\to Y$ in the
theorem to a homotopy of holomorphic sections $\xi_t:U\to f^*E|_U$,
with $\xi_0$ the zero section, such that
$\tilde s\circ \xi_t = f_t$ for $t\in[0,1]$.
Assume now that $p:E\to Y$ is a trivial algebraic vector bundle.
Then, every iterate of $(E,p,s)$ is also a trivial bundle. Since
$f:\Omega\to Y$ is fibrewise algebraic,
the pullback bundle $f^*E\to\Omega$ is fibrewise algebraically
trivial. Hence, sections of $f^*E\to \Omega$
can be identified with maps from $\Omega$ to the fibre,
which is a Euclidean space. Thus,
Theorem \ref{th:OkaWeil} lets us approximate every section
$\xi_t$ uniformly on $K$ by a holomorphic fibrewise algebraic section
$\tilde \xi_t:\Omega\to f^*E$. The map
$F_t=\tilde s \circ \tilde \xi_t:\Omega \to Y$ is then
of class $\Ascr(\Omega,Y)$ and it approximates $f_t$ on $K$
for every $t\in [0,1]$. The approximation can be made uniform in $t\in [0,1]$
by including $[0,1]$ in a standard way into $\C$ and
applying Theorem \ref{th:OkaWeil} with the compact
holomorphically convex set $K\times [0,1]$ in $\Omega\times\C$.
(The homotopy $\{\xi_t\}_{t\in[0,1]}$
must first be approximated on $K\times [0,1]$ by a holomorphic
map in a neighbourhood of $K\times [0,1]$ in $\Omega\times\C$,
which is an application of Mergelyan's theorem.)
This gives a holomorphic map $F:\Omega\times \C \to Y$
satisfying the conclusion of the theorem
which is algebraic on $\Omega_x\times \C$ for every $x\in X$.
\end{proof}
\begin{remark}
The relative algebraic Oka principle in
\cite[Theorem 3.1]{Forstneric2006AJM} holds for all algebraically
elliptic target manifolds $Y$. (It was stated for the ostensibly
bigger class of algebraically subelliptic manifolds, but it has
recently been shown by Kaliman and Zaidenberg
\cite{KalimanZaidenberg2024FM}
that these two classes of manifolds coincide. See also
\cite[Theorem 6.2]{Forstneric2023Indag} for a more complete
list of equivalent algebraic ellipticity conditions.)
We do not know whether Theorem \ref{th:AOP} holds
for this bigger class of manifolds.
\end{remark}
An example of a flexible manifold, which is of particular
interest in the theory of minimal surfaces in Euclidean
spaces $\R^n$, $n\ge 3$, is the punctured null quadric
\[
\boldA = \{z=(z_1,\ldots,z_n)\in \C^n\setminus \{0\}:
z_1^2 + z_2^2 + \cdots + z_n^2=0\}.
\]
(See \cite[Proposition 1.15.3]{AlarconForstnericLopez2021}
and \cite{Andrist2023}.)
Algebraic 1-forms with values in $\bold A$ on finitely punctured compact
Riemann surfaces, which satisfy the real period vanishing
conditions on closed curves in $M$, gives rise to conformal minimal
surfaces $M\to \R^n$ of finite total curvature
via the Enneper--Weierstrass representation formula, and vice versa.
(See \cite{Osserman1986} and
\cite[Chap.\ 4]{AlarconForstnericLopez2021} for a discussion
of this topic.) Hence, the fact that the Teichm\"uller
family $V(g,n)$ for $n\ge 1$ admits nontrivial fibrewise algebraic maps
$V(g,n)\to \boldA$ raises the question whether
there one can realise the fibres of $V(g,n)$ as a family of minimal surfaces
of finite total curvature depending holomorphically on the parameter
$t\in T(g,n)$.
%
%
% ACKNOWLEDGEMENTS
%
%
\medskip
\noindent {\bf Acknowledgements.}
Research was supported by the European Union
(ERC Advanced grant HPDR, 101053085) and grants P1-0291 and
N1-0237 from ARIS, Republic of Slovenia.
I wish to thank Finnur L\'arusson for asking the question
answered by Theorem \ref{th:main} and for
proposing Corollary \ref{cor:2} in a private communication,
and Daniel Huybrechts and Vladimir Markovi\'c for useful
remarks and suggestions.
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%%
%%%%%%%%%% THE BIBLIOGRAPHY
%%%%%%%%%%
%%%%%%%%%%
%{\bibliographystyle{abbrv} \bibliography{references}} \begin{comment}
\begin{thebibliography}{10}
\bibitem{ACampoJiPapadopoulos2016}
N.~A'Campo, L.~Ji, and A.~Papadopoulos.
