Hypocoercivity in algebraically constrained partial differential equations with application to Oseen equations
Extracted source text · arXiv 2212.06631
%Version 2.1 April 2023
% See section 11 of the User Manual for version history
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% %%
%% Please do not use \input{...} to include other tex files. %%
%% Submit your LaTeX manuscript as one .tex document. %%
%% %%
%% All additional figures and files should be attached %%
%% separately and not embedded in the \TeX\ document itself. %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%\documentclass[referee,sn-basic]{sn-jnl}% referee option is meant for double line spacing
%%=======================================================%%
%% to print line numbers in the margin use lineno option %%
%%=======================================================%%
%%\documentclass[lineno,sn-basic]{sn-jnl}% Basic Springer Nature Reference Style/Chemistry Reference Style
%%======================================================%%
%% to compile with pdflatex/xelatex use pdflatex option %%
%%======================================================%%
%%\documentclass[pdflatex,sn-basic]{sn-jnl}% Basic Springer Nature Reference Style/Chemistry Reference Style
%%Note: the following reference styles support Namedate and Numbered referencing. By default the style follows the most common style. To switch between the options you can add or remove Numbered in the optional parenthesis.
%%The option is available for: sn-basic.bst, sn-vancouver.bst, sn-chicago.bst, sn-mathphys.bst. %
\RequirePackage{amsthm}
%%\documentclass[sn-nature]{sn-jnl}% Style for submissions to Nature Portfolio journals
\documentclass[sn-basic,Numbered]{sn-jnl}% Basic Springer Nature Reference Style/Chemistry Reference Style
%%\documentclass[sn-mathphys,Numbered]{sn-jnl}% Math and Physical Sciences Reference Style
%%\documentclass[sn-aps]{sn-jnl}% American Physical Society (APS) Reference Style
%%\documentclass[sn-vancouver,Numbered]{sn-jnl}% Vancouver Reference Style
%%\documentclass[sn-apa]{sn-jnl}% APA Reference Style
%%\documentclass[sn-chicago]{sn-jnl}% Chicago-based Humanities Reference Style
%%\documentclass[default]{sn-jnl}% Default
%%\documentclass[default,iicol]{sn-jnl}% Default with double column layout
%%%% Standard Packages
%%<additional latex packages if required can be included here>
\usepackage{graphicx}%
\usepackage{multirow}%
\usepackage{amsmath,amssymb,amsfonts}%
\usepackage{amsthm}%
\usepackage{mathrsfs}%
\usepackage[title]{appendix}%
\usepackage{xcolor}%
\usepackage{textcomp}%
\usepackage{manyfoot}%
\usepackage{booktabs}%
\usepackage{algorithm}%
\usepackage{algorithmicx}%
\usepackage{algpseudocode}%
\usepackage{listings}%
%%%%
% additional packages
%\usepackage{tikz}
\usepackage[shortlabels]{enumitem}
\usepackage{mathtools} % for using \underbracket
\usepackage{verbatim}% to use \begin{comment} ... \end{comment}
%
%
%-------------------arXiv header-----------------------------------------
\begin{comment}
\documentclass[11pt,a4paper]{article}
\usepackage{stix}
%\usepackage{verbatim}
\usepackage{refcheck} % remove in final version
\usepackage[T1]{fontenc}
\usepackage[a4paper,includeheadfoot,margin=2.54cm]{geometry}
\usepackage{xfrac}
\usepackage{amsthm, amsmath}
\theoremstyle{plain} %default
\newtheorem{theorem}{Theorem}
\newtheorem{corollary}{Corollary}
\newtheorem{lemma}{Lemma}
\newtheorem{proposition}{Proposition}
\newtheorem{claim}{Claim}
\theoremstyle{definition}
\newtheorem{definition}{Definition}
\newtheorem{remark}{Remark}
\newtheorem{notation}{Notation}
\newtheorem{eexample}{Example}
\newtheorem*{assumptions}{Assumptions}
\end{comment}
%-------------------arXiv header-----------------------------------------
%%%%%=============================================================================%%%%
%%%% Remarks: This template is provided to aid authors with the preparation
%%%% of original research articles intended for submission to journals published
%%%% by Springer Nature. The guidance has been prepared in partnership with
%%%% production teams to conform to Springer Nature technical requirements.
%%%% Editorial and presentation requirements differ among journal portfolios and
%%%% research disciplines. You may find sections in this template are irrelevant
%%%% to your work and are empowered to omit any such section if allowed by the
%%%% journal you intend to submit to. The submission guidelines and policies
%%%% of the journal take precedence. A detailed User Manual is available in the
%%%% template package for technical guidance.
%%%%%=============================================================================%%%%
%\jyear{2021}%
%% as per the requirement new theorem styles can be included as shown below
\theoremstyle{thmstyleone}%
\newtheorem{theorem}{Theorem}% meant for continuous numbers
%%\newtheorem{theorem}{Theorem}[section]% meant for sectionwise numbers
%% optional argument [theorem] produces theorem numbering sequence instead of independent numbers for Proposition
\newtheorem{proposition}[theorem]{Proposition}%
%%\newtheorem{proposition}{Proposition}% to get separate numbers for theorem and proposition etc.
\newtheorem{corollary}{Corollary}
\newtheorem{lemma}{Lemma}
\newtheorem{claim}{Claim}
\theoremstyle{thmstyletwo}%
\newtheorem{example}{Example}%
%\newtheorem{remark}{Remark}%
\theoremstyle{thmstylethree}%
\newtheorem{definition}{Definition}%
\newtheorem{remark}{Remark}
\newtheorem{notation}{Notation}
\newtheorem{eexample}{Example}
\newtheorem*{assumptions}{Assumptions}
\raggedbottom
%%\unnumbered% uncomment this for unnumbered level heads
%\newtheorem{eexample}{Example}{\bf}{\rm}
%\newtheorem*{notation}{Notation}{\bf}{\rm}
%\newtheorem*{assumptions}{Assumptions}{\bf}{\rm}
% Franz
%\usepackage{algorithm}
%\usepackage{algorithmic}
%%\floatname{algorithm}{Procedure}
%\renewcommand{\algorithmicrequire}{\textbf{Input:}}
\algrenewcommand\algorithmicrequire{\textbf{Input:}}
\algrenewcommand\algorithmicensure{\textbf{Output:}}
% https://tex.stackexchange.com/questions/33866/algorithm-tag-and-page-break/33869#33869
\begin{comment}
\makeatletter
\newenvironment{breakablealgorithm}
{% \begin{breakablealgorithm}
\begin{center}
\refstepcounter{algorithm}% New algorithm
\hrule height.8pt depth0pt \kern2pt% \@fs@pre for \@fs@ruled
\renewcommand{\caption}[2][\relax]{% Make a new \caption
{\raggedright\textbf{\fname@algorithm~\thealgorithm} ##2\par}%
\ifx\relax##1\relax % #1 is \relax
\addcontentsline{loa}{algorithm}{\protect\numberline{\thealgorithm}##2}%
\else % #1 is not \relax
\addcontentsline{loa}{algorithm}{\protect\numberline{\thealgorithm}##1}%
\fi
\kern2pt\hrule\kern2pt
}
}{% \end{breakablealgorithm}
\kern2pt\hrule\relax% \@fs@post for \@fs@ruled
\end{center}
}
\makeatother
\end{comment}
\makeatletter
\newenvironment{breakablealgorithm}
{
\begin{center}
\refstepcounter{algorithm}
\renewcommand{\caption}[1]
{
\addcontentsline{loa}{algorithm}{\protect\numberline{\thealgorithm}##1}
\parbox{\textwidth}
% Makes a unbreakable block and can also be done by `minipage`.
{
\hrule height.8pt depth0pt \kern2pt
{\raggedright\textbf{\fname@algorithm~\thealgorithm} ##1\par}
\kern2pt\hrule\kern2pt
}
}
}
{
\kern2pt\hrule\relax
\end{center}
}
\makeatother
\newcommand{\makegray}[1]{{\color{gray}#1}}
\newcommand{\makegreen}[1]{{\color{green}#1}}
\newcommand{\makered}[1]{{\color{red}#1}}
\newcommand{\makeblue}[1]{{\color{blue}#1}}
\newcommand{\makecyan}[1]{{\color{cyan}#1}}
\newcommand{\makeorange}[1]{{\color{black}#1}}
\newcommand{\new}[1]{{\color{blue}#1}}
\newcommand{\ii}{\mathrm{i}}
\newcommand{\C}{\mathbb{C}}
\newcommand{\N}{\mathbb{N}}
\newcommand{\R}{\mathbb{R}}
\newcommand{\T}{\mathbb{T}}
\newcommand{\X}{\mathbb{X}}
\newcommand{\Z}{\mathbb{Z}}
\newcommand{\Zo}{\mathbb{Z}\setminus\{0\}}
\newcommand{\Rnn}{\R^{n\times n}}
\newcommand{\Cnn}{\C^{n\times n}}
\newcommand{\Cn}{\C^n}
\newcommand{\bigO}{{\mathcal{O}}}
\newcommand{\dd}[1][x]{\,\operatorname{d}\!#1}
\newcommand{\dt}{\,\dd[t]}
\newcommand{\ddt}{\frac{\dd[]}{\dd[t]}}
\newcommand{\ddtddt}{\frac{\dd[]^2}{\dd[t^2]}}
\newcommand{\ddtddtddt}{\frac{\dd[]^3}{\dd[t^3]}}
\newcommand{\intRd}{\int_{\R^d}}
\newcommand{\Rd}{\R^d}
\newcommand{\eps}{\varepsilon}
\newcommand{\spn}{\operatorname{span}}
\newcommand{\ran}{\operatorname{ran}}
\newcommand{\corank}{\operatorname{corank}}
\newcommand{\He}{\operatorname{He}}
\newcommand{\tr}{\operatorname{Tr}}
% all matrices bold face (but not operators)
\newcommand{\mA}{\mathbf{A}}
\newcommand{\mB}{\mathbf{B}}
\newcommand{\mC}{\mathbf{C}}
\newcommand{\mD}{\mathbf{D}}
\newcommand{\mE}{\mathbf{E}}
\newcommand{\mF}{\mathbf{F}}
\newcommand{\mG}{\mathbf{G}}
\newcommand{\mH}{\mathbf{H}}
\newcommand{\mI}{\mathbf{I}}
\newcommand{\mJ}{\mathbf{J}}
\newcommand{\mK}{\mathbf{K}}
\newcommand{\mL}{\mathbf{L}}
\newcommand{\mM}{\mathbf{M}}
\newcommand{\mN}{\mathbf{N}}
\newcommand{\mO}{\mathbf{O}}
\newcommand{\Null}{\mathbf{0}}
\newcommand{\mP}{\mathbf{P}}
\newcommand{\mQ}{\mathbf{Q}}
\newcommand{\mQlin}{\mathbf{Q_{lin}}}
\newcommand{\mR}{\mathbf{R}}
\newcommand{\mS}{\mathbf{S}}
\newcommand{\mT}{\mathbf{T}}
\newcommand{\mU}{\mathbf{U}}
\newcommand{\mV}{\mathbf{V}}
\newcommand{\mW}{\mathbf{W}}
\newcommand{\mX}{\mathbf{X}}
\newcommand{\mY}{\mathbf{Y}}
\newcommand{\mZ}{\mathbf{Z}}
\newcommand{\msqrtE}{\mE^{1/2}}
\newcommand{\msqrtEinv}{(\msqrtE)^{-1}}
\newcommand{\tA}{\widetilde \mA}
\newcommand{\tB}{\widetilde \mB}
\newcommand{\tD}{\widetilde \mD}
\newcommand{\tE}{\widetilde \mE}
\newcommand{\tG}{\widetilde \mG}
\newcommand{\tJ}{\widetilde \mJ}
\newcommand{\tN}{\widetilde \mN}
\newcommand{\tPi}{\widetilde\Pi}
\newcommand{\tR}{\widetilde \mR}
\newcommand{\tS}{\widetilde \mS}
\newcommand{\tSigma}{\widetilde \Sigma}
\newcommand{\tX}{\widetilde \mX}
\newcommand{\tY}{\widetilde \mY}
%% code from mathabx.sty and mathabx.dcl
\DeclareFontFamily{U}{mathx}{\hyphenchar\font45}
\DeclareFontShape{U}{mathx}{m}{n}{
<5> <6> <7> <8> <9> <10>
<10.95> <12> <14.4> <17.28> <20.74> <24.88>
mathx10
}{}
\DeclareSymbolFont{mathx}{U}{mathx}{m}{n}
\DeclareFontSubstitution{U}{mathx}{m}{n}
\DeclareMathAccent{\widecheck}{0}{mathx}{"71}
\DeclareMathAccent{\wideparen}{0}{mathx}{"75}
\newcommand{\chA}{\widecheck \mA}
\newcommand{\chJ}{\widecheck \mJ}
\newcommand{\chR}{\widecheck \mR}
\newcommand{\hA}{\widehat \mA}
\newcommand{\hJ}{\widehat \mJ}
\newcommand{\hR}{\widehat \mR}
\newcommand{\hS}{\widehat \mS}
\newcommand{\tn}{\tilde n}
\newcommand{\mAH}{\mA_H}
\newcommand{\mAS}{\mA_S}
\newcommand{\mAI}{\mA^{-1}}
\newcommand{\mAIH}{(\mA^{-1})_H}
\newcommand{\mAIS}{(\mA^{-1})_S}
\newcommand{\mBH}{\mB_H}
\newcommand{\mBS}{\mB_S}
\newcommand{\mCH}{\mC_H}
\newcommand{\mCS}{\mC_S}
\newcommand{\mDH}{\mD_H}
\newcommand{\mDS}{\mD_S}
\newcommand{\mHC}{m_{HC}}
\newcommand{\mHCmAI}{\widetilde{m}_{HC}}
\newcommand{\bu}{{\bf u}}
\newcommand{\bw}{{\bf w}}
\newcommand{\xk}{{w_k}}
\newcommand{\x}{w}
% Matrices in the associated discrete-time system
\newcommand{\mAd}{\mathsf{A}} %{\widetilde{\mA}}
\newcommand{\mBd}{\mathsf{B}} %{\widetilde{\mB}}
\newcommand{\mCd}{\mathsf{C}}
\newcommand{\mPd}{\mathsf{P}}
\newcommand{\mQd}{\mathsf{Q}}
\newcommand{\bbH}{\mathbb{H}}
\newcommand{\Hn}{\mathbb{H}_n} % Hermitian Matrices
\newcommand{\PDHn}{\mathbb{H}_n^{>}} % Positive Definite Hermitian Matrices
\newcommand{\PSDHn}{\mathbb{H}_n^{\geq}} % Positive SemiDefinite Hermitian Matrices
% Franz
\newcommand{\ee}{\mathrm e}
\newcommand{\topH}{\mathsf{H}} %
\newcommand{\ip}[2]{\langle {#1}, {#2} \rangle}
\newcommand{\ipBig}[2]{\Big\langle {#1}, {#2} \Big\rangle}
\newcommand{\norm}[1]{\| {#1} \|}
\newcommand{\normBig}[1]{\Big\| {#1} \Big\|}
\newcommand{\normP}[1]{\| {#1} \|_{\mP}}
\DeclareMathOperator*{\argmax}{arg\,max}
\DeclareMathOperator{\const}{const}
\DeclareMathOperator{\diver}{div}
\DeclareMathOperator{\diag}{diag}
\DeclareMathOperator{\kernel}{ker}
\DeclareMathOperator{\range}{range}
\DeclareMathOperator{\rank}{rank}
\DeclareMathOperator{\rot}{rot}
\DeclareMathOperator{\sgn}{sgn}
\newcommand{\Rp}{\R^+}
\newcommand{\Rpo}{\R^+_0}
\newcommand{\hRzz}{\hR_{2,2}}
\newcommand{\VhRzz}{\big(\hR_{2,2}\big)^{1/2}}
\newcommand{\hJzz}{\hJ_{2,2}}
\newcommand{\hJkzz}{(\hJ_{k_1})_{2,2}}
% calligraphic
\newcommand{\cB}{\mathcal{B}}
\newcommand{\cD}{\mathcal{D}}
\newcommand{\cE}{\mathcal{E}}
\newcommand{\cG}{\mathcal{G}}
\newcommand{\cH}{\mathcal{H}}
\newcommand{\cK}{\mathcal{K}}
\newcommand{\cL}{\mathcal{L}}
\newcommand{\cQ}{\mathcal{Q}}
\newcommand{\cV}{\mathcal{V}}
\newcommand{\cW}{\mathcal{W}}
\newcommand{\wcH}{\widetilde{\cH}}
\newcommand{\fS}{\mathcal{S}}
\newcommand{\fT}{\mathcal{T}}
\newcommand{\fU}{\mathcal{U}}
\newcommand{\fV}{\mathcal{V}}
\newcommand{\fW}{\mathcal{W}}
% Comparing our construction in [AAC20] with Van Dooren's Algorithm 1 we change the notation as follows:
\newcommand{\fX}{\mathsf{A}} %{\mathcal{X}}
\newcommand{\fY}{\mathsf{Z}} %{\mathcal{Y}}
\newcommand{\fZ}{\mathsf{B}} %{\mathcal{Z}}
\newcommand{\bx}{x}
% Franz
\newcommand{\SG}{\mT} % semigroup
\newcommand{\uI}{u_1}
\newcommand{\ukI}{\phi_{k,1}}
\newcommand{\uII}{u_2}
\newcommand{\ukII}{\phi_{k,2}}
%--------------------------------------------------------
\begin{document}
%--------------------Springer version-----------------------------------
\title[Hypocoercivity in algebraically constrained partial differential equations]{Hypocoercivity in algebraically constrained partial differential equations with application to Oseen equations}
%%=============================================================%%
%% Prefix -> \pfx{Dr}
%% GivenName -> \fnm{Joergen W.}
%% Particle -> \spfx{van der} -> surname prefix
%% FamilyName -> \sur{Ploeg}
%% Suffix -> \sfx{IV}
%% NatureName -> \tanm{Poet Laureate} -> Title after name
%% Degrees -> \dgr{MSc, PhD}
%% \author*[1,2]{\pfx{Dr} \fnm{Joergen W.} \spfx{van der} \sur{Ploeg} \sfx{IV} \tanm{Poet Laureate}
%% \dgr{MSc, PhD}}\email{iauthor@gmail.com}
%%=============================================================%%
\author[1]{\fnm{Franz} \sur{Achleitner}}\email{franz.achleitner@tuwien.ac.at}
\equalcont{All authors contributed equally to this work.}
\author*[1]{\fnm{Anton} \sur{Arnold}}\email{anton.arnold@tuwien.ac.at}
\equalcont{All authors contributed equally to this work.}
\author[2]{\fnm{Volker} \sur{Mehrmann}}
\email{mehrmann@math.tu-berlin.de}
\equalcont{All authors contributed equally to this work.}
\affil*[1]{\orgdiv{Technische Universit\"at Wien (TU Wien)}, \orgname{Institute of Analysis and Scientific Computing}, \orgaddress{\street{Wiedner Hauptstr. 8-10}, \city{Wien}, \postcode{A-1040}, %\state{State},
\country{Austria}}}
\affil[2]{\orgdiv{Technische Universit\"at Berlin (TU Berlin)}, \orgname{Institut f.~Mathematik MA 4-5}, \orgaddress{\street{Stra\ss{}e des 17.~Juni 136}, \city{Berlin}, \postcode{D-10623}, %\state{State},
\country{Germany}}}
%%==================================%%
%% sample for unstructured abstract %%
%%==================================%%
%\abstract{The abstract serves both as a general introduction to the topic and as a brief, non-technical summary of the main results and their implications. Authors are advised to check the author instructions for the journal they are submitting to for word limits and if structural elements like subheadings, citations, or equations are permitted.}
\abstract{
The long-time behavior of solutions to different versions of Oseen equations of fluid flow on the 2D torus is analyzed using the concept of hypocoercivity.
The considered models are isotropic Oseen equations where the viscosity acts uniformly in all directions and anisotropic Oseen-type equations with different viscosity directions.
The hypocoercivity index is determined (if it exists) and it is shown that similar to the finite dimensional case of ordinary differential equations and differential-algebraic equations it characterizes its decay behavior.
}
\keywords{hypocoercivity (index), dissipative systems, constrained PDEs, Oseen equation}
\pacs[MSC Classification]{Primary 34A30, Secondary 34C11, 47F06, 35E05}
%\thanks{The first author (FA) was supported by the Austrian Science Fund (FWF) via the FWF-funded SFB \# F65. The second author (AA) was supported by the Austrian Science Fund (FWF), partially via the FWF-doctoral school "Dissipation and dispersion in non-linear partial differential equations'' (\# W1245) and the FWF-funded SFB \# F65. The third author (VM) was supported by Deutsche Forschungsgemeinschaft (DFG) via the DFG-funded SFB \# 910.}
%\thanks{Grants or other notes about the article that should go on the front page should be placed here. General acknowledgments should be placed at the end of the article.}
\maketitle
%--------------------Springer version-----------------------------------
%-----------------arXiv version--------------------------------------
\begin{comment}
\title{Hypocoercivity in algebraically constrained partial differential equations with application to Oseen equations}
\author{Franz Achleitner\thanks{Vienna University of Technology, Institute of Analysis and Scientific Computing, Wiedner Hauptstr. 8-10, A-1040 Wien, Austria, franz.achleitner@tuwien.ac.at},
Anton Arnold\thanks{Vienna University of Technology, Institute of Analysis and Scientific Computing, Wiedner Hauptstr. 8-10, A-1040 Wien, Austria, anton.arnold@tuwien.ac.at},
and Volker Mehrmann\thanks{Technische Universit\"at Berlin, Institut f.~Mathematik, MA 4-5, Stra\ss{}e des 17.~Juni 136, D-10623 Berlin, mehrmann@math.tu-berlin.de}
}
\begin{abstract}
The long-time behavior of solutions to different versions of Oseen equations of fluid flow on the 2D torus is analyzed using the concept of hypocoercivity. The considered models are isotropic Oseen equations where the viscosity acts uniformly in all directions and anisotropic Oseen-type equations with different viscosity directions. The hypocoercivity index is determined (if it exists) and it is shown that similar to the finite dimensional case of ordinary differential equations and differential-algebraic equations it characterizes its decay behavior.
\end{abstract}
{\bf Key words.} hypocoercivity (index), dissipative systems, constrained PDEs, Oseen equation
%\subjclass{Primary 34A30; Secondary 34C11, 47F06, 35E05}
{\bf MSC 2010 subject classification.} Primary 34A30; Secondary 34C11, 47F06, 35E05
\end{comment}
%-----------------arXiv version--------------------------------------
\maketitle
\tableofcontents
%--------------------------------------------------------------
\section{Introduction}
\label{sec:intro}
This paper is concerned with the long-time behavior and hypocoercivity structure of (an)isotropic Oseen equations from fluid dynamics. The Oseen equations describe the flow of a viscous and incompressible fluid at low Reynolds numbers and they have the form
%
\begin{equation} \label{eq:Oseen1}
\begin{cases}
u_t &=-(b\cdot\nabla)u- \nabla p +\nu \Delta u\,, \quad t>0\,, \\
0&=-\diver u \,.
\end{cases}
\end{equation}
%
The Oseen equations arise when one linearizes the incompressible or nearly incompressible {\em Navier-Stokes equations} describing the flow of a Newtonian fluid,
%
\begin{align*}
{u}_t &= - (u\cdot\nabla) u - \nabla p+\nu \Delta u\,, \quad t>0\,,\\
0&=-\diver u \,,
\end{align*}
%
around a prescribed vector field $b$, that is independent of space and time, see e.g. \cite{Shi73}.
As it also includes a (linear) convective term, it can be seen as an improvement of the flow description by the %\makered{linearized Navier-Stokes equations and} %\makecyan{(Franz: Ich w\"urde den roten Text streichen, da %Stokes und Oseen beides linearisierte Navier-Stokes %equations sind. Pa\ss{}t das f\"ur euch?)}
Stokes equations, see \cite[\S4.10]{Ba2000}, \cite[Chap.\ 2, \S11]{La1996}.
