Translators to Higher Order Mean Curvature Flows in $\mathbb R^n\times\mathbb R$ and $\mathbb H^n\times\mathbb R$
Extracted source text · arXiv 2211.03918
\documentclass{amsart}[12pt]
\usepackage[hidelinks]{hyperref}
\usepackage{verbatim}
\usepackage{yfonts} %
\usepackage{amssymb} %
\usepackage{amsthm}
\usepackage{array}
\usepackage{booktabs}%
\usepackage{hhline}%
\usepackage{xy} %
\usepackage{epsfig}%
\usepackage{color}%
\usepackage{upgreek}
\usepackage[english]{babel}
\usepackage{epigraph}%
\usepackage{fancybox}%
\setcounter{totalnumber}{2}
\usepackage{shadow}
\usepackage{afterpage}
\usepackage{mathrsfs}
\usepackage{enumitem}
\usepackage{tabularx}
\usepackage{subcaption}
\usepackage{graphicx}
%\usepackage{epstopdf}
%\usepackage[crop=off]{auto-pst-pdf}
\usepackage{type1cm}
\usepackage{eso-pic}
\usepackage{color}
\usepackage{upgreek}
\usepackage[foot]{amsaddr}
\usepackage{bigints}
%\usepackage[pagewise]{lineno}\linenumbers
\newtheorem{theorem}{Theorem}
\newtheorem{lemma}{Lemma}
\newtheorem{proposition}{Proposition}
\theoremstyle{definition}
\newtheorem{definition}{Definition}
\newtheorem{remark}{Remark}
\newtheorem{example}{Example}
\newtheorem{acknowledgments}{Acknowledgments}
\theoremstyle{plain}
\newtheorem{corollary}{Corollary}
\newtheorem*{dCW}{Theorem}
\newtheorem*{sacks}{Sacksteder Rigidity Theorem}
\newtheorem*{nt}{Theorem 1'\,}
\newtheorem*{alexander}{Theorem}
\newtheorem*{MCP}{Maximum-Continuation Principle}
\newcommand{\otoprule}{\midrule[\heavyrulewidth]}
\newcommand{\red}{\textcolor{red}}
\newcommand{\blue}{\textcolor{blue}}
\newcommand{\vt}{\vspace{.1cm}}
\newcommand{\vtt}{\vspace{.2cm}}
\newcommand{\R}{\mathbb{R} }
\newcommand{\q}{\mathbb{Q} }
\newcommand{\Z}{\mathbb{Z} }
%\newcommand{\C}{\mathbb{C} }
\newcommand{\N}{\mathbb{N} }
\newcommand{\h}{\mathbb{H}}
\newcommand{\ha}{\mathscr{H} }
\newcommand{\hr}{\h^n\times\R}
\newcommand{\harr}{\ha^n\times\R_1}
\newcommand{\hf}{\mathbb{H}_{\mathbb F}^m}
\newcommand{\hfr}{\mathbb{H}_{\mathbb F}^m\times\R}
\newcommand{\g}{{\rm grad}}
\newcommand{\n}{\partial_t}
\newcommand{\s}{\mathbb{S}}
\renewcommand{\thefootnote}{\scriptsize{(\roman{footnote})}}
\renewcommand{\rho}{\varrho}
\renewcommand{\theta}{\varTheta}
\renewcommand{\Theta}{\varTheta}
\renewcommand{\Sigma}{\varSigma}
\renewcommand{\Omega}{\varOmega}
\renewcommand{\Lambda}{\varLambda}
\renewcommand{\tau}{\uptau}
\captionsetup[subfigure]{labelfont=rm}
\renewcommand{\thesubfigure}{\alph{subfigure}}
\usepackage{amsmath}% http://ctan.org/pkg/amsmath
\newcommand{\myfrac}[3][0pt]{\genfrac{}{}{}{}{\raisebox{#1}{$#2$}}{\raisebox{-#1}{$#3$}}}
\newcommand{\overbar}[1]{\mkern 1.5mu\overline{\mkern-1.5mu#1\mkern-1.5mu}\mkern 1.5mu}
\newcommand{\mr}{\overbar M^n\times\R}
%\renewcommand{\overbar{M}\times_\rho\,\R}{\R\times_\rho\overbar{M}}
\newcommand{\ssr}{\scriptscriptstyle{R}}
\newcommand{\ssl}{\scriptscriptstyle{L}}
\newcommand{\qr}{\q_\epsilon^n\times\R}
\newcommand{\transv}{\mathrel{\text{\tpitchfork}}}
\makeatletter
\newcommand{\tpitchfork}{%
\vbox{
\baselineskip\z@skip
\lineskip-.52ex
\lineskiplimit\maxdimen
\m@th
\ialign{##\crcr\hidewidth\smash{$-$}\hidewidth\crcr$\pitchfork$\crcr}
}%
}
\makeatother
\begin{document}
\title[]
{Translators to Higher Order Mean Curvature Flows in $\R^n\times\R$ and
$\h^n\times\R$}
\author{Ronaldo F. de Lima \and Giuseppe Pipoli.}
\address[A1]{Departamento de Matem\'atica - Universidade Federal do Rio Grande do Norte}
\email{ronaldo.freire@ufrn.br}
\address[A2]{Department of Information Engineering, Computer Science and Mathematics, Università degli Studi
dell’Aquila.}
\email{giuseppe.pipoli@univaq.it}
\subjclass[2020]{53E40 (primary), 53C42, 53C45 (secondary).}
\keywords{translators -- higher order mean curvature -- product space.}
\maketitle
\begin{abstract}
We consider translators to the extrinsic flows defined by the higher order mean curvatures $H_r$
($r$-MCF, for short) in $\mathbb R^n\times\mathbb R$ and
$\mathbb H^n\times\R,$ where $n\ge 2$ and $r\in\{1,\dots,n-1\}.$
We show that there exist rotational bowl-type and translating catenoid-type
translators to $r$-MCF in both $\mathbb R^n\times\mathbb R$ and
$\mathbb H^n\times\mathbb R,$ and also that there exist
parabolic and hyperbolic translating catenoid-type translators
to $r$-MCF in $\mathbb H^n\times\mathbb R.$
We also establish the uniqueness (together with the Euclidean and
hyperbolic grim reapers, and the vertical totally geodesic hyperplanes) of
these translators among those which are invariant by either rotations or translations
(Euclidean, parabolic or hyperbolic).
Our main results extend to the context of $r$-MCF in
$\mathbb R^n\times\mathbb R$ and $\mathbb H^n\times\mathbb R$
the existence and uniqueness theorems by Altschuler--Wu (of the bowl soliton) and
Clutterbuck--Schn\"urer--Schulze (of the translating catenoids) in Euclidean space.
\end{abstract}
\section{Introduction}
Extrinsic flows of hypersurfaces in Riemannian manifolds is
a most prominent topic in submanifold theory. Such a flow is generated by
a hypersurface moving in the direction of its normal vector
with speed given by a smooth symmetric function of its principal curvatures.
When this movement constitutes a continuous translation
in a fixed direction, this hypersurface is called a \emph{translating soliton}
(or simply a \emph{translator}) to the flow.
The mean curvature flow (MCF, for short), that is, the extrinsic flow determined
by the mean curvature function, is certainly the most studied extrinsic flow.
Indeed, there is a vast literature on MCF in Euclidean
space and, in particular, on translators to MCF. In this context,
it is well known that translators appear naturally as type II singularities (cf. \cite{huisken-sinestrari}).
The rotationally symmetric translators to MCF in Euclidian space $\R^{n+1}=\R^n\times\R$ are completely classified.
They constitute the entire graph obtained by Altschuler and Wu \cite{altschuler-wu} known as the
\emph{bowl soliton}, and the one-parameter family of annuli obtained by
Clutterbuck, Schn\"urer, and Schulze \cite{schulzeetal} known as \emph{translating catenoids}.
(See \cite[Section 13.1]{andrewsetal} and the references therein for a detailed account on translators to
MCF in Euclidean space.)
Translators are naturally conceived in product spaces $M\times\R,$ where $M$ is a Riemannian manifold.
On this matter, Bueno \cite{bueno} managed to construct bowl-type and translating catenoid-type
translators to MCF in $\h^2\times\R.$ Also, in \cite{lira-martin}, Lira and Martín considered translators to MCF in products
$M\times\R,$ where $M$ is a Hadamard manifold endowed with a rotationally invariant metric.
There, they constructed bowl-type and translating catenoid-type rotational translators, as well as
translators which are invariant by either parabolic or hyperbolic translations of $M\times\R$
(see Section \ref{sec-uniqueness} for definitions). They also classified translators to MCF which are invariant
by either rotations or translations. However, their list of translators
having this property seem to be incomplete (see Remark \ref{rem-liramartin} in Section \ref{sec-uniqueness}).
In this paper, we consider translators
to the extrinsic flows defined by the higher order mean curvatures $H_r,$ $r\in\{1,\dots,n-1\},$ in $\R^n\times\R$ and
$\h^n\times\R.$ For short, we write $r$-MCF for such a flow. More precisely,
we address the problem of constructing and classifying translators to $r$-MCF which are
invariant by either rotations or translations (Euclidean, parabolic or hyperbolic).
Then, we show that there exist rotational bowl-type and translating catenoid-type
translators to $r$-MCF in both $\R^n\times\R$ and $\h^n\times\R,$ and also that there exist
parabolic and hyperbolic translating catenoid-type translators to $r$-MCF in $\h^n\times\R.$
It should be mentioned that, for $r>1,$ some of the translating catenoid-type translators
we obtain here have singular sets of null measure. We also establish the uniqueness
(together with the Euclidean and
hyperbolic grim reapers, and the vertical totally geodesic hyperplanes) of
these translators among those which are invariant by either rotations or translations.
We point out that our main results extend to the context of $r$-MCF
the aforementioned theorems by Altschuler--Wu and Clutterbuck--Schn\"urer--Schulze, as well as
the ones by Bueno. In fact, our methods applied to the case $r=1$ provide new
proofs for these theorems.
The paper is organized as follows. In Section \ref{sec-preliminaries}, we set some notation and introduce the
notion of graph in $\qr$ over parallel hypersurfaces of $\q_\epsilon^n,$ where
$\q_\epsilon^n$ is either $\R^n$ ($\epsilon=0$) or $\h^n$ ($\epsilon=-1$)
and the parallels are totally umbilical in $\q_\epsilon^n.$ In Section \ref{sec-translators}, we discuss translators
to $r$-MCF in $\qr,$ establishing
a key lemma. In Section \ref{sec-rotationaltranslators}, we prove the
existence of rotational bowl-type and translating catenoid-type
translators to $r$-MCF in $\qr.$ The parabolic and hyperbolic versions of these
results for translators in $\h^n\times\R$
are obtained in Sections \ref{sec-parabolic} and \ref{sec-hyperbolic},
respectively. In Section \ref{sec-uniqueness}, we
classify the translators to $r$-MCF in $\qr$ which are invariant by
either rotations or translations, and establish other uniqueness results
regarding such translators. Finally, in Section \ref{sec-nonexistence},
we prove that complete $r$-mean convex translators to $r$-MCF in $\qr$
are not cylindrically bounded.
\section{Preliminaries} \label{sec-preliminaries}
\subsection{Hypersurfaces of $\qr$}
We shall consider oriented hypersurfaces in the product
$\qr$ endowed with its standard product metric,
where $\q_\epsilon^n$ denotes the simply connected space form of
constant sectional curvature $\epsilon\in\{0,-1\},$ i.e., Euclidean space
$\R^n$ or hyperbolic space $\h^n.$
Given an oriented hypersurface $\Sigma$ of $\qr,$
set $N$ for its unit normal field and $A$ for its shape operator with respect to
$N,$ so that
\[
AX=-\overbar\nabla_XN, \,\, X\in T\Sigma,
\]
where $\overbar\nabla$ is the Levi-Civita connection of $\qr,$ and $T\Sigma$ is the tangent bundle of $\Sigma$.
The principal curvatures of $\Sigma,$ that is, the eigenvalues of the shape operator
$A,$ will be denoted by $k_1\,, \dots ,k_n$.
We define the \emph{height function} $\phi$ and the \emph{angle function} $\Theta$
of $\Sigma$ as:
\[
\phi:=\pi_{\scriptscriptstyle\R}|_\Sigma \quad\text{and}\quad \Theta:=\langle N,\partial_t\rangle,
\]
where $\partial _t$ denotes the gradient of the projection $\pi_{\scriptscriptstyle\R}$ of
$\qr$ on its second factor $\R.$ Notice that $\partial_t$ is a Killing field on $\qr.$ So,
denoting by $\nabla u$ the gradient of a smooth function $u$ on $\Sigma$ and writing
$T:=\nabla\phi,$ we have that the identities
\[
T=\partial_t-\theta N\quad\text{and}\quad AT=-\nabla\theta
\]
hold everywhere on $\Sigma.$ From the first of them, one has:
\[
\|T\|^2=1-\theta^2.
\]
Given an integer $r\in\{1,\dots,n\},$ recall that the (non normalized)
$r$-th \emph{mean curvature} $H_r$ of a hypersurface $\Sigma$ of $\qr$
is the function:
\[
H_r:=\sum_{i_1<\cdots <i_r}k_{i_1}\dots k_{i_r}\,.
\]
Notice that $H_1$ and $H_n$ are the non normalized mean curvature and
the Gauss-Kronecker curvature functions of $\Sigma,$ respectively.