\newblock On {Grothendieck}'s construction of {Teichm{\"u}ller} space.
\newblock In {\em Handbook of Teichm\"uller theory. Volume VI}, pages 35--69.
Z{\"u}rich: European Mathematical Society (EMS), 2016.
\bibitem{Ahlfors1960}
L.~V. Ahlfors.
\newblock The complex analytic structure of the space of closed {Riemann}
surfaces.
\newblock {\em Princeton Math. Ser.}, 24:45--66, 1960.
\bibitem{AhlforsBers1960}
L.~V. Ahlfors and L.~Bers.
\newblock Riemann's mapping theorem for variable metrics.
\newblock {\em Ann. of Math. (2)}, 72:385--404, 1960.
\bibitem{AlarconForstnericLopez2021}
A.~Alarc\'{o}n, F.~Forstneri\v{c}, and F.~J. L\'{o}pez.
\newblock {\em Minimal surfaces from a complex analytic viewpoint}.
\newblock Springer Monographs in Mathematics. Springer, Cham, 2021.
\bibitem{AlarconForstnericLopez2019JGEA}
A.~Alarc\'{o}n, F.~Forstneri\v{c}, and F.~J. L\'{o}pez.
\newblock Every meromorphic function is the {G}auss map of a conformal minimal
surface.
\newblock {\em J. Geom. Anal.}, 29(4):3011--3038, 2019.
\bibitem{AlarconLarusson2025Crelle}
A.~Alarc{\'o}n and F.~L{\'a}russon.
\newblock Regular immersions directed by algebraically elliptic cones.
\newblock {\em J. Reine Angew. Math.}, 823:113--135, 2025.
\bibitem{AlarconLopez2022APDE}
A.~Alarc\'{o}n and F.~J. L\'{o}pez.
\newblock Algebraic approximation and the {M}ittag-{L}effler theorem for
minimal surfaces.
\newblock {\em Anal. PDE}, 15(3):859--890, 2022.
\bibitem{Andrist2023}
R.~B. Andrist.
\newblock Integrable generators of {Lie} algebras of vector fields on
$\mathrm{SL}_2(\mathbb{C})$ and on $xy = z^2$.
\newblock {\em J. Geom. Anal.}, 33(8):18, 2023.
\newblock Id/No 240.
\bibitem{ArzhantsevFlennerKalimanKutzschebauchZaidenberg2013DMJ}
I.~Arzhantsev, H.~Flenner, S.~Kaliman, F.~Kutzschebauch, and M.~Zaidenberg.
\newblock Flexible varieties and automorphism groups.
\newblock {\em Duke Math. J.}, 162(4):767--823, 2013.
\bibitem{Bers1972}
L.~Bers.
\newblock Uniformization, moduli, and {Kleinian} groups.
\newblock {\em Bull. Lond. Math. Soc.}, 4:257--300, 1972.
\bibitem{BersEhrenpreis1964}
L.~Bers and L.~Ehrenpreis.
\newblock Holomorphic convexity of {Teichm{\"u}ller} spaces.
\newblock {\em Bull. Am. Math. Soc.}, 70:761--764, 1964.
\bibitem{Chirka1989}
E.~M. Chirka.
\newblock {\em Complex analytic sets}, volume~46 of {\em Mathematics and its
Applications (Soviet Series)}.
\newblock Kluwer Academic Publishers Group, Dordrecht, 1989.
\newblock Translated from the Russian by R. A. M. Hoksbergen.
\bibitem{deBruyn2018}
R.~van Dobben~de Bruyn, 2018.
\newblock
\url{https://mathoverflow.net/questions/298710/when-a-family-of-curve-is-an-affine-morphism}.
\bibitem{DocquierGrauert1960}
F.~Docquier and H.~Grauert.
\newblock Levisches {P}roblem und {R}ungescher {S}atz f\"ur {T}eilgebiete
{S}teinscher {M}annigfaltigkeiten.
\newblock {\em Math. Ann.}, 140:94--123, 1960.
\bibitem{EarleKra1976}
C.~J. Earle and I.~Kra.
\newblock On sections of some holomorphic families of closed {Riemann}
surfaces.
\newblock {\em Acta Math.}, 137:49--79, 1976.
\bibitem{EliashbergGromov1992}
Y.~Eliashberg and M.~Gromov.