The Oseen equations are a typical example of an operator differential-algebraic equation (DAEs) of the form
%
\begin{equation}\label{opDAE}
\begin{bmatrix} \mI & 0 \\ 0 & 0 \end{bmatrix}
\frac{d}{dt} \begin{bmatrix} u \\ p \end{bmatrix}
= -\begin{bmatrix} \mC & -\mD^T \\ \mD & 0 \end{bmatrix} \begin{bmatrix} u \\ p \end{bmatrix}
\ ,\quad t>0\,,
\end{equation}
%
with the unbounded operators $\mC=b\cdot\nabla -\nu\Delta$ and $\mD=\diver$, see~\cite[p. 466]{EM13}.
DAEs of the form~\eqref{opDAE} also arise when the Oseen system is semi-discretized in space e.g. via a finite element discretization. This constitutes what is often called a vertical method of lines approach, see e.g., \cite{Jo16,Lay08,Ran00}.
%Operator DAEs of the form \eqref{opDAE} also arise when the Oseen system is semi-discretized in space via a finite element discretization or the method of lines, see e.g., \cite{Jo16,Lay08,Ran00}.
In all described cases the equations have to be supplemented by suitable initial and boundary conditions. While in applications, \makeorange{see e.g. \cite{La1996,BaJi08,Gal11}}, the Oseen equation is typically considered on subsets of $\R^d$ or on unbounded exterior domains, to keep the presentations and the technicalities of our analysis simple, we analyze its long-time behavior here on the torus $\T^2 := (0,2\pi)^2$, \makeorange{similar to~\cite{FMRT01,T95,T97}}.
We perform the analysis using the concept of \emph{hypocoercivity} which was introduced %by Villani
in \cite{Vi09} in the study of unconstrained evolution equations (mostly partial differential equations) of the form $\ddt{x}=-\mC x$ on some Hilbert space~$\cH$, where the (possibly unbounded) operator~$-\mC$ generates a uniformly exponentially stable $C_0$-semigroup $(e^{-\mC t })_{t\geq 0}$.
% See \cite[\S V.1 Eq.~(1.9)]{EnNa00}
More precisely, for hypocoercive operators~$\mC$ there exist constants $\mu>0$ and $c\ge1$, such that
%
\begin{equation}\label{exp-decay}
\|e^{-\mC t} x_0\|_{\widetilde \cH}
\le c\,e^{-\mu t} \|x_0\|_{\widetilde \cH}
\qquad
\mbox{\rm for all}\,x_0\in \widetilde \cH\,,
\qquad
t \geq 0\,,
\end{equation}
%
where $\widetilde \cH$ is another Hilbert space, densely embedded in $(\ker \mC)^\perp\subset \cH$.
Often, the evolution equation~$\ddt{x}=-\mC x$ is also called hypocoercive.
For infinitesimal generators $-\mC$, an estimate~\eqref{exp-decay} with $c=1$ holds if (and only if) $\mC$ is coercive.
The long-time behavior of many systems exhibiting hypocoercivity has been studied frequently in recent years, including Fokker--Planck equations~\cite{AASt15,ArEr14,Vi09}, kinetic equations~\cite{DoMoSc09,DoMoSc15}, and reaction-transport equations of BGK-type~\cite{AAC16,AAC18}. In these works, in particular in \cite{AAC16,AAC18,ArEr14}, the issue was to determine the sharp (i.e.\ maximal) exponential decay rate~$\mu$, while to determine at the same time the smallest multiplicative constant~$c\ge1$ is a rather recent topic, e.g. see~\cite{AAS19}.
Also the short-time behavior of linear evolution equations and its link to the \emph{hypocoercivity index} was recently discussed for systems of ordinary differential equations in \cite{AAC22,AAM22b} and for Fokker--Planck equations \cite[Th.\ 3.6]{ASS20}.
In this paper we consider (unbounded) operators $\mC$ on a %separable
Hilbert space~$\cH$ such that the operator $\mC$ is \emph{accretive}, i.e. $\mC$ has a nonnegative self-adjoint part, and that $-\mC$ generates a \emph{contraction semigroup}, i.e., %~$(e^{-\mC t})_{t\geq 0}$ on~$\cH$.
it satisfies $\|e^{-\mC t}\|_\cH \leq 1$ for all $t\geq0$.
We characterize those accretive operators~$\mC$ which are hypocoercive, i.e., those for which $-\mC$ generates a uniformly exponentially stable $C_0$-semigroup.
Furthermore, based on our characterization and following the Lyapunov theory on Hilbert spaces, see e.g. \cite{Da70,Gr90,HaWe97}, we will construct appropriate strict Lyapunov functionals.
The remainder of this paper is structured as follows:
In \S\ref{sec:hypocoercivity} we review key notions from hypocoercivity for finite dimensional ODEs and DAEs, and then extend it to the Hilbert space case.
In \S\ref{ssec:Stokes} we apply these techniques to analyze Oseen equations on the 2D torus where the viscosity acts uniformly in all directions.
In Section~\S\ref{ssec:Oseen:anisotropic} we then study two anisotropic Oseen-type equations on the 2D torus, where the viscosity is different in the different space directions and the drift is either constant or space dependent.
\subsection*{Notation}
The conjugate transpose (transpose) of a matrix $\mC$ is denoted by $\mC^H$ ($\mC^\top$).
The set of Hermitian matrices in $\Cnn$ is denoted by $\bbH_n$.
Positive definiteness (semi-definiteness) of $\mC\in\bbH_n$ is denoted by $\mC>0$ ($\mC\geq 0$).
%The set of all positive definite (semi-definite) Hermitian matrices in $\Cnn$ is denoted by $\bbH_n^>$ ($\bbH_n^\geq$).
The unique \makeorange{positive semi-definite} square root of a positive semi-definite Hermitian matrix~$\mR$ is denoted by $\mR^{1/2}$ and the real part of a complex number $z$ is denoted by $\Re(z)$.
For linear operators on Hilbert spaces we use the following notation.
%which is unfortunately not used consistently in the literature,
%see e.g.~\cite{Kato}, \makered{as summarized in~\cite[\S A.3.2]{CuZw20}, see also \cite[\S VI.4]{RS-I}.}
%{For completeness} we recall the definition of %\emph{positive} self-adjoint operators since it is not %used consistently in the literature, compare \cite[Definition A.3.80]{CuZw20} and \cite[\S VI.4]{RS-I}:
%
%\begin{definition}\label{def:selfadj}
The set of (possibly unbounded) linear operators from a Hilbert space $\mathcal H$ (with inner product $\ip{\cdot}{\cdot}$) to itself is denoted by ${\mathcal L}({\mathcal H})$, the subset of bounded linear operators by ${\mathcal B}({\mathcal H})$.
For a linear operator $\mC\in{\cL}({\cH})$ with domain $\cD(\mC)$, $\mC^*$ denotes the adjoint operator of $\mC$.
A self-adjoint operator $\mC\in {\mathcal L}({\mathcal H})$ is called \emph{nonnegative} ($\mC\geq 0$) if $\ip{\mC x}{x}=\ip{x}{\mC^*x}\geq 0$ for all $x\in\cD(\mC)$;
$\mC$ is called \emph{positive} ($\mC>0$) if $\ip{\mC x}{x}> 0$ for all $x\in\cD(\mC)\setminus\{0\}$.
We consider spaces of square-summable doubly-infinite sequences taking values in $\C^n$, $n=1,2,3$ and denote them as $\ell^2(\Z;\C^n)$; if $n=1$ then we will use the short-hand notation $\ell^2(\Z)=\ell^2(\Z;\C)$.
$\dot H^1_{per}(\T^2)$ denotes the homogeneous Sobolev space $\{f\in H^1_{per}(\T^2)\,|\, \int_{\T^2} f \dd[x]=0\}$ of periodic functions on the 2D-torus $\T^2$, and $\dot H^{-1}_{per}(\T^2)$ is its dual space.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Hypocoercivity for finite and infinite dimensional evolution equations}
\label{sec:hypocoercivity}
%In this section we recall the concepts of \emph{hypocoercivity and the hypocoercivity index} for finite and infinite dimensional (constrained) evolution equations.
In this section we recall the concepts of \emph{hypocoercivity and the hypocoercivity index} for (constrained) evolution equations. Our discussion will be in three steps, starting with finite dimensional cases, then infinite dimensional operator evolution equations with bounded generators, and finally some cases of unbounded generators.
We begin with the class of ordinary differential equation (ODE) systems
%
\begin{equation}\label{ODE:A}
\dot{x}(t) = \mA x(t)\,, \qquad t\geq 0\,,
\end{equation}
%
with some function $x:[0,\infty)\to\C^n$ and a constant matrix~$\mA\in\Cnn$. The second class are differential-algebraic equation (DAE) systems
%
\begin{equation}\label{DAE:EA}
\mE \dot{x}(t) = \mA x(t)\,, \qquad t\geq 0\,,
\end{equation}
%
for a pair~$(\mE,\mA)$ of constant matrices~$\mE,\mA\in\Cnn$ with $\mE=\mE^H$ positive semi-definite.
%
Note that if~$\mE\in\Hn$ is positive definite, then it has a positive definite matrix square root~$\msqrtE\in\Hn$, and by a change of variables~$y:= \msqrtE x$ and by scaling the equation by $\msqrtEinv$, the DAE~\eqref{DAE:EA} takes the form
%
\begin{equation}\label{ODE:A-tilde}
\dot{y}(t) =\tA y
\qquad\text{where }
\tA :=\msqrtEinv\mA \msqrtEinv \ .
\end{equation}
%
However, if $\mE$ is singular then the behavior of the two systems~\eqref{ODE:A}--\eqref{DAE:EA} is fundamentally different.
Writing a matrix $\mA\in\Cnn$ as the sum of its Hermitian part $\mAH =(\mA +\mA^H)/2$ and skew-Hermitian part $\mAS =(\mA -\mA^H)/2$, we have the following definition.
%
\begin{definition}[{Definition 4.1.1 of~\cite{Be18}}] \label{def:dissipative:finiteDim}
A matrix~$\mA\in\Cnn$ is called~\emph{dissipative} (resp. \emph{semi-dissipative}) if the Hermitian part~$\mAH$ is negative definite (resp. negative semi-definite).
For a (semi-)dissipative matrix~$\mA\in\Cnn$, the associated ODE~\eqref{ODE:A} is called \emph{(semi-)dissipative Hamiltonian ODE}.
A DAE ~\eqref{DAE:EA} with (semi-)dissipative matrix $\mA\in\Cnn$ and positive semi-definite Hermitian matrix~$\mE\in\Cnn$ is called~ \emph{(semi-)dissipative Hamiltonian DAE}.
\end{definition}
%
The notion (semi-)dissipative Hamiltonian is motivated by the fact that if $\mAH=0$ and $\mE$ is the identity, then~\eqref{ODE:A} is a \emph{Hamiltonian system} with \emph{Hamiltonian} $\mathsf H=(x^H \mE x)/2$, see~\cite{MehMW18} and also
%and~\eqref{DAE:EA} is a Hamiltonian
%Note that %since $\mE=\mE^H\geq 0$, note that
%the above definition for the DAE~\eqref{DAE:EA} is %independent of~$\mE$ (see
\cite[Remark 1]{AAM21} and~\cite[Theorem 3(E1)]{AAM21}.
%
In the following (to avoid too many indices), we often write semi-dissipative matrices~$\mA$ in the form~$\mA =\mJ -\mR$ with a skew-Hermitian matrix~$\mJ=\mAS$ and a positive semi-definite Hermitian matrix $\mR =-\mAH$.
For ODE systems, hypocoercivity and the hypocoercivity index are defined as follows:
%
\begin{definition}[{\cite{AAC22}}] \label{def:matrix:hypocoercive}
A matrix $\mC\in\Cnn$ is called~\emph{coercive} if its Hermitian part $\mCH$ is positive definite, and it is called~\emph{hypocoercive} if the spectrum of~$\mC$ lies in the \emph{open} right half plane.
%For practical reasons, a matrix~$\mA\in\Cnn$ is called \emph{negative hypocoercive} if the spectrum of~$\mA$ lies in the \emph{open} left half plane. [AA: ever needed??]
%\end{definition}
%\begin{definition}[{\cite{AAC22}}]
%\label{def:matrix:HC-index}
Let $\mJ,\mR\in\Cnn$ satisfy $\mR=\mR^H\geq 0$ and $\mJ=-\mJ^H$.
The~\emph{hypocoercivity index (HC-index)~$m_{HC}$ of the matrix~$\mC=\mR-\mJ$} is defined as the smallest integer~$m\in\N_0$ (if it exists) such that
\begin{equation}\label{Tm:J-R}
%\mT_m :=
\sum_{j=0}^m \mJ^j \mR (\mJ^H)^j > 0
\end{equation}
holds.
\end{definition}
\begin{remark}\label{rem:short-t-decay}
\begin{enumerate}
\item[(1)] Clearly, \eqref{Tm:J-R} is equivalent to the condition
\begin{equation}\label{Tm:J-R-strict}
\sum_{j=0}^m \mJ^j \mR (\mJ^H)^j \ge \kappa\mI
\end{equation}
for some $\kappa>0$, where $\mI$ denotes the identity matrix.
This variant will be needed in the infinite dimensional case below.
Condition~\eqref{Tm:J-R-strict} was also used in \cite{AAC18}.
\item[(2)] By a non-trivial result, the HC-index characterizes the short-time decay of semi-dissipative Hamiltonian ODE systems $\dot x(t)=-\mC x(t)$: Its system matrix $\mC$ has a HC-index $m_{HC}\in\N_0$ if and only if
\begin{equation} \label{short-t-decay}
\|e^{-\mC t}\|_2 = 1-ct^{2m_{HC}+1} + \mathcal O(t^{2m_{HC}+2}) \quad \mbox{for }t\to 0^+,
\end{equation}
for some $c>0$. Here we used that $\|e^{-\mC t}\|_2$ is a real analytic function on some (small) time interval $[0,t_0)$, see Theorem 2.7(a) in \cite{AAC22}.
This classification can be extended to the DAE case by considering the HC-index of the dynamical part, see Proposition \ref{prop:DAE+HC-decay} in the Appendix
\ref{app:DAEs:hypocoercivity}.
\end{enumerate}
\end{remark}
Condition \eqref{Tm:J-R} is equivalent to the well known \emph{Kalman rank condition}:
%
\begin{lemma}[{\cite[Proposition 1, Remark 4]{AAC18}}] \label{lem:Equivalence}
Let $\mJ,\mR\in\Cnn$ satisfy $\mR=\mR^H\geq 0$ and $\mJ=-\mJ^H$. %, and let $\sqrt(\mR)$ be the positive semi-definite square root of $\mR$.
%
Then the following conditions are equivalent:
%
\begin{enumerate}[(B1)]
\item \label{B:KRC}
There exists $m\in\N_0$ such that
%
\begin{equation}\label{condition:KalmanRank}
\rank[{\mR},\mJ{\mR},\ldots,\mJ^m {\mR}]=n \,.
\end{equation}
%
\item \label{B:KRC'}
There exists $m\in\N_0$ such that
%
\begin{equation}\label{condition:KalmanRank'}
\bigcap_{j=0}^m \ker\big(\mR^{1/2}\ \mJ^j\big)
= \{0\} \,.
\end{equation}
%
\item \label{B:Tm}
There exists $m\in\N_0$ such that \eqref{Tm:J-R} holds.
%
%\begin{equation}\label{Tm:J-R}
% \mT_m :=\sum_{j=0}^m \mJ^j \mR (\mJ^H)^j > 0 \,.
%\end{equation}
\end{enumerate}
%
Moreover, the smallest possible~$m\in\N_0$ in~\ref{B:KRC}, \ref{B:KRC'}, and~\ref{B:Tm} coincide.
\end{lemma}
%
A similar equivalence result (on a subspace describing the dynamics of the system) has also been shown for DAE systems in \cite{AAM21}, see Appendix~\ref{app:DAEs:hypocoercivity} here.\\
The characterization of hypocoercive matrices is related to results in control theory (as discussed in~\cite[Remark 2]{AAM21}):
\begin{remark}[Connection to control theory: finite dimensional setting] \label{rem:controlTheory}
Consider a state-space system
\begin{equation}\label{ODE:AxBu}
\dot x(t) =\mA x +\mB u
\end{equation}
for constant matrices $\mA,\mB\in\Cnn$. % (not necessarily of the same dimensions).
%
A pair~$(\mA,\mB)$ of square matrices~$\mA,\mB\in\Cnn$ is called \emph{controllable} if the controllability matrix~$[\mB, \mA\mB, \mA^2 \mB,\ldots,\mA^{n-1} \mB]$ has full rank.
For a controllable pair~$(\mA,\mB)$, the smallest possible integer~$k$ such that the controllability (sub)matrix $[\mB, \mA\mB, \mA^2 \mB,\ldots,\mA^{k-1} \mB]$ has full rank, is called the \emph{controllability index}, see e.g.~\cite[\S 6.2.1]{Ch99}, \cite[\S 6.2.1]{Ka80}, or~\cite[\S 5.7]{Wo85}.
%In the discrete-time setting, %$\sum_{j=0}^\infty \mA^j \mB\mB^H %(\mA^H)^j$ is known as %\emph{controllability Gramian}.
\begin{itemize}
%
\item
For semi-dissipative matrices~$\mJ-\mR$, the HC-index of $\mR-\mJ$ is one less than the controllability index of~$(\mA,\mB)=(\mJ,\mR)$.
%
\item
Moreover, for general input-output systems there is also the dual concept of \emph{observability}.
Conditions like~\eqref{condition:KalmanRank'} are most often formulated in the context of observability, but not %only
frequently in the context of controllability.
%Therefore, the equivalence of conditions in Lemma~\eqref{lem:Equivalence} is well-known, but discussed separately in continuous-time/discrete-time and controllability/observability settings with/without mentioning the controllability/observability index.
%
\item
Whereas the controllability/observability indices have a clear interpretation in the discrete-time setting, see e.g.~\cite[p.171]{Ch99}; their interpretation in the continuous-time setting is not so clear.
In particular, we are not aware of a characterization which is comparable to~\eqref{short-t-decay} in Remark~\ref{rem:short-t-decay}.
\end{itemize}
\end{remark}
While the two conditions \eqref{condition:KalmanRank} and \eqref{condition:KalmanRank'}
are clearly equivalent in the finite dimensional ODE case, only the latter one may also be used for linear operators.
In fact, the above results are easily extended to
%In this section we recall some basics for
evolution equations of the form
%
\begin{equation}\label{C-eq}
\ddt{x}=-\mC x\,,\quad t>0,
\end{equation}
%
with $\mC$ a bounded or unbounded operator
on some infinite dimensional %, separable
Hilbert space~$\cH$:
\begin{definition}[{\cite[\S V.3.10]{Kato}}]
A linear operator $\mC$ on a Hilbert space $\cH$, with domain $\cD(\mC)$, is said to be \emph{accretive} if the numerical range of $\mC$ is a subset of the right half plane, that is, if $\Re\ip{\mC x}{x} \geq 0$ for all $x\in\cD(\mC)$.
In this case $-\mC$ is said to be \emph{dissipative}.
And $\mC$ is called \emph{coercive} if there exists $\gamma>0$ such that $\ip{\mC x}{x}\geq \gamma\|x\|^2$ for all $x\in\cD(\mC)$.
\end{definition}
%
Note that in this definition we follow the convention in semigroup theory, see e.g.~\cite[Proposition 3.23]{EnNa00}; whereas to be consistent with Definition~\ref{def:dissipative:finiteDim} we would have to call such an operator \emph{semi-dissipative}.
% [Pazy, Corollary 4.4] of Lumer-Phillips on %Banach spaces
%, where the linear operator~$\mC\in {\mathcal %L}({\mathcal H})$ has the domain $\cD(\mC)$, %has trivial kernel, is not coercive, but the %solutions of\eqref{C-eq} exhibit exponential %decay in time, as in \eqref{exp-decay}.
% \makered{Since $\cH$ is isometrically %isomorphic to $\ell^2(\N)$, we shall, for %notational simplicity, only consider the latter %here.
% We shall therefore denote in %\S\ref{ssec:nonconst-drift} the application of %operators to an element of $\ell^2(\N)$ as a %matrix-vector multiplication (both being of %infinite dimension).}
%A dissipative operator $\mA$ on a Hilbert space~$\cH$ which is \emph{densely defined}, i.e. for which $\overline{\cD(\mA)}=\cH$, generates a \emph{contraction semigroup} $(\SG(t))_{t\geq 0}$, i.e., it satisfies $\|\SG(t)\|\leq 1$ for all $t\geq0$. See e.g. \cite{CuZw20}.
%\makered{[Das reicht nicht für Lumer-Phillips.]}
The classical Lyapunov characterization of (uniformly) exponentially stable semigroups on finite-dimensional Hilbert spaces easily extends to infinite dimensional settings:
%
\begin{theorem}[{\cite[Theorem 4.1.3]{CuZw20}},\cite{Da70}]
Suppose that $\mA$ is the infinitesimal generator of the $C_0$-semigroup $\mT(t)$ on the Hilbert space~$\cH$.
Then $\mT(t)$ is uniformly exponentially stable if and only if there exists a bounded positive operator $\mP\in\cB(\cH)$ such that
\begin{equation} \label{LMI:H}
\ip{\mA x}{\mP x} +\ip{\mP x}{\mA x} =-\ip{x}{x} \qquad\text{for all } x\in\cD(\mA)\ .
\end{equation}
Equation~\eqref{LMI:H} is called \emph{Lyapunov equation}.
If $\mT(t)$ is uniformly exponentially stable, then the unique self-adjoint solution of~\eqref{LMI:H} is given by
\begin{equation}
\mP x =\int_0^{\infty} \mT(s)^* \mT(s) x \dd[s]
\qquad\text{for } x\in\cH\ .
\end{equation}
\end{theorem}
%
Next we recall the definition of hypocoercive operators in~\cite{Vi09}, which generalizes Definition~\ref{def:matrix:hypocoercive}:
% to linear operators:
% see also \cite[\S V.1 Eq.~(1.9)]{EnNa00}.
%
\begin{definition}[{\cite[\S I.3.2]{Vi09}}] \label{def:hypoinfdim}
Let~$\mC$ be a (possibly unbounded) operator on a Hilbert space~$\cH$ with kernel~$\ker\mC$.
Let~$\widetilde\cH$ be a Hilbert space, which is continuously and densely embedded in $(\ker\mC)^\perp$, endowed with a scalar product~$\ip{\cdot}{\cdot}_{\widetilde\cH}$ and norm $\|\cdot\|_{\widetilde\cH}$.
%Let $-\mC$ generate a continuous semigroup $(\ee^{-t\mC})_{t\geq 0}$ on~$\cH$. \\
The operator $\mC$ is called \emph{hypocoercive} on $\widetilde\cH$ if $-\mC$ generates a uniformly exponentially stable $C_0$-semigroup $(e^{-\mC t })_{t\geq 0}$ on~$\widetilde\cH\hookrightarrow (\ker\mC)^\perp$, i.e.~\eqref{exp-decay} holds.
\end{definition}
Next we shall generalize the notion \emph{hypocoercivity index} to the infinite dimensional Hilbert space case. As in the finite dimensional case, we consider operators of the form $\mC=\mR-\mJ$ with $\mR$ self-adjoint and nonnegative, and $\mJ$ skew-adjoint. For technical reasons (related to the operator domain) we shall first discuss bounded operators $\mC$ and then some special situations of unbounded operators $\mC$.
%In order to have a characterization of hypocoercivity as in the finite dimensional case, we consider first bounded operators that are the sum of a self-adjoint nonnegative and skew-adjoint operator.
%
\begin{definition}
\label{def:op:HC-index}
Consider bounded operators $\mR,\,\mJ\in\cB(\cH)$ on a Hilbert space~$\cH$ such that $\mR$ is self-adjoint and nonnegative, and $\mJ$ is skew-adjoint, i.e., $\mJ^*=-\mJ$.
The~\emph{hypocoercivity index (HC-index)~$m_{HC}$ of the (accretive) operator~$\mC:=\mR-\mJ\in{\mathcal B}({\mathcal H})$ } is defined as the smallest integer~$m\in\N_0$ (if it exists) such that
\begin{equation}\label{Op:J-R}
\sum_{j=0}^m \mJ^j \mR (\mJ^*)^j \ge\kappa\mI
% \sum_{j=0}^m \mJ^j \mR (\mJ^*)^j >0
\end{equation}
for some $\kappa>0$.