A hypersurface $\Sigma$ whose $r$-th mean curvature function is positive
will be called $r$-\emph{mean convex}.
\begin{remark} \label{rem-r-meancurvaturevector}
Setting $\mathbf{k}:=(k_1,\dots ,k_n),$ it is easily seen that the $r$-th mean curvature
function $H_r$ satisfies $H_r(-\mathbf{k})=-H_r(\mathbf{k})$ when $r$ is odd. In this case,
given an orientable hypersurface $\Sigma\subset\qr$ and a unit normal field $N$
on $\Sigma,$ the $r$-\emph{mean curvature vector}
$\mathbf{H_r}:=H_rN$
is well defined, that is, it is independent of the orientation $N.$
On the other hand, for $r$ even, one has $H_r(-\mathbf{k})=H_r(\mathbf{k}),$
so that the $r$-mean curvature vector is not defined.
\end{remark}
\subsection{Graphs on parallel hypersurfaces}
Let $\mathscr F:=\{M_s\subset\q_\epsilon^n\,;\, s\in I\}$
be a family of parallel totally umbilical
hypersurfaces of $\q_\epsilon^n,$ where $I\subset\R$ is an open interval.
Given a smooth function $\phi$ on $I,$ let
$$f\colon M_{s_0}\times I\rightarrow\qr, \,\,\, s_0\in I,$$
be the immersion given by
\[
f(p,s):=(\exp_p(s\eta_{s_0}(p),\phi(s)), \,\,\, (p,s)\in M_{s_0}\times I,
\]
where $\exp$ denotes the exponential map of $\q_\epsilon^n,$ and $\eta_0$ is the unit
normal of $M_{s_0}.$
The hypersurface $\Sigma=f(M_{s_0}\times I)$ is a vertical graph over an open set
of $\q_\epsilon^n$ whose level hypersurfaces are the parallels $M_s$ to
$M_{s_0}.$
\begin{definition}
With the above notation, we shall call $\Sigma$ an $(M_s,\phi)$-\emph{graph}.
\end{definition}
As proved in \cite{delima-manfio-santos}, all the principal curvatures of
an $(M_s,\phi)$-graph $\Sigma$ (when endowed with the metric induced by $f$) are functions of $s$ alone, and are given by:
\begin{equation} \label{eq-principalcurvatures}
k_i(s)=-\alpha(s)\rho(s), \,\,\, i=1,\dots, n-1, \quad\text{and}\quad k_n(s)=\rho'(s),
\end{equation}
where $\alpha(s)$ is the principal curvature function of $M_s,$ and $\rho$ is the function:
\begin{equation}\label{eq-rho}
\rho:=\frac{\phi'}{\sqrt{1+(\phi')^2}}\cdot
\end{equation}
By integrating \eqref{eq-rho}, we conclude that $\rho$ determines the height function
$\phi$ up to a constant.
More precisely:
\begin{equation}\label{eq-phi10}
\phi(s)=\int_{s_0}^{s}\frac{\rho(u)}{\sqrt{1-\rho^2(u)}}du+\phi(s_0), \,\,\, s_0, s\in I.
\end{equation}
It can also be proved that the equality
\begin{equation} \label{eq-theta2}
\theta=\sqrt{1-\rho^2}
\end{equation}
holds everywhere on $\Sigma.$ %In particular, $\rho=\|T\|.$
(See Section 3 of \cite{delima-manfio-santos} for details and proofs.)
\subsection{Trigonometric functions}
We shall adopt a unified notation for Euclidean and
hyperbolic trigonometric functions. Namely, we will
write
\[
\tan_\epsilon=\frac{\sin_\epsilon}{\cos_\epsilon}\quad\text{and}\quad \cot_\epsilon=\frac{1}{\tan_\epsilon}\,,
\]
where $\cos_\epsilon$ and $\sin_\epsilon$ are defined as in
Table \ref{table-trigfunctions}.
\begin{table}[thb]%
\centering %
\begin{tabular}{ccc}
\toprule %
{{\small\rm Function}} & $\epsilon=0$ & $\epsilon=-1$ \\\otoprule
$\cos_\epsilon (s)$ & $1$ & $\cosh s$ \\\midrule
$\sin_\epsilon (s)$ & $s$ & $\sinh s$ \\\bottomrule
\end{tabular}
\vtt
\caption{Definition of $\cos_\epsilon$ and $\sin_\epsilon.$}
\label{table-trigfunctions}
\end{table}
\section{Translators to the $r$-th Mean Curvature Flow} \label{sec-translators}
Given positive integers $n\ge 2$ and $r\in\{1,\dots, n\},$
we say that an oriented hypersurface $\Sigma$ of $\qr$
\emph{moves under} $H_r$-flow if there exists a one-parameter
family of immersions $F\colon M_0\times[0,u_0)\rightarrow M\times\R,$ $u_0\le+\infty,$
such that
\begin{equation} \label{eq-Hrflow}
\left\{
\begin{array}{l}
\frac{\partial F}{\partial u}^\perp(p,u)=H_r(p,u)N(p,u).\\[1ex]
F(M_0\,, 0)=\Sigma,
\end{array}
\right.
\end{equation}
where $N(p,u)$ is the
unit normal to the hypersurface $F_u:=F(.\,, u),$
$H_r(p,u)$ is the $r$-th mean curvature
of $F_u$ with respect to $N_u:=N(.\,, u),$ and
$\frac{\partial F}{\partial u}^\perp$ denotes the normal component of
$\frac{\partial F}{\partial u},$ i.e.,
\[
\frac{\partial F}{\partial u}^\perp=\left\langle\frac{\partial F}{\partial u},N_u\right\rangle N_u\,.
\]
In particular, the first equality in \eqref{eq-Hrflow} is equivalent to
\begin{equation} \label{eq-condition}
\left\langle\frac{\partial F}{\partial u}(p,u),N(p,u)\right\rangle=H_r(p,u).
\end{equation}
We call such a map $F$ an $H_r$-\emph{flow} in $M\times\R.$
Denote by $\exp$ the exponential map of $M\times\R$ and
consider an isometric immersion $F_0\colon M_0\rightarrow M\times\R$.
Define then the map
\[
F(p,u):=\exp_{F_0(p)}(u\partial_t), \,\,(p,u)\in M_0\times [0,+\infty),
\]
and notice that, for each $u\in (0,+\infty),$ the hypersurface $F(M_0\,, u)$ is nothing
but an upwards vertical translation of $\Sigma:=F(M_0,0).$ Since vertical translations are isometries of
$M\times\R$ keeping the second factor invariant,
we have that $\Sigma$ and $F(M_0\,, u)$
are congruent, so that their angle functions and $r$-th
mean curvature functions coincide, that is,
\begin{equation} \label{eq-angles}
\theta(p,u)=\theta(p,0) \quad\text{and}\quad H_r(p,u)=H_r(p,0)\,\,\, \forall (p,u)\in M_0\times [0,u_0).
\end{equation}
Now, differentiating $F$ with respect to $u,$ we have
\begin{equation} \label{eq-partialt}
\frac{\partial F}{\partial u}(p,u)=(d\exp_{F_0(p)})(u\partial_t)\partial_t=\partial_t\,.
\end{equation}
From \eqref{eq-angles} and \eqref{eq-partialt}, we have that
$F$ satisfies \eqref{eq-condition} if and only if
the equality
\[
\theta (p,0)=H_r(p,0)
\]
holds for all $p\in M_0.$ This fact motivates the following concept.
\begin{definition}
Given positive integers $n\ge 2$ and $r\in\{1,\dots, n\},$
we say that a hypersurface $\Sigma$ of $\qr$ is a \emph{translating soliton}
to the $r$-th mean curvature flow ($r$-MCF, for short), if the equality
$H_r=\theta$ holds everywhere on $\Sigma.$ Translating solitons are also called \emph{translators}.
\end{definition}
Let $\Gamma\subset\q_\epsilon^n$ be a totally geodesic hyperplane of
$\q_\epsilon^n.$ Then, $\Sigma=\Gamma\times\R$ is a totally geodesic hypersurface
of $\qr$ which we call a \emph{vertical hyperplane.} On such a $\Sigma,$
$H_r=\theta=0,$ which implies that
$\Sigma$ is translator to $r$-MCF for all $r\in\{1,\dots,n\}.$ Moreover,
from the first equality in \eqref{eq-Hrflow}, all points of $\Sigma$ are
stationary under $r$-MCF.
In total contrast to vertical hyperplanes, $r$-mean convex translators to $r$-MCF in
$\qr$ have no stationary points. On such a translator, one has
$\langle N,\partial_t\rangle=\theta=H_r>0,$ so that the unit normal
$N$ always points upwards.
\begin{remark}
We shall consider $n$-submanifolds $\Sigma$ of
$\qr$ which are of class at least $C^2,$ except on a set of null measure $\Lambda\subset\Sigma,$
where $\Sigma$ is of class $C^{k-1},$ $k\in\{1,2\}.$
In this case, we shall say that $\Sigma$ is $C^k$-\emph{singular} on $\Lambda.$ If the equality
$H_r=\theta$ holds on $\Sigma-\Lambda,$ by abuse of terminology,
we still call $\Sigma$ a \emph{translator} to $r$-MCF
(see Remark \ref{rem-catenoidsmove} in Section \ref{sec-rotationaltranslators}).
\end{remark}
Let $\Phi$ be the reflection over a horizontal hyperplane $\Pi:=\q_\epsilon^n\times\{t\}$
in $\qr.$ Suppose that $\Sigma$ is a translator to $r$-MCF in $\qr$ with unit normal $N,$ %and that $r$ is even.
and call $\overbar\Sigma$ the hypersurface $\Phi(\Sigma)$ with unit normal
$\overbar N\circ\Phi:=-\Phi_*N.$ Then, if $r$ is even, $\overbar\Sigma$ is
a translator to $r$-MCF as well. Indeed, in this case, the $r$-mean curvature function is invariant by
change of orientation. This, together with the fact that $\Phi$ is an isometry, gives that the
$r$-mean curvature $H_r$ of $\Sigma$ at a point $p$ coincides with the
$r$-mean curvature $\overbar H_r$ of $\overbar\Sigma$ at $\Phi(p).$ Therefore,
%denoting by $\theta$ and $\overbar\theta$ the angle functions of $\Sigma$ and
%$\overbar\Sigma,$ respectively, one has
\[
\overbar H_r\circ\Phi=H_r=\langle N,\partial_t\rangle=\langle \Phi_*N,\Phi_*\partial_t\rangle=\langle\overbar N\circ\Phi,\partial_t\rangle,
\]
which gives that $\overbar\Sigma$ is a translator to $r$-MCF.
In the above setting, assume that $\partial\Sigma$ is nonempty and constitutes
a hypersurface of $\Pi.$ Assume further that $\Sigma$ is at least $C^2$ up to its boundary,
and that the angle function $\theta$ of $\Sigma$ extends $C^1$-smoothly to $\partial\Sigma$ and
satisfies either:
\begin{itemize}
\item[i)] $\theta|_{\partial\Sigma}=0.$
\item[ii)] $\theta|_{\partial\Sigma}=1.$
\end{itemize}
Under these conditions, we have that
$\Sigma\cup\Phi(\Sigma)$
is a hypersurface of $\qr$ (to be called a \emph{translator with horizontal symmetry}) which
is at least $C^2$ on $\partial\Sigma$ if (i) occurs, and is $C^1$-singular on $\partial\Sigma$ if (ii) occurs
(see Fig.~\ref{fig-translatingcatenoid}).
%Notice that there is no such a translator to $r$-MCF when $r$ is odd.
\subsection{Graphs on parallels as translators}
Taking the identities \eqref{eq-principalcurvatures} and writing
\[
H_r = \sum_{i_1<\cdots <i_r\ne n}k_{i_1}\dots k_{i_r}+\sum_{i_1<\cdots <i_{r-1}}k_{i_1}\dots k_{i_{r-1}}k_{n}\,,
\]
we have that the $r$-th mean curvature of an $(M_s,\phi)$-graph $\Sigma$ is
a function of $s$ alone, which is given by
\[
H_r={{n-1}\choose{r}}(-\alpha\rho)^r+{{n-1}\choose{r-1}}(-\alpha\rho)^{r-1}\rho'.%=\sqrt{1-\rho^2(s)}.
\]
This last equality, together with \eqref{eq-theta2}, gives the following
\begin{lemma} \label{lem-main}
An $(M_s,\phi$)-graph $\Sigma$ in $\qr$ is a translator to $r$-{\rm MCF} if and only if
its associated $\rho$-function satisfies:
\begin{equation} \label{eq-rMCFodegeneral}
{{n-1}\choose{r}}(-\alpha\rho)^r+{{n-1}\choose{r-1}}(-\alpha\rho)^{r-1}\rho'=\sqrt{1-\rho^2}.
\end{equation}
\end{lemma}
As a first application of Lemma \ref{lem-main}, we shall recover
a classical translator to MCF in Euclidean space.
\begin{example}[\emph{grim reaper}] \label{exam-grimreaper}
Considering Lemma \ref{lem-main} for $\epsilon=0,$ $r=1,$ and $\{M_s\,;\, s\in\R\}$ a family of parallel
totally geodesic hyperplanes in $\R^n,$ we have that \eqref{eq-rMCFodegeneral} reduces to
\[
\rho'=\sqrt{1-\rho^2},
\] which gives $\rho(s)=\sin(s).$ Consequently, the height function
of the corresponding $(M_s,\phi)$-graph $\Sigma$ is (assuming $\phi(0)=0$):
\[
\phi(s)=\int_{0}^{s}\frac{\rho(u)}{\sqrt{1-\rho^2(u)}}du=\int_{0}^{s}\tan(u)du=-\log(\cos s),
\]
so that $\Sigma$ is the solution to MCF in $\R^{n}\times\R$ known as the \emph{grim reaper} (Fig. \ref{fig-grimreaper}).