\newblock Embeddings of {S}tein manifolds of dimension {$n$} into the affine
space of dimension {$3n/2+1$}.
\newblock {\em Ann. of Math. (2)}, 136(1):123--135, 1992.
\bibitem{Forstneric2003AM}
F.~Forstneri\v{c}.
\newblock Noncritical holomorphic functions on {S}tein manifolds.
\newblock {\em Acta Math.}, 191(2):143--189, 2003.
\bibitem{Forstneric2006AJM}
F.~Forstneri\v{c}.
\newblock Holomorphic flexibility properties of complex manifolds.
\newblock {\em Amer. J. Math.}, 128(1):239--270, 2006.
\bibitem{Forstneric2017E}
F.~Forstneri\v{c}.
\newblock {\em Stein manifolds and holomorphic mappings (The homotopy principle
in complex analysis)}, volume~56 of {\em Ergebnisse der Mathematik und ihrer
Grenzgebiete. 3. Folge}.
\newblock Springer, Cham, second edition, 2017.
\bibitem{Forstneric2023Indag}
F.~Forstneri{\v{c}}.
\newblock Recent developments on {Oka} manifolds.
\newblock {\em Indag. Math., New Ser.}, 34(2):367--417, 2023.
\bibitem{Forstneric2025ICM}
F.~Forstneri{\v c}.
\newblock From {S}tein manifolds to {O}ka manifolds: the h-principle in complex
analysis.
\newblock {\em To appear in Proc. ICM 2026}.
\newblock \url{https://arxiv.org/abs/2509.21197}.
\bibitem{Forstneric2024Runge}
F.~Forstneri\v{c}.
\newblock Runge and {M}ergelyan theorems on families of open {R}iemann
surfaces.
\newblock {\em arXiv e-prints}, 2024.
\newblock \url{https://arxiv.org/abs/2412.04608}.
\bibitem{ForstnericLarusson2011}
F.~Forstneri\v{c} and F.~L{\'a}russon.
\newblock Survey of {O}ka theory.
\newblock {\em New York J. Math.}, 17A:11--38, 2011.
\bibitem{ForstnericSigurdardottir2025}
F.~Forstneri{\v c} and {\'A}.~E.~Sigur{\dh}ard{\'o}ttir.
\newblock The Oka principle for tame families of Stein manifolds.
\newblock {\em arXiv e-prints}, 2025.
\url{https://arxiv.org/abs/2510.25887}
\bibitem{Grauert1958MA}
H.~Grauert.
\newblock Analytische {F}aserungen \"uber holomorph-vollst{\"a}ndigen
{R}{\"a}umen.
\newblock {\em Math. Ann.}, 135:263--273, 1958.
\bibitem{Grauert1958AM}
H.~Grauert.
\newblock On {L}evi's problem and the imbedding of real-analytic manifolds.
\newblock {\em Ann. of Math. (2)}, 68:460--472, 1958.
\bibitem{GrauertRemmert1979}
H.~Grauert and R.~Remmert.
\newblock {\em Theory of {S}tein spaces}, volume 236 of {\em Grundlehren Math.
Wiss.}
\newblock Springer-Verlag, Berlin-New York, 1979.
\newblock Translated from the German by Alan Huckleberry.
\bibitem{GrauertRemmert1984}
H.~Grauert and R.~Remmert.
\newblock {\em Coherent analytic sheaves}, volume 265 of {\em Grundlehren Math.
Wiss.}
\newblock Springer-Verlag, Berlin, 1984.
\bibitem{Gromov1989}
M.~Gromov.
\newblock Oka's principle for holomorphic sections of elliptic bundles.
\newblock {\em J. Amer. Math. Soc.}, 2(4):851--897, 1989.
\bibitem{Grothendieck1962}
A.~Grothendieck.
\newblock Techniques de construction en g\'eom\'etrie analytique. {I.
Description axiomatique de l'espace de Teichm\"uller}.
\newblock In {\em S\'eminaire Henri Cartan, tome 13, no. 1 (1960-1961),
expos\'es no.\ 7 et 8, 1--33}. 1962.
\bibitem{GunningNarasimhan1967}
R.~C. Gunning and R.~Narasimhan.
\newblock Immersion of open {R}iemann surfaces.
\newblock {\em Math. Ann.}, 174:103--108, 1967.
\bibitem{HoringPeternell2024}
A.~H{\"o}ring and T.~Peternell.
\newblock Stein complements in compact {K{\"a}hler} manifolds.
\newblock {\em Math. Ann.}, 390(2):2075--2111, 2024.