%in the sense that
%\begin{equation}\label{Op:J-Rx}
% \sum_{j=0}^m \big\langle x, \mJ^j \mR (\mJ^*)^j x\big\rangle %_{\cH}
% >0 \quad \mbox{for all}\, x\in\cH\setminus\{0\}\,.
%\end{equation}
\end{definition}
%
Here we generalized the uniform condition \eqref{Tm:J-R-strict}, since the example $\mC=\mR=\diag(1/j;\,j\in\N)$ on $\cH=\ell^2(\N)$ would satisfy the non-uniform condition \eqref{Tm:J-R}. However, the corresponding semigroup satisfies $\|e^{-\mC t}\|=1,\,t\ge0$, hence $\mC$ is not hypocoercive.
%Note that in contrast to the finite dimensional counterpart, \eqref{Tm:J-R}, the left hand side of \eqref{Op:J-Rx} does not have to be bounded away from zero uniformly on the unit ball in $\cH$.
%\makered{We remark that the correspondence \eqref{short-t-decay} between the hypocoercivity index of $\mC$ and the short-time behavior of the propagator norm $\|e^{-\mC t}\|$ can be generalized to generators $\mC\in{\mathcal B}(\mathcal H)$, see the follow-up paper \cite{AAMN22}.\\ }
%{\color{magenta} Volker, ich denke das sollten wir nicht schreiben und auch nicht auf ein noch zu schreibendes paper zizieren}
\begin{comment}
%
\begin{lemma} \label{lem:Op-Equivalence}
Consider bounded operators $\mR,\,\mJ\in\cB(\cH)$ on a Hilbert space~$\cH$ such that $\mR$ is self-adjoint and nonnegative, and $\mJ$ is skew-adjoint, i.e., $\mJ^*=-\mJ$.
Then the following two conditions are equivalent:
\begin{enumerate}[(B1')]
\setcounter{enumi}{1}
\item \label{B:KRC'op}
There exists $m\in\N_0$ such that
%
\[
\bigcap_{j=0}^m \ker\big(\mR^{1/2}\ \mJ^j\big)
= \{0\} \,.
\]
%
\item \label{B:Tmop}
There exists $m\in\N_0$ such that \eqref{Op:J-R} holds. \end{enumerate}
Moreover, the smallest possible~$m\in\N_0$ coincides in both cases (if it exists).
\end{lemma}
\begin{proof}
Since we have for all $x\in\cH$
\[
\sum_{j=0}^m \big\langle x,\mJ^j \mR (\mJ^*)^j x\big\rangle %_{\cH}
= \sum_{j=0}^m \| \mR^{1/2}\ \mJ^j x\|^2 %_{\cH}^2\,,
\]
the result is immediate.
\end{proof}
\end{comment}
When defining next the hypocoercivity index for unbounded operators, we do not aim at the largest generality of equations \eqref{C-eq}, but rather we present a framework that covers the Oseen equations in \S\ref{ssec:Oseen:anisotropic} below.
In our extension, the unbounded operator $\mC=\mR-\mJ$ is accretive
with the following assumptions:
\begin{assumptions}\
\begin{enumerate}[label=(A\arabic*)]
\item \label{prop:R}
The unbounded operator $\mR$ with dense domain $\mathcal D(\mR)\subset \cH$ is self-adjoint (and hence closed) and nonnegative in $\cH$.
The operator $\mJ$ is bounded and skew-adjoint on $\cH$.
\item \label{prop:J}
% The operator $\mJ$ is bounded and skew-adjoint on $\cH$, and
$\mJ$ satisfies $\mJ(\mathcal D(\mR))\subset \mathcal D(\mR)$.
%
\item \label{prop:R1/2}
For the self-adjoint (and hence closed) operator $\mR^{1/2}$ defined on $\mathcal D(\mR^{1/2})\supset\mathcal D(\mR)$ assume that $\mJ(\mathcal D(\mR^{1/2}))\subset \mathcal D(\mR^{1/2})$.
\end{enumerate}
\end{assumptions}
Under these assumptions, standard arguments from semigroup theory show that $-\mR$ and hence also $-\mC=-\mR+\mJ$ are dissipative with domain~$\cD(\mR)$ and, hence, infinitesimal generators of $C_0$-semigroups of contractions on $\cH$, see e.g. \S1.4 of \cite{Pazy}.
Moreover these semigroups are analytic, see~\cite[Theorem III.2.10]{EnNa00}:
Due to~\cite[Corollary II.4.7]{EnNa00}, the operator $-\mR$ generates a bounded analytic semigroup on~$\cH$, and $\mJ$ is $\mR$-bounded with $\mR$-bound $a_0=0$, see~\cite[Definition III.2.1]{EnNa00}.
%Thus, due to~\cite[Theorem III.2.10]{EnNa00}, the operator $\mJ-\mR$ generates an analytic semigroup on~$\cH$.
For unbounded operators, the relation between the domains of the operator, its self-adjoint part and its skew-adjoint part can be subtle.
Therefore, different extensions of Definition~\ref{def:op:HC-index} are reasonable.
For example, under the assumptions~\ref{prop:R}--\ref{prop:J}, each term of the sum~\eqref{Op:J-R} is well-defined on $\mathcal D(\mR)$.
However, the following extension to $\cD(\mR^{1/2})$ is more convenient for the subsequent lemma:
\begin{definition}\label{def:unb-op:HC-index}
Let the (accretive) operator~$\mC=\mR-\mJ$ satisfy the Assumptions~\ref{prop:R}, \ref{prop:R1/2}.
Then the~\emph{hypocoercivity index} (HC-index)~$m_{HC}$ of~$\mC$ is defined as the smallest integer~$m\in\N_0$ (if it exists) such that
\begin{equation} \label{Op:J-R:unbdd}
\sum_{j=0}^m \| \mR^{1/2}\ (\mJ^*)^j x\|^2
\geq \kappa \|x\|^2 \qquad \mbox{for all } \,x\in\cD(\mR^{1/2})\ ,
\end{equation}
for some $\kappa>0$.
\end{definition}
In both of the above settings (Definitions~\ref{def:op:HC-index} and \ref{def:unb-op:HC-index}) we have the following infinite dimensional analog of Lemma \ref{lem:Equivalence}.
\begin{comment}
\makeblue{Compare with~\cite[Example 6.27g on p.287 ff]{CuZw20}, see also~\cite{CuPr78,Kl91}.} \makered{[AA: Wenn wir diese blauen Referenzen angeben, sollten wir explizit erklären, was der Leser aus diesem Vergleich lernen soll.]}
\end{comment}
%
\begin{lemma} \label{lem:Op-Equivalence}
Let the operators $\mR,\,\mJ\in\cL(\cH)$ on a Hilbert space~$\cH$ satisfy either the assumptions in Definition \ref{def:op:HC-index} (if $\mC=\mR-\mJ$ is bounded) or the assumptions in Definition \ref{def:unb-op:HC-index} (if $\mC$ is unbounded).
Then the following three conditions are equivalent:
\begin{enumerate}[(B1')]
\setcounter{enumi}{0}
%
\item \label{B:G*_surjective}
There exists $m\in\N_0$ such that
\[
\spn \Bigg(\bigcup_{j=0}^m \range\big(\mJ^j \mR^{1/2}\big)\Bigg) =\cH\ .
\]
%
\item \label{B:KRC'op}
There exists $m\in\N_0$ such that
%
\[
\bigcap_{j=0}^m \ker\big(\mR^{1/2}\ \mJ^j\big)
= \{0\} \,, \qquad
\text{and }
\spn \Bigg(\bigcup_{j=0}^m \range\big(\mJ^j \mR^{1/2}\big)\Bigg)\ \text{ is closed.}
\]
%
\item \label{B:Tmop}
There exists $m\in\N_0$ such that \eqref{Op:J-R:unbdd} holds for some $\kappa>0$.
\end{enumerate}
Moreover, the smallest possible~$m\in\N_0$ coincides in all cases (if it exists).
\end{lemma}
\begin{proof}
To prove Lemma~\ref{lem:Op-Equivalence} we will make use of the (equivalent) characterizations of surjective operators in~\cite[Theorem 2.21]{Br11}.
To this end, we will introduce the following Hilbert spaces and operators.
For $m\in\N_0$, the direct sum $\cH^{m+1}:=\bigoplus_{j=0}^m \cH =\{(y_0,\ldots,y_m)|\ y_j\in\cH,\ j=0,\ldots,m\}$ endowed with the inner product $\ip{x}{y}_{\cH^{m+1}}:=\sum_{j=0}^m \ip{x_j}{y_j}$; $x,y\in\cH^{m+1}$ is again a Hilbert space.
Define the linear operator
\[
\mG:\
\cD(\mR^{1/2}) \subset\cH \to \cH^{m+1}\,, \qquad
x \mapsto \begin{bmatrix} \mR^{1/2}x \\ \mR^{1/2}\mJ^* x \\ \vdots \\ \mR^{1/2} (\mJ^*)^m x \end{bmatrix}\ .
\]
Since $\mR^{1/2}$ is closed and $\mJ$ is bounded, also the operators $\mR^{1/2}(\mJ^*)^j$ are closed. Hence the operator~$\mG$ is densely defined (in both cases of assumptions $\cD(\mG):=\cD(\mR^{1/2})$ is dense in $\cH$) and closed.
Its adjoint reads
\begin{align*}
\mG^*:\
\cD(\mG^*) \subset\cH^{m+1} &\to \cH\,, \quad\mbox{with } \cD(\mG^*) \supset \big(\cD(\mR^{1/2})\big)^{m+1}\,, \\
y &\mapsto \sum_{j=0}^m \mJ^j \mR^{1/2} y_j\ .
\end{align*}
Next, we identify the operator $\mG$ and its adjoint $\mG^*$ in our statement with the ones in~\cite[Theorem 2.21]{Br11}:
First,
\begin{equation}\label{rangeG*}
\range(\mG^*)
=\spn \Bigg(\bigcup_{j=0}^m \range\big(\mJ^j \mR^{1/2}\big)\Bigg) \stackrel{!}{=} \cH
\end{equation}
gives the equivalence of the conditions \ref{B:G*_surjective} and (a) in \cite[Theorem 2.21]{Br11}, where ``$\stackrel{!}{=}$'' indicates the condition to be satisfied.
Note that in \eqref{rangeG*} each $\range\big(\mJ^j \mR^{1/2}\big)$ has to be evaluated on the $j$-th component of $\cD(\mG^*)$, which may indeed be a proper superset of $\cD(\mR^{1/2})$.
Second,
\[
\ker(\mG)
= \bigcap_{j=0}^m \ker\big(\mR^{1/2} (\mJ^*)^j\big)
= \bigcap_{j=0}^m \ker\big(\mR^{1/2} \mJ^j\big)
\stackrel{!}{=} \{0\}\ .
\]
Moreover, due to the assumptions, $\range(\mG)$ is closed if and only if $\range(\mG^*)$ is closed.
This gives the equivalence of the conditions \ref{B:KRC'op} and (c) in \cite[Theorem 2.21]{Br11}. %, where ``$\stackrel{!}{=}$'' indicates the condition to be satisfied.
Third,
\begin{multline*}
\| \mG x\|_{\cH^{m+1}}^2
= \sum_{j=0}^m \| (\mG x)_j\|^2
= \sum_{j=0}^m \| \mR^{1/2}\ (\mJ^*)^j x\|^2
\stackrel{!}{\ge} \kappa \|x\|^2
\\
\text{for all } x\in \cD(\mR^{1/2})=\cD(\mG) \ .
\end{multline*}
This gives the equivalence of the conditions \ref{B:Tmop} and (b) in \cite[Theorem 2.21]{Br11}. %, where ``$\stackrel{!}{\ge}$'' indicates the condition to be satisfied.
The three equivalences established in Theorem~\cite[Theorem 2.21]{Br11} thus imply the three equivalences of Lemma~\ref{lem:Op-Equivalence}.
\end{proof}
\begin{remark}[Connection to control theory: infinite dimensional setting]
Consider the control system~\eqref{ODE:AxBu} where, for simplicity, $\mA,\mB\in\cL(\cH)$ operate on the same Hilbert space~$\cH$.
In the infinite dimensional setting many more concepts of controllability exist, %/observability
e.g.~depending on the operator domains~$\cD(\mA),\cD(\mB)\subset\cH$.
In \cite[Theorem 3.18]{CuPr78} and \cite[Theorem 6.2.27]{CuZw20} it is observed that for $\mA,\mB\in\cB(\cH)$, system~\eqref{ODE:AxBu} is \emph{exactly controllable} if and only if Condition \ref{B:G*_surjective} in Lemma \ref{lem:Op-Equivalence} holds.
%Moreover, the controllability %/observability
%index is seldom explicitly discussed [in textbooks such as~\cite{CuZw20,St05}], but see~\cite[Theorem 3.18]{CuPr78}.
%However, the equivalence of the conditions in Lemma~\ref{lem:Op-Equivalence} \emph{with $\mJ,\mR\in\cB(\cH)$ and $m=\infty$} is discussed e.g. in~\cite[Example 6.27f]{CuZw20} in the discrete-time setting.
\end{remark}
After recalling
the basic hypocoercivity concepts, in the next two sections we apply the techniques for the analysis of different variants of the Oseen equations.
We consider the Oseen equations as a constrained PDE model on a torus. Since this allows for a modal decomposition, it reduces to an infinite system of DAEs.
Depending on the detailed shape of the drift and diffusion terms in the Oseen equation, these models exhibit a wide range of hypocoercivity phenomena:
it may be coercive (see \S\ref{ssec:Stokes}), hypocoercive with index 1 (see \S\ref{ssec:nonconst-drift}), or not hypocoercive (see \S\ref{ssec:const-drift}).
%---------------------------------------------------------------------------
\section{Isotropic Oseen equation on the 2D torus} \label{ssec:Stokes}
%\makered{[AA: Kann die modale Zerlegung hier und in \S\ref{ssec:const-drift} auch f\"ur beliebige Gebiete $\Omega$ mit no-slip RB gemacht werden? Entkoppeln die Moden dann weiterhin?] see~\cite{FMRT01} }
As first model we consider the time-dependent, incompressible Oseen equation of fluid dynamics with \emph{isotropic} viscosity on the 2D torus $\T^2 := (0,2\pi)^2$,
\begin{subequations}\label{eq:Oseen}
\begin{align}
u_t
&=-(b\cdot\nabla)u -\nabla p +\nu \Delta u\,,
&& t>0\,, \quad\text{on }\T^2\,, \label{eq:Oseen:1a}
\\
0
&=-\diver u \,,
&& t\geq 0\,, \label{eq:Oseen:1b}
\end{align}
\end{subequations}
for the vector-valued velocity field~$u=u(x,t)$ and the scalar pressure~$p =p(x,t)$ in the space variable~$x\in\T^2$ and the time variable~$t\geq 0$.
The constant $\nu>0$ denotes the viscosity coefficient and $b\in\R^2$ is the constant drift velocity.
Note that since in~\eqref{eq:Oseen} diffusion acts uniformly in all directions, we call the Oseen model~\eqref{eq:Oseen} \emph{isotropic}.
For \eqref{eq:Oseen} we assume periodic boundary conditions in both $u$ and $p$.
%Due to the periodic boundary conditions in~\eqref{eq:Oseen}
Hence, this model actually could be simplified right away:
Taking the divergence of the first equation in~\eqref{eq:Oseen} yields
$\Delta p(\cdot,t) =0$ and hence~$p(\cdot,t)$ is constant in~$x$.
It also shows that the vector-valued transport-diffusion equation
\begin{equation}\label{trans-diff}
u_t=-(b\cdot\nabla)u +\nu \Delta u
\end{equation}
preserves the incompressibility if the initial condition satisfies $\diver u(0)=0$, which is assumed in the sequel.
Since it is known that in this case the normal component of $u$ has a periodic extension \cite[\S IX.1.2+3]{DL3}, \cite[\S II.5]{FMRT01},
it follows that, for any initial condition
%
\[
u(0)\in H_{per}(\diver 0, \T^2):=\{u\in (L^2(\T^2))^2\, |\,\diver u=0\},
\]
%
equation \eqref{trans-diff} and hence \eqref{eq:Oseen} has a unique smooth solution for $t>0$, and its explicit Fourier representation can be obtained from the first equation of the Fourier expansion \eqref{eq:Stokes:Fourier} below.
Here, in order to pave the way for more general applications, we proceed differently and employ negative hypocoercivity of matrix pencils in semi-dissipative Hamiltonian DAEs. So we ignore this possible simplification and rather follow our discussion from~\cite[\S3]{AAM21}.
The following analysis is an extension of \S4.1 in \cite{AAM21}, which considered the Stokes equation.
Due to the periodic setting, we consider the Fourier expansion of~\eqref{eq:Oseen} with
%
\begin{equation}\label{Fourier}
u(x,t)
= \sum_{k\in\Z^2} \phi_k(t) e^{i k\cdot x} \,, \qquad
p(x,t)
= \sum_{k\in\Z^2} p_k(t) e^{i k\cdot x} \,.
\end{equation}
%
Since $u$ and $p$ are real valued, their Fourier coefficients~$\phi_k(t)\in\C^2$, $p_k(t)\in\C$, $k\in\Z^2$, obey the symmetry $\bar\phi_k=\phi_{-k}$, $\bar p_k=p_{-k}$ and they satisfy the decoupled evolution equations
%
\begin{equation}\label{eq:Stokes:Fourier}
\begin{cases}
\ddt \phi_k &= -i k p_k -i(b\cdot k)\phi_k -\nu |k|^2 \phi_k \ , \quad t>0 \ , \\
0 &=-i k\cdot \phi_k \ .
\end{cases}
\end{equation}
The mode~$k=0$ satisfies $\phi_0(t)=$ const (corresponding to momentum conservation) and~$p_0(t)=$ arbitrary.
To enforce unique solvability of~\eqref{eq:Oseen}, we require the pressure as $p_0(t)\equiv 0$. Hence, this flow is a \emph{transversal wave} \cite[p.73]{La1996}.
For~$k\ne 0$ we write~\eqref{eq:Stokes:Fourier} as a system of decoupled DAEs.
%, each having DAE-index~$2$:
%
\begin{equation} \label{DAE:Stokes}
\mE \dot{w}_k (t) =\mA_k \xk \ , \quad t\geq 0\ ,
\end{equation}
%
for~$\xk :=[\ukI,\ukII,p_k]^\top \in\C^3$ with the matrices~$\mE:=\diag(1,1,0)$ and
%
\begin{equation} \label{DAE:Stokes:A}
\mA_k :=
\begin{bmatrix}
-i(b\cdot k)-\nu |k|^2 & 0 & -i k_1 \\
0 & -i(b\cdot k)-\nu |k|^2 & -i k_2 \\
-i k_1 & -i k_2 & 0
\end{bmatrix} \ .
\end{equation}
The modal functions~$\xk(t)$, $k\in\Z^2$ correspond to the function~$x(t)$ in~\S\ref{sec:hypocoercivity} and \cite{AAM21}, since~$x=[x_1, x_2]^\top$ is used here for the spatial variable.
%
Following the notation from~\S\ref{sec:hypocoercivity}, we decompose $\mA_k$ as $\mA_k =\mJ_k -\mR_k$ with $\mR_k :=\diag(\nu |k|^2,\nu |k|^2, 0)$ and
%
\begin{equation} \label{DAE:Stokes:J}
\mJ_k :=
\begin{bmatrix}
-i(b\cdot k) & 0 & -i k_1 \\
0 & -i(b\cdot k) & -i k_2 \\
-i k_1 & -i k_2 & 0
\end{bmatrix} \ .
\end{equation}
%
In order to determine the hypocoercivity index of~\eqref{DAE:Stokes} we employ the unitary transformation of~\eqref{DAE:Stokes} to staircase form via the Algorithm in~\cite[Lemma 5]{AAM21}, which we recall as Lemma~\ref{lem:tSF} in Appendix~\ref{app:DAEs:hypocoercivity}.
%which actually
Applying the staircase algorithm to the triple~$(\mE,\mJ_k,\mR_k)$ yields
%the following (unitary) congruence %transformation. We remark that the Staircase %Algorithm actually constitutes the proof of %Lemma \ref{lem:tSF},} \new{and is reproduced in %.}
\begin{subequations}\label{checkJk}
\begin{align} %\label{checkJk}
\widecheck \mJ_k
&=\mP_k \mJ_k \mP_k^H
=\begin{bmatrix}
-i(b\cdot k) & 0 & -|k| \\
0 & -i(b\cdot k) & 0 \\
|k| & 0 & 0
\end{bmatrix} \ , \quad
%
\\
\widecheck \mR_k
&=\mP_k \mR_k \mP_k^H
=\mR_k \ , \qquad
%
\widecheck \mE
=\mP_k \mE \mP_k^H
=\mE \ ,
\end{align}
\end{subequations}
%with constants $n_1 =n_2 =n_4=1$, $n_3 =n_5 =0$ %
where
%
\begin{equation} \label{ex:Stokes:Pk}
\mP_k
=\tfrac1{|k|}
\begin{bmatrix}
k_1 & k_2 & 0 \\
-k_2 & k_1 & 0 \\
0 & 0 & i|k|
\end{bmatrix} \ .
\end{equation}
%Due to these values of the constants $n_j$ we %shall use (particularly in %\S\ref{ssec:nonconst-drift}) the indices %%$1,\,2,\,4$ for the three rows and columns in the above $3\times3$ matrices. \\
The evolution of~\eqref{DAE:Stokes} translates via $y_k =[y_{k,1},y_{k,2},y_{k,3}]^\top :=\mP_k \xk$
%(note that $y_{k,3}$ is void)
into the staircase form
%
\begin{equation} \label{DAE:Stokes:SF}
\widecheck \mE \dot y_k (t) =(\widecheck \mJ_k -\widecheck \mR_k) y_k(t) \ , \quad t\geq 0 \ .
\end{equation}
%
Note that in the staircase form~\eqref{staircase:EJR} the third and fifth blocks are missing, i.e. $n_3=0$ and $n_5=0$.
%\smallskip
%From (P1), (P2) in~\cite[Corollary 1]{AAM21} we %see that all pencils $\lambda \widecheck \mE %-(\widecheck \mJ_k -\widecheck \mR_k)$, $k\ne %0$ are regular of DAE-index~$2$.
Setting $y_k:=\mP_k w_k$, $y_k=:[y_{k,1}, y_{k,2}, y_{k,3}]^\top$ and using~\eqref{DAE:Stokes:SF}, we obtain
%
\[
y_{k,1} =\tfrac1{|k|} k\cdot \phi_k =0 \,, \qquad
y_{k,3} = i p_k =0 \qquad
\text{for all } k\ne 0 .
\]
%
Hence, following \eqref{underlyingODE:y} in Appendix~\ref{app:DAEs:hypocoercivity}, the dynamic part of
the DAE~\eqref{DAE:Stokes:SF} is given as
%\[
%\mE_{2,2} \dot y_{k,2}
%=
%\widehat \mA_{k,2,2} y_{k,2} \ ,
%\quad
%t\geq 0 \ ,
%\]
%with $\mE_{2,2}=1$, $\widehat \mA_{k,2,2} %=-i(b\cdot k)-\nu |k|^2$; \new{such that}
\begin{equation} \label{DAE:Stokes:y2}
\dot y_{k,2} =-i(b\cdot k) y_{k,2}-\nu |k|^2 y_{k,2} \ , \quad k\ne 0 \ .
\end{equation}
Therefore, the DAE systems~\eqref{DAE:Stokes:SF}, for $k\ne 0$, exhibit non-trivial dynamics with HC-index $0$, see Definition~\ref{def:DAE:mHC}.
Equation~\eqref{DAE:Stokes:y2} is the modal decomposition of the (dissipative) transport-diffusion equation \eqref{trans-diff} on $\T^2$.
Hence, the solution $(u(\cdot,t),\,p(\cdot,t))$ of the Oseen equation~\eqref{eq:Oseen} converges, as $t\to \infty$, to the constant (in $x$ and $t$) equilibrium $(\phi_0,p_0)\in\R^3$, for $p_0=0$ with the exponential decay rate~$\mu=\min_{k\ne 0} (\nu|k|^2) =\nu$.