Notice that, for $r>1,$ \eqref{eq-rMCFodegeneral} reduces to
$\sqrt{1-\rho^2}=0,$ giving that $\rho(s)=1$ for all $s\in(-\infty,+\infty).$ In this case, the
corresponding $(M_s,\phi)$-graph degenerates into a vertical totally geodesic hyperplane of
$\R^{n}\times\R.$
\end{example}
\begin{figure}
\centering
\includegraphics[scale=.8]{grimreaper.pdf}
\caption{\small The grim reaper.}
\label{fig-grimreaper}
\end{figure}
\subsection{Tangency principle for $r$-mean convex translators} \label{subsec-tangencyprinciple}
Let $\Sigma_1$ and $\Sigma_2$ be two $r$-mean convex translators to $r$-MCF in
$\qr$ with possibly nonempty boundary. Assume that %$r$ is odd and that
$\Sigma_1$ and $\Sigma_2$ are tangent at a point
$p$ which is interior to both of them. In this case,
the angle function, and so the $r$-th mean curvature, of
$\Sigma_1$ and $\Sigma_2$ coincide at $p,$ for both unit normals
of these translators point upwards.
Assume further that,when expressed as local graphs over an open set of their common tangent
space at $p,$ $\Sigma_1$ is above $\Sigma_2.$ In this setting, the
interior tangency principle as stated in \cite{fontenele-silva} applies, provided
that $\Sigma_2$ is strictly convex at one of its points
(see also \cite[Section 6]{elbert-saearp}). As a consequence,
$\Sigma_1$ coincides with $\Sigma_2$ in a neighborhood of $p.$ Considering also
the continuation principle for elliptic equations (see \cite{protter}), we conclude that
the interior of one of these translators is necessarily contained in the other.
\begin{remark}
Our methods for constructing translators to $r$-MCF as graphs on parallel hypersurfaces
are not quite effective for $r=n,$ so that this case will not be considered in our main results.
Hence, in order to avoid repetitions, we shall assume once and for all that
\emph{the $r$-mean curvature flows considered here are those for which $1\le r <n.$}
\end{remark}
\section{Rotational Translators to $r$-MCF} \label{sec-rotationaltranslators}
In this section, we concern rotational translators
$\Sigma$ to $r$-MCF in $\qr$ which are vertical
graphs over open sets of $\q_\epsilon^n.$ Such a $\Sigma$
is necessarily an $(M_s,\phi)$-graph, where
$\{M_s\,;\, s\in I\subset(0,+\infty)\}$ is a family
of concentric geodesic spheres of $\q_\epsilon^n$ centered
at a point $o\in\q_\epsilon^n.$
For each $s\in I,$ we shall consider the outward orientation of
$M_s,$ so that its principal curvature function is
$\alpha(s)=-\cot_\epsilon(s).$
In the above setting, we have from \eqref{eq-principalcurvatures} that the principal curvatures of $\Sigma$ are
\begin{equation} \label{eq-principalcurvatures-rotational}
k_i(s)=\cot_\epsilon(s)\rho(s), \,\,\, i=1,\dots, n-1, \quad\text{and}\quad k_n(s)=\rho'(s),
\end{equation}
and that equation \eqref{eq-rMCFodegeneral} is:
\begin{equation} \label{eq-MCFode1}
{{n-1}\choose{r}}\cot_\epsilon^r(s)\rho^r(s)+{{n-1}\choose{r-1}}\cot_\epsilon^{r-1}(s)\rho^{r-1}(s)\rho'(s)=\sqrt{1-\rho^2(s)}.
\end{equation}
Setting $\tau=\rho^r,$ equation \eqref{eq-MCFode1} can be written as
\begin{equation} \label{eq-MCFode2}
\tau'(s)=C\sqrt{1-\tau^{2/r}(s)}\tan_\epsilon^{r-1}(s)-(n-r)\cot_\epsilon(s)\tau(s),
\end{equation}
where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
Equation \eqref{eq-MCFode2} suggests the consideration of the following Cauchy problem:
\begin{equation} \label{eq-IVP}
\left\{
\begin{array}{l}
y'(s)=F(s,y(s))\\[1ex]
y(s_0)=y_0,
\end{array}
\right.
\end{equation}
where $(s_0,y_0)\in\Omega:=(0,+\infty)\times[-1,1],$ and $F=F_{(\epsilon,n,r)}$ is the function:
\begin{equation} \label{eq-F}
F(s,y):=C\sqrt{1-y^{2/r}}\tan_\epsilon^{r-1}(s)-(n-r)\cot_\epsilon(s)y, \,\,\, (s,y)\in\Omega.
\end{equation}
Since $F$ is $C^\infty$ in the interior of $\Omega$
and continuous at the points $(s_0,\pm1)\in\partial\Omega,$
the orbits of the slope field determined by $F$ constitute
a foliation of $\Omega$ by the graphs of the (smooth) solutions of \eqref{eq-IVP} (Fig. \ref{fig-plot}).
Consequently, the end points of such graphs are boundary points of $\Omega.$
\begin{proposition} \label{prop-solutions}
Given $s_0>0,$ let $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ be the solutions of \eqref{eq-IVP}
for $y_0=-1$ and $y_0=1,$ respectively. Then, $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ are both defined in $[s_0,+\infty),$ and there
exists $L=L(\epsilon,n,r)\in(0,1],$ independent of $s_0,$ such that
\begin{equation} \label{eq-L}
\lim_{s\rightarrow+\infty}\tau_{s_0}^-(s)=\lim_{s\rightarrow+\infty}\tau_{s_0}^+(s)=L.
\end{equation}
In addition, the following hold:
\begin{itemize}[parsep=1ex]
\item[\rm i)] $\tau_{s_0}^-$ is increasing in $[s_0,+\infty).$ In particular, $\tau_{s_0}^-$ has one and only one
zero $s_1\in(s_0,+\infty).$
%%%%%%%%%%%%%%%%
\item[\rm ii)] $\tau_{s_0}^+$ is positive %and bounded away from zero
in $[s_0,+\infty).$
\item[\rm iii)] $0<L<1$ if $\epsilon=-1,$ and $L=1$ if $\epsilon=0.$
\end{itemize}
\end{proposition}
\begin{figure}
\centering
\includegraphics[scale=.29]{Hr-plot.png}
\hfill
\includegraphics[scale=.29]{Hr-plote=0.png}
\caption{\small Orbits of the slope field of $F_{(\epsilon, n,r)}$ for $(\epsilon, n,r)=(-1,4,3)$ (left),
and $(\epsilon, n,r)=(0,4,3)$ (right).}
\label{fig-plot}
\end{figure}
The green and blue curves in Figure \ref{fig-plot} are the graphs of $\tau_{s_0}^-$ and $\tau_{s_0}^+,$
respectively, for $s_0\approx 0.2.$
The red curve separating the graphs of $\tau_{s_0}^-$ and $\tau_{s_0}^+$
is the graph of the solution $\tau_0$ defined in Proposition \ref{prop-solutionatzero} below.
\begin{proof}[Proof of Proposition \ref{prop-solutions}]
It follows from \eqref{eq-MCFode2} that $(\tau_{s_0}^-)'(s_0)=(n-r)\cot_\epsilon(s_0)>0,$
so that $\tau_{s_0}^-$ is increasing near $s_0.$ It is also clear from \eqref{eq-MCFode2} that $\tau_{s_0}^-$
will be increasing as long as it stays negative. So, we have two possibilities, $\tau_{s_0}^-$ vanishes at some point
$s_1>s_0,$ or it is defined in $[s_0,+\infty),$ being negative and increasing in this interval. In the latter case,
we must have $(\tau_{s_0}^-)'(s)\rightarrow 0$ as $s\rightarrow+\infty.$ However, under that hypothesis, we would
also have from \eqref{eq-MCFode2} that
\[
0<\lim_{s\rightarrow+\infty}(\tau_{s_0}^-)'(s)\le+\infty,
\]
which is clearly a contradiction. Hence, $\tau_{s_0}^-(s_1)=0$ for some $s_1>s_0.$
Since $(\tau_{s_0}^-)'(s_1)=C\tan_\epsilon(s_1)>0,$ we have that $\tau_{s_0}^-$ is increasing
in a neighborhood of $s_1.$ Assume that $\tau_{s_0}^-$ has a first critical point $s^*>s_0,$
which is necessarily a local maximum, that is,
$(\tau_{s_0}^-)'(s^*)=0\ge(\tau_{s_0}^-)''(s^*).$ From
\eqref{eq-MCFode2}, one has
\[
(\tau_{s_0}^-)''(s^*)=C(r-1)\sqrt{1-\tau^{2/r}(s^*)}\tan_\epsilon^{r-2}(s^*)\,{\rm sec}_\epsilon^2(s^*)+(n-r)\,{\rm csc}_\epsilon^2(s^*)\tau(s^*)>0,
\]
which is a contradiction. Therefore, $\tau_{s_0}^-$ is increasing.
Assume now that there exists $s_2>s_1$ such that $\tau_{s_0}^-(s_2)=1.$ Then,
$(\tau_{s_0}^-)'(s_2)=-(n-r)\cot_\epsilon(s_2)<0,$ which contradicts that $\tau_{s_0}^-$ is increasing.
Therefore, $\tau_{s_0}^-$ is defined in $[s_0,+\infty).$
In particular,
\begin{equation} \label{eq-Ls0}
L_{s_0}^-:=\lim_{s\rightarrow+\infty}\tau_{s_0}^-(s)
\end{equation}
is well defined and satisfies $0<L_{s_0}^-\le 1.$
At the same time, we have that $(\tau_{s_0}^-)'(s)\rightarrow 0$
as $s\rightarrow+\infty,$ which implies that
\begin{equation} \label{eq-limF}
\lim_{s\rightarrow+\infty}F(s,\tau_{s_0}^-(s))=0.
\end{equation}
For $\epsilon=-1,$ we have that
$$F(s,\tau_{s_0}^-(s))=C\sqrt{1-(\tau_{s_0}^-)^{2/r}(s)}\tanh^{r-1}(s)-(n-r)\coth(s)\tau_{s_0}^-(s).$$
From this, and equalities \eqref{eq-Ls0} and \eqref{eq-limF}, one has that $L_{s_0}^-$ satisfies:
\[
C\sqrt{1-(L_{s_0}^-)^{2/r}}-(n-r)L_{s_0}^-=0,
\]
which implies that $L^-:=L_{s_0}^-$ is indeed independent
of $s_0$ and that $0<L^-<1.$
For $\epsilon=0,$ one has
$$F(s,\tau_{s_0}^-(s))=C\sqrt{1-(\tau_{s_0}^-)^{2/r}(s)}\,s^{r-1}-\frac{n-r}{s}\tau_{s_0}^-(s),$$
which, together with \eqref{eq-Ls0} and \eqref{eq-limF}, implies that
$L^-:=L_{s_0}^-=1.$
Regarding $\tau_{s_0}^+,$ we have from \eqref{eq-MCFode2} that
$(\tau_{s_0}^+)'(s_0)=-(n-r)\cot_\epsilon(s_0)<0,$
so that $\tau_{s_0}^+$ is decreasing near $s_0.$ Also,
by the above considerations, the graphs of solutions
to \eqref{eq-IVP} which cross the $s$-axis are increasing. Hence, $\tau_{s_0}^+$ is positive.
Suppose that $s^*>s_0$ is a critical point of $\tau_{s_0}^+.$
Then, as for $\tau_{s_0}^-,$ one has $(\tau_{s_0}^+)''(s^*)>0,$ which implies that
$s^*$ is a strict local minimum. In particular, if $s^*$ exists, it is unique. If so,
$\tau_{s_0}^+$ is increasing in any interval $I\subset(s^*,+\infty)$ where it is defined.
However, as it was for $\tau_{s_0}^-,$ there is no $s_2>s^*$ such that $\tau_{s_0}^+(s_2)=1,$
which implies that $\tau_{s_0}^+$ is defined in $[s_0,+\infty)$ if $s^*$ exists, being increasing
in $[s^*,+\infty).$
If $s^*$ does not exist, then $\tau_{s_0}^+$ is decreasing and well defined in $[s_0,+\infty).$
In this case, $\tau_{s_0}^+$ is bounded away from zero, for the graphs of $\tau_{s_0}^-$ and $\tau_{s_0}^+$
are disjoint. In any case, one has that the limit
\[
L_{s_0}^+:=\lim_{s\rightarrow+\infty}\tau_{s_0}^+(s)
\]
is well defined and satisfies $0<L_{s_0}^+\le 1.$
Proceeding as for $\tau_{s_0}^-,$ we easily conclude that
$L^+:=L_{s_0}^+$ is independent of $s_0$ and that, in fact, the equality
$L^+=L^-$ holds. This finishes the proof.
\end{proof}
\begin{proposition} \label{prop-solutionatzero}
Given $s_0>0,$ let $\tau_{s_0}^-,$ $\tau_{s_0}^+,$ and $L\in(0,1]$ be as in
Proposition \ref{prop-solutions}.