\bibitem{Hormander1990}
L.~H{\"o}rmander.
\newblock {\em An introduction to complex analysis in several variables},
volume~7 of {\em North-Holland Mathematical Library}.
\newblock North-Holland Publishing Co., Amsterdam, third edition, 1990.
\bibitem{Hubbard1972}
J.~Hubbard.
\newblock Sur la non-existence de sections analytiques {\`a} la courbe
universelle de {Teichm{\"u}ller}. ({On} the non-existence of analytic
sections of an universal {Teichm{\"u}ller} curve).
\newblock {\em C. R. Acad. Sci., Paris, S{\'e}r. A}, 274:978--979, 1972.
\bibitem{Hubbard1976}
J.~H. Hubbard.
\newblock {\em Sur les sections analytiques de la courbe universelle de
{Teichm{\"u}ller}}, volume 166 of {\em Mem. Am. Math. Soc.}
\newblock Providence, RI: American Mathematical Society (AMS), 1976.
\bibitem{KalimanZaidenberg2024FM}
S.~Kaliman and M.~Zaidenberg.
\newblock Gromov ellipticity and subellipticity.
\newblock {\em Forum Math.}, 36(2):373--376, 2024.
\bibitem{Lempert1981}
L.~Lempert.
\newblock La m\'{e}trique de {K}obayashi et la repr\'{e}sentation des domaines
sur la boule.
\newblock {\em Bull. Soc. Math. France}, 109(4):427--474, 1981.
\bibitem{Markovic2018}
V.~Markovi{\'c}.
\newblock Carath{\'e}odory's metrics on {Teichm{\"u}ller} spaces and
{{\(L\)}}-shaped pillowcases.
\newblock {\em Duke Math. J.}, 167(3):497--535, 2018.
\bibitem{Nag1988}
S.~Nag.
\newblock {\em The complex analytic theory of {Teichm{\"u}ller} spaces}.
\newblock Can. Math. Soc. Ser. Monogr. Adv. Texts. New York etc.: Wiley, 1988.
\bibitem{Osserman1986}
R.~Osserman.
\newblock {\em A survey of minimal surfaces}.
\newblock Dover Publications, Inc., New York, second edition, 1986.
\bibitem{Remmert1957}
R.~Remmert.
\newblock Holomorphe und meromorphe {A}bbildungen komplexer {R}\"aume.
\newblock {\em Math. Ann.}, 133:328--370, 1957.
\bibitem{Schurmann1997}
J.~Sch{\"u}rmann.
\newblock Embeddings of {S}tein spaces into affine spaces of minimal dimension.
\newblock {\em Math. Ann.}, 307(3):381--399, 1997.
\bibitem{Serre1955AIF}
J.-P. Serre.
\newblock G\'eom\'etrie alg\'ebrique et g\'eom\'etrie analytique.
\newblock {\em Ann. Inst. Fourier, Grenoble}, 6:1--42, 1955--1956.
\bibitem{Siu1976}
Y.~T. Siu.
\newblock Every {S}tein subvariety admits a {S}tein neighborhood.
\newblock {\em Invent. Math.}, 38(1):89--100, 1976/77.
\bibitem{stacks}
{The Stacks Project Authors}.
\newblock \textit{Stacks Project}.
\newblock \url{https://stacks.math.columbia.edu}, 2018.
\bibitem{Teichmuller1940}
O.~Teichm{\"u}ller.
\newblock Extremal quasiconformal maps and quadratic differentials.
\newblock {\em Abh. Preu{{\ss}}. Akad. Wiss., Math.-Naturw. Kl.},
1940(22):1--197, 1940.
\bibitem{Teichmuller1943}
O.~Teichm{\"u}ller.
\newblock Bestimmung der extremalen quasikonformen {Abbildungen} bei
geschlossenen orientierten {Riemannschen} {Fl{\"a}chen}.
\newblock {\em Abh. Preu{{\ss}}. Akad. Wiss., Math.-Naturw. Kl.}, 1943(4):42
s., 1943.
\bibitem{Teichmuller1944}
O.~Teichm{\"u}ller.
\newblock Ver{\"a}nderliche {Riemannsche} {Fl{\"a}chen}.
\newblock {\em Deutsche Math.}, 7:344--359, 1944.
\bibitem{Wolpert1987}
S.~A. Wolpert.
\newblock Geodesic length functions and the {Nielsen} problem.
\newblock {\em J. Differ. Geom.}, 25:275--296, 1987.
\end{thebibliography}
%\end{comment}
\end{document}