The transformation $y_k :=\mP_k w_k$ is related to the Helmholtz-Leray decomposition of vector fields on bounded domains, see~\cite[\S II.3]{FMRT01}:
\begin{remark}[Helmholtz-Leray decomposition]
Note that space-periodic vector fields $u\in H_{per}(\diver,\T^2)$ with $H_{per}(\diver, \T^2):=\{u\in (L^2(\T^2))^2\, |\,\diver u\in L^2(\T^2)\}$ can be written as
\begin{equation} \label{Helmholtz-Leray}
u =\nabla q +v \qquad\text{with } v\in H_{per}(\diver 0,\T^2)\ ,\quad q\in H^1(\T^2)\ .
\end{equation}
For $u\in H_{per}(\diver,\T^2)$, the expression~\eqref{Helmholtz-Leray} is called the \emph{Helmholtz-Leray decomposition} of~$u$, and~$q$ is unique (up to an additive constant).
Moreover, the map $\Pi_L: H_{per}(\diver,\T^2)\to H_{per}(\diver 0,\T^2)$, $u\mapsto v[u]$ is well-defined and a projection onto the divergence-free vector fields.
The map $\Pi_L$ is called the \emph{Leray projector} [for space-periodic vector fields].
\newline
To determine the Leray projector for $u\in H_{per}(\diver,\T^2)$ in the Fourier representation~\eqref{Fourier}, we extract the leading $2\times2$-subblock from $\mP_k$, $k\in\Z^2\setminus\{0\}$ and define
\begin{equation} %\label{ex:Stokes:Pk}
\widecheck\mP_k
:=\tfrac1{|k|}
\begin{bmatrix}
k_1 & k_2 \\
-k_2 & k_1 \\
\end{bmatrix} \ , \qquad
k\in\Z^2\setminus\{0\}\ .
\end{equation}
Then, for the modes $k\in\Z^2\setminus\{0\}$, the Leray projector $\Pi_L$ is given as
\[
\begin{bmatrix} \phi_{k,1} \\ \phi_{k,2} \end{bmatrix}
\mapsto
\widecheck\mP_k^*
\begin{bmatrix} 0 & 0 \\ 0 & 1 \end{bmatrix}
\widecheck\mP_k
\begin{bmatrix} \phi_{k,1} \\ \phi_{k,2} \end{bmatrix}\ .
\]
\end{remark}
\makeorange{
\begin{remark}[Possible extensions]
\begin{enumerate}
\item
Our hypocoercivity analysis can be extended to isotropic Oseen equations on the 3D torus:
A modal decomposition yields again a family of decoupled DAE-systems (but) with more complicated system matrices.
See also the hypocoercivity analysis of BGK-type equations on 1D, 2D, and 3D tori in~\cite{AAC18}.
\item
In case of the isotropic Oseen equation~\eqref{eq:Oseen} in the whole space and in exterior domains, solutions exhibit only algebraic-in-time decay, see~\cite{BaJi08}.
For kinetic equations in whole space and without confinement, the hypocoercivity analysis has been adapted to prove the expected algebraic [instead of exponential] temporal decay, see~\cite{BoDoMiMoSc20,ArDoScWo21}.
%see~\cite{BoDoMiMoSc20,ArJiWo20,ArDoScWo21}.
Similarly, the hypocoercivity analysis can be extended to Oseen equations in whole space.
\end{enumerate}
\end{remark}
}%-makeorange
When modifying~\eqref{eq:Oseen} into an anisotropic Oseen equation with viscosity only in the $x_2$-direction, the dynamics becomes more interesting:
Depending on the prescribed convection field~$b$, the dynamics may be hypocoercive or not.
This is the topic of the next section.
%-----------------------------------------------------------
\section{Anisotropic Oseen-type equations on the 2D torus}
\label{ssec:Oseen:anisotropic}
\makeorange{Motivated by rotating fluids in geophysics}, the 3D Navier-Stokes equation with anisotropic viscosity, where the vertical viscosity was reduced or even zero, has been studied, \makeorange{see~\cite{CDGG,Pa,Pa05periodic,CheDesGalGre06}}.
In analogy, we shall study now Oseen-type models~with anisotropic viscosity on the torus (\makeorange{similar to \cite{TW96,CheDesGalGre06,Mik23}}).
We consider these equations as a useful simplified mathematical model to illustrate (hypo)coercivity, rather than considering them for their physical interest (similar to~\cite{FMRT01,T95,T97}).
For simplicity of the presentation we again confine ourselves here to the 2D case, i.e. to $\T^2 :=(0,2\pi)^2$:
The model
\begin{subequations}\label{model:Oseen:anisotropic}
\begin{align}
u_t
&=-(b(x)\cdot \nabla) u -\nabla p +\nu \partial_{x_2}^2 u \,,
&& t>0 ,\ \text{on }\T^2 , \label{model:Oseen:anisotropic:1a} \\
%
0
&=-\diver u\ ,
&& t\geq 0\ , \label{model:Oseen:anisotropic:1b}
\end{align}
\end{subequations}
is subject to periodic boundary conditions;
it has prescribed transport with a drift velocity vector $b(x)\in\R^2$ which may depend on $x\in\T^2$, and diffusion only in~$x_2$ (hence, we call the model~\eqref{model:Oseen:anisotropic} \emph{anisotropic}).
\makeorange{
\begin{remark}
Navier-Stokes equation with anisotropic viscosity in~3D, are used to model rotating fluids in geophysics~\cite{CDGG,Pa,Pa05periodic,CheDesGalGre06}; but also to model \emph{anisotropic fluids}, e.g.~with
relaxed ellipticity condition on the viscosity tensor~\cite{Mik23}.
Linearizing the Navier-Stokes equation with anisotropic viscosity around a constant vector field~$b\in\R^2$ yields the anisotropic Oseen equation~\eqref{model:Oseen:anisotropic}.
Moreover, Oseen equations~\eqref{eq:Oseen} with space-dependent vector field $b=b(x)$ have been derived to describe the motion of a Navier–Stokes liquid around a rotating rigid body, see~\cite[\S VIII]{Gal11}.
This is strictly speaking not an Oseen equation anymore, since the derivation of Oseen needs that $b$ is constant in space.
The periodic setting may be used to model fluid flow in periodic composite structures, and is a basic setting in homogenisation theories for inhomogeneous fluids, see~\cite{Mik23}.
\end{remark}
}%end-makeorange
We analyze this model in two variants: For a constant convection field~$b\in\R^2$, the modes still decouple but the generator of the evolution is (depending on $b$) either coercive or not even hypocoercive.
If the convection field is non-constant in space, e.g. if
$b=[\sin (x_2),0]^\top$, then
%with $U_\infty(x_2)=\sin x_2$,}
the spatial modes are coupled and the generator of the (infinite dimensional) problem becomes hypocoercive.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Oseen-type equation with constant drift velocity}
\label{ssec:const-drift}
Let us first consider \eqref{model:Oseen:anisotropic} as an evolution equation on the space of divergence-free vector fields.
In the case of a constant drift velocity vector~$b\in\R^2$, we can argue like in~\S\ref{ssec:Stokes} to find $p(\cdot,t)=\textrm{const}$, and we again normalize the pressure as $p\equiv 0$.
Then, the (linear) generator of~\eqref{model:Oseen:anisotropic} takes the shape $-\mC =-b\cdot \nabla +\nu \partial_{x_2}^2$ and it acts identically on the two components $\uI$ and $\uII$.
Proceeding as in \S\ref{ssec:Stokes} and using the analog of the modal evolution equation \eqref{eq:Stokes:Fourier} shows that, for any initial condition $u(0)\in H_{per}(\diver 0, \T^2)$, Equation \eqref{model:Oseen:anisotropic} with $b\in\R^2$ has a unique mild solution for $t>0$.
Hence, the operator~$-\mC$ generates a $C_0$-semigroup on~$\cH :=H_{per}(\diver 0, \T^2)$, see also \cite{EnNa00,Pazy}.
But due to the degenerate, anisotropic diffusion and the lack of hypocoercivity (see Proposition \ref{prop:Oseen:constantU:notHC} below), solutions are in general not smooth here.
This is illustrated by an example, see~\eqref{counterex:Oseen:constantU:notHC} below.
In order to check the hypocoercivity of~$\mC$ as characterized in~\eqref{exp-decay}, we determine the kernel of $\mC$ in~$\cH$:
\begin{lemma}\label{lem:Oseen:constantU:kernel}
Let $b=[b_1,b_2]^\top\in\R^2$ be constant in~\eqref{model:Oseen:anisotropic}.
\begin{itemize}
\item
If $b_1\ne 0$ then $\ker\mC =\R^2$, i.e. the constant-in-$x$ flows.
Moreover, the $L^2$-orthogonal complement of $\ker\mC$ in~$\cH$ is given by \[ \{u\in H_{per}(\diver 0, \T^2) \ |\ \int_{\T^2} u\dd[x]=0\} \ .\]
\item
If $b_1=0$ then the kernel of~$\mC$ in~$\cH$ is given as
\[
\ker\mC =\{ u\in\cH:\ u_1=\const\ ,\quad u_2=u_2(x_1)\}.
\]
Moreover, the $L^2$-orthogonal complement of $\ker\mC$ in~$\cH$ is given by
\begin{equation} \label{tildeH:b_1:0}
\{u\in H_{per}(\diver 0, \T^2) \ |\ \int_{\T^2} u\dd[x]=0\ ,\quad u_2=u_2(x_2) \}.
\end{equation}
\end{itemize}
\end{lemma}
\begin{proof}
If $b_1\ne0$ then $\mC u=0$ reads $b\cdot \nabla u_j =\nu \partial_{x_2}^2 u_j$; $j=1,2$.
The space-``time'' periodic parabolic equation $b_1 \partial_{x_1} v =-b_2 \partial_{x_2} v +\nu \partial_{x_2}^2 v$ admits only constant periodic solutions.
The Hilbert space~$\cH$ is endowed with an $L^2$-inner product, which implies the final statement.
If $b_1=0$ then $\mC u=0$ reads $0 =-b_2 \partial_{x_2} u_j +\nu \partial_{x_2}^2 u_j$; $j=1,2$.
The elliptic equation $0 =-b_2 \partial_{x_2} v +\nu \partial_{x_2}^2 v$ admits only periodic solutions which are constant in $x_2$ and $2\pi$-periodic in $x_1$.
Moreover, $0 =\diver u =\partial_{x_1}u_1(x_1) +\partial_{x_2}u_2(x_1) =\partial_{x_1}u_1(x_1)$ implies that $u_1 =\const$.
The Hilbert space~$\cH$ is endowed with an $L^2$-inner product, which implies the final statement.
\end{proof}
We continue to analyze the anisotropic Oseen-type model~\eqref{model:Oseen:anisotropic} with $b\in\R^2$ and $b_1\ne 0$.
Following the characterization of hypocoercivity in~\eqref{exp-decay} and Lemma~\ref{lem:Oseen:constantU:kernel}, we also introduce the Hilbert space
\[
\wcH :=\{ u\in\cH\ |\ \int_{\T^2} u \dd[x]=0 \},
\]
endowed with the $L^2$-inner product. Since the condition $\int_{\T^2} u \dd[x]=0$ is preserved under the flow, $-\mC$ also generates a $C_0$-semigroup on $\wcH$, see e.g. \cite{Pazy,EnNa00}.
But related to its long-time behavior we have the following result:
%
\begin{proposition}\label{prop:Oseen:constantU:notHC}
Let $b\in\R^2$ be constant with $b_1\ne 0$.
Then, the operator~$\mC =b\cdot\nabla -\nu \partial_{x_2}^2$ is neither coercive nor hypocoercive in~$\wcH$.
\end{proposition}
%
\begin{proof}
First we introduce the maximal domain of $\mC$ in $\wcH$: $\cD(\mC)=\{u\in\wcH\ | \ \mC u\in (L^2(\T^2))^2\}$. Since $\mC$ has constant coefficients, $u\in \cD(\mC)$ implies $\mC u\in\wcH$. For future reference we also give their characterization in Fourier space (cf.\ \eqref{Fourier} and~\cite[\S II.5]{FMRT01}):
\begin{align}\label{D(C)-fourier}
u\in \wcH \quad &\Leftrightarrow \quad
\{\phi_k\} \in \ell^2(\Z^2;\C^2)\;\mbox{ with } \phi_0=0,\; k\cdot \phi_k=0\;
\mbox{for all}\ k\in\Z^2, \nonumber
\\
u\in \cD(\mC) \quad &\Leftrightarrow \quad
\{\phi_k\},\; \big\{\big(i(b\cdot k)+\nu k_2^2\big)\phi_k\big\} \in \ell^2(\Z^2;\C^2)\; \nonumber
\\
&\hspace{8em} \mbox{ with } \phi_0=0,\; k\cdot \phi_k=0\;
\mbox{for all}\ k\in\Z^2.
\end{align}
To investigate the coercivity of $\mC$, we compute for $u\in\cD(\mC)$:
\begin{equation}\label{C-coercive}
\ip{u}{\mC u}_{\wcH}
=\nu \int_{\T^2} |\partial_{x_2} u|^2 \dd[x]
=\nu \int_{\T^2} \big[ (\partial_{x_2} \uI)^2 +\tfrac12 (\partial_{x_1} \uI)^2 +\tfrac12 (\partial_{x_2} \uII)^2 \big] \dd[x],
\end{equation}
where we have used that $\diver u=0$.
The last integral does not involve $\partial_{x_1} \uII$.
Hence, $u:=[0,\sin (x_1)]^\top$ is a counterexample to coercivity on~$\wcH$ since $\ip{u}{\mC u}_{\wcH} =0$.
If~$\mC$ was hypocoercive, the trivial solution~$u=0$ would be asymptotically stable on~$\wcH$ due to the decay estimate~\eqref{exp-decay}.
But for any $v\in H^1_{per}(\T)$ with $\int_\T v(x_1)\dd[x_1]=0$, the (undamped) traveling wave
\begin{equation} \label{counterex:Oseen:constantU:notHC}
u(x,t) =\begin{bmatrix} 0 \\ v(x_1 -b_1 t) \end{bmatrix}
\end{equation}
solves~\eqref{model:Oseen:anisotropic} in~$\wcH$, in violation of~\eqref{exp-decay}.
\end{proof}
%
This lack of hypocoercivity in the model~\eqref{model:Oseen:anisotropic} with constant $b\in\R^2$ and $b_1\ne 0$ can be understood quite easily:
It includes drift and diffusion in the $x_2$-direction, but the uniform transport in the $x_1$-direction does not entail mixing between the different vertical layers.
If the flow field has only a vertical component~$\uII$, possibly different in each vertical layer (as shown in~\eqref{counterex:Oseen:constantU:notHC}), then the flow field gets transported in the $x_1$-direction and remains incompressible.
The lack of hypocoercivity can also be verified by considering \eqref{model:Oseen:anisotropic} as a constrained partial differential equation (PDAE), and bringing its modal representation into staircase form, see Example~\ref{ex:Oseen:constantU} in Appendix~\ref{app:DAEs:hypocoercivity}.
\begin{remark}
In contrast to the above case $b_1\ne0$, for the anisotropic Oseen-type model~\eqref{model:Oseen:anisotropic} with $b\in\R^2$ and $b_1 =0$, we have to define the Hilbert space $\wcH$ as in~\eqref{tildeH:b_1:0} endowed with the $L^2$-inner product.
Using that $u\in\wcH$ satisfies $u_2=u_2(x_2)$ and using~\eqref{C-coercive}, this shows that $\ip{u}{\mC u}_{\wcH} \geq \nu \|u\|_{\dot{H}^1}^2 /2$ for all $u\in\cD(\mC)=\{u\in\wcH\,|\, \partial_{x_2}^2u\in (L^2(\T^2))^2\}$.
Due to the Poincar\'e inequality $\|u\|_{\dot{H}^1}\geq \|u\|_{L^2}$ for all $u\in\wcH\cap H^1(\T^2)$ (trivially obtained from the Fourier representation \eqref{Fourier}; see also~\cite{Ev10,FMRT01}), $\mC$ is coercive on (the smaller) Hilbert space $\wcH$ here.
\end{remark}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Oseen-type equation with non-constant drift velocity}
\label{ssec:nonconst-drift}
As second example we consider the anisotropic equation~\eqref{model:Oseen:anisotropic} with a non-constant drift field $b(x) =[b_1(x_2),0]^\top$ such that
\begin{subequations}\label{model:Oseen:anisotropic_2}
\begin{align}
u_t
&=-b_1(x_2)\ \partial_{x_1} u -\nabla p +\nu \partial_{x_2}^2 u \,,
&& t>0 ,\ \text{on }\T^2 , \label{model:Oseen:anisotropic_2:1a} \\
%
0
&=-\diver u\ ,
&& t\geq 0\ . \label{model:Oseen:anisotropic_2:1b}
\end{align}
\end{subequations}
In order to simplify the detailed hypocoercivity analysis below,
we choose $b_1(x_2) =\sin (x_2)$; in this case only neighboring $k_2$-modes are coupled.
We conjecture that choosing another non-constant $x_2$-periodic drift field $b_1$ would lead to an analogous result, but it would need a more cumbersome analysis.
%{}[Beweis haben wir daf\"ur aber bisher nicht. Wichtig ist jedenfalls, dass die nicht abklingende $(k_1,0)$-Mode an eine abklingende $(k_1,k_2)$-Mode gekoppelt ist.]
\makeorange{In this direction, the existence of stationary solutions for the isotropic Oseen equation~\eqref{eq:Oseen} with space-dependent drift~$b=b(x)$ and $x\in\R^3$ has been studied, see~\cite{AmrCon11}.}
In contrast to \S\ref{ssec:const-drift}, this model does not preserve vertical layers of the flow field but rather mixes them, hence giving rise to hypocoercivity.
While the pressure could have been eliminated from the beginning in the model with constant drift term, this is not possible here.
Hence, \eqref{model:Oseen:anisotropic_2} with $b_1(x_2)$ has to be considered as a (true) constrained PDE.\\
\noindent
{\bf Analytic framework and well-posedness:}
For the analysis of this case we apply the divergence to the evolution equation~\eqref{model:Oseen:anisotropic_2:1a}, leading to the following condition on the pressure:
\begin{equation}\label{model:divOseen}
\begin{cases}
\Delta p =-\cos(x_2)\ \partial_{x_1} \uII , \quad \text{on } \T^2 , \\
\text{periodic boundary conditions for $p$} , \\
\int_{\T^2} p \dd[x] =0 ,
\end{cases}
\end{equation}
where the last condition was added to ensure uniqueness of~$p$.
For a given inhomogeneity $u\in (L^2(\T^2))^2$, Equation \eqref{model:divOseen} has a unique weak solution in the homogeneous Sobolev space $\dot H^1_{per}(\T^2)$, which we denote by $p =p[u]$. A standard elliptic estimate shows
\begin{equation}\label{p-u-estimate}
\|p\|_{\dot H^1_{per}(\T^2)} \le \|u_2\|_{L^2(\T^2)/ \R} \le \|u\|_{(L^2(\T^2)/ \R)^2}.
\end{equation}
The generator of the flow~\eqref{model:Oseen:anisotropic_2} with $b_1(x_2) =\sin (x_2)$ takes the form
\begin{equation}\label{Oseen:generator:sin}
-\mC u
=-\sin(x_2)\ \partial_{x_1} u -\nabla p[u] +\nu \partial_{x_2}^2 u .
\end{equation}
Note that $\diver u=0$ and $p[u]$ from \eqref{model:divOseen} imply that $\diver(\mC u)=0$. The assertion $\int \mC u\dd[x]=0$ follows trivially from $\int u \dd[x]=0$.
As before, we choose the Hilbert spaces
\[
\cH :=H_{per}(\diver 0, \T^2) ,\qquad
\wcH :=\{ u\in\cH\ |\ \int_{\T^2} u \dd[x]=0 \},
\]
and the maximal domain of $\mC$ is $\cD(\mC)=\{u\in\wcH\ | \ \sin(x_2)\partial_{x_1} u-\nu\partial^2_{x_2}u \in (L^2(\T^2))^2\}$, since the linear map $u\mapsto \nabla p$ from \eqref{model:divOseen} is bounded on $(L^2(\T^2))^2$.
With this framework we have the following result:
\begin{lemma}\label{lem:Oseen:sin}
Let $b_1(x_2)=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2}.
Then:
\begin{enumerate}[(1)]
\item \label{Oseen:Ux2:kernel}
$\ker\mC =\R^2$, i.e. the constant-in-$x$ flows; and the $L^2$-orthogonal complement of $\ker\mC$ is $\wcH$.
%
\item \label{Oseen:Ux2:notC}
The operator~$\mC$ is not coercive on~$\wcH$.
\end{enumerate}
\end{lemma}
%
\begin{proof}
\ref{Oseen:Ux2:kernel}
For $u\in\cD(\mC)$ we compute:
\[
0
=\int_{\T^2} u\cdot \mC u\dd[x]
=\int_{\T^2} \big(\tfrac12 \sin(x_2)\ \partial_{x_1} |u|^2 +\nu |\partial_{x_2} u|^2\big) \dd[x]
=\nu \int_{\T^2} |\partial_{x_2} u|^2 \dd[x] ,
\]
to obtain $u=u(x_1)$, and hence $0=\diver u=\partial_{x_1} \uI$.
So, using $\uI=\textrm{const}$ in~\eqref{Oseen:generator:sin} yields $p=p(x_2)$ and
\[
\sin(x_2)\ \partial_{x_1} \uII(x_1) +\partial_{x_2} p(x_2) =0 ,
\]
or equivalently for $x_2\not\in\{0,\pi\}$:
\[
\partial_{x_1} \uII(x_1)
=-\frac{\partial_{x_2} p(x_2)}{\sin (x_2)}
=\lambda
\qquad
\text{with some } \lambda\in\R .
\]
The periodicity of $\uII$ implies $\lambda=0$ and hence $\uII=\textrm{const}$, $p=0$.
\ref{Oseen:Ux2:notC}
As in \S\ref{ssec:const-drift}, $u:=[0, \sin (x_1)]^\top \in \cD(\mC)$ is a counterexample to coercivity on~$\wcH$, since $\ip{u}{\mC u}_{\wcH} =0$, by noting that $p[u]=\frac12 \cos(x_2)\cos(x_1)$ solves \eqref{model:divOseen}.
\end{proof}
Next we discuss the existence and uniqueness of a mild solution to the initial value problem \eqref{model:Oseen:anisotropic_2} when eliminating the pressure via \eqref{model:divOseen}. The resulting PDE-evolution problem for $u$ only, with $u(0)\in\wcH$, will be analyzed with semigroup theory, see e.g.\ \cite{Pazy}.
\begin{proposition}
On $\wcH$, the operator $-\mC$ from \eqref{Oseen:generator:sin} satisfies the following properties:
\begin{enumerate}
\item[(1)] It is dissipative, densely defined, and closed.
\item[(2)] It generates a $C_0$-semigroup of contractions on $\wcH$.
\end{enumerate}
\end{proposition}
\begin{proof}
(1) A similar computation as in \eqref{C-coercive} shows $\langle u,-\mC u\rangle_{\wcH} \le 0\;\mbox{ for all}\, u \in\cD(\mC)$, i.e.\ the dissipativity of~$-\mC$.
For the density of $\cD(\mC)$ in $\wcH$ we first consider
\[
\widetilde\cD := \{u\in \wcH\, | \,\partial_{x_1}u,\,\partial_{x_2}^2u \in (L^2(\T^2))^2\} \subset\cD(\mC).
\]
Truncating the Fourier representation of any $u\in\widetilde\cD$ (in analogy to \eqref{D(C)-fourier}) one easily finds that $\widetilde\cD$ is $L^2$-dense in $\wcH$, and hence also $\cD(\mC)$ in $\wcH$.
Using Theorem 1.4.5(c) of \cite{Pazy} we find that $-\mC$ is closable. Moreover, $-\overline\mC=-\mC$, since $\cD(\mC)$ was chosen as the maximal domain.