Then, there exists a solution $\tau_0:[0,+\infty)\rightarrow[0,L)$ of
\eqref{eq-IVP} such that:
\begin{itemize}[parsep=1ex]
\item [\rm i)] $\tau_0(0)=0.$
\item [\rm ii)] $\tau_0'(0)=1$ if $r=1,$ and \,$\tau_0'(0)=0$ if $1<r\le n-1.$
\item [\rm iii)] $\tau_0$ is positive and increasing in $(0,+\infty).$
\item [\rm iv)] $\displaystyle \lim_{s\rightarrow+\infty}\tau_0(s)=L.$
\item [\rm v)] For any $s_0>0,$ the inequalities $\tau_{s_0}^-<\tau_0<\tau_{s_0}^+$ hold on $[s_0,+\infty).$
\end{itemize}
\end{proposition}
\begin{proof}
Given $s_0>0,$ let $s_1=s_1(s_0)$ be such that $\tau_{s_0}^-(s_1)=0$ and set
$$\hat\tau_{s_0}=\tau_{s_0}^-|_{[s_1,+\infty)}.$$
Then, $\hat\tau_{s_0}'(s_1)=C\tan_\epsilon^{r-1}(s_1).$ Also,
from Proposition \ref{prop-solutions}, $\hat\tau_{s_0}([s_1,+\infty))=[0,L).$
Hence, since $s_1(s_0)\rightarrow 0$ as $s_0\rightarrow 0,$ we have that
$\hat\tau_{s_0}$ converges uniformly (as $s_0\rightarrow 0$) to a solution
$\tau_0\colon[0,+\infty)\rightarrow[0,L)$ of \eqref{eq-IVP},
which satisfies:
\begin{equation} \label{eq-tau00}
\tau_0(0)=0 \quad\text{and}\quad \tau_0'(0)=\left\{
\begin{array}{ccl}
1 & \text{if} & r=1,\\
0 & \text{if} & 1<r\le n-1.
\end{array}
\right.
\end{equation}
The property of $\tau_0$ stated in (iii) follows from Proposition \ref{prop-solutions}-(ii) and
the positivity of $\hat\tau_{s_0}$ in
$[s_1,+\infty).$ Property (iv) follows from the first limit in \eqref{eq-L}.
Finally, denoting by $\mathcal G_0, \mathcal G_{s_0}^-,$ and $\mathcal G_{s_0}^+$
the graphs of $\tau_0, \, \tau_{s_0}^-,$ and $\tau_{s_0}^+,$ respectively, we have that
$\mathcal G_0$ separates $\Omega$ into two connected components,
one below $\mathcal G_0,$ say $\Omega^-,$ and one above $\mathcal G_0,$ $\Omega^+.$
As can be easily checked, for any $s_0>0,$ one has
$\mathcal G_{s_0}^-\subset\Omega^-$ and
$\mathcal G_{s_0}^+\subset\Omega^+.$ This clearly implies (v) and finishes
our proof.
\end{proof}
\begin{proposition} \label{prop-noblow}
Let $\tau_0$ be as in Proposition \ref{prop-solutionatzero}.
Then, for $\rho_0=\tau_0^{1/r},$ one has that the limits
\[
L_1:=\lim_{s\rightarrow 0}(\cot_\epsilon(s)\rho_0(s))
\quad\text{and}\quad L_2:=\lim_{s\rightarrow 0}\rho_0'(s)
\]
are both finite.
\end{proposition}
\begin{proof}
We have that equality \eqref{eq-MCFode1} holds for $\rho=\rho_0.$ Hence,
$L_1$ and $L_2$ cannot be both infinite, since
$\sqrt{1-\rho_0^2(s)}\rightarrow 1$ as $s\rightarrow 0.$ In addition,
\[
\lim_{s\rightarrow 0}(\cot_\epsilon(s)\rho_0(s))=\lim_{s\rightarrow 0}\left(\frac{\rho_0(s)}{\tan_\epsilon(s)}\right)
=\lim_{s\rightarrow 0}(\rho_0'(s)\cos_\epsilon^2(s)),
\]
which clearly implies that $L_1$ and $L_2$ are both finite.
\end{proof}
\begin{figure}[htb]
\centering
\includegraphics[width=5.5cm]{b-plot.png}
\hfill
\includegraphics[width=6.1cm]{r-bowlsoliton.pdf}
\caption{\small The graph of $\tau_0$ (left) and the
$r$-bowl soliton obtained from it (right).}
\label{fig-bowlsoliton}
\end{figure}
Now, we are in position to state and prove our first main result.
\begin{theorem} \label{th-existence}
Given $r\in\{1,\dots ,n-1\},$ the following assertions hold:
\begin{itemize}[parsep=1ex]
\item[\rm i)] There exists a rotational strictly convex translator $\Sigma_0$ to $r$-{\rm MCF}
in $\qr$ (to be called the $r$-\emph{bowl soliton}) which is an entire vertical graph
contained in the closed half-space $\q_\epsilon^n\times[0,+\infty)$ with unbounded height.
\item[\rm ii)] If \,$r$ is odd, there exists a one-parameter family
$\mathscr C_r=\{\Sigma_\lambda\,;\, \lambda>0\}$ of
properly embedded rotational translators to $r$-{\rm MCF} in $\qr$
(to be called $r$-\emph{translating catenoids})
which are all homeomorphic to an annulus. In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda>0,$ $\Sigma_\lambda$ is the union of two graphs $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+$ over the complement of a ball $B_\lambda(o)\subset\q_\epsilon^n$ which have unbounded height
and satisfy $\partial\Sigma^{\pm}=\partial B_\lambda(o).$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ Each $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$ is contained in a half-space of \,$\qr,$ and its set of
points of minimal height is an $(n-1)$-sphere centered at the axis of rotation which is contained in
a horizontal hyperplane below $\q_\epsilon^n\times\{0\}.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For $r>1,$ any $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$
is $C^2$-singular along its $(n-1)$-sphere of minimal height.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda>0,$ the graphs $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+$ have the same asymptotic behavior of
the $r$-bowl soliton $\Sigma_0.$ More precisely, the angle functions
$\theta^-,$ $\theta^+,$ and\, $\theta_0,$ of\, $\Sigma_\lambda^-,$
$\Sigma_\lambda^+$ and $\Sigma_0,$ respectively, satisfy:
\begin{equation} \label{eq-thetas}
\lim_{s\rightarrow+\infty}\theta^-(s)=\lim_{s\rightarrow+\infty}\theta^+(s)=\lim_{s\rightarrow+\infty}\theta_0(s)=\sqrt{1-L^2},
\end{equation}
where $L=L(\epsilon,n,r)$ equals $1$ if $\epsilon=0,$ and satisfies
$$C\sqrt{1-L^{2/r}}-(n-r)L=0$$ if $\epsilon=-1,$ where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\item[\rm iii)] If \,$r$ is even, there are two one-parameter families
$\mathscr C_r^i=\{\Sigma_\lambda^i\,;\, \lambda>0\},$ $i=1,2,$ of
properly embedded rotational translators to $r$-{\rm MCF} in $\qr$
(to be called $r$-\emph{translating catenoids}) with horizontal symmetry,
which are all homeomorphic to an annulus. In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda>0,$ $\Sigma_\lambda^i$ is the union of a graph
on the complement of a ball $B\subset\q_\epsilon^n\times\{0\}$ (with radius depending on $\lambda$)
with its reflection about
$\q_\epsilon^n\times\{0\}.$ Moreover,
each such graph is contained in a
half-space of $\q_\epsilon^n\times\{0\}$ with
unbounded height, and has $\partial B$ as its boundary.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ The $r$-translating catenoids of $\mathscr C_r^1$
are $C^1$-singular on their intersections with $\q_\epsilon^n\times\{0\},$
whereas the ones of \,$\mathscr C_r^2$ are $C^2$ hypersurfaces of $\qr.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda>0,$ the angle functions
$\theta^i$ and\, $\theta_0,$ of the upper graph of $\Sigma_\lambda^i$
and the bowl soliton $\Sigma_0,$ respectively, satisfy:
\begin{equation} \label{eq-thetaseven}
\lim_{s\rightarrow+\infty}\theta^i(s)=\lim_{s\rightarrow+\infty}\theta_0(s)=\sqrt{1-L^2},
\end{equation}
where $L=L(\epsilon,n,r)$ equals $1$ if $\epsilon=0,$ and satisfies
$$C\sqrt{1-L^{2/r}}-(n-r)L=0$$ if $\epsilon=-1,$ where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
\end{itemize}
\end{theorem}
\begin{proof}
Let $\rho_0$ be as in Proposition \ref{prop-noblow}. Then, by
\eqref{eq-MCFode1}, \eqref{eq-MCFode2}, and Lemma \ref{lem-main},
the rotational graph $\Sigma_0$ with $\rho$-function $\rho_0$
and height function
\[
\phi_0(s)=\int_{0}^{s}\frac{\rho_0(u)}{\sqrt{1-\rho_0^2(u)}}du
\]
is a translator to $r$-MCF in $\qr.$
Setting $\{o\}\times\R,$ $o\in\q_\epsilon^n,$ for the axis of rotation of
$\Sigma_0,$ we have from \eqref{eq-principalcurvatures-rotational} and Proposition \ref{prop-noblow}
that the principal curvatures
of $\Sigma_0$ at the point $o\in\Sigma_0$ are well defined, so that $\Sigma_0$ is at least
$C^2$ at $o.$ Since $0=\phi_0(0)<\phi_0(s)$ for all $s>0,$ we have that
$\Sigma_0$ is contained in the half-space $\q_\epsilon^n\times [0,+\infty),$
and is tangent to $\q_\epsilon^n\times\{0\}$ at $o.$ In particular, $\Sigma_0$ is strictly convex
at $o.$
By Proposition \ref{prop-solutionatzero}, $\tau_0,$ and so $\rho_0,$ is defined in $[0,+\infty),$ which implies that
$\Sigma_0$ is an entire graph over $\q_\epsilon^n\times\{0\}$ (Fig. \ref{fig-bowlsoliton}).
Furthermore, its height function $\phi_0$ is unbounded.
Indeed, since $\tau_0,$ and so $\rho_0,$ is increasing in $(0,+\infty),$ for any $a>0,$ one has
\[
\phi_0(s)>\int_{a}^{s}\frac{\rho_0(u)}{({1-(\rho_0(u))^2})^{1/2}}du\ge
\int_{a}^{s}\rho_0(u)du\ge \rho_0(a)(s-a),
\]
which clearly implies that $\phi_0$ is unbounded. In addition,
it follows from \eqref{eq-principalcurvatures-rotational} that all principal
curvatures $k_i(s)$ of $\Sigma_0$ are positive for $s>0,$ which gives
that $\Sigma_0$ is strictly convex. This proves (i).
To prove (ii), consider $s_0=\lambda>0,$ and let $\tau_\lambda^-$
and $\tau_\lambda^+$ be as in Proposition \ref{prop-solutions}.
Denote by $\Sigma_\lambda^-$ and $\Sigma_\lambda^+$ the rotational graphs whose $\rho$-functions
are $\rho_\lambda^-=(\tau_\lambda^-)^{1/r}$ and $\rho_\lambda^+=(\tau_\lambda^+)^{1/r},$ and
whose height functions are
\[
\phi_\lambda^-(s)=\int_{\lambda}^{s}\frac{\rho_\lambda^-(u)}{({1-(\rho_\lambda^-(u))^2})^{1/2}}du \quad\text{and}\quad
\phi_\lambda^+(s)=\int_{\lambda}^{s}\frac{\rho_\lambda^+(u)}{({1-(\rho_\lambda^+(u))^2})^{1/2}}du,
\]
respectively.
Assuming $r$ odd, let us write $s(\lambda)$ for the point at which
$\rho_\lambda^-$ vanishes, so that
$\rho_\lambda^-$ is negative in the interval $(\lambda,s(\lambda)),$ and positive in $(s(\lambda),+\infty).$
Then, $\phi_\lambda^-(s)$ is decreasing in $(\lambda,s(\lambda)),$
and increasing in $(s(\lambda),+\infty).$ Furthermore, the tangent spaces of
the closure of $\Sigma_\lambda^-$ in $\qr$ along its boundary are all well defined and vertical, for
$\rho_\lambda^-(\lambda)=-1.$
Since
$\tau_{\lambda}^-(s(\lambda))=0<(\tau_{\lambda}^-)'(s(\lambda))$ and
$$(\rho_{\lambda}^-)'(s)=\frac{1}{r}(\tau_{\lambda}^-(s))^{\frac{1-r}{r}}(\tau_{\lambda}^-)'(s),$$
if $r>1,$ one has $(\rho_{\lambda}^-)'(s)\rightarrow +\infty$ as $s\rightarrow s(\lambda).$
From this and \eqref{eq-principalcurvatures-rotational},
we conclude that, for $r>1,$ the second fundamental form of $\Sigma_\lambda^-$ blows up at all points
of its $(n-1)$-sphere of (minimal) height $\phi(s(\lambda)),$ i.e., $\Sigma_\lambda^-$ is
$C^2$-singular at these points.
Clearly, $\rho_\lambda^+$ is positive in $[\lambda,+\infty),$ which implies that
the same is true for $\phi_\lambda^+.$ As for $\Sigma_\lambda^-,$ the tangent spaces of
the closure of $\Sigma_\lambda^+$ in $\qr$ along its boundary are all well defined and vertical.
Since $\partial\Sigma_\lambda^-=\partial\Sigma_\lambda^+=\partial B_\lambda(o),$
it follows from the above considerations that
$$\Sigma_\lambda:={\rm closure}\,(\Sigma_\lambda^-)\cup{\rm closure}\,(\Sigma_\lambda^+)$$ is a
translator to $r$-MCF (Fig. \ref{fig-wing}). Furthermore, just as we did for $\phi_0,$ we
can prove that $\phi_\lambda^-$ and $\phi_\lambda^+$ are both unbounded.
We also point out that $\Sigma_\lambda$ is $C^2$-smooth on the common boundary
$\partial\Sigma_\lambda^\pm$ of $\Sigma_\lambda^\pm.$ To see this, first observe that
the principal curvatures on $\partial\Sigma_\lambda^-$ are (recall that
$\rho_\lambda^-(\lambda)=-\rho_\lambda^+(\lambda)=-1$ and that $r$ is odd):
$$k_i=-\alpha(\lambda)\rho_\lambda^-(\lambda)=\alpha(\lambda)=-\cot_\epsilon(\lambda)<0,$$ and
$$k_n=(\rho_\lambda^-)'(\lambda)=\frac{1}{r}(\tau_\lambda^-(\lambda))^{\frac{1-r}{r}}(\tau_\lambda^-)'(\lambda)=\frac{1}{r}(\tau_\lambda^-)'(\lambda)>0,$$
whereas the ones of $\partial\Sigma_\lambda^+$ are:
$$k_i=-\alpha(\lambda)\rho_\lambda^+(\lambda)=-\alpha(\lambda)=\cot_\epsilon(\lambda)>0,$$
and
$$k_n=(\rho_\lambda^+)'(\lambda)=\frac{1}{r}(\tau_\lambda^+(\lambda))^{\frac{1-r}{r}}(\tau_\lambda^+)'(\lambda)=\frac{1}{r}(\tau_\lambda^+)'(\lambda)<0.$$
However, by (10), $(\tau_\lambda^+)(\lambda)'=-(\tau_\lambda^-)(\lambda)'.$ Hence, after a change of orientation of either
$\Sigma_\lambda^-$ or $\Sigma_\lambda^+,$ we conclude from the above equalities that $\Sigma_\lambda$ is $C^2$-smooth on
$\partial\Sigma_\lambda^\pm.$
\begin{figure}
\centering
\includegraphics[width=5.5cm]{plot-wing0.png}
\hfill
\includegraphics[width=6.1cm]{wing.pdf}
\caption{\small The graphs of $\tau_\lambda^-$ and $\tau_\lambda^+$ (left) and the
$r$-translating catenoid $\Sigma_\lambda$ obtained from them (right). For $r>1,$ $\Sigma_\lambda$ is $C^2$-singular
on the horizontal $(n-1)$-sphere of minimal height.}
\label{fig-wing}
\end{figure}
Now, by Proposition \ref{prop-solutionatzero}-(v), we have that the inequalities
$\rho_\lambda^-(s)<\rho_0(s)<\rho_\lambda^+(s)$
hold for all $s\in[\lambda,+\infty).$ Thus,
$$\phi_\lambda^-(s)<\phi_0(s)<\phi_\lambda^+(s) \,\,\, \forall s\in[\lambda,+\infty).$$
In particular, $\Sigma_\lambda$ is properly embedded.
Finally, equality \eqref{eq-thetas} follows directly from the relation
$\theta^2=1-\rho^2,$ equality \eqref{eq-L}, and Proposition \ref{prop-solutionatzero}-(iv).
This concludes the proof of (ii).
To prove (iii), assume that $r$ is even. In this case, keeping the notation above,
we have to disregard the negative part of $\tau_\lambda^-,$ since
$(\rho_\lambda^-)^r\ge0,$ that is,
\[
\rho_\lambda^-=\pm(\hat\tau_\lambda^-)^{1/r}\quad\text{and}\quad \rho_\lambda^+=\pm(\tau_\lambda^+)^{1/r},
\]
where $\hat\tau_\lambda^-:=\tau_\lambda^-|_{[s(\lambda),+\infty)}>0.$
(Notice that, for $r$ even, $-\rho$ is a solution to \eqref{eq-rMCFodegeneral} whenever $\rho$ is a solution.)
Regarding the choice of signs for $\rho_\lambda^-,$ it is easily seen that, after a vertical translation,
the rotational graph obtained from
the negative $\rho_\lambda^-=-(\hat\tau_\lambda^-)^{1/r}$ is the reflection $\Phi(\Sigma_\lambda^-)$
(over the horizontal hyperplane $\q_\epsilon^n\times\{0\}$) of
the rotational graph $\Sigma_\lambda^-$ obtained from the positive $\rho_\lambda^-=(\hat\tau_\lambda^-)^{1/r}.$
Since $\rho_\lambda^-(s(\lambda))=\tau_\lambda^-(s(\lambda))=0,$ we have that $\theta_\lambda^-(s(\lambda))=1,$
that is, $\theta_\lambda^-|_{\partial\Sigma_\lambda^-}=1.$
This gives that
\[
\Sigma_\lambda^1:={\rm closure}\,(\Sigma_\lambda^-)\cup{\rm closure}\,(\Phi(\Sigma_\lambda^-))
\] is a translator to $r$-MCF with
horizontal symmetry, and that $\Sigma$ is $C^1$-singular on $\Sigma\cap(\q_\epsilon^n\times\{0\})$
(see Section \ref{sec-translators}). We denote by $\mathscr C_r^1$ the one-parameter
family (with parameter $\lambda\in(0,+\infty)$) of all such translators (Fig.~\ref{fig-translatingcatenoid}-left).
The reasoning of the preceding paragraph applies to $\rho_\lambda^+.$ The only difference is that
$\rho_\lambda^+(\lambda)=\tau_\lambda^+(\lambda)=1,$
which yields $\theta_\lambda^+|_{\partial\Sigma_\lambda^+}=0.$ As a consequence, we have that
\[
\Sigma_\lambda^2:={\rm closure}\,(\Sigma_\lambda^+)\cup{\rm closure}\,(\Phi(\Sigma_\lambda^+))
\]
is a $C^2$ hypersurface of $\qr$ for, as in the proof of (ii), the principal curvatures of
$\Sigma_\lambda^+$ extend $C^2$-smoothly to $\partial\Sigma_\lambda^+.$ Therefore,
$\Sigma_\lambda^2$ is a $C^2$ translator to $r$-MCF with horizontal symmetry.
We denote by $\mathscr C_r^2$ the one-parameter
family (with parameter $\lambda\in(0,+\infty)$) of all such translators (Fig. \ref{fig-translatingcatenoid}-right).
The proofs that the height functions of the $r$-translating catenoids in
$\mathscr C_r^i$ are unbounded are analogous to the ones given in the proof of (ii).
The same applies to equality \eqref{eq-thetaseven}. This finishes our proof.
\end{proof}
\begin{figure}
\centering
\includegraphics[width=6cm]{translatingcatenoid0.pdf}
\hspace{.6cm}
\includegraphics[width=4cm]{translatingcatenoid.pdf}
\caption{\small $r$-translating catenoids with horizontal symmetry.}
\label{fig-translatingcatenoid}
\end{figure}
\begin{remark} \label{rem-catenoidsmove}
In what concerns $r(>1)$-translating catenoids
as in Theorem \ref{th-existence}-(ii), it should be noticed that, on each of them,
the $r$-th mean curvature $H_r$ extends $C^1$-smoothly to
the $C^2$-singular $(n-1)$-sphere of minimal height, and equals $1$ there. Hence, despite the fact
that $r(>1)$-translating catenoids have $C^2$-singular sets, they move under the $H_r$-flow,
that is, they are genuine translators to $r$-MCF.
The same goes for the other $r(>1)$-translating catenoids with
$C^2$-singular sets we shall obtain in the next sections.
\end{remark}
\begin{remark} \label{rem-catenoids}
In the above setting, it is not hard to see that, for a given $s_0>0,$
the restriction of $\tau_{s_0}^+$ to $(s_0,+\infty)$ converges uniformly to
$\tau_0|_{(0,+\infty)}$ as $s_0\rightarrow 0.$ By the definition of $\tau_0,$ the same is true
for the restriction of $\tau_{s_0}^-$ to the interval where it is positive. As a consequence, we have that
\emph{the rotational $r$-translating catenoids $\Sigma_\lambda\in\mathscr C_r$ (for $r$ odd),
as well as the upper graphs of the $r$-translating catenoids of $\mathscr C_r^i$ (for $r$ even),
converge (in compact sets) to
$\Sigma_0-\{o\}$ as $\lambda\rightarrow 0,$} where
$\Sigma_0$ is the $r$-bowl soliton of $\qr.$
\end{remark}
\section{Parabolic Translators to $r$-MCF in $\h^n\times\R$} \label{sec-parabolic}
We shall consider now translators
$\Sigma$ to $r$-MCF in $\h^n\times\R$ which are foliated by
horospheres of $\h^n.$ More precisely, such a $\Sigma$
will be obtained from $(\mathscr H_s,\phi)$-graphs, where
$\{\mathscr H_s\,;\, s\in I\subset(-\infty,+\infty)\}$ is a family
of horospheres of $\h^n$ centered
at a fixed point $p_\infty$ of the ideal boundary $\partial_\infty\h^n$ of
$\h^n.$ For each $s\in(-\infty,+\infty),$ we shall consider the orientation of
$\mathscr H_s$ in such a way that its principal curvature function is $\alpha(s)=-1.$
From \eqref{eq-principalcurvatures}, the principal curvatures of $\Sigma$ are
\begin{equation} \label{eq-principalcurvatures-parabolic}
k_i(s)=\rho(s), \,\,\, i=1,\dots, n-1, \quad\text{and}\quad k_n(s)=\rho'(s),
\end{equation}
and so equation \eqref{eq-rMCFodegeneral} is:
\begin{equation} \label{eq-MCFode-parabolic}
{{n-1}\choose{r}}\rho^r(s)+{{n-1}\choose{r-1}}\rho^{r-1}(s)\rho'(s)=\sqrt{1-\rho^2(s)}.
\end{equation}
As before, setting $\tau=\rho^r,$ equation \eqref{eq-MCFode-parabolic} takes the form
\begin{equation} \label{eq-parabolicMCFode2}
\tau'(s)=C\sqrt{1-\tau^{2/r}(s)}-(n-r)\tau(s),
\end{equation}
where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
Hence, the Cauchy problem we consider now is:
\begin{equation} \label{eq-IVP-parabolic}
\left\{
\begin{array}{l}
y'(s)=F(y(s))\\[1ex]
y(s_0)=y_0,
\end{array}
\right.
\end{equation}
where $(s_0,y_0)\in\Omega:=(-\infty,+\infty)\times[-1,1],$ and $F=F_{(n,r)}$ is the function:
\begin{equation} \label{eq-Fparabolic}
F(y):=C\sqrt{1-y^{2/r}}-(n-r)y, \,\,\, (s,y)\in\Omega.
\end{equation}
\begin{figure}
\centering
\includegraphics[scale=.3]{parabolicplot1.png}
\caption{\small Orbits of the slope field of $F_{(n,r)}$ for $(n,r)=(4,3).$}
\label{fig-parabolicplot1}
\end{figure}
The next result is an analogue of Proposition \ref{prop-solutions}.
\begin{proposition} \label{prop-parabolicsolutions}
Given $s_0\in(-\infty,+\infty),$ let $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ be the solutions of \eqref{eq-IVP-parabolic}
for $y_0=-1$ and $y_0=1,$ respectively. Then, $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ are both defined in $[s_0,+\infty)$ and satisfy:
\begin{equation} \label{eq-parabolicL}
\lim_{s\rightarrow+\infty}\tau_{s_0}^-(s)=\lim_{s\rightarrow+\infty}\tau_{s_0}^+(s)=L,
\end{equation}
where $L\in(0,1)$ is the positive root of the equation $F(y)=0.$
In addition, the following hold:
\begin{itemize}[parsep=1ex]
\item [\rm i)] The constant function $\tau_0(s)=L,$ $s\in(-\infty,+\infty),$ is a solution of \eqref{eq-IVP-parabolic}.
\item[\rm ii)] $\tau_{s_0}^-$ is increasing in $[s_0,+\infty).$ In particular,
$\tau_{s_0}^-$ has one and only one zero $s_1\in(s_0,+\infty).$
%%%%%%%%%%%%%%%%
\item[\rm iii)] $\tau_{s_0}^+$ is positive
in $[s_0,+\infty).$
\end{itemize}
\end{proposition}
\begin{proof}
Assertion (i) is trivial. The proofs of (ii), (iii), and equality \eqref{eq-parabolicL}
are analogous to the corresponding ones in the proof of Proposition \ref{prop-solutions}.
\end{proof}
\begin{figure}
\centering
\includegraphics[width=6cm]{parabolicplot2.png}
\hspace{1cm}
\includegraphics[width=4.5cm]{parabolicgraph.pdf}
\caption{\small The graph of $\tau_0$ (left) and the
parabolic $r$-bowl soliton obtained from it (right).}
\label{fig-parabolicbowl}
\end{figure}
\begin{theorem} \label{th-existence-parabolic}
Let
$\mathscr F:=\{\mathscr H_s\,;\, s\in(-\infty,+\infty)\}$
be a family of parallel horospheres of \,$\h^n.$
Then, the following assertions hold:
\begin{itemize}[parsep=1ex]
\item[\rm i)] There exists a convex (and nowhere strictly convex) translator $\Sigma_0$ to $r$-{\rm MCF}
in $\h^n\times\R$ (to be called the \emph{parabolic} $r$-\emph{bowl soliton}) which is a constant angle entire vertical graph
foliated by (vertical translations of) the horospheres of $\mathscr F,$ whose height function is unbounded
above and below.