(2) For the second statement we easily compute
$$
-\mC^* u
=\sin(x_2)\ \partial_{x_1} u -\nabla p[u] +\nu \partial_{x_2}^2 u,
$$
which is defined at least on $\widetilde\cD$ and also dissipative. Hence, Corollary 1.4.4 (to the Lumer-Phillips Theorem) from \cite{Pazy} implies that $-\mC$ is the infinitesimal generator of a $C_0$-semigroup of contractions on~$\wcH$.
\end{proof}
Next we shall illustrate the short-time behavior of $\|e^{-\mC t}\|$ (in the spirit of~\eqref{short-t-decay}) by analyzing the Taylor expansion of the norm for a single trajectory at $t=0$. Using the same initial condition as in the proof of Lemma \ref{lem:Oseen:sin}, i.e.\ $u(0)=[0,\sin(x_1)]^\top$ which is a linear combination of the modes $\binom{\pm1}{0}$~, we compute:
\begin{align*}
\mC u(0)&=-\frac12 \begin{bmatrix} \sin(x_1)\cos(x_2) \\ -\cos(x_1)\sin(x_2) \end{bmatrix}
\in \cD(\mC),\\
\mC^2 u(0)&=-\frac12 \begin{bmatrix} \frac35 \cos(x_1)\sin(2x_2)+\nu\sin(x_1)\cos(x_2) \\ \sin(x_1)\big(\frac12-\frac3{10}\cos(2x_2)\big)-\nu\cos(x_1)\sin(x_2) \end{bmatrix}
\in \cD(\mC),\\
\mC^3 u(0)&= \begin{bmatrix} \sin(x_1) \big(\frac{19}{80}\cos(x_2)-\frac{63}{400}\cos(3x_2)\big) \\ \cos(x_1)\big(\frac{21}{400}\sin(3x_2)-\frac{19}{80}\sin(x_2) \big) \end{bmatrix}\\
&\qquad +\nu \begin{bmatrix} -\frac32 \cos(x_1)\sin(2x_2) \\ \sin(x_1)\big(-\frac14+\frac34\cos(2x_2)\big) \end{bmatrix}
+\frac{\nu^2}2 \begin{bmatrix} -\sin(x_1)\cos(x_2) \\ \cos(x_1)\sin(x_2) \end{bmatrix},
\end{align*}
where we used the following solutions to the corresponding Poisson equation~\eqref{model:divOseen}:
\begin{align*}
p[\mC u(0)]&=-\frac1{20} \sin(x_1)\sin(2x_2),\\
p[\mC^2 u(0)]&=-\frac1{80} \cos(x_1)\big(7\cos(x_2)-\frac35\cos(3x_2)\big)-\frac\nu{20} \sin(x_1)\sin(2x_2).
\end{align*}
Hence we obtain
\begin{align*}
\|u(0)\|^2_{\wcH} &=2\pi^2,\\
\ddt \|u(t)\|^2_{\wcH}\,\big|_{t=0} &= -2\langle u(0),\mC u(0)\rangle_{\wcH}=0,\\
\ddtddt \|u(t)\|^2_{\wcH}\,\big|_{t=0} &= 2\langle \mC u(0),\mC u(0)\rangle_{\wcH} +2\langle u(0),\mC^2 u(0)\rangle_{\wcH}=0,\\
\ddtddtddt \|u(t)\|^2_{\wcH}\,\big|_{t=0} &= -6\langle \mC u(0),\mC^2 u(0)\rangle_{\wcH} -2\langle u(0),\mC^3 u(0)\rangle_{\wcH}=-2\nu\pi^2<0.
\end{align*}
This implies the following lower bound on the propagator norm:
\begin{equation}\label{prop-norm-est}
\|e^{-\mC t}\| := \sup_{\|v\|_{\wcH}=1} \|e^{-\mC t}v\|_{\wcH}\ge1 -\frac1{12}\nu\,t^3+\mathcal O(t^4) \quad \mbox{for } t\to 0^+.
\end{equation}
The first non-constant term in the Taylor expansion of $\|u(t)\|^2_{\wcH}$ has exponent~$a=3$.
In Proposition \ref{prop:Oseen:y2} below, we show that the modal expansion of~\eqref{model:Oseen:anisotropic_2} with~$b_1(x_2)=\sin (x_2)$,
has hypocoercivity index $m_{HC}=1$.
Although the analog of the result \eqref{short-t-decay} has not yet been established for unbounded generators~$-\mC$, we note that the exponent $a=3$ in the Taylor series of~$\|u(t)\|^2_{\wcH}$ satisfies again $a=2\mHC+1$.\\
%We illustrated this correspondence for the Oseen-type evolution of this subsection, although the analog of the result \eqref{short-t-decay} has not yet been established for unbounded generators~$-\mC$.\\
\noindent
{\bf Modal decomposition:}
To analyze the hypocoercivity of~\eqref{model:Oseen:anisotropic_2} with $b_1(x_2) =\sin (x_2)$, we perform a modal decomposition of this model as in~\S\ref{ssec:Stokes}.
But due to the non-constant coefficient $b_1(x_2)$, these modes decouple now only w.r.t.~$k_1$, but not w.r.t.~$k_2$.
As the analog of~\eqref{eq:Stokes:Fourier} we obtain for $k\in\Z^2$ the DAE system
%
\begin{equation}\label{eq:Oseen:Fourier}
\begin{cases}
\ddt \phi_k
&= \tfrac{k_1}{2}(\phi_{k+e_2} -\phi_{k-e_2}) -i k p_k -\nu k_2^2 \phi_k \ ,
\quad
t>0 \ , \\
%
0& =-i k\cdot \phi_k \ ,
\end{cases}
\end{equation}
with $e_2:=[0,\, 1]^\top$.
All modes with $k_1=0$ are fully decoupled, showing that $\phi_0(t)=\const$ and one has exponential decay with rate at least~$\mu=\nu$ for all modes with $k=(0,k_2)\ne 0$.
More precisely, proceeding as for \eqref{eq:Stokes:Fourier} with $b=0$ one obtains: $\phi_{(0,k_2),2}(t)=0,\,p_{(0,k_2)}(t)=0$ for $k_2\ne0$, as well as
\begin{equation}\label{k10-decay}
|\phi_{(0,k_2),1}(t)| \le |\phi_{(0,k_2),1}(0)|\,e^{-t\nu k_2^2}\,,
\quad k_2\ne0\,.
\end{equation}
But for $k_1\ne 0$ the modes are only semi-decoupled, i.e., for each fixed $k_1\ne 0$, \eqref{eq:Oseen:Fourier} forms an infinite dimensional DAE system with $k_2\in\Z$.
Using the Fourier decomposition it is
%\footnote{Due to the Fourier decomposition it is
notationally simpler to consider $\ell^2(\Z)$ rather than $\ell^2(\N)$.
Then the doubly infinite complex vector
\begin{align*}
w_{k_1}
&:=
\big\{[\phi_{(k_1,k_2),1}, \phi_{(k_1,k_2),2}, p_{(k_1,k_2)}]^\top \big\}_{k_2\in\Z} \\
& = [...\,|\, \phi_{(k_1,-1),1}, \phi_{(k_1,-1),2}, p_{(k_1,-1)}\,|\,\phi_{(k_10),1}, \phi_{(k_1,0),2}, p_{(k_1,0)}\,|\,\phi_{(k_1,1),1}, \phi_{(k_1,1),2}, p_{(k_1,1)}\,|\,...]^\top \\
&\in\ell^2(\Z;\C^3)
\end{align*}
%
satisfies the DAE
%
\begin{equation} \label{DAE:Oseen}
\mE \dot{w}_{k_1}(t)
=
(\mJ_{k_1} -\mR) w_{k_1}(t) ,
\qquad
t\geq 0 ,
\end{equation}
with doubly infinite diagonal matrices $\mE, \mR\in\cL(\ell^2(\Z;\C^3))$ given as
%
\begin{align}
\mE
&:= \diag(\ \diag([1,1,0]);\, k_2\in\Z) \nonumber
\\
&\qquad = \diag(\ldots\,|\, 1,1,0\,|\, 1,1,0\,|\, 1,1,0\,|\, \ldots)\ , \label{DAE:Oseen:E}
\\
\mR
&:= \nu \diag(\ \diag([k_2^2,k_2^2,0]);\, k_2\in\Z) \nonumber
\\
&\qquad = \nu \diag(\ldots\,|\, 4,4,0\,|\, 1,1,0\,|\, 0,0,0\,|\, 1,1,0\,|\, 4,4,0\,|\, \ldots)\ ,
\label{DAE:Oseen:R}
\end{align}
%
and the doubly infinite block-tridiagonal matrix $\mJ_{k_1}\in\cL(\ell^2(\Z;\C^3))$ with $3\times3$-blocks given as
%In the 7-diagonal and block tri-diagonal matrix
\begin{equation}\label{DAE:Oseen:J}
\begin{split}
&\mJ_{k_1}
:=
\\
&\left[ \begin{array}{ccc|ccc|ccc|ccc|ccc}
& \ddots & & & & & & & & & & & & & \\
\hline
-\tfrac{k_1}2 & & & 0 & 0 & -i k_1 & \tfrac{k_1}2 & & & & & & & & \\
& -\tfrac{k_1}2 & & 0 & 0 & i & & \tfrac{k_1}2 & & & & & & & \\
& & 0 & -i k_1 & i & 0 & & & 0 & & & & & & \\
\hline
& & & -\tfrac{k_1}2 & & & 0 & 0 & -i k_1 & \tfrac{k_1}2 & & & & & \\
& & & & -\tfrac{k_1}2 & & 0 & 0 & 0 & & \tfrac{k_1}2 & & & & \\
& & & & & 0 & -i k_1 & 0 & 0 & & & 0 & & & \\
\hline
& & & & & & -\tfrac{k_1}2 & & & 0 & 0 & -i k_1 & \tfrac{k_1}2 & & \\
& & & & & & & -\tfrac{k_1}2 & & 0 & 0 & -i & & \tfrac{k_1}2 & \\
& & & & & & & & 0 & -i k_1 & -i & 0 & & & 0 \\
\hline
& & & & & & & & & & & & & \ddots &
\end{array}\right]
{\scriptsize
\begin{array}{l}
k_2=-1 \\[3em] k_2=0 \\[3em] k_2=1
\end{array} } \\
& {\scriptstyle\hspace{2em} \widetilde k_2=-2 \hspace{4em} \widetilde k_2=-1 \hspace{5em} \widetilde k_2=0 \hspace{5em} \widetilde k_2=1 \hspace{4em} \widetilde k_2=2 }
%
\end{split},
\end{equation}
%
where missing elements % in~\eqref{DAE:Oseen:J}
are understood to be zero.
Here we label the $3\times 3$ submatrices $\mJ_{k_1}(k_2,\widetilde k_2)$ by the corresponding modal numbers of $k_2$.
In particular, the diagonal blocks have the general form
%
\begin{equation}\label{DAE:Oseen:Jkk}
\mJ_{k_1} (k_2,k_2)
=
\begin{bmatrix}
0 & 0 & -i k_1 \\
0 & 0 & -i k_2 \\
-i k_1 & -i k_2 & 0
\end{bmatrix} .
\end{equation}
%
\medskip
\noindent
{\bf Staircase transformation:}
To bring~\eqref{DAE:Oseen} into staircase form
%.
%But for the infinite matrices %in~\eqref{DAE:Oseen}
we have to modify Step~1 in~\cite[Algorithm 5]{AAM21}, see e.g.~Appendix~\ref{app:DAEs:hypocoercivity}, and hence also the presentation of the result in Lemma~\ref{lem:tSF}:
On the one hand the matrix~$\mE$ is already diagonal, but on the other hand it has both an infinite dimensional kernel and range.
Hence, it would be impractical to separate the corresponding eigenvalues.
Instead we shall leave them interlaced as in~\eqref{DAE:Oseen:E}.
This has the following (purely notational) consequence on partitioning the matrices $\mJ_{k_1}$, $k_1\ne 0$.
With the notation from~\cite[Algorithm 5: Step 1.2]{AAM21}, see also Appendix~\ref{app:DAEs:hypocoercivity}, we partition every $3\times 3$ submatrix of $\mJ$ into blocks of size~2 and~1.
In \cite[Algorithm 5: Step 1.2]{AAM21} the finite dimensional matrix $\mJ$ is partitioned as $\begin{bmatrix} \widetilde\mJ_{1,1} & -\widetilde\mJ_{2,1}^H \\ \widetilde\mJ_{2,1} & \widetilde\mJ_{2,2} \end{bmatrix}$.
%where the four submatrices are (as usual) non-overlapping.
Just for notational simplicity we suppressed here and in the sequel the index $k_1$ and write $\mJ:=\mJ_{k_1}$. Next, we decompose the space of sequences $\ell^2(\Z;\C^3)$ as
\begin{equation}
\ell^2(\Z;\C^3) =\ell^2(\Z;\spn\{e_1,e_2\}) \oplus \ell^2(\Z;\spn\{e_3\}) ,
\end{equation}
where $e_1=[1,0,0]^\top$, $e_2=[0,1,0]^\top$, $e_3=[0,0,1]^\top$; and use that these subspaces are isomorphic to
\begin{equation}
\ell^2(\Z;\spn\{e_1,e_2\}) \simeq \ell^2(\Z;\C^2)\ , \qquad
\ell^2(\Z;\spn\{e_3\}) \simeq \ell^2(\Z;\C)\ .
\end{equation}
In analogy to Algorithm~\ref{algorithm:staircase:E_J-R}:~Step~1.2, we decompose the operator~$\mJ\in\cL(\ell^2(\Z;\C^3))$ into the linear operators
\begin{align*}
\tJ_{1,1}\in \cL\big(\ell^2(\Z;\C^2)\big) , \quad
\tJ_{2,1}\in \cL\big(\ell^2(\Z;\C^2);\ell^2(\Z;\C)\big) , \quad
\tJ_{2,2}\equiv 0\in \cL\big(\ell^2(\Z;\C)\big) ,
\end{align*}
represented by the doubly infinite block-tridiagonal matrices with $2\times2$-blocks
%
\begin{equation}
\begin{split}\label{DAE:Oseen:J11}
\tJ_{1,1}
&:= \frac{k_1}2
\left[ \begin{array}{c|cc|cc|cc|cc|cc|c}
\ddots & & \ddots & & \ddots & & & & & & & \\
\hline
& -1 & & 0 & 0 & 1 & & & & & & \\
& & -1 & 0 & 0 & & 1 & & & & & \\
\hline
& & & -1 & & 0 & 0 & 1 & & & & \\
& & & & -1 & 0 & 0 & & 1 & & & \\
\hline
& & & & & -1 & & 0 & 0 & 1 & & \\
& & & & & & -1 & 0 & 0 & & 1 & \\
\hline
& & & & & & & \ddots & & \ddots & & \ddots
\end{array}\right]
{\scriptsize
\begin{array}{l}
\\[0.5em] k_2=-1 \\[2em] k_2=0 \\[2em] k_2=1 \\[1em]
\end{array} } \ , \\
& {\scriptstyle\hspace{5.8em} \widetilde k_2=-2 \hspace{0.9em} \widetilde k_2=-1 \hspace{1em} \widetilde k_2=0 \hspace{1.3em} \widetilde k_2=1 \hspace{0.5em} \widetilde k_2=2 }
\end{split}
\end{equation}
%
and the doubly infinite block-tridiagonal matrices with $1\times2$-blocks
%
\begin{equation}
\begin{split}\label{DAE:Oseen:J21}
\tJ_{2,1}
:= \diag( [-i k_1\ -i k_2];\ k_2\in\Z)
&=\! \left[ \begin{array}{c|cc|cc|cc|c}
\ddots & & & & & & & \\
\hline
& -i k_1 & i & & & & & \\
\hline
& & & -i k_1 & 0 & & & \\
\hline
& & & & & -i k_1 & -i & \\
\hline
& & & & & & & \ddots
\end{array}\right]
{\scriptsize
\begin{array}{l}
\\[-0.4em] k_2=-1 \\[0.6em] k_2=0 \\[0.6em] k_2=1 \\[0.5em]
\end{array} } . \\
& {\scriptstyle\hspace{4em} \widetilde k_2=-1 \hspace{1em} \widetilde k_2=0 \hspace{2em} \widetilde k_2=1 \hspace{2.5em} }
\end{split}
\end{equation}
\begin{comment}
Here, however, these four submatrices will be overlapping, but have ``holes'' that are then mutually filled by elements of one of the other submatrices. More precisely, in the following representations of $\tJ_{1,1}$ and $\tJ_{2,1}$ ``*'' denotes such ``holes'' (but not zeros) in the matrices.
%
\newcommand{\mDone}{\begin{array}{cc} \tfrac{-k_1}2 & \\ & \tfrac{-k_1}2 \end{array}}
\newcommand{\mDtwo}{\begin{array}{cc} \tfrac{k_1}2 & \\ & \tfrac{k_1}2 \end{array}}
\newcommand{\mDzero}{\begin{array}{cc} 0 & 0 \\ 0 & 0 \end{array}}
%
\begin{equation}\label{DAE:Oseen:J11}
\begin{split}
&\tJ_{1,1}
:=
\\
&\left[ \begin{array}{c|c||c|c||c|c||c|c||c|c}
& \ddots & & & & & & & & \\
\hline \hline
\mDone & * & \mDzero & * & \mDtwo & * & & * & & * \\
\hline
* & * & * & * & * & * & * & * & * & * \\
\hline \hline
& * & \mDone & * & \mDzero & * & \mDtwo & * & & * \\
\hline
* & * & * & * & * & * & * & * & * & * \\
\hline \hline
& * & & * & \mDone & * & \mDzero & * & \mDtwo & * \\
\hline
* & * & * & * & * & * & * & * & * & * \\
\hline \hline
& & & & & & & & \ddots & \\
\end{array}\right]
\begin{array}{l}
k_2=-1 \\[3em] k_2=0 \\[3em] k_2=1
\end{array} , \\
& \hspace{2em} \widetilde k_2=-2 \hspace{4.5em} \widetilde k_2=-1 \hspace{4em} \widetilde k_2=0 \hspace{4em} \widetilde k_2=1 \hspace{4em} \widetilde k_2=2
\end{split}
\end{equation}
and
%
\begin{equation}\label{DAE:Oseen:J21}
\begin{split}
\tJ_{2,1}
:=
&\left[ \begin{array}{c|c||c|c||c|c||c|c||c|c}
& \ddots & & & & & & & & \\
\hline \hline
* & * & * & * & * & * & * & * & * & * \\
\hline
& * & -i k_1 \quad i & * & & * & & * & & * \\
\hline \hline
* & * & * & * & * & * & * & * & * & * \\
\hline
& * & & * & -i k_1 \quad 0 & * & & * & & * \\
\hline \hline
* & * & * & * & * & * & * & * & * & * \\
\hline
& * & & * & & * & -i k_1 \quad -i & * & & * \\
\hline \hline
& & & & & & & & \ddots & \\
\end{array}\right]
\begin{array}{l}
\\[-1.5em] k_2=-1 \\[1.5em] k_2=0 \\[1.5em] k_2=1
\end{array} . \\
& \widetilde k_2=-2 \hspace{2em} \widetilde k_2=-1 \hspace{2em} \widetilde k_2=0 \hspace{3.5em} \widetilde k_2=1 \hspace{2.5em} \widetilde k_2=2
\end{split}
\end{equation}
For each fixed index pair $(k_2,\,\widetilde k_2)$, note that $\tJ_{2,1}$ fills one of the three holes in the $(k_2,\,\widetilde k_2)$-subblock of $\tJ_{1,1}$. Moreover, $\tJ_{2,1}^H$ fills the second hole, and $\tJ_{2,2}=0$ (see \eqref{DAE:Oseen:J}).
\end{comment}
When partitioning $\tR$ like $\tJ$, we find from \eqref{DAE:Oseen:R} that $\tR_{2,2}=0$. Hence Step~2 of Algorithm~\ref{algorithm:staircase:E_J-R} is trivial.
Since the subblocks of $\tJ_{2,1}$ are identical to the $(2,1)$-subblock of~\eqref{DAE:Stokes:J}, with $k_1\ne 0$ fixed and $k_2\in\Z$, their singular value decompositions (SVDs) also coincide.
Thus, the unitary transformation matrix $\mP_{k_1,k_2}$ (pertaining to each $3\times 3$ submatrix~\eqref{DAE:Oseen:Jkk}) is given by~\eqref{ex:Stokes:Pk}.
Hence, the (global) unitary transformation matrix for the DAE systems~\eqref{DAE:Oseen} with $k_1\in\Zo$ is given by the block diagonal matrix
\begin{equation}\label{model:Oseen:Pk1}
\mP_{k_1}
=
\diag\Bigg(
\tfrac1{|k|}
\begin{bmatrix}
k_1 & k_2 & 0 \\
-k_2 & k_1 & 0 \\
0 & 0 & i|k|
\end{bmatrix} ;\
k_2\in\Z \Bigg) \ ,
\end{equation}
which represents a bounded unitary operator on $\ell^2(\Z;\C^3)$. Also the leading $2\times2$-subblocks~\eqref{ex:Stokes:Pk} form a bounded unitary operator on $\ell^2(\Z;\C^2)$. For each fixed $k_1\ne0$ in~\eqref{DAE:Oseen}, this is an infinite dimensional analog of the staircase transformation in Lemma \ref{lem:tSF}. Hence it translates~\eqref{DAE:Oseen}, via $y_{k_1} :=\mP_{k_1} w_{k_1} \in\ell^2(\Z;\C^3)$, into the staircase form
\begin{equation}\label{DAE:Oseen:staircase}
\widecheck\mE \dot{y}_{k_1}(t)
=
(\widecheck{\mJ}_{k_1} -\widecheck\mR) y_{k_1}(t) ,
\qquad
t\geq 0 .
\end{equation}
Here, the matrices are given by $\widecheck\mE =\mP_{k_1} \mE\mP_{k_1}^H =\mE$, $\widecheck{\mR} =\mP_{k_1} \mR\mP_{k_1}^H =\mR$, and $\widecheck{\mJ}_{k_1}=\mP_{k_1}\mJ_{k_1}\mP_{k_1}^H$ is a block tridiagonal matrix with the blocks
%
\begin{align*}
\widecheck{\mJ}_{k_1}(k_2,k_2)
&=
\begin{bmatrix}
0 & 0 & -|k| \\
0 & 0 & 0 \\
|k| & 0 & 0
\end{bmatrix} ,
\\
\widecheck{\mJ}_{k_1}(k_2,k_2\pm 1)
&=
\pm \frac{k_1}{2 |k|\ |k\pm e_2|}
\begin{bmatrix}
|k|^2 \pm k_2 & \mp k_1 & 0 \\
\pm k_1 & |k|^2\pm k_2 & 0 \\
0 & 0 & 0
\end{bmatrix} .
\end{align*}
In analogy to Definition \ref{def:DAE:mHC}, the non-trivial dynamics of~\eqref{DAE:Oseen:staircase} is given by \eqref{underlyingODE:y} as the infinite ODE
\begin{equation}\label{ODE:Oseen:y2}
\mE_{2,2} \big(\dot{y}_{k_1}\big)_2
=
\big(\widehat{\mJ}_{k_1} -\widehat\mR\big)_{2,2} \big(y_{k_1}\big)_2 .
\end{equation}
Here, $\big(y_{k_1}\big)_2=\big\{\frac{1}{|k|}\big(-k_2\ukI+k_1\ukII\big)\big\}_{k_2\in\Z} \in\ell^2(\Z;\C)$ is the projection of $y_{k_1} \in\ell^2(\Z;\C^3)$, where only the second element of each 3-block of $y_{k_1}$ is kept.
This second component is closely related to the modal representation of the vorticity $\rot u$, see the proof of Theorem \ref{th-u-decay} and Remark \ref{rem:isometry-rotu} below for details.
Analogously, the three matrices in \eqref{ODE:Oseen:y2} are obtained by keeping only the $(2,2)$-element of each submatrix, and they act on $\ell^2(\Z;\C)$. Hence, $\mE_{2,2}=\mI$.
The infinite matrix
\begin{equation}\label{DAE:Oseen:staircase:R}
\widehat{\mR}_{2,2}
=\nu \diag(k_2^2;\ k_2\in\Z)
=\nu \diag(\ldots,\ 9,4,1,0,1,4,9,\ \ldots)
\end{equation}
is positive semi-definite but not coercive.