\item[\rm ii)] If \,$r$ is odd, there exists a one-parameter family
$\mathscr C_r=\{\Sigma_\lambda\,;\, \lambda\in(-\infty,+\infty)\}$ of
properly embedded translators to $r$-{\rm MCF} in $\h^n\times\R$
(to be called \emph{parabolic} $r$-\emph{translating catenoids})
which are all homeomorphic to Euclidean space $\R^n.$ In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda\in(-\infty,+\infty),$ $\Sigma_\lambda$ is the union of two graphs $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+$ over the complement of the horoball bounded by $\mathscr H_\lambda\subset\h^n,$ both unbounded from above,
such that $\partial\Sigma^{\pm}=\mathscr H_\lambda.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ Each parabolic $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$
is contained in a half-space of \,$\h^n\times\R,$
and its set of points of minimal height is a vertically translated horosphere of $\mathscr F$
contained in a horizontal hyperplane below \,$\h^n\times\{0\}.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For $r>1,$ any parabolic $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$
is $C^2$-singular along its horosphere of minimal height.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda\in(-\infty,+\infty),$ the graphs $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+$ have the same asymptotic behavior of
the constant angle parabolic $r$-bowl soliton $\Sigma_0.$ More precisely, the angle functions
$\theta_\lambda^-,$ $\theta_\lambda^+,$ and\, $\theta_0,$ of\, $\Sigma_\lambda^-,$
$\Sigma_\lambda^+$ and $\Sigma_0,$ respectively, satisfy:
\begin{equation} \label{eq-parabolicthetas}
\lim_{s\rightarrow+\infty}\theta_\lambda^-(s)=\lim_{s\rightarrow+\infty}\theta_\lambda^+(s)=\theta_0\,.
\end{equation}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\item[\rm iii)] If \,$r$ is even, there are two one-parameter families
$\mathscr C_r^i=\{\Sigma_\lambda^i\,;\, \lambda\in\R\},$ $i=1,2,$ of
properly embedded translators to $r$-{\rm MCF} in $\h^n\times\R$
(to be called \emph{parabolic} $r$-\emph{translating catenoids}) with horizontal symmetry,
which are all homeomorphic to Euclidean space $\R^n.$ In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda\in(-\infty,+\infty),$ $\Sigma_\lambda^i$ is the union of a graph
on the complement of a horoball $B\subset\q_\epsilon^n\times\{0\}$ (depending on $\lambda$)
with its reflection about $\h^n\times\{0\}.$ Moreover,
each such graph is contained in a
half-space of $\h^n\times\{0\}$ with
unbounded height, and has $\partial B$ as its boundary.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ The $r$-translating catenoids of $\mathscr C_r^1$
are $C^1$-singular on their intersections with $\h^n\times\{0\},$
whereas the ones of \,$\mathscr C_r^2$ are $C^2$ hypersurfaces of $\h^n\times\R.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda\in\R,$ the angle function
$\theta_\lambda^i$ of the upper graph of $\Sigma_\lambda^i\in\mathscr C_r^i$ satisfies:
\begin{equation} \label{eq-thetasevenparabolic}
\lim_{s\rightarrow+\infty}\theta_\lambda^i(s)=\theta_0(s).
\end{equation}
% where $L=L(\epsilon,n,r)$ equals $1$ if $\epsilon=0,$ and satisfies
%$$C\sqrt{1-L^{2/r}}-(n-r)L=0$$ if $\epsilon=-1,$ where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
\end{itemize}
\end{theorem}
\begin{figure}
\centering
\includegraphics[width=5.5cm]{parabolicplot3.png}
\hfill
\includegraphics[width=5cm]{parabolicwing.pdf}
\caption{\small The graphs of $\tau_\lambda^-$ and $\tau_\lambda^+$ (left) and the
parabolic $r$-translating catenoid $\Sigma_\lambda$ obtained from them (right). For $r>1,$ $\Sigma_\lambda$ is $C^2$-singular
on the horizontal horosphere of minimal height.}
\label{fig-parabolicwing}
\end{figure}
\begin{proof}
Let $\rho_0=\tau_0^{1/r},$ where $\tau_0=L$ is the constant function defined in
Proposition \ref{prop-parabolicsolutions}.
Then, by \eqref{eq-MCFode-parabolic}, \eqref{eq-parabolicMCFode2},
and Lemma \ref{lem-main},
the $(\mathscr H_s,\phi)$-graph $\Sigma_0$ with $\rho$-function $\rho_0$
and height function
\[
\phi_0(s)=\int_{0}^{s}\frac{\rho_0(u)}{\sqrt{1-\rho_0^2(u)}}du=\frac{\rho_0}{\sqrt{1-\rho_0^2}}s, \,\,\, s\in(-\infty,+\infty),
\]
is a translator to $r$-MCF in $\h^n\times\R.$
Since $\phi_0$ is a linear function on $\R,$ and $\rho_0$ is constant,
$\Sigma_0$ is an entire graph over $\q_\epsilon^n\times\{0\}$ whose angle
function $\theta_0$ is constant (for $\theta_0^2=1-\rho_0^2$),
and whose height function is unbounded from above and from below
(Fig. \ref{fig-parabolicbowl}).
In addition,
it follows from \eqref{eq-principalcurvatures-parabolic} that the principal
curvatures $k_i$ of $\Sigma_0$ are all constant and positive, except for $k_n,$ which vanishes
everywhere. In particular, $\Sigma_0$ is convex and nowhere strictly convex. This proves (i).
The proofs of (ii) and (iii) are completely analogous to the ones given for
assertions (ii) and (iii) of Theorem \ref{th-existence}.
\end{proof}
\section{Hyperbolic Translators to $r$-MCF in $\h^n\times\R$} \label{sec-hyperbolic}
In this section, we shall consider translators
$\Sigma$ to $r$-MCF in $\h^n\times\R$ which are invariant by
hyperbolic horizontal translations of $\h^n\times\R.$
Such a $\Sigma$ will be constructed from $(\mathscr E_s,\phi)$-graphs, where
$\{\mathscr E_s\,;\, s\in I\subset(-\infty,+\infty)\}$ is a family
of equidistant hypersurfaces of $\h^n$ which are equidistant from the totaly geodesic
hyperplane $\mathscr E_0\subset\h^n.$
For each $s\in(-\infty,+\infty),$ we shall consider the orientation of
$\mathscr E_s$ in such a way that its principal curvature function is
$\alpha(s)=-\tanh s.$
\begin{figure}
\centering
\includegraphics[width=4cm]{parabolictranslatingcatenoid0.pdf}
\hspace{1cm}
\includegraphics[width=4cm]{parabolictranslatingcatenoid.pdf}
\caption{\small Parabolic $r$-translating catenoids with horizontal symmetry.}
\label{fig-parabolictranslatingcatenoid}
\end{figure}
From \eqref{eq-principalcurvatures}, the principal curvatures of $\Sigma$ are
\begin{equation} \label{eq-principalcurvatures-parabolic}
k_i(s)=\rho(s)\tanh s, \,\,\, i=1,\dots, n-1, \quad\text{and}\quad k_n(s)=\rho'(s),
\end{equation}
so that equation \eqref{eq-rMCFodegeneral} now is:
\begin{equation} \label{eq-MCFode-hyperbolic}
{{n-1}\choose{r}}\tanh^r(s)\rho^r(s)+{{n-1}\choose{r-1}}\tanh^{r-1}(s)\rho^{r-1}(s)\rho'(s)=\sqrt{1-\rho^2(s)}.
\end{equation}
Setting $\tau=\rho^r,$ equation \eqref{eq-MCFode-hyperbolic} takes the form
\begin{equation} \label{eq-hyperbolicMCFode2}
\tau'(s)=C\sqrt{1-\tau^{2/r}(s)}\coth^{r-1}(s)-(n-r)\tanh(s)\tau(s),
\end{equation}
where $C=C(n,r)=r{{n-1}\choose{r-1}}^{-1}.$
Now, define
\[
I:=\left\{
\begin{array}{lcc}
(-\infty,+\infty) & \text{if} & r=1,\\[1ex]
(0,+\infty) & \text{if} & r>1,
\end{array}
\right.
\]
and consider the Cauchy problem:
\begin{equation} \label{eq-IVP-hyperbolic}
\left\{
\begin{array}{l}
y'(s)=F(s,y(s))\\[1ex]
y(s_0)=y_0,
\end{array}
\right.
\end{equation}
where $(s_0,y_0)\in\Omega:=I\times\R$ and $F=F_{(n,r)}$ is the function:
\begin{equation} \label{eq-Fhyperbolic}
F(s,y):=C\sqrt{1-y^{2/r}}\coth^{r-1}(s)-(n-r)\tanh(s)y, \,\,\, (s,y)\in\Omega.
\end{equation}
\begin{figure}[htb]
\centering
\includegraphics[scale=.26]{hyperbolicplot1.png}
\hspace{1cm}
\includegraphics[scale=.26]{hyperbolicplot2.png}
\caption{\small Orbits of the slope field of $F_{(4,1)}$ (left) and $F_{(4,3)}$ (right).}
\label{fig-hyperbolicfigure1}
\end{figure}
\begin{proposition} \label{prop-hyperbolicsolutions}
Given $s_0>0,$ let $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ be the solutions of \eqref{eq-IVP-hyperbolic}
for $y_0=-1$ and $y_0=1,$ respectively.
Then, $\tau_{s_0}^-$ and \,$\tau_{s_0}^+$ are both defined in $[s_0,+\infty),$ and there
exists $L=L(n,r)\in(0,1),$ independent of $s_0,$ such that
\begin{equation} \label{eq-hyperbolicL}
\lim_{s\rightarrow+\infty}\tau_{s_0}^-(s)=\lim_{s\rightarrow+\infty}\tau_{s_0}^+(s)=L.
\end{equation}
In addition, the following hold:
\begin{itemize}[parsep=1ex]
\item[\rm i)] $\tau_{s_0}^-$ has one and only one zero $s_1\in(s_0,+\infty).$
%%%%%%%%%%%%%%%%
\item[\rm ii)] $\tau_{s_0}^+$ is positive in $[s_0,+\infty).$
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\item [\rm iii)] For $r=1$ and $\lambda\in(-1,1),$ the solution $\tau_\lambda$ of \eqref{eq-IVP-hyperbolic}
satisfying $\tau_\lambda(0)=\lambda$ is defined on $(-\infty,+\infty),$ and
$\tau_\lambda(s)\to\pm L$ as ${s\to\pm\infty}.$
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\end{itemize}
\end{proposition}
\begin{proof}
Reasoning as in the proof of Proposition \ref{prop-solutions}-(i), we conclude that there exists
$s_1\in(s_0,+\infty)$ such that $\tau_{s_0}^-(s_1)=0.$ At $s_1,$ we have
from \eqref{eq-hyperbolicMCFode2} that the derivative of $\tau_{s_0}^-$
is positive. Thus, $s_1$ is the only zero of $\tau_{s_0}^-.$ Moreover, since
$F(s,1)<0$ for all $s>0$ ($F$ as in \eqref{eq-Fhyperbolic}), we have that there is no $s\in(s_0,+\infty)$ such that
$\tau_{s_0}^-(s)=1,$ which implies that $\tau_{s_0}^-$ is defined in $[s_0,+\infty).$
We also have from \eqref{eq-hyperbolicMCFode2} that $(\tau_{s_0}^+)'(s_0)<0,$ that is,
$\tau_{s_0}^+$ is decreasing near $s_0.$ In addition, the arguments in the
preceding paragraph also show that
$\tau_{s_0}^+$ cannot have zeros, and that there is no $s\in(s_0,+\infty)$ such that
$\tau_{s_0}^+(s)=1.$ Hence, $\tau_{s_0}^+$ is positive and is defined in $[s_0,+\infty).$
The proof of \eqref{eq-hyperbolicL} is analogous to the one given for \eqref{eq-L}.
It remains to prove (iii). To this end, notice that the function $F$ for $r=1$ is
\[
F(s,y)=\sqrt{1-y^{2}}-(n-1)\tanh(s)y,
\]
which implies that, for any $\lambda\in(-1,1),$
$\tau_\lambda$ is well defined in an open interval $I_0:=(-\delta,\delta)\owns 0.$ In addition,
$\tau_\lambda'(\lambda)=\sqrt{1-\lambda^{2}}>0,$ so that $\tau_\lambda$ is increasing
in $I_0$ if $\delta$ is sufficiently small. Since $F(s_0,1)<0<F(s_0,-1)$ (resp. $F(s_0,-1)<0<F(s_0,1)$)
for $s_0>0$ (resp. $s_0<0),$ we have that $\tau_\lambda(s)\ne\pm1,$ which implies that
$\tau_\lambda$ is defined on $(-\infty,+\infty).$
Finally, writing
\[
L_{\pm}^\lambda:=\lim_{s\to\pm\infty}\tau_\lambda(s),
\]
and recalling that $L$ satisfies
$\sqrt{1-L^{2}}-(n-1)L=0,$
we have from $$0=\lim_{s\to\pm\infty}\tau_\lambda'(s)=\lim_{s\to\pm\infty}F(s,\tau_\lambda(s))$$
that
\[
\sqrt{1-(L_{\pm}^\lambda)^{2}}\mp(n-1)L_{\pm}^\lambda=0,
\]
which gives $-L_-^\lambda=L_+^\lambda=L$ (Fig. \ref{fig-hyperbolicfigure1}).