Moreover it represents an unbounded self-adjoint operator on $\ell^2(\Z;\C)$ with dense domain
\begin{equation} \label{domain:R_22}
\mathcal D(\widehat{\mR}_{2,2})=\{z\in \ell^2(\Z;\C)\, |\, \sum_{j\in\Z} j^4\,|z(j)|^2<\infty\}\ .
\end{equation}
Furthermore,
\begin{multline} \label{VhRzz} %\sqrt{\hRzz}
(\widehat{\mR}_{2,2})^{1/2}=\sqrt\nu \diag(|k_2|;\ k_2\in\Z)\ ,
\\
\cD\big((\widehat{\mR}_{2,2})^{1/2}\big) =
\{z\in \ell^2(\Z;\C)\, |\, \sum_{j\in\Z} j^2\,|z(j)|^2<\infty\}\ .
\end{multline}
The operator $(\widehat{\mJ}_{k_1})_{2,2} \in\cL(\ell^2(\Z;\C))$ can be represented as the skew-adjoint tridiagonal matrix
\begin{equation}\label{DAE:Oseen:staircase:J}
\begin{split}
&\big(\widehat{\mJ}_{k_1}\big)_{2,2}
\\
&=
{\scriptsize
\tfrac{k_1}2 \!
\begin{bmatrix}
\ddots & \ddots & \ddots & & & &
\\
& -\frac{k_1^2 +2}{\sqrt{k_1^2+1}\sqrt{k_1^2+4}} & 0 & \frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|} & & &
\\
& & -\frac{k_1^2}{|k_1| \sqrt{k_1^2+1}} & 0 & \frac{k_1^2}{|k_1|\ \sqrt{k_1^2+1}} & &
\\
& & & -\frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|} & 0 & \frac{k_1^2+2}{\sqrt{k_1^2+1}\ \sqrt{k_1^2+4}} &
\\
& & & & \ddots & \ddots & \ddots
\end{bmatrix}
\begin{array}{l}
\\ k_2=-1 \\[1em] k_2=0 \\[1em] k_2=1 \\[1em]
\end{array} }
\\
& {\scriptstyle\hspace{7em} \widetilde k_2=-2 \hspace{3em} \widetilde k_2=-1 \hspace{3.25em} \widetilde k_2=0 \hspace{3.5em} \widetilde k_2=1 \hspace{3.5em} \widetilde k_2=2}
\end{split}\! .
\end{equation}
Its off-diagonal terms have the general form
\[
\big(\widehat{\mJ}_{k_1}\big)_{2,2} (k_2,k_2\pm 1)
=
\pm \frac{k_1}2 \frac{|k|^2 \pm k_2}{|k|\ |k\pm e_2|} ,
\]
hence, for each fixed $k_1\ne0$, $\big(\widehat{\mJ}_{k_1}\big)_{2,2}$ is a bounded operator on $\ell^2(\Z;\C)$.
Moreover, one easily sees that
\[
\big(\widehat{\mJ}_{k_1}\big)_{2,2} \big(\mathcal D(\widehat{\mR}_{2,2})\big)\subset \mathcal D(\widehat{\mR}_{2,2}) \ ,\quad
\big(\widehat{\mJ}_{k_1}\big)_{2,2} \big(\cD\big((\widehat{\mR}_{2,2})^{1/2}\big)\big)\subset \cD\big((\widehat{\mR}_{2,2})^{1/2}\big) \ .
\]
Thus, the operators $\big(\widehat{\mJ}_{k_1}\big)_{2,2}$ and $\widehat{\mR}_{2,2}$ satisfy the assumptions~\ref{prop:R}, \ref{prop:R1/2} in the Definition \ref{def:unb-op:HC-index} of the HC-index.
Hence, for fixed $k_1\ne 0$, $(\widehat{\mJ}_{k_1}-\widehat{\mR})_{2,2}$ in~\eqref{ODE:Oseen:y2} generates an analytic semigroup of contractions on~$\ell^2(\Z;\C)$. Moreover,
either of the three conditions from Lemma \ref{lem:Op-Equivalence} can be used to determine the HC-index (see Proposition \ref{prop:Oseen:y2} below).\\
%\makeblue{
%To show that, for fixed $k_1\ne 0$, $(\widehat{\mJ}_{k_1}-\widehat{\mR})_{2,2}$ in~\eqref{ODE:Oseen:y2} generates an analytic semigroup on~$\ell^2(\Z;\C)$ we use a result in perturbation theory of semigroups, see~\cite[Theorem III.2.10]{EnNa00}:
%\begin{remark}[analytic semigroup]
%The operator $-\widehat{\mR}_{2,2}$ with domain~$\mathcal D(\widehat{\mR}_{2,2})$ on the Hilbert space~$\ell^2(\Z;\C)$ is self-adjoint and dissipative.
%Due to~\cite[Corollary II.4.7]{EnNa00}, the operator $-\widehat{\mR}_{2,2}$ generates a bounded analytic semigroup on~$\ell^2(\Z;\C)$.
%
%Moreover, for fixed $k_1\ne 0$, the operator~$(\widehat{\mJ}_{k_1})_{2,2}$ is bounded on~$\ell^2(\Z;\C)$, hence, it is $\widehat{\mR}_{2,2}$-bounded with $\widehat{\mR}_{2,2}$-bound $a_0=0$, see~\cite[Definition III.2.1]{EnNa00}.
%
Altogether, due to~\cite[Theorem III.2.10]{EnNa00}, the operator $(\widehat{\mJ}_{k_1}-\widehat{\mR})_{2,2}$ with domain~$\cD(\widehat{\mR}_{2,2})$ generates an analytic semigroup on~$\ell^2(\Z;\C)$.
%\end{remark}
%}%end-makeblue
\medskip
\noindent
{\bf Exponential decay of the dynamical part:}
Using DAE concepts we separated the dynamical part from the algebraic constraint. The following proposition studies the hypocoercivity of the dynamical part. More precisely,
with the above setup we can now establish the exponential decay of $\big(y_{k_1}\big)_2\,$, i.e., the modal decomposition of the vorticity $\rot u$ (in fact in $\dot H^{-1}_{per}(\T^2)$, see Remark \ref{rem:isometry-rotu} for details):
%
\begin{proposition}\label{prop:Oseen:y2}
Let $b_1=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2}.
Then, for each $k_1\in\Zo$, the modal dynamics~\eqref{ODE:Oseen:y2} is hypocoercive in the sense of~\eqref{exp-decay} in~$\ell^2(\Z;\C)$.
Moreover, its HC-index $\mHC=1$.
\end{proposition}
%
We remark that HC-index $\mHC=1$ was already illustrated on one example in the estimate \eqref{prop-norm-est}.
In the following, we will use the canonical unit vectors $(e_j)_{j\in\Z}$ as orthonormal basis for
% the infinite dimensional Hilbert space
~$\ell^2(\Z)$% of square-summable sequences
, which are defined for $j,k\in\Z$ as
\begin{equation} \label{ONB:l2}
e_j^k
=
\begin{cases}
1 & \text{for } k=j, \\
0 & \text{for } k\ne j.
\end{cases}
\end{equation}
%
\begin{proof}
To determine the HC-index of~$\big(\hJ_{k_1} -\hR\big)_{2,2}$ in~\eqref{ODE:Oseen:y2}, we use Condition \ref{B:G*_surjective} in Lemma \ref{lem:Op-Equivalence}:
The HC-index of $\big(\hJ_{k_1} -\hR\big)_{2,2}$ cannot be zero, since the selfadjoint part $\hRzz$ has a nontrivial kernel $\ker(\hRzz) =\spn\{e_0\}$.
We prove that its HC-index is $1$ by verifying Condition~\ref{B:G*_surjective} with $m=1$ in Lemma~\ref{lem:Op-Equivalence}:
For any $h\in\cH$, we determine $v,w\in\cD\big(\VhRzz\big)$ such that
\begin{equation} \label{h=Rv+JRw}
h =\VhRzz v +\hJkzz\VhRzz w \ .
\end{equation}
To do this, we compute
\begin{align*}
\big(\widehat{\mJ}_{k_1}\big)_{2,2}\
\big(\widehat{\mR}_{2,2}\big)^{1/2}
=
-\sqrt{\nu} \frac{k_1}2
&\begin{bmatrix}
\ddots & \ddots & -\frac{k_1^2+2}{|k|\ |k-e_2|} & & & &
\\
& \ddots & 0 & 0 & & &
\\
& 0 & \frac{k_1^2}{|k|\ |k+e_2|} & 0 & -\frac{k_1^2}{|k|\ |k-e_2|} & 0 &
\\
& & & 0 & 0 & \ddots &
\\
& & & & \frac{k_1^2+2}{|k|\ |k+e_2|} & \ddots & \ddots
\end{bmatrix}
\begin{array}{l}
\\[-0.4em] {\scriptstyle k_2=-1} \\[0.4em] {\scriptstyle k_2=0} \\[0.4em] {\scriptstyle k_2=1} \\[1em]
\end{array} ,
\\
& {\scriptstyle\hspace{4.6em} %\widetilde{k}_2=-2 \hspace{1.5em}
\widetilde{k}_2=-1 \hspace{1em} \widetilde{k}_2=0 \hspace{1.3em} \widetilde{k}_2=1 }
\end{align*}
see also~\eqref{hJzz} below.
Since $\big(\VhRzz v\big)_0 =0$, we observe that
\begin{align*}
h_0
&
%=-\sqrt{\nu} \frac{k_1}2 \Big(\frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|} w_{-1} -\frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|} w_1\Big)
=-\sqrt{\nu} \frac{k_1}2 \frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|} (w_{-1} -w_1) \ ,
\intertext{and choosing $w_{-1}=2w_1$ yields,}
h_0
&=\underbracket{-\sqrt{\nu} \frac{k_1}2 \frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|}}_{=:\alpha} w_1 \ ,
\end{align*}
such that $w_1 =\alpha^{-1} h_0$.
In fact, defining $w$ as
\[
w_j :=
\begin{cases}
\alpha^{-1} h_0 &\text{if $j=1$}\ , \\
2\alpha^{-1} h_0 &\text{if $j=-1$}\ , \\
0 &\text{else}\ ,
\end{cases}
\]
yields $w\in\cD\big(\VhRzz\big)$.
Then $h-\hJkzz\VhRzz w\in\range\VhRzz$ such that
\[
v
:= %\Big(\VhRzz\Big|_{\range\VhRzz}\Big)^{-1}
\Big(\VhRzz\Big)^+
\big(h-\hJkzz\VhRzz w\big)
\in\cD\big(\VhRzz\big) \ ,
\]
where $\Big(\VhRzz\Big)^+=\frac1{\sqrt\nu} \diag\big(...,\frac13,\frac12,1,0,1,\frac12,\frac13,...\big)$ denotes the Moore-Penrose inverse.
Thus, for all $h\in\cH$, there exist $v,w\in\cD\big(\VhRzz\big)$ such that~\eqref{h=Rv+JRw} holds. This proves Condition~\ref{B:G*_surjective} with $m=1$ in Lemma~\ref{lem:Op-Equivalence} and finishes the proof of the statement about the HC-index.
We remark that determining the hypocoercivity index of
$-{\mA}_{k_1}:=(\hR-\hJ_{k_1})_{2,2}$ directly via the coercivity condition in Definition~\ref{def:unb-op:HC-index} would be much more tedious, since both the matrix $-{\mA}_{k_1}$ and the corresponding left-hand-side of \eqref{Op:J-R:unbdd} (with $m=1$) are 5-diagonal, see Appendix~\ref{app:QuantitativeEstimate}.
\bigskip
In order to derive a hypocoercivity estimate~\eqref{exp-decay} for the evolution equation \eqref{ODE:Oseen:y2} we proceed similarly to~\S4.3 in~\cite{AAC16} (there for a linear transport-reaction equation in 1D) and construct a strict Lyapunov functional~$\|y\|_{\mX_{k_1}}^2 := \ip{y}{\mX_{k_1} y}_{\ell^2(\Z)}$ for each $k_1\in\Zo$.
In the following, we will use the canonical unit vectors $(e_j)_{j\in\Z}$ as orthonormal basis for
% the infinite dimensional Hilbert space
~$\ell^2(\Z)$% of square-summable sequences
, which are defined for $j,k\in\Z$ as
\[
e_j^k
=
\begin{cases}
1 & \text{for } k=j, \\
0 & \text{for } k\ne j.
\end{cases}
\]
We construct (an ansatz for) a strict Lyapunov functional using~\cite[Algorithm 3]{AAM21}, see Appendix~\ref{app:algorithm:1Pi}, starting with $\Pi_0=\mI$, $\tA_0=-\big(\widehat{\mJ}_{k_1}\big)_{2,2}$, and $\tB_0=\widehat{\mR}_{2,2}$.
The orthogonal projection~$\tPi_1$ onto $\ker(\tB_0\tB_0)$ is given as $\tPi_1=e_0 e_0^\top$, such that $\Pi_1:=\tPi_1 =e_0 e_0^\top$.
Here, by abuse of notation, sequences (such as $e_0$) are interpreted as infinite column vectors and $e_0 e_0^\top$ is the outer product of two (infinite) vectors such that $e_0 e_0^\top \in\cB(\ell^2(\Z))$ with
\[
\big(e_0 e_0^\top\big)_{j,k\in\Z}
=\begin{cases}
1 & \text{for } k=j=0\ , \\
0 & \text{else } \ .
\end{cases}
\]
Then,
\[
\tA_1
:= \Pi_1\tA_0\Pi_1
= -e_0 \underbracket{e_0^\top \big(\widehat{\mJ}_{k_1}\big)_{2,2} e_0}_{=0} e_0^\top
\]
and
\begin{align*}
\tB_1
&:=
\Pi_1 \tA_0 (\mI-\Pi_1)
=
\tfrac{k_1}2
\begin{bmatrix}
\ddots & \ddots & \ddots & & & &
\\
& 0 & 0 & 0 & & &
\\
& & \frac{k_1^2}{|k_1| \sqrt{k_1^2+1}} & 0 & -\frac{k_1^2}{|k_1|\ \sqrt{k_1^2+1}} & &
\\
& & & 0 & 0 & 0 &
\\
& & & & \ddots & \ddots & \ddots
\end{bmatrix}
{\scriptsize
\begin{array}{l}
\\ k_2=-1 \\[0.5em] k_2=0 \\[0.5em] k_2=1 \\[1em]
\end{array} } ,
\\
&=
\tilde{\alpha}
\begin{bmatrix}
\ddots & \ddots & \ddots & & & &
\\
& 0 & 0 & 0 & & &
\\
& & 1 & 0 & -1 & &
\\
& & & 0 & 0 & 0 &
\\
& & & & \ddots & \ddots & \ddots
\end{bmatrix}
=
\tilde{\alpha} e_0 \underbracket{[\ldots,0,1,0,-1,0,\ldots]}_{=(e_{-1} -e_1)^\top} ,
\end{align*}
where $\tilde{\alpha} =\tfrac{k_1}2 \frac{k_1^2}{|k_1| \sqrt{k_1^2+1}}$.
In the next iteration,
\begin{align*}
\tB_1 \tB_1^H
&=
\tilde{\alpha}^2 e_0 (e_{-1} -e_1)^\top \big(e_0 (e_{-1} -e_1)^\top\big)^H
\\
&=
\tilde{\alpha}^2 e_0 (e_{-1} -e_1)^\top \; (e_{-1} -e_1) e_0^\top
\\
%&=
% \tilde{\alpha}^2 (e_0 e_0^\top) \big( (e_{-1} -e_1)^\top (e_{-1} -e_1)\big)
%\\
&=
2 \tilde{\alpha}^2 (e_0 e_0^\top) .
\end{align*}
Hence, the orthogonal projection~$\tPi_2$ onto $\ker \big(\tB_1\tB_1^H\big)$ is given as $\tPi_2 =\mI -e_0 e_0^\top =\mI -\Pi_1$.
Due to $\Pi_2 :=\tPi_2 \Pi_1 =0$, the iteration terminates after one iteration in agreement with $\mHC=1$.
Finally, the ansatz for the weight matrix $\mX_{k_1}$ is chosen as
\begin{equation}\label{model:Oseen:X}
\mX_{k_1}
:= \Pi_0+\epsilon_1 (\tA_0\Pi_1 + \Pi_1 \tA_0^H) =
\mI +\underbracket{\epsilon_1 \frac{k_1^3}{2\sqrt{k_1^2+1}\ |k_1|}}_{=:\epsilon_{k_1}\in\R}
\underbracket{\left[ \begin{array}{c|ccc|c}
& & & & \\
\hline
& 0 &-1 & 0 & \\
&-1 & 0 & 1 & \\
& 0 & 1 & 0 & \\
\hline
& & & &
\end{array}\right]}_{=:\mY} \ ,
\end{equation}
for some $\epsilon_1>0$ to be determined. We remark that all blank elements in the matrix in \eqref{model:Oseen:X} are zero.
In the finite-dimensional setting, for sufficiently small $\epsilon_1>0$, the squared weighted norm $\|\cdot\|_{\mX_{k_1}}^2$ yields a strict Lyapunov functional, see~\cite{AAC22b}.
In this infinite dimensional example, we shall verify this statement directly.
%For $\epsilon_{k_1} :=\epsilon_1 \frac{k_1^3}{2\sqrt{k_1^2+1}\ |k_1|}$,
The infinite matrix~$\mX_{k_1}$ is positive definite for $|\epsilon_{k_1}|<1/{\sqrt2}$.
\begin{comment}
For $k_1\in\Zo$, choosing $\epsilon_1:=\epsilon/k_1$ yields
\[
\epsilon_{k_1}
=
\epsilon\ \frac{k_1^2}{2\sqrt{k_1^2+1}\ |k_1|}
\leq \frac{\epsilon}2 \ ,
\]
since the function $\beta_1: \Z\ni k \mapsto k^2/\big(2\sqrt{k^2 +1}\ |k|\big)$ satisfies $0\leq \beta_1(k)\leq 1/2$ for all $k\in\Z$.
Consequently, the choice $\epsilon_1:=\epsilon/k_1$ for $\epsilon\in(-\sqrt{2},\sqrt{2})$ ensures that $|\epsilon_{k_1}|<\sfrac1{\sqrt2}$ such that the infinite matrix~$\mX_{k_1}$ is positive definite.
\end{comment}
Moreover,
\begin{equation}\label{DAE:Oseen:X}
(1-\sqrt{2} |\epsilon_{k_1}|)\mI
\leq
\mX_{k_1}
\leq
(1+\sqrt{2} |\epsilon_{k_1}|)\mI .
\end{equation}
\medskip
Finally, the coefficients $\epsilon_{k_1},\, k_1\ne0$ %(resp. the coefficient~$\epsilon\in(-\sqrt{2},\sqrt{2})$)
should be chosen such that the Lyapunov matrix inequalities (LMIs)
\begin{equation}\label{DAE:Oseen:LMI}
\mA_{k_1}^H \mX_{k_1} +\mX_{k_1}\mA_{k_1} +2\mu_{k_1}\mX_{k_1}
\leq
0\,,\quad k_1\ne0
\end{equation}
hold with $\mu_{k_1}$ as large as possible.
%Here we used $\mA_{k_1} :=(\widehat{\mJ}_{k_1})_{2,2} -\widehat{\mR}_{2,2}$.
Due to Lemma~\ref{lemma:Oseen:LMI} (below), there exists an $\alpha_{\min}>0$ such that for all $\alpha\in(0,\alpha_{\min})$, the matrices $\mX_{k_1}$ defined in~\eqref{model:Oseen:X} with $\epsilon_{k_1}:=\alpha/{k_1}$ satisfy the LMIs~\eqref{DAE:Oseen:LMI}, where the decay rates~$\mu_{k_1}\ge\lambda_{1,\min}/4>0$ are uniformly bounded from below for all $k_1\in\Zo$; the bound $\lambda_{1,\min}=\lambda_{1,\min}(\alpha)$ is defined in~\eqref{model:Oseen:lambda1}.
Consequently, for $\alpha\in(0,\alpha_{\min})$, the squared weighted norms $\|(y_{k_1})_2\|_{\mX_{k_1}}^2,\, k_1\in\Zo$
%\begin{equation}
% \|y_{k_1,2}\|_{\mX_{k_1}}^2 :=
% \ip{y_{k_1,2}}{\mX_{k_1} y_{k_1,2}} \ ,
% \qquad \text{for all } k_1\in\Zo \ ,
%\end{equation}
are strict Lyapunov functionals for the dynamics of each $(y_{k_1})_2$. More precisely,
\begin{multline}\label{model:Oseen:decay:y2}
\|(y_{k_1})_2(t)\|_{\ell^2(\Z)}
\leq
\sqrt{\kappa(\mX_{k_1})} e^{-t \lambda_{1,\min}/4} \|(y_{k_1})_2(0)\|_{\ell^2(\Z)}
\\
\leq
\sqrt{\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} e^{-t \lambda_{1,\min}/4} \|y_{k_1,2}(0)\|_{\ell^2(\Z)} \ ,\quad t \geq 0\,,
\end{multline}
where $\alpha<\alpha_{\min}\leq 1/\sqrt 2$ and $\kappa(\mX_{k_1}) =\|\mX_{k_1}\|_2\ \|\mX_{k_1}^{-1}\|_2 = \frac{1+\sqrt2 \alpha/|k_1|}{1-\sqrt2 \alpha/|k_1|}$ is the condition number of~$\mX_{k_1}$.
This finishes the proof (of the first statement) of Proposition~\ref{prop:Oseen:y2}.
\end{proof}
%
%
The following lemma was used in the above proof.
There, we did not intend to find the optimal decay rate of each system in~\eqref{ODE:Oseen:y2} (and similarly of~\eqref{DAE:Oseen:staircase}). Hence, we shall (only) prove that there exists an $\epsilon_1>0$ such that $\mQ_{k_1} :=-(\mA_{k_1}^H \mX_{k_1} +\mX_{k_1}\mA_{k_1})$ is positive definite; afterwards we shall determine $\mu_{k_1}>0$ such that~\eqref{DAE:Oseen:LMI} holds.
%
\begin{lemma}\label{lemma:Oseen:LMI}
Consider $\mA_{k_1} :=(\widehat{\mJ}_{k_1})_{2,2} -\widehat{\mR}_{2,2}$ with the matrices defined in~\eqref{DAE:Oseen:staircase:R}--\eqref{DAE:Oseen:staircase:J}.
For any fixed $\nu>0$, there exists $\alpha_{\min}>0$ (defined in~\eqref{cond:Oseen:alpha}) such that for all $\alpha\in(0,\alpha_{\min})$, the matrices $\mX_{k_1}$ defined in~\eqref{model:Oseen:X} with $\epsilon_{k_1}:=\alpha/k_1$ satisfy the Lyapunov matrix inequality~\eqref{DAE:Oseen:LMI}, where $\mu_{k_1}>0$ is uniformly bounded from below by $\lambda_{1,\min}/4>0$ (defined in~\eqref{model:Oseen:lambda1}) for all $k_1\in\Zo$.
\end{lemma}
%
The technical proof of this lemma is deferred to Appendix~\ref{appendix_three}.\\
%
\noindent
{\bf Exponential decay of the full system:
}Combining the modal decay of Proposition \ref{prop:Oseen:y2} with the decay of the $(0,k_2)$-modes we shall obtain next the hypocoercivity estimate of the anisotropic Oseen equation \eqref{model:Oseen:anisotropic_2} in
state domain.
%Relating to the notation from \S\ref{ssec:sepHilbert} e choose the following Hilbert spaces:
\begin{align*}
u(t)
&\in \cH := H_{per}(\diver 0, \T^2)\\
u(t)-u_\infty
&\in \wcH:=\{ u\in\cH\ |\ \int_{\T^2} u \dd[x]=0 \}\ ,
%:= H_{per}(\diver 0, \T^2) \makered{\big/\R}\, ,
\end{align*}
where the steady state $u_\infty:=\frac{1}{4\pi^2}\int_{\T^2} u(x,0)\dd[x]$ equals the constant-in-$t$ mode $\phi_0$.
Moreover, both $\cH$ and $\wcH$ are closed subspaces of $(L^2(\T^2))^2$. For the Fourier decomposition \eqref{Fourier} we have in $\wcH$
\begin{equation}\label{parseval}
\|u(t)-u_\infty\|_{\wcH}^2
= \|u(t)-u_\infty\|_{(L^2(\T^2))^2}^2
=\frac{1}{4\pi^2} \sum_{k\ne0} |\phi_k(t)|^2\,.