\end{proof}
\begin{figure}
\centering
\includegraphics[width=5.2cm]{hyperbolicplot4.png}
\hfill
\includegraphics[width=6cm]{hyperbolicgrimreaper.pdf}
\caption{\small The graph of $\tau_0$ (left) and half of the
hyperbolic grim reaper $\Sigma_0$ obtained from it (right).}
\label{fig-hypgrimreaper}
\end{figure}
\begin{theorem} \label{th-existence-hyperbolic}
Let $\mathscr F:=\{\mathscr E_s\,;\, s\in(-\infty,+\infty)\}$
be a family of equidistant hypersurfaces of \,$\h^n.$
Then, the following assertions hold:
\begin{itemize}[parsep=1ex]
\item[\rm i)] For $r=1,$ there exists a one-parameter family
$\mathscr G=\{\Sigma_\lambda\,;\, \lambda\in(-1,1)\}$ of entire graphs
(to be called \emph{hyperbolic grim reapers}) whose level hypersurfaces
are the elements of $\mathscr F.$ Each $\Sigma_\lambda$ is unbounded and is contained in the
half-space $\h^n\times[0,+\infty),$ being tangent to an equidistant hypersurface
$\mathscr E_{s(\lambda)}\subset\h^n\times\{0\}.$
\item[\rm ii)] If \,$r$ is odd, there exists a one-parameter family
$\mathscr C_r=\{\Sigma_\lambda\,;\, \lambda\in(0,+\infty)\}$ of
properly embedded translators to $r$-{\rm MCF} in $\h^n\times\R$
(to be called \emph{hyperbolic} $r$-\emph{translating catenoids})
which are all homeomorphic to Euclidean space $\R^n.$ In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda\in(0,+\infty),$ $\Sigma_\lambda$ is the union of two graphs $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+,$ both unbounded from above, over one of the connected components of
the complement of the convex region of $\h^n$ bounded by $\mathscr E_0$ and $\mathscr E_\lambda.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ Each hyperbolic $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$
is contained in a half-space of \,$\h^n\times\R,$
and its set of points of minimal height is a vertically translated equidistant hypersurface of $\mathscr F$
contained in a horizontal hyperplane below $\h^n\times\{0\}.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For $r>1,$ any hyperbolic $r$-translating catenoid $\Sigma_\lambda\in\mathscr C_r$
is $C^2$-singular along its equidistant hypersurface of minimal height.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda\in(0,+\infty),$ the angle functions
$\theta^-$ and \,$\theta^+$ of $\Sigma_\lambda^-$ and
$\Sigma_\lambda^+,$ respectively, satisfy:
\begin{equation} \label{eq-parabolicthetas}
\lim_{s\rightarrow+\infty}\theta^-(s)=\lim_{s\rightarrow+\infty}\theta^+(s)=\sqrt{1-L^{2/r}},
\end{equation}
where $L$ satisfies
$C\sqrt{1-L^{2/r}}-(n-r)L=0,$ being $C=r{{n-1}\choose{r-1}}^{-1}.$
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\item[\rm iii)] If \,$r$ is even, there are two one-parameter families
$\mathscr C_r^i=\{\Sigma_\lambda^i\,;\, \lambda>0\},$ $i=1,2,$ of
properly embedded translators to $r$-{\rm MCF} in $\h^n\times\R$
(to be called \emph{hyperbolic} $r$-\emph{translating catenoids}) with horizontal symmetry,
which are all homeomorphic to Euclidean space $\R^n.$ In addition, one has that:
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For each $\lambda>0,$ $\Sigma_\lambda^i$ is the union of a graph
over one of the connected components of
the complement of the convex region $V$ of $\h^n$ bounded by $\mathscr E_0$ and an equidistant
$\mathscr E_{\bar\lambda},$ $\bar\lambda=\bar\lambda(\lambda),$
with its reflection about $\h^n\times\{0\}.$ Moreover,
each such graph is contained in a
half-space of $\h^n\times\{0\}$ with
unbounded height, and has $\partial V$ as its boundary.
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ The hyperbolic $r$-translating catenoids of $\mathscr C_r^1$
are $C^1$-singular on their intersections with $\h^n\times\{0\},$
whereas the ones of \,$\mathscr C_r^2$ are $C^2$ hypersurfaces of $\h^n\times\R.$
%%%%%%%%%%%%%%%%%%%%
\subitem $\bullet$ For any $\lambda>0,$ the angle function
$\theta_\lambda^i$ of the upper graph of $\Sigma_\lambda^i\in\mathscr C_r^i$ satisfies:
\begin{equation} \label{eq-thetasevenhyperbolic}
\lim_{s\rightarrow+\infty}\theta_\lambda^i(s)=\sqrt{1-L^{2/r}},
\end{equation}
where $L$ satisfies
$C\sqrt{1-L^{2/r}}-(n-r)L=0,$ being $C=r{{n-1}\choose{r-1}}^{-1}.$
\end{itemize}
\end{theorem}
\begin{figure}
\centering
\includegraphics[width=5.5cm]{hyperbolicplot3.png}
\hfill
\includegraphics[width=5cm]{hyperbolicwing.pdf}
\caption{\small The graphs of $\tau_\lambda^-$ and $\tau_\lambda^+$ (left) and the
hyperbolic $r$-translating catenoid $\Sigma_\lambda$ obtained from them (right). For $r>1,$ $\Sigma_\lambda$ is $C^2$-singular
on the horizontal equidistant hypersurface of minimal height.}
\label{fig-hyperbolicwing}
\end{figure}
\begin{proof}
Given $\lambda\in(-1,1),$ let $\tau_\lambda\colon(-\infty,+\infty)\rightarrow\R$ be as in
Proposition \ref{prop-hyperbolicsolutions}-(iii). Set
$\Sigma_\lambda$ for the $(\mathscr E_s,\phi_\lambda)$-graph
with $\rho$-function $\rho_\lambda=\tau_\lambda$ and height function
\[
\phi_\lambda(s)=\int_{s(\lambda)}^{s}\frac{\rho_\lambda(u)}{\sqrt{1-\rho_\lambda^2(u)}}du, \,\,\, s\in(-\infty,+\infty),
\]
where $s(\lambda)$ satisfies $\rho_\lambda(s(\lambda))=0.$
Then, $\Sigma$ is an entire graph over $\h^n$ and, by Lemma \ref{lem-main}, is a translator to MCF
in $\h^n\times\R.$ Also, since $\rho_\lambda$ is negative in $(-\infty,s(\lambda))$ and positive in $(s(\lambda),+\infty),$
we have that $\phi_\lambda(s)>0$ for all $s\ne s(\lambda),$ which implies that $\Sigma_\lambda$ is contained in the
half-space $\h^n\times[0,+\infty),$ and is tangent to the equidistant hypersurface
$\mathscr E_{s(\lambda)}\subset\h^n\times\{0\},$ for
$\phi_\lambda(s(\lambda))=\phi_\lambda'(s(\lambda))=0$ (see Fig. \ref{fig-hypgrimreaper}).
To prove that $\phi_\lambda$ is unbounded, notice that the
function $\rho_\lambda/\sqrt{1-\rho_\lambda^2}$ is bounded below by a positive constant $C_0$
in any interval $(a,+\infty)$ with $a>s(\lambda)$ sufficiently large, for
\[
\lim_{s\rightarrow+\infty}\frac{\rho_\lambda(s)}{\sqrt{1-\rho_\lambda^2(s)}}=\frac{L}{\sqrt{1-L^2}}>0.
\]
Then, for any $s\in (a,+\infty),$ one has
\[
\phi_\lambda(s)>\int_{a}^{s}\frac{\rho_0(u)}{\sqrt{1-\rho_0^2(u)^2}}du\ge C_0(s-a),
\]
which implies that $\phi_\lambda$ is unbounded. This proves (i).
The proofs of (ii) and (iii) are completely analogous to the ones given for
assertions (ii) and (iii) of Theorem \ref{th-existence}.
\end{proof}
\begin{remark} \label{rem-symmetryequation}
The function $F$ defined in \eqref{eq-Fhyperbolic}
satisfies $F(s,y)=F(-s,-y)\,\forall(s,y)\in\Omega.$ Hence, the trajectories of
$\tau_{-s_0}^{\pm}$ for $s_0>0$ are the reflections about
the origin $(0,0)$ of the $(s,\tau)$-plane of the trajectories
of $\tau_{s_0}^{\pm}.$ In particular, the corresponding hyperbolic
$r$-translating catenoids obtained from them are congruent. For that reason, only the case
$s_0>0$ was considered in the statement of Proposition \ref{prop-hyperbolicsolutions}.
\end{remark}
\begin{figure}
\centering
\includegraphics[width=4.75cm]{hyperbolictranslatingcatenoid0.pdf}
\hspace{1cm}
\includegraphics[width=4.5cm]{hyperbolictranslatingcatenoid.pdf}
\caption{\small Hyperbolic $r$-translating catenoids with horizontal symmetry.}
\label{fig-hyperbolictranslatingcatenoid}
\end{figure}
\section{Uniqueness Results} \label{sec-uniqueness}
In hyperbolic space $\h^n,$ there are three special
types of one-parameter families of
isometries. They are the rotations around a fixed point (\emph{elliptic isometries}),
the translations along horocycles sharing the same point at infinity (\emph{parabolic isometries}),
and the translations along a fixed geodesic (\emph{hyperbolic isometries}).
In Euclidean space, the one-parameter
groups of rotations and translations in a fixed direction are well known.
Notice that each one of the aforementioned groups of isometries of $\q_\epsilon^n$
fixes a family of parallel totally umbilical hypersurfaces. Indeed, an elliptic isometry fixes
a family of parallel spheres, whereas a parabolic (resp. hyperbolic) isometry
fixes a family of parallel horospheres (resp. equidistant hypersurfaces). A translation
in a fixed direction in $\R^n$ fixes a family of parallel hyperplanes.
These isometries of $\q_\epsilon^n$
extend naturally to isometries of $\q_\epsilon^n\times\R$ which fix the factor $\R$ pointwise.
Therefore, a hypersurface $\Sigma\subset\qr$ which is invariant by such an isometry is
foliated by vertical translations of its corresponding parallel totally umbilical hypersurfaces.
We shall call such a $\Sigma$ a \emph{symmetric} hypersurface of $\qr.$
\begin{definition}
A totally geodesic vertical hyperplane of $\qr,$ the grim reaper in $\R^n\times\R,$
or any of the properly embedded translators to
$r$-MCF of Theorems \ref{th-existence}, \ref{th-existence-parabolic},
and \ref{th-existence-hyperbolic} will be called a \emph{fundamental translator to $r$-{\rm MCF}}
in $\qr.$ Any vertical translation of such a translator will also be called a fundamental translator to $r$-{\rm MCF}.
\end{definition}
Notice that all fundamental translators to $r$-MCF are symmetric. In what follows, we shall prove
that they are unique with respect to this property.
\begin{lemma} \label{lem-uniquenessgraph}
Let $\Sigma'$ be a connected symmetric translator to $r$-{\rm MCF}
which is a vertical graph in $\qr$ over an open set of \,$\q_\epsilon^n.$
Then, $\Sigma'$ is an open set of one of the fundamental translators to $r$-{\rm MCF}.
\end{lemma}
\begin{proof}
Being symmetric, $\Sigma'$ is an $(M_s,\phi)$-graph such that the parallel family
$\{M_s\}$ is totally umbilical in $\q_\epsilon^n.$ Hence, by Lemma \ref{lem-main},
the $\rho$-function of $\Sigma'$ satisfies \eqref{eq-rMCFodegeneral}.
If $M_s$ is a family of parallel hyperplanes in $\R^n,$ as we have seen
in Example \ref{exam-grimreaper}, we must have $r=1,$ in which case
$\Sigma'$ is an open set of the grim reaper. Otherwise,
we have that $\tau=\rho^r$ is a solution to one of the Cauchy problems \eqref{eq-IVP},
\eqref{eq-IVP-parabolic} or \eqref{eq-IVP-hyperbolic}.
However, for each of these problems, the trajectories of the $\tau$-functions of the fundamental
translators foliate the corresponding region $\Omega$ on which $F$ is defined.
Thus, whenever $\tau$ is defined, it must coincide with the
$\tau$-function associated to one of the fundamental translators $\Sigma,$ which implies
that $\Sigma'\subset\Sigma.$
\end{proof}
\begin{theorem} \label{th-uniquenesssymetric}
Let $\Sigma$ be a connected symmetric translator to $r$-{\rm MCF}
in \,$\qr.$ Then, $\Sigma$ is an open set of a fundamental translator to $r$-{\rm MCF}.
\end{theorem}
\begin{proof}
Let us suppose first that $\theta$ never vanishes on $\Sigma.$ In this case,
this hypersurface is given by a union of symmetric vertical graphs.
By Lemma \ref{lem-uniquenessgraph}, each such graph is contained in one and only one
of the fundamental translators to $r$-MCF. Then, since $\Sigma$ is connected,
the same is true for $\Sigma.$
Suppose now that $\theta$ vanishes on an open set of $\Sigma.$ Then, since $\Sigma$ is
symmetric and connected, it must be contained in a vertical
hyperplane of $\qr.$
Finally, assume that the set $\mathcal O\subset\Sigma$ on which $\theta$ never vanishes
is open and dense in $\Sigma.$ Then, from the first part of the proof,
any connected component of $\mathcal O$ is contained
in one and only one fundamental translator to $r$-{\rm MCF}.
The result, then, follows from the connectedness of $\Sigma.$
\end{proof}
\begin{remark} \label{rem-liramartin}
In \cite{lira-martin}, Lira and Martín considered translators to MCF in products
$M\times\R,$ where $M$ is a Hadamard manifold endowed
with a rotationally invariant metric. In their Theorem 13, they aim to list
all possible symmetric translators to MCF in such products. However,
the hyperbolic translating catenoids seem to be missing in the statement.