\end{equation}
%
\begin{theorem}\label{th-u-decay}
Let $b_1=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2} and $u(0)\in H_{per}(\diver 0, \T^2)$.
Then
\begin{multline}\label{model:Oseen:L2-decay}
\|\nabla p(t)\|_{(L^2(\T^2))^2}\le \|u(t)-u_\infty\|_{(L^2(\T^2))^2}
\\
\leq
\sqrt{\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} e^{-t \min(\nu,\,\lambda_{1,\min}/4)} \|u(0)-u_\infty\|_{(L^2(\T^2))^2} \ ,\quad t \geq 0\,,
\end{multline}
where $\alpha$ and $\lambda_{1,\min}$ are as in \eqref{model:Oseen:decay:y2}.
\end{theorem}
%
%
\begin{proof}
We first prove that
\begin{equation}\label{y-norm}
\|\big(y_{k_1}\big)_2\|_{\ell^2(\Z)} =
\|\phi_{(k_1,\cdot)}\|_{\ell^2(\Z;\C^2)}\,,\quad \mbox{for all}\,k_1\ne0\,
\end{equation}
for divergence-free flow fields $u\in\wcH$, and that
\begin{equation}\label{y-form}
\big(y_{k_1}\big)_2=\big\{\frac{1}{|k|}\big(-k_2\ukI+k_1\ukII\big)\big\}_{k_2\in\Z}\,,\quad k_1\ne0\,
\end{equation}
is a bijection from the divergence-free subspace of $\ell^2(\Z;\C^2)$, i.e., for $ik\cdot \phi_k=0$, to $\ell^2(\Z)$; see also the remark on the leading $2\times2$-subblock of $\mP_{k_1}$ after \eqref{model:Oseen:Pk1}.
To this end we first note that the right hand side of \eqref{y-form} is
related to the vorticity of $u$, since the modal representation of $\rot u$ (in 2D) is $i(-k_2\ukI+k_1\ukII)$.
Given a divergence-free $\phi_{(k_1,\cdot)} \in \ell^2(\Z;\C^2)$ with $k_1\ne0$ we compute with \eqref{y-form}:
\begin{align*}
& \|\big(y_{k_1}\big)_2\|_{\ell^2(\Z)}^2
\\
&= \sum_{k_2\in\Z} \frac{1}{|k|^2} \big(k_2^2|\ukI|^2 -k_1k_2 \overline{\ukI} \, \ukII -k_1k_2 \ukI \, \overline{\ukII}+k_1^2|\ukII|^2\big)
\\
&= \sum_{k_2\in\Z} \frac{1}{|k|^2} \big(k_2^2|\ukI|^2 +k_1^2|\ukI|^2+k_2^2|\ukII|^2 +k_1^2|\ukII|^2\big) =
\|\phi_{(k_1,\cdot)}\|_{\ell^2(\Z;\C^2)}^2\,,
\end{align*}
where we used $k\cdot \phi_k=0$ twice.
For the other direction, given a $\big(y_{k_1}\big)_2 \in \ell^2(\Z)$ with $k_1\ne0$ we define the divergence-free flow field
\[
\ukI:=-\frac{k_2}{|k|} \big(y_{k_1}\big)_2\,,\quad
\ukII:=\frac{k_1}{|k|} \big(y_{k_1}\big)_2\,,\quad
k_2\in\Z\,,
\]
which is compatible with \eqref{y-form}. Then we have
\[
\|\phi_{(k_1,\cdot)}\|_{\ell^2(\Z;\C^2)}^2
= \sum_{k_2\in\Z} \Big(\frac{k_2^2}{|k|^2} + \frac{k_1^2}{|k|^2} \Big)|\big(y_{k_1}\big)_2|^2
= \|\big(y_{k_1}\big)_2\|_{\ell^2(\Z)}^2\,,
\]
and this proves the isometry and bijectivity.
Finally, we sum up the (square of the) modal inequalities \eqref{model:Oseen:decay:y2} for $k_1\ne0$ and the inequalities \eqref{k10-decay} for $k_1=0$ but $k_2\ne0$. This yields the claimed decay estimate \eqref{model:Oseen:L2-decay} for $u(t)-u_\infty$ with using \eqref{parseval}. The decay of $\|\nabla p(t)\|_{(L^2(\T^2))^2}$ then follows from the estimate \eqref{p-u-estimate}.
\end{proof}
%
\begin{remark}\label{rem:isometry-rotu}
The modal isometry \eqref{y-norm} can be extended to physical space by combining all modes $\big(y_{k_1}\big)_2$.
Recalling from \eqref{y-form} the dependence of $y_2=y_2[u]$ on $u$, let $y_2:=\big\{\big(y_{k}\big)_2\big\}_{k\in\Z^2\setminus \{0\}}$. For scalar functions $f$ on $\T^2$ we define the following homogeneous Sobolev space via the Fourier decomposition of $f$:
\[
f\in \dot H_{per}^{-1}(\T^2) \quad :\Leftrightarrow \quad \|f\|_{\dot H_{per}^{-1}(\T^2)}^2 :=
\frac1{4\pi^2} \sum_{k\ne0} \frac1{|k|^2}|f_k|^2 <\infty\,.
\]
Then we have the following relation for divergence-free flow fields $u$ on $\T^2$ with vanishing average, i.e.\ $\int_{\T^2} u(x)\dd[x]=0$:
The space $\dot H_{per}^{-1}(\T^2)$\; [for $\rot u=-\partial_{x_2}\uI+\partial_{x_1}\uII$] is isometrically isomorphic to %\makered{$\wcH=H_{per}(\diver \,0,\T^2)\big/\R$}
\[\wcH =\{ u\in H_{per}(\diver \,0,\T^2)\ |\ \int_{\T^2} u \dd[x]=0 \}\;\]
[for $u$].
More precisely we have
\begin{multline*}
\frac1{4\pi^2} \|y_2[u]\|^2_{\ell^2(\Z^2\setminus\{0\})}
= \frac1{4\pi^2} \sum_{k\ne0} |(y_k)_2|^2
= \frac1{4\pi^2} \sum_{k\ne0} \Big|\frac1{|k|}\big(-k_2\ukI+k_1\ukII\big)\Big|^2
\\
= \|\rot u\|^2_{\dot H_{per}^{-1}(\T^2)}
= \frac1{4\pi^2} \|\phi_k\|^2_{(\ell^2(\Z^2\setminus\{0\}))^2}
= \|u\|^2_{(L^2(\T^2))^2}\,,
\end{multline*}
where we used \eqref{y-norm}. This relation shows that $\|u(t)\|_{(L^2(\T^2))^2}$ describes all combined modes of the dynamical part in \eqref{ODE:Oseen:y2} by summing over $k_1\in\Z$.
\end{remark}
\medskip
In Proposition \ref{prop:Oseen:y2} we established the exponential decay of the dynamical mode component $(y_{k_1})_2(t)$.
Next we shall extend this result by including the enslaved components $(y_{k_1})_1\equiv0$ and $(y_{k_1})_3(t)=i\,p_{k_1}(t)$.
%
\begin{proposition}\label{prop:Oseen:y}
Let $b_1=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2}.
Then, for each $k_1\in\Zo$, the modal dynamics~\eqref{DAE:Oseen:staircase} is hypocoercive in the sense of~\eqref{exp-decay} in~$\ell^2(\Z;\C^3)$.
% Moreover, its HC-index $\mHC=1$.
\end{proposition}
%
\begin{proof}
Noting again that the third block in the staircase form~\eqref{DAE:Oseen:staircase} is void, it follows that
\begin{align*}
y_{k_1}
&=
\mP_{k_1} w_{k_1}
=
\big[ \mP_{(k_1,k_2)} [\phi_{(k_1,k_2),1}, \phi_{(k_1,k_2),2}, p_{(k_1,k_2)}]^\top ;\ k_2\in\Z \big]
\\
&=:
\big[ [y_{(k_1,k_2),1}, y_{(k_1,k_2),2}, y_{(k_1,k_2),3}]^\top ;\ k_2\in\Z \big]
\in \ell^2(\Z;\C^3).
\end{align*}
%since $y_{k_1,3}$ is void.
In our infinite dimensional model, we deduce (in analogy to the results of~\cite[Corollary 1]{AAM21}) that
\begin{equation}\label{DAE:Oseen:y}
\begin{split}
\big(y_{k_1}\big)_1 &= 0 \ ,
\\
\big(y_{k_1}\big)_2 &\text{ is governed by~\eqref{ODE:Oseen:y2}} \ ,
\\
\big(y_{k_1}\big)_3
&=
(\widecheck{\mJ}_{k_1})_{3,1}^{-H} (-(\widecheck{\mJ}_{k_1})_{2,1}^H -\mR_{2,1}^H) \big(y_{k_1}\big)_2 \ ,
\end{split}
\end{equation}
where $(\widecheck{\mJ}_{k_1})_{3,1} =\diag(|k|=\sqrt{k_1^2 +k_2^2}; \ k_2\in\Z)$, $\mR_{2,1}=0$, and $(\widecheck{\mJ}_{k_1})_{2,1}$ is a symmetric tridiagonal matrix whose diagonal elements are zero and the off-diagonal elements are $(\widecheck{\mJ}_{k_1})_{2,1}(k_2,k_2\pm 1) =k_1^2 /(2\ |k|\ |k\pm e_2|)$ with $e_2:=[0,\, 1]^\top$.
Then, for fixed $k_1\ne 0$, the third component of \eqref{DAE:Oseen:y} reads explicitly
\begin{equation}\label{y2-y3-relation}
\big(y_{k_1}\big)_3=
\big(y_{k}\big)_3= -\frac{k_1^2}{2|k|^2}\left(\frac{\big(y_{k-e_2}\big)_2}{|k-e_2|}+\frac{\big(y_{k+e_2}\big)_2}{|k+e_2|}\right) , \quad k=[k_1,\,k_2]^\top,\quad k_2\in\Z.
\end{equation}
%We observe that $y_{k_1,3}$ ``behaves like'' $y_{k_1,2}$, such that
Since $(y_{k_1})_3$ is a linear combination of $(y_{k_1})_2$ with $k_2$-uniformly bounded coefficients, solutions $y_{k_1}(t)$ for consistent initial data will converge to $0$ with uniform exponential rate $\lambda_{1,\min}/4>0$ for all $k_1\in\Zo$.
Since the multiplying factors in the modal relation \eqref{y2-y3-relation} are $\mathcal O(1/|k|)$ and by using \eqref{y-norm}, \eqref{parseval}, we see that the decay of $p(t)$ is in a Sobolev space one level higher than the decay of $u(t)-u_\infty$. With different tools, this was already reflected in \eqref{model:Oseen:L2-decay}.
A consistent initial value $y_{k_1}(0)$ again has to satisfy~\eqref{DAE:Oseen:y} such that
\begin{equation}\label{DAE:Oseen:y0}
\begin{split}
\big(y_{k_1}\big)_1(0) &= 0 \ ,
\\
\big(y_{k_1}\big)_2(0) &\in\ell^2(\Z) \ ,
\\
\big(y_{k_1}\big)_3(0)
&=
(\widecheck{\mJ}_{k_1})_{3,1}^{-H} (-(\widecheck{\mJ}_{k_1})_{2,1}^H -\mR_{2,1}^H) \big(y_{k_1}\big)_2(0) \ .
\end{split}
\end{equation}
Then, the associated solution $y_{k_1}(t)$ of~\eqref{DAE:Oseen:staircase} satisfies
\begin{equation} \label{model:Oseen:decay:y}
\begin{split}
\| y_{k_1}(t)\|^2
&=
\| \big(y_{k_1}\big)_2(t)\|^2 +\| \big(y_{k_1}\big)_3(t)\|^2
\\
&=
\| \big(y_{k_1}\big)_2(t)\|^2 +\| (\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\, \big(y_{k_1}\big)_2(t)\|^2
\\
&\leq
\Big(1 +\big\|(\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\big\|^2 \Big) \| \big(y_{k_1}\big)_2(t)\|^2
\\
&\leq
\Big(1 +\big\|(\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\big\|^2 \Big) {\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} \| \big(y_{k_1}\big)_2(0)\|^2 e^{-\lambda_{1,\min} t/2}
\\
&\leq
\Big(1 +\big\|(\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\big\|^2 \Big) {\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} \| y_{k_1}(0)\|^2 e^{-\lambda_{1,\min} t/2} \ ,
\end{split}
\end{equation}
due to~\eqref{model:Oseen:decay:y2}, where $\alpha<\alpha_{\min}\leq 1/\sqrt 2$, $\lambda_{1,\min}=\lambda_{1,\min}(\alpha)$ is defined in~\eqref{model:Oseen:lambda1}, and $\|\mT\|$ denotes the operator norm of a bounded operator~$\mT\in\mathcal B(\ell(\Z))$.
This finishes the proof of Proposition~\ref{prop:Oseen:y}.
\end{proof}
%
As a final step we note that the (inverse) staircase transformation also implies exponential decay of the variable $\xk :=[\ukI,\ukII,p_k]^\top$ which is the modal representation of $[u,\, p]^\top$:
%
\begin{proposition}\label{prop:Oseen:w}
Let $b_1=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2}.
Then, for each $k_1\in\Z$, the modal dynamics~\eqref{DAE:Oseen} is hypocoercive in the sense of~\eqref{exp-decay} in~$\ell^2(\Z;\C^3)$.
% Moreover, its HC-index $\mHC=1$.
\end{proposition}
%
\begin{proof}
For $k_1=0$, the (evolution equations of the) modes~$w_{k_1}$ are decoupled.
Following the analysis of the family of decoupled DAEs~\eqref{DAE:Stokes} (with $k_1=0$), and in particular \eqref{k10-decay} shows that $w_0$ converges to the infinite complex vector
\[
w_{0}^\infty
:=
[\ldots;\, 0, 0, 0;\, \phi_{(0,0),1}, \phi_{(0,0),2}, p_{(0,0)};\, 0, 0, 0;\, \ldots]^\top
\ ,
\]
(which corresponds to the constant equilibrium $(\phi_0,p_0)$) with the exponential decay rate $\nu$($=\min_{k_2\ne 0} (\nu k_2^2)$).
For $k_1\in\Zo$, consistent initial data~$w_{k_1}(0)$ of system~\eqref{DAE:Oseen} has the form $w_{k_1}(0)=\mP_{k_1}^H y_{k_1}(0)$ where $\mP_{k_1}$ is defined in~\eqref{model:Oseen:Pk1} and $y_{k_1}(0)$ is given as~\eqref{DAE:Oseen:y0}.
Then, the associated solution~$w_{k_1}(t)$ of system~\eqref{DAE:Oseen} satisfies
\begin{align*}
\| w_{k_1}(t)\|^2
&=
\| \mP_{k_1}^H y_{k_1}(t)\|^2
=
\| y_{k_1}(t)\|^2 \\
&\leq
\Big(1 +\big\|(\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\big\|^2 \Big) {\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} \| y_{k_1}(0)\|^2 e^{-\lambda_{1,\min} t/2} \\
&=
\Big(1 +\big\|(\widecheck{\mJ}_{k_1})_{3,1}^{-1} (\widecheck{\mJ}_{k_1})_{2,1}\big\|^2 \Big) {\tfrac{1+\sqrt{2}\ \alpha}{1-\sqrt{2}\ \alpha}} \| w_{k_1}(0)\|^2 e^{-\lambda_{1,\min} t/2} \ ,
\end{align*}
due to~\eqref{model:Oseen:decay:y}, where $\alpha<\alpha_{\min}\leq 1/\sqrt 2$, $\lambda_{1,\min}=\lambda_{1,\min}(\alpha)$ is defined in~\eqref{model:Oseen:lambda1}, and $\|\mT\|$ denotes the operator norm of a bounded operator~$\mT\in\mathcal B(\ell(\Z))$.
Altogether, for consistent initial data, solutions of system~\eqref{DAE:Oseen} (and resp.~\eqref{eq:Oseen:Fourier}) converge to the constant equilibrium with a uniform exponential rate.
\end{proof}
%-----------------------------------------------------------
\section{Conclusions}\label{sec:conclusions}
After extending the notion of hypocoercivity index to evolution equations in (infinite dimensional) Hilbert spaces, we have performed the analysis of the long-time decay behavior of three variants of isotropic and anisotropic Oseen-type equations from fluid dynamics (for simplicity on a 2D torus). Due to the torus setting we used DAE theory in Fourier space to classify the hypocoercivity index. These equations are either coercive, hypocoercive with index 1, or even not hypocoercive (the latter showing exponential convergence only to a traveling wave solution).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{appendices}
\section{Appendix}\label{sec:appendix}
\subsection{Hypocoercivity in linear semi-dissipative DAEs}
\label{app:DAEs:hypocoercivity}
Here, we recall the basic theory of hypocoercivity for finite-dimensional linear semi-dissipative Hamiltonian DAEs~\cite{AAM21}:
%
\begin{definition}[{\cite[Definition~4]{AAM21}}]\label{def:hypoDAE}
A matrix pencil $\lambda \mE-\mA$ is called~\emph{negative hypocoercive} if the pencil is regular, of DAE-index at most two and the finite eigenvalues of the pencil $\lambda \mE-\mA$ have negative real part.
\end{definition}
%
We note that a regular pencil might not have \emph{any} finite eigenvalues, in which case the last condition would be void.
Due to \cite[Theorem~3]{AAM21}, a linear semi-dissipative Hamiltonian DAE system~\eqref{DAE:EA} with a regular pencil~$\lambda \mE-\mA$ only has finite eigenvalues with non-positive real part.
%
The definition of the hypocoercivity index for DAEs is based on a staircase form of DAEs, see \cite[Lemma 5]{AAM21}:
%
\begin{lemma}[Staircase form for triple $(\mE,\mJ,\mR)$] \label{lem:tSF}
Let $\mE,\mJ,\mR\in\Cnn$ satisfy $\mE=\mE^H\geq 0$, $\mR=\mR^H\geq 0$ and $\mJ=-\mJ^H$.
Then there exists a unitary matrix $\mP\in\Cnn$, such that $\widecheck \mE :=\mP\ \mE\ \mP^H$, $\widecheck \mJ :=\mP\ \mJ\ \mP^H$ and $\widecheck \mR :=\mP\ \mR\ \mP^H$ satisfy
%
\begin{equation}\label{staircase:EJR}
\widecheck \mE %:=\mP\ \mE\ \mP^H
=:\begin{bmatrix}
\mE_{1,1} & \mE_{2,1}^H & 0 & 0 & 0 \\
\mE_{2,1} & \mE_{2,2} & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0
\end{bmatrix}\! ,
\ %& %
\widecheck \mJ %:=\mP\ \mJ\ \mP^H
=:\begin{bmatrix}
\mJ_{1,1} & -\mJ_{2,1}^H & -\mJ_{3,1}^H & -\mJ_{4,1}^H & 0\\
\mJ_{2,1} & \mJ_{2,2} & -\mJ_{3,2}^H& 0 & 0\\
\mJ_{3,1} & \mJ_{3,2} & \mJ_{3,3}& 0 & 0\\
\mJ_{4,1} & 0& 0 & 0 &0\\
0 & 0 & 0 & 0 & 0
\end{bmatrix}\! ,
\ %& %
\widecheck \mR %:=\mP\ \mR\ \mP^H
=:\begin{bmatrix}
\mR_{1,1} & \mR_{2,1}^H & \mR_{3,1}^H& 0 & 0 \\
\mR_{2,1} & \mR_{2,2} & \mR_{3,2}^H &0 & 0 \\
\mR_{3,1} & \mR_{3,2} & \mR_{3,3} &0 & 0 \\
0 & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0
\end{bmatrix}\! .
\end{equation}
%
These three matrices are partitioned in the same way, with (square) diagonal block matrices of sizes $n_1,n_2,n_3,n_4=n_1,n_5\in\N_0$.
If the block matrices $\mE_{1,1}$, $\mE_{2,2}$ (as well as $\mE_{2,1}$) are present, then the matrices $\mE_{1,1}$, $\mE_{2,2}$ (as well as $\begin{bmatrix} \mE_{1,1} & \mE_{2,1}^H \\ \mE_{2,1} & \mE_{2,2} \end{bmatrix}$) are positive definite.
If the block matrices $\mJ_{4,1}$, $\mJ_{3,3} -\mR_{3,3}$ are present, then the matrices $\mJ_{4,1}$, $\mJ_{3,3}-\mR_{3,3}$ are invertible.
\end{lemma}
The proof is given as a constructive algorithm, see~\cite[Algorithm 5]{AAM21}, which is reproduced here as Algorithm~\ref{algorithm:staircase:E_J-R} below.
\smallskip
% To reproduce 'Algorithm 5', set the counter 'algorithm' to 4 since it will be increased by calling the environment 'breakablealgorithm':
\setcounter{algorithm}{4}
\begin{breakablealgorithm}
\caption{Staircase Algorithm for triple~$(\mE,\mJ,\mR)$}
\label{algorithm:staircase:E_J-R}
\begin{algorithmic}[1]
\Statex ----------- \textit{Step 1} -----------
\State Perform a spectral decomposition of~$\mE$ such that
\[
\mE =\mU_E \begin{bmatrix} \tE_{1,1} & 0 \\ 0 & 0 \end{bmatrix} \mU_E^H,
\]
with $\mU_E\in\Cnn$ unitary, $\tE_{1,1}\in \C^{\tilde n_1, \tilde n_1}$ positive definite or $\tilde n_1=0$.
%
\State Set
\begin{align*}
\mP
&:= \mU^H_E , \qquad
\tJ
:= \mU^H_E\ \mJ\ \mU_E
=\begin{bmatrix}
\tJ_{1,1} & -\tJ_{2,1}^H \\
\tJ_{2,1} & \tJ_{2,2}
\end{bmatrix} ,
\\ %
\tR
&:= \mU^H_E\ \mR\ \mU_E
=\begin{bmatrix}
\tR_{1,1} & \tR_{2,1}^H \\
\tR_{2,1} & \tR_{2,2} \end{bmatrix} , \qquad
%
\tE :=\mU^H_E\ \mE\ \mU_E .
\end{align*}
%
\Statex ----------- \textit{Step 2} -----------
\If{$\tilde n_1<n$}
\State Apply~\cite[Lemma 2]{AAM21} to $\tJ_{2,2}-\tR_{2,2}\in\C^{(n-\tilde n_1)\times (n-\tilde n_1)}$ such that
\[
\mP_{2,2}\ (\tJ_{2,2}-\tR_{2,2})\ \mP_{2,2}^H
=\begin{bmatrix} \tSigma_{2,2} & 0 \\ 0 & 0\end{bmatrix},
\]
with $\tSigma_{2,2}\in \mathbb C^{\tilde n_2,\tilde n_2}$ invertible or $\tilde n_2=0$.
\EndIf
\State Set
\[
\mP_2:=\begin{bmatrix} \mI& 0 \\ 0 & \mP_{2,2} \end{bmatrix} \in\Cnn,\qquad \mP:=\mP_2 \mP.
\]
\State Set $\tE:=\mP_2\ \tE\ \mP_2^H$,
\[
\def\arraystretch{1.4}
\tJ := \mP_2\ \tJ\ \mP_2^H
=: \left[ \begin{array}{c|cc}
\tJ_{1,1} & -\tJ_{2,1}^H & -\tJ_{3,1}^H \\
\hline %\\[-3.6mm]
\tJ_{2,1} & \tJ_{2,2} & 0 \\
\tJ_{3,1}& 0 & 0
\end{array}\right],\qquad
%
\tR := \mP_2\ \tR\ \mP_2^H
=: \left[ \begin{array}{c|cc}
\tR_{1,1} & \tR_{2,1}^H & 0 \\
\hline %\\[-3.6mm]
\tR_{2,1} & \tR_{2,2} & 0 \\
0 & 0 & 0
\end{array}\right],\quad
\]
with $\tJ_{2,2}-\tR_{2,2}=\tSigma_{2,2}$.
(The lines indicate the partitioning of the block matrices~$\tJ$ and~$\tR$ in the previous step.