\end{remark}
Given a hypersurface $\Sigma\subset\qr,$ we will call a transversal intersection
$$\Sigma^t:=\Sigma\transv(\q_\epsilon^n\times\{t\})$$
a \emph{horizontal section} of $\Sigma.$
An evident property of any fundamental translator to $r$-MCF is that its angle function
is constant along its horizontal sections. In the next two theorems, we shall see that this property
essentially characterizes these translators.
\begin{theorem} \label{th-characterization}
Let $\Sigma\subset\qr$ be a connected non totally geodesic translator to
$r$-{\rm MCF} with the following properties:
\begin{itemize}[parsep=.3ex]
\item[\rm i)] The angle function $\theta$ of $\Sigma$ is constant on each horizontal section $\Sigma^t\subset\Sigma.$
\item[\rm ii)] There is a totally umbilical (as a hypersurface of \,$\q_\epsilon^n$) horizontal section of $\Sigma$
on which the angle function $\theta$ is nonzero.
\end{itemize}
Then, $\Sigma$ is an open set of one of the fundamental translators to $r$-{\rm MCF}.
\end{theorem}
\begin{proof}
Reasoning as in the proof of Theorem \ref{th-uniquenesssymetric},
we can assume without loss of generality that $\theta T$ does not vanish on $\Sigma.$
(Notice that $T$ cannot vanish on an open set of a translator in $\qr.$)
Since $\theta$ is constant on each $\Sigma^t,$ we have that $\nabla\theta$ is parallel to $T$
on $\Sigma.$ This, together with the identity $AT=-\nabla\theta,$ gives that $T$
is a principal direction of $\Sigma.$
Hence, by \cite[Theorem 6]{delima-roitman}, $\Sigma$ is given locally by
a graph over an open set of $\q_\epsilon^n$ whose level hypersurfaces are parallel.
Clearly, the level hypersurfaces
of two distinct local graphs in $\Sigma$ which intersect must belong to the same family of parallels. Since
$\Sigma$ is connected, we conclude that there exists one and
only one family $\mathscr F$ of parallel hypersurfaces of $\q_\epsilon^n$ such
that the level hypersurfaces of any local graph in $\Sigma$ are elements of $\mathscr F.$ But, from (ii), $\mathscr F$ contains
a totally umbilical hypersurface, which implies that all elements of $\mathscr F$ are totally umbilical. Thus,
$\Sigma$ is symmetric. The result, then, follows from Theorem \ref{th-uniquenesssymetric}.
\end{proof}
Recall that a family of parallel hypersurfaces $\{M_s\,;\, s\in I\}$ of
\,$\q_\epsilon^n$ is called \emph{isoparametric}
if each hypersurface $M_s$ has constant mean curvature (possibly depending on $s$).
If so, each hypersurface $M_s$ is also called
\emph{isoparametric.} It is well known that the totally umbilical hypersurfaces
of $\q_\epsilon^n$ are all isoparametric. Besides them, the only isoparametric
hypersurfaces of $\q_\epsilon^n$ are the \emph{generalized cylinders}, which are the hypersurfaces of the form
$\q_\epsilon^{n-1-k}\times\s^k,$ where $\q_\epsilon^{n-1-k}$ is a totally geodesic hypersurface
of $\q_\epsilon^{n}$ of dimension $n-1-k<n-1,$ and $\s^k$ is the $k$-dimensional geodesic sphere of $\q_\epsilon^{n}.$
\begin{theorem} \label{th-characterizationr=1}
Let $\Sigma\subset\qr$ be a connected non totally geodesic translator to
{\rm MCF} such that:
\begin{itemize}[parsep=.3ex]
\item[\rm i)] The angle function $\theta$ of $\Sigma$ is constant on each horizontal section $\Sigma^t\subset\Sigma.$
\item[\rm ii)] Not all horizontal sections of $\Sigma$ are generalized cylinders of \,$\q_\epsilon^n.$
\end{itemize}
Then, $\Sigma$ is an open set of one of the fundamental translators to {\rm MCF}.
\end{theorem}
\begin{proof}
As in the proof of Theorem \ref{th-characterization}, it follows from (i) that
$\Sigma$ is a union of local graphs whose level hypersurfaces are all in the same family
$\mathscr F$ of parallel hypersurfaces of \,$\q_\epsilon^n.$ So, it remains to prove that
the hypersurfaces of $\mathscr F$ are totally umbilical. Since, by (ii), $\mathscr F$ is not a family
of generalized cylinders, it suffices to prove that $\mathscr F$ is isoparametric.
Given a horizontal section $\Sigma^t\subset\Sigma,$ \cite[Lemma 1]{delima-roitman}
gives that the mean curvature $H_t$ of $\Sigma^t$ (as a hypersurface of $\q_\epsilon^n$) and the mean
curvature $H$ of $\Sigma$ relate as
\begin{equation} \label{eq-Ht}
H_t=-\frac{1}{\sqrt{1-\theta^2}}\left(H-\frac{1}{\|T\|^2}\langle AT,T\rangle\right).
\end{equation}
In addition, we have from \cite[Theorem 6]{delima-roitman} that $\langle AT,T\rangle/\|T\|^2$ is constant on $\Sigma^t.$
Since $H=\theta$ on $\Sigma,$ it follows from \eqref{eq-Ht} that $H_t$ is constant
on $\Sigma^t.$ Therefore, the family $\mathscr F$ is isoparametric, as we wished to prove.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Nonexistence Results} \label{sec-nonexistence}
In this final section, we shall prove that
there exist no complete $r$-mean convex translators to $r$-MCF which are confined
in a cylinder $B\times\R\subset\qr,$ where $B$ is either a geodesic ball of
$\q_\epsilon^n$ or a horoball of $\h^n.$ For that, we shall employ
the standard method of using $r$-translating catenoids as barriers
(see \cite[Theorem 5.9]{bueno}, \cite[Theorem 18]{lira-martin}).
%We conclude by discussing some aspects of translators to the Gauss-Kronecker curvature $H_n.$
\begin{theorem} \label{th-nocilindricallybounded}
There is no complete $r$-mean convex translator to $r$-{\rm MCF} in $\qr$ which is contained in a
cylinder $\mathcal C=B_R(o)\times\R$ over a geodesic open ball $B_R(o).$
\end{theorem}
\begin{proof}
Suppose, by contradiction, that there exists a complete translator
$\Sigma$ to $r$-MCF which is contained in a cylinder $\mathcal C=B_R(o)\times\R.$
Assuming first that $r$ is odd,
consider the family $\mathscr C_r:=\{\Sigma_\lambda\,;\, \lambda\in(0,+\infty)\}$ of
rotational $r$-translating catenoids with axis $\{o\}\times\R$ as in Theorem \ref{th-existence}-(ii).
After a vertical translation, we can assume that $\Sigma\cap(\q_\epsilon^n\times\{0\})\ne\emptyset.$
In this setting, we have that $\Sigma\cap\Sigma_\lambda=\emptyset$ for all $\lambda>R,$ and that
$\Sigma\cap\Sigma_\lambda\ne\emptyset$ for all sufficiently small $\lambda>0.$ Therefore, there exists
$\lambda_*>0$ such that $\Sigma_{\lambda_*}$ has a first contact with $\Sigma.$ By the asymptotic behavior of
$\Sigma_{\lambda_*},$ this contact cannot be ``at infinity'', so that
$\Sigma_{\lambda_*}$ and $\Sigma$ are tangent at a point $p_*\in\Sigma_{\lambda_*}\cap\Sigma.$
Suppose that $p_*$ is on the horizontal $(n-1)$-sphere $S_*$ of $\Sigma_{\lambda_*}$ of minimal height. In this case,
near $p_*,$ $\Sigma$ is a graph $\mathcal G$ over an open ball $B_*,$ centered at $p_*,$
of the horizontal hyperplane of $\qr$ through $p_*.$
Choosing $\lambda<\lambda_*$ sufficiently close to $\lambda_*,$ we have that the
horizontal $(n-1)$-sphere of $\Sigma_{\lambda}$ of minimal height
is sufficiently close to $S_*,$ so that $\Sigma_\lambda$ is transversal to $B_*.$
This, however, implies that
$\Sigma_\lambda$ intersects $\mathcal G,$ which
contradicts the fact that, among the $r$-translating
catenoids of $\mathscr C_r,$ $\Sigma_{\lambda_*}$ establishes the first contact with $\Sigma.$
It follows from the above that $p_*$ is an interior point of the connected component
$\Sigma_{\lambda_*}'$ of $\Sigma_{\lambda_*}-S_*$ which contains $p_*.$
Since $\Sigma$ and $\Sigma_{\lambda_*}'$ are both $r$-mean convex, their unit normals
at $p_*$ coincide. In addition, $\Sigma_{\lambda_*}'$ is $C^2$ and is
strictly convex on an open set. Therefore, we can apply the tangency principle to conclude that
an open subset $\Sigma'$ of $\Sigma$ containing $p_*$ coincides with $\Sigma_{\lambda_*}'$
(see Section \ref{subsec-tangencyprinciple}). This is a contradiction, since $\Sigma'$ is
contained in $\mathcal C$ and $\Sigma_{\lambda_*}'$ is not.
Assume now that $r$ is even and consider the family $\mathscr C_r^2$ of complete properly embedded
$C^2$-smooth rotational $r$-translating catenoids as in Theorem \ref{th-existence}-(iii). Arguing as above,
we get that $\Sigma$ has a first contact with an $r$-translating catenoid $\Sigma_{\lambda_*}^2\in\mathscr C_r^2$
at a point $p_*\in\Sigma\cap\Sigma_{\lambda_*}^2.$
Since $\Sigma$ is $r$-mean convex, $p_*$ cannot be on the $(n-1)$-sphere $S_*$ of $\Sigma_{\lambda_*}^2$ in
$\q_\epsilon^n\times\{0\}.$ Hence, $p_*$ lies on either the upper graph or the lower graph of $\Sigma_{\lambda_*}^2-S_*.$
(In the latter case, we take on the lower graph the orientation that makes its angle function positive.)
From the tangency principle, an open set of $\Sigma$ should coincide with either
of these graphs, which is a contradiction.
\end{proof}
Considering Theorem \ref{th-existence-parabolic}, we can easily adapt the reasoning in the above proof
to obtain the following result, which is in fact stronger than
Theorem \ref{th-nocilindricallybounded} for $\epsilon=-1.$
\begin{theorem} \label{th-parabolicnocilindricallybounded}
There is no complete
translator to $r$-{\rm MCF} in $\h^n\times\R$ which is contained in a
cylinder $\mathcal C=B\times\R$ over a horoball $B\subset\h^n.$
\end{theorem}
\begin{thebibliography}{99}
\bibitem{andrewsetal} B. Andrews, B. Chow, C. Guenther, M. Langford: Extrinsic Geometric Flows.
Graduate studies in mathematics {\bf 206}, American Mathematical Society (2020).
\vt
\bibitem{altschuler-wu} S. J. Altschuler, L. F. Wu: Translating surfaces of the non-parametric
mean curvature flow with prescribed contact angle. Calc. Var. Partial Differ. Equ. 2(1), 101--111 (1994).
\vt
\bibitem{bueno} A. Bueno: Translating solitons of the mean curvature flow in the space $\h^2\times\R.$ J. Geom. 109, 42 (2018).
\vt
\bibitem{schulzeetal} J. Clutterbuck, O. C. Schnürer, F. Schulze: Stability of translating solutions to mean curvature flow.
Calc. Var. 29, 281--293 (2007).
\vt
\bibitem{delima-manfio-santos} R. F. de Lima, F. Manfio, J. P. dos Santos: Hypersurfaces of constant
higher order mean curvature in $M\times\R.$ Ann. Mat. Pura Appl. {\bf 201}, 2979--3028 (2022).
\vt
\bibitem{delima-roitman} R. F. de Lima, P. Roitman : Helicoids and catenoids in $M\times\R.$
Ann. Mat. Pura Appl. {\bf 200}, 2385--2421 (2021).
\vt
\bibitem{elbert-saearp} M. F. Elbert, R. Sa Earp: Constructions of $H_r$-hypersurfaces, barriers and Alexandrov theorem in
$\h^n\times\R$. Ann. Mat. Pura Appl. {\bf 194}, 1809--1834 (2015).
\vt
\bibitem{fontenele-silva} F. Fontenele, S. Silva: A tangency principle and applications. Illinois J. of Math.
{\bf 54}, 213--228 (2001).
\vt
%\bibitem{lawn-ortega} M.A. Lawn, M. Ortega, Translating solitons from semi-Riemannian submersions,
%Preprint arXiv :1607.04571.
\bibitem{huisken-sinestrari} G. Huisken, C. Sinestrari: Convexity estimates for mean curvature flow and singularities of mean
convex surfaces. Acta Math. {\bf 183} (1), 45--70 (1999).
\bibitem{lira-martin} J. H. S. Lira, F. Martín: Translating solitons in Riemannian products.
J. Differential Equations 266, 7780--7812 J. Differential Equations (2019).
\vt
%\bibitem{martinetal} F. Martín, A. Savas-Halilaj, K. Smoczyk: On the topology of translating
%solitons of the mean curvature flow. Calculus of Variations and Partial Differential
%Equations 54 (2015), no. 3, 2853--2882.
\bibitem{protter} M. Protter: Unique continuation for elliptic equations, Trans. Amer. Math. Soc.,
{\bf 95}, 81--91 (1960).
\end{thebibliography}
\end{document}
%https://homepages.bluffton.edu/~nesterd/apps/slopefields.html
%(2/3)(1-y)^(1/2)tanh(x)-2coth(x)y