Note that the positive semi-definiteness of the Hermitian matrix~$\mR$ implies the~$0$ structure in~$\tR$.)
%
\Statex ----------- \textit{Step 3} -----------
\State Define $\tilde n_3 :=n -\tilde n_1 -\tilde n_2$.
\If{$\tilde n_3>0$}
\State Perform an SVD of $\tJ_{3,1}$ such that
\[
\tJ_{3,1} = \mU_{3,1} \begin{bmatrix} \tSigma_{3,1} & 0\\ 0 & 0 \end{bmatrix} \mV^H_{3,1} \in \C^{\tilde n_3 \times \tilde n_1}\,,
\]
\hspace{12pt} with~$\tSigma_{3,1}\in\R^{n_1 \times n_1}$ nonsingular diagonal or $n_1 =0$.
\EndIf
\State Set
\[
\mP_3 :=\begin{bmatrix}
\mV_{3,1}^H & & \\
& \mI & \\
& & \mU_{3,1}^H
\end{bmatrix} \in\Cnn,\qquad
\mP:= \mP_3 \mP.
\]
\State Set $\widecheck \mE
:= \mP_3\ \tE\ \mP_3^H$, $\widecheck\mJ
:= \mP_3\ \tJ\ \mP_3^H$, $\widecheck\mR
:= \mP_3\ \tR\ \mP_3^H$ such that
\begin{align*}
\widecheck \mE
%:= \mP_3\ \tE\ \mP_3^H
&=: \left[ \begin{array}{cc|c|cc}
\mE_{1,1} & \mE_{2,1}^H & 0 &0 & 0 \\
\mE_{2,1} & \mE_{2,2} & 0 & 0 & 0 \\
\hline
0 & 0 & 0 & 0 & 0 \\
\hline
0 & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0
\end{array}\right],
\\
\widecheck\mJ
%:= \mP_3\ \tJ\ \mP_3^H
&=:\left[ \begin{array}{cc|c|cc}
\mJ_{1,1} & -\mJ_{2,1}^H & -\mJ_{3,1}^H & -\mJ_{4,1}^H & 0 \\
\mJ_{2,1} & \mJ_{2,2} & -\mJ_{3,2}^H & 0 & 0 \\
\hline
\mJ_{3,1} & \mJ_{3,2} & \mJ_{3,3} & 0 & 0 \\
\hline
\mJ_{4,1} & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0
\end{array}\right],
\qquad %
\widecheck\mR
%:= \mP_3\ \tR\ \mP_3^H
=:\left[ \begin{array}{cc|c|cc}
\mR_{1,1} & \mR_{2,1}^H & \mR_{3,1}^H & 0 & 0 \\
\mR_{2,1} & \mR_{2,2} & \mR_{3,2}^H & 0 & 0 \\
\hline
\mR_{3,1} & \mR_{3,2} & \mR_{3,3} & 0 & 0 \\
\hline
0 & 0 & 0 & 0 & 0 \\
0 & 0 & 0 & 0 & 0
\end{array}\right],
\end{align*}
%
which are of the desired form with $n_2:=\tilde n_1 -n_1$, $n_3 :=\tilde n_2$, $n_4:=n_1$, $n_5:=\tilde n_3 -n_4$.
The matrices~$\mJ_{4,1}:=\tSigma_{3,1}$ and $\mJ_{3,3}-\mR_{3,3} =\tJ_{2,2}-\tR_{2,2} =\tSigma_{2,2}$ are invertible.
\end{algorithmic}
\end{breakablealgorithm}
%The matrices~$\mJ_{4,1}:=\tSigma_{3,1}$ and $\mJ_{3,3}-\mR_{3,3} =\tJ_{2,2}-\tR_{2,2} =\tSigma_{2,2}$ are invertible.
%\end{proof}
%
The pencil~$\lambda \mE-(\mJ-\mR)$ is associated to the DAE~\eqref{DAE:EA} with~$\mA=\mJ-\mR$. Using the above lemma, it can be transformed into a DAE in staircase form,
%
\begin{equation}\label{DAE:EA:hat:0}
\widecheck\mE\dot y =(\widecheck\mJ -\widecheck\mR)y\ ,
\qquad\text{with } y:=\mP x\ .
\end{equation}
In analogy to \eqref{staircase:EJR}, the vector $y$ can be partitioned as $y=(y_1,...,y_5)^T$ with subvectors of the respective length $n_1,...,n_5$.
%
For systems~\eqref{DAE:EA:hat:0} with~$n_5=0$, the underlying implicit ODE systems are given by the system in~$y_2$ that are obtained by eliminating all other variables.
For example, if $n_2>0$ and $n_3\geq 0$, then this yields systems of the form
%
\begin{equation}\label{underlyingODE:y}
\mE_{2,2} \dot y_2
=\widehat \mA_{2,2} y_2
=(\widehat \mJ_{2,2}-\widehat \mR_{2,2}) y_2,
\end{equation}
%
with $\mE_{2,2}=\widehat\mE_{2,2}$ Hermitian positive definite and $\widehat \mA_{2,2}$ semi-dissipative.
Here
%
\begin{multline*}
\widehat\mJ_{2,2} :=(\widehat\mA_{2,2})_S \ , \quad
\widehat\mR_{2,2} :=-(\widehat\mA_{2,2})_H \ ,
\\
\text{where }
\widehat\mA_{2,2}
=\begin{cases}
\widecheck\mA_{2,2} \ , & \text{if } n_3=0\ , \\
\widecheck\mA_{2,2} -\widecheck\mA_{2,3}\widecheck\mA_{3,3}^{-1}\widecheck\mA_{3,2} \ , & \text{if } n_3>0\ .
\end{cases}
\end{multline*}
This staircase form now allows to define the hypocoercivity index also for DAEs:
\begin{definition}[{\cite[Definition 5]{AAM21}}] \label{def:DAE:mHC}
Consider a linear semi-dissipative Hamiltonian DAE system~\eqref{DAE:EA} with a regular pencil~$\lambda \mE-\mA$ and the unitarily congruent DAE~\eqref{DAE:EA:hat:0} in staircase form~\eqref{staircase:EJR}.
If the underlying implicit ODE~\eqref{underlyingODE:y} is missing (present) then system~\eqref{DAE:EA} is said to exhibit \emph{(non-)trivial dynamics}.
In case of non-trivial dynamics, the~\emph{HC-index~$m_{HC}$} of~$\lambda \mE-\mA$ is defined as the HC-index of the system matrix~$(\mE_{2,2}^{1/2})^{-1} \widehat{\mA}_{2,2} (\mE_{2,2}^{1/2})^{-1}$ of~\eqref{underlyingODE:y} (in the sense of Definition \ref{def:matrix:hypocoercive}), otherwise it is defined as~$0$.
\end{definition}
%
The following proposition %is the DAE-counterpart of Theorem~\ref{th:HC-decay}.
%It
states that the HC-index characterizes the short time behavior of its solution propagator, but restricted to the dynamical subspace.
%
\begin{proposition}[{\cite[Proposition 3]{AAM21}}] \label{prop:DAE+HC-decay}
Consider the semi-dissipative Hamiltonian DAE~\eqref{DAE:EA} with a regular, negative hypocoercive pencil~$\lambda\mE-\mA$, DAE-index at most two, non-trivial dynamics, and consistent initial condition~$x(0)$.%, see Remark~\ref{rem:IC:consistency}.
Then its (finite) HC-index is $\mHC\in\N_0$, if and only if
%
\[
\|S(t)\|_{\mE}
=
1 -c t^a +\bigO(t^{a+1})
\qquad
\text{for } t\to 0^+ \ ,
\]
where $c>0$ and $a=2\mHC+1$, and the propagator (semi-)norm pertaining to the evolution of~\eqref{DAE:EA} reads
%
\[
\|S(t)\|_{\mE}
:=
\sup_{\stackrel{\|x(0)\|_{\mE}=1}{\text{for consistent } x(0)}}
\|x(t)\|_{\mE} \ ,
\qquad
t\geq 0 \ .
\]
\end{proposition}
The anisotropic Oseen model~\eqref{model:Oseen:anisotropic} with constant $b\in\R^2$ and $b_1\ne 0$ is not hypocoercive, see Proposition~\ref{prop:Oseen:constantU:notHC}.
This lack of hypocoercivity can also be verified by considering \eqref{model:Oseen:anisotropic} with $b\in\R^2$ and $b_1\ne 0$ as a partial differential-algebraic equation (PDAE), and bringing its modal representation into staircase form:
\begin{eexample}\label{ex:Oseen:constantU}
Consider the anisotropic Oseen model~\eqref{model:Oseen:anisotropic} with constant vector $b\in\R^2$.
Proceeding as in~\S\ref{ssec:Stokes} yields with $\mP_k$ from~\eqref{ex:Stokes:Pk}:
%
\[
\widecheck \mE\ \dot y_k (t)
=(\widecheck \mJ_k -\widecheck \mR_k)\ y_k(t)
\ , \qquad %
t\geq 0 \ ,
\]
%
with $\widecheck\mE =\diag(1,1,0)$, $\widecheck\mR_k =\diag(\nu k_2^2,\nu k_2^2, 0)$, and $\widecheck \mJ_k$ as in~\eqref{checkJk}.
%
The modes with $k_2=0$ (and $k_1\ne 0$) imply $y_{k,1}(t)= \ukI (t)\,{k_1}/{|k_1|} =0$, which is also a consistency condition on the initial value~$\ukI(0)$, i.e.\ the divergence-free condition of these initial modes.
Since the corresponding $\widecheck\mR_k=0$, the modes with $k=(k_1,0)$ are not hypocoercive.
In fact, they are purely oscillatory and have no damping.
This modal approach also shows that general solutions to~\eqref{model:Oseen:anisotropic} in~$\wcH$ converge with rate~$\mu=\nu$ to traveling waves like~\eqref{counterex:Oseen:constantU:notHC}.
\end{eexample}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Review of Algorithm 3 from \cite{AAM21}}
\label{app:algorithm:1Pi}
The purpose of Algorithm 3 from \cite{AAM21} is to construct an ansatz for strict Lyapunov functionals for semi-dissipative Hamiltonian ODEs~\eqref{ODE:A} with negative hypocoercive matrix $\mA\in\Cnn$.
In~\cite{AAC22b}, explicit restrictions on~$\epsilon_j$ (relative to other parameters) were derived such that a suitable choice of $\epsilon_j$ turns the ansatz in Step~10 of Algorithm~\ref{algorithm:1Pi} into a strict Lyapunov functional.
Consider a semi-dissipative matrix~$\mA=\mJ-\mR$ with finite HC-index, then Algorithm~3 reads as follows:
\smallskip
% To reproduce 'Algorithm 3', set the counter 'algorithm' to 2 since it will be increased by calling the environment 'breakablealgorithm':
\setcounter{algorithm}{2}
\begin{breakablealgorithm}
\caption{Construction of a strict Lyapunov functional}
\label{algorithm:1Pi}
\begin{algorithmic}[1]
\Require $\Pi_0:=\mI$, $\tA_0:=-\mJ$, $\tB_0:=\mR$, $j:=1$
\State Construct an orthogonal projection~$\widetilde{\Pi}_j$ onto $\ker \big(\tB_{j-1} \tB_{j-1}^H\big)$.
\State $\Pi_j :=\widetilde{\Pi}_j \Pi_{j-1}$
\While{$\Pi_j \ne0$}
\State Set $\tA_j :=\Pi_j \tA_{j-1} \Pi_j$, $\tB_j :=\Pi_j \tA_{j-1} (\Pi_{j-1} -\Pi_j)$.
\State $j := j+1$
\State Construct an orthogonal projection~$\widetilde{\Pi}_j$ onto $\ker \big(\tB_{j-1} \tB_{j-1}^H\big)$.
\State $\Pi_j :=\widetilde{\Pi}_j \Pi_{j-1}$
\EndWhile
\State $\mHC :=j-1$
\State Set $\mX :=\Pi_0 +\sum_{j=1}^{\mHC} \epsilon_j \big(\tA_{j-1}\Pi_j +\Pi_j \tA_{j-1}^H\big)$ for sufficiently small $\epsilon_j>0$.
\Ensure $\|\cdot\|_\mX^2 :=\ip{\cdot}{\mX\cdot}$
\end{algorithmic}
\end{breakablealgorithm}
%
%%%%%%%%%%%%%%%%%%%
\section{Auxiliary results for the infinite dimensional ODE (\ref{ODE:Oseen:y2})}
\subsection{Quantitative estimate of~$\kappa$ for the evolution generators from Proposition~\ref{prop:Oseen:y2}}
\label{app:QuantitativeEstimate}
\newcommand{\hj}{\widehat{j}}
\newcommand{\hk}{\widehat{k}}
\newcommand{\hkm}{\hk_-}
\newcommand{\hkp}{\hk_+}
\newcommand{\hl}{\widehat{\ell}}
\newcommand{\normalized}[1]{\frac{#1}{|#1|}}
%
Here we compute the HC-index of the system matrix~$(\hR-\hJ_{k_1})_{2,2}$ in the modal dynamics~\eqref{ODE:Oseen:y2} using directly Definition~\ref{def:unb-op:HC-index}, and derive a quantitative estimate for $\kappa>0$ in~\eqref{Op:J-R:unbdd}.
This is an alternative approach to the proof given for Proposition~\ref{prop:Oseen:y2}.
%
\begin{proposition}\label{prop:Oseen:y2:kappa}
Let $b_1=\sin (x_2)$ in~\eqref{model:Oseen:anisotropic_2}.
Then, for each $k_1\in\Zo$, the system matrix~$-{\mA}_{k_1}:=(\hR-\hJ_{k_1})_{2,2}$ in the modal dynamics~\eqref{ODE:Oseen:y2} has HC-index $\mHC=1$, and satisfies~\eqref{Op:J-R:unbdd} for some $\kappa\geq \nu/100>0$.
\end{proposition}
%
\begin{proof}
We show that there exists $\kappa>0$ such that~\eqref{Op:J-R:unbdd} with $m=1$, $\mJ=\hJzz:=\hJkzz$ and $\mR=\hRzz$ holds.
Let $x\in\cD(\VhRzz)$ be a unit vector, and decompose $x=v+w$ where $v\in(\ker(\VhRzz))^\perp$ and $w\in\ker(\VhRzz)$.
Define $\alpha:=\|v\|_2$, and then supposing $0<\alpha<1$, define the unit vector $u:=v/\alpha$ and note that $w/\sqrt{1-\alpha^2}=e_0$, cf.~\eqref{DAE:Oseen:staircase:R}.
We now compute
\begin{equation} \label{est:xRx}
\normBig{\VhRzz x}^2
=\normBig{\VhRzz v}^2
=\alpha^2 \normBig{\VhRzz u}^2
\geq \alpha^2 \nu
\end{equation}
using~\eqref{DAE:Oseen:staircase:R}.
In the same way, we find
\begin{equation} \label{est:xJRJx}
\begin{split}
&\normBig{\VhRzz\hJzz^* x}^2 \\
&=\ipBig{\VhRzz\hJzz^* (v+w)}{\VhRzz\hJzz^* (v+w)} \\
&=\normBig{\VhRzz\hJzz^* v}^2 +2\Re\Big(\ipBig{\VhRzz\hJzz^* v}{\VhRzz\hJzz^* w}\Big) +\normBig{\VhRzz\hJzz^* w}^2 \\
&=\alpha^2 \normBig{\VhRzz\hJzz^* u}^2 +\alpha\sqrt{1-\alpha^2} 2\Re\Big(\ipBig{\VhRzz\hJzz^* u}{\VhRzz\hJzz^* e_0} \Big)
\\
&\qquad +(1-\alpha^2) \normBig{\VhRzz\hJzz^* e_0}^2 \ .
\end{split}
\end{equation}
\medskip\noindent
\underline{Step 1:} We now estimate from below the three terms of~\eqref{est:xJRJx}.
For $k_1\ne 0$, the operator $\hJzz:=(\hJ_{k_1})_{2,2}$ given in~\eqref{DAE:Oseen:staircase:J} can be written as
\begin{multline} \label{hJzz}
\hJzz (k_2,\ell)
= \ip{e_{k_2}}{\hJzz e_{\ell}}
=\frac{k_1}{2}
\begin{cases}
\pm \frac{\hk}{|\hk|}\cdot\frac{\hl}{|\hl|}\ , &\text{for } \ell=k_2\pm 1\ , \\
0\ , &\text{else,}
\end{cases}
\\
\text{where }
\hk:=\begin{bmatrix} k_1 \\ k_2 \end{bmatrix}, \quad
\hl:=\begin{bmatrix} k_1 \\ \ell \end{bmatrix} ,
\end{multline}
such that $\hJzz =-\hJzz^*$ and
\begin{comment}
\begin{equation*} %\label{hJzz*}
\hJzz^* (k_2,\ell)
= \ip{e_{k_2}}{\hJzz^* e_{\ell}}
=\frac{k_1}{2}
\begin{cases}
\pm \frac{\hk}{|\hk|}\cdot\frac{\hl}{|\hl|}\ , &\text{for } \ell=k_2\mp 1\ , \\
0\ , &\text{else.}
\end{cases}
\end{equation*}
Consequently,
\begin{equation} \label{hJzzVhRzz}
\hJzz (\hRzz)^{1/2} (j,k_2)
= \ip{e_j}{\hJzz (\hRzz)^{1/2} e_{k_2}}
=\sqrt{\nu}\ \frac{k_1}{2}
\begin{cases}
\pm |k_2| \frac{\hj}{|\hj|}\cdot\frac{\hk}{|\hk|}\ , &\text{for } k_2=j\pm 1\ , \\
0\ , &\text{else,}
\end{cases}
\end{equation}
and
\end{comment}
\begin{equation*} %\label{hJzz*}
\Big((\hRzz)^{1/2} \hJzz^*\Big) (k_2,\ell)
= \ip{e_{k_2}}{(\hRzz)^{1/2} \hJzz^* e_{\ell}}
=\sqrt{\nu}\ \frac{k_1}{2}
\begin{cases}
\pm |k_2| \frac{\hk}{|\hk|}\cdot\frac{\hl}{|\hl|}\ , &\text{for } \ell=k_2\mp 1 , \\
0\ , &\text{else.}
\end{cases}
\end{equation*}
In particular, we deduce
\begin{equation}
\begin{split}
\Big(\VhRzz\hJzz^*\Big) e_0 (k_2)
=\sqrt{\nu}\ \frac{k_1}{2}
\begin{cases}
\pm \frac{k_1^2}{\sqrt{k_1^2+1}\ |k_1|}\ , &\text{for } k_2=\pm 1\ , \\
0\ , &\text{else,}
\end{cases}
\end{split}
\end{equation}
such that
\begin{subequations} \label{est:xJRJx_parts}
\begin{equation}
\normBig{\VhRzz\hJzz^* e_0}^2
=\frac{\nu}2 \frac{k_1^4}{(k_1^2+1)} \ .
\end{equation}
\medskip\noindent
Continuing with the second term of~\eqref{est:xJRJx}, and with $k_1\ne 0$, we derive using $u_0=0$ that
\begin{equation}
\begin{split}
&2\Re\Big(\ipBig{\VhRzz\hJzz^* u}{\VhRzz\hJzz^* e_0}\Big)
\\
&=-\nu \frac{k_1^2}{4} \frac{(k_1^2+2)\ k_1^2}{(k_1^2+1)\ |k_1|\ \sqrt{k_1^2+4}} 2 \big(\Re(u_{-2}) +\Re(u_2) \big)
\\
&\geq -\nu \frac{k_1^2}{4} \frac{(k_1^2+2)\ k_1^2}{(k_1^2+1)\ |k_1|\ \sqrt{k_1^2+4}} 2 \big(|\Re(u_{-2})| +|\Re(u_2)| \big) \ .
\end{split}
\end{equation}
\medskip\noindent
For the first term of~\eqref{est:xJRJx}, and with $k_1\ne 0$, we compute using $u_0=0$ that
\begin{equation} \label{est:uJRJu}
\begin{split}
\big\|\VhRzz\hJzz^* u \big\|^2
%
&=\nu \frac{k_1^2}4 \sum_{k_2=-\infty}^{\infty} \Big| -|k_2| \normalized{\hk}\cdot\normalized{\hkm} u_{k_2-1} +|k_2| \normalized{\hk}\cdot\normalized{\hkp} u_{k_2+1} \Big|^2 \\
%
&=\nu \frac{k_1^2}4 \sum_{k_2=-\infty}^{\infty} |k_2|^2\ \Big| -\normalized{\hk}\cdot\normalized{\hkm} u_{k_2-1} +\normalized{\hk}\cdot\normalized{\hkp} u_{k_2+1} \Big|^2 \\
%
&\geq \nu \frac{k_1^2}4 \sum_{k_2=-1,1} |k_2|^2\ \Big| -\normalized{\hk}\cdot\normalized{\hkm} u_{k_2-1} +\normalized{\hk}\cdot\normalized{\hkp} u_{k_2+1} \Big|^2 \\
%
%&= \nu \frac{k_1^2}4 \Big( \Big| -\normalized{\hk}\cdot\normalized{\hkm} u_{-2} \Big|^2 + \Big| \normalized{\hk}\cdot\normalized{\hkp} u_{2} \Big|^2 \Big) \\
%
&= \nu \frac{k_1^2}4 \frac{(k_1^2+2)^2}{(k_1^2+1)\ (k_1^2 +4)} \Big( |u_{-2}|^2 + |u_{2}|^2 \Big) \\
%
&\geq \nu \frac{k_1^2}4 \frac{(k_1^2+2)^2}{(k_1^2+1)\ (k_1^2 +4)} \Big( (\Re(u_{-2}))^2 + (\Re(u_{2}))^2 \Big)
\end{split}
\end{equation}
where
\[
\hk =\begin{bmatrix} k_1 \\ k_2 \end{bmatrix}, \qquad
\hkm :=\begin{bmatrix} k_1 \\ k_2-1 \end{bmatrix}, \qquad
\hkp :=\begin{bmatrix} k_1 \\ k_2+1 \end{bmatrix}. \qquad
\]
\end{subequations}
\medskip\noindent
Combining~\eqref{est:xJRJx} and~\eqref{est:xJRJx_parts} yields
\begin{equation}\label{est:xJRJx_final}
\begin{split}
&\normBig{\VhRzz\hJzz^* x}^2
\\
&\geq \alpha^2 \nu \frac{k_1^2}4 \frac{(k_1^2+2)^2}{(k_1^2+1)\ (k_1^2 +4)} \Big( (\Re(u_{-2}))^2 + (\Re(u_{2}))^2 \Big) \\
& \quad
-\alpha\sqrt{1-\alpha^2} \nu \frac{k_1^2}{2} \frac{(k_1^2+2)\ k_1^2}{(k_1^2+1)\ |k_1|\ \sqrt{k_1^2+4}} \big(|\Re(u_{-2})| +|\Re(u_2)| \big)
+(1-\alpha^2) \frac{\nu}2 \frac{k_1^4}{(k_1^2+1)}
\\
%&= \nu \frac{k_1^2}4 \Bigg( \alpha^2 \frac{(k_1^2+2)^2}{(k_1^2+1)\ (k_1^2 +4)} \Big( (\Re(u_{-2}))^2 + (\Re(u_{2}))^2 \Big)
%-\alpha\sqrt{1-\alpha^2} 2 \frac{(k_1^2+2)\ k_1^2}{(k_1^2+1)\ |k_1|\ \sqrt{k_1^2+4}} \big(|\Re(u_{-2})| +|\Re(u_2)| \big)
%+(1-\alpha^2) 2 \frac{k_1^2}{(k_1^2+1)} \Bigg) \\
&= \nu \frac{k_1^2}4 \Bigg( \alpha \frac{k_1^2+2}{\sqrt{k_1^2+1}\ \sqrt{k_1^2 +4}} |\Re(u_{-2})|
-\sqrt{1-\alpha^2} \frac{|k_1|}{\sqrt{k_1^2+1}} \Bigg)^2 \\
& \quad
+\nu \frac{k_1^2}4 \Bigg( \alpha \frac{k_1^2+2}{\sqrt{k_1^2+1}\ \sqrt{k_1^2 +4}} |\Re(u_2)|
-\sqrt{1-\alpha^2} \frac{|k_1|}{\sqrt{k_1^2+1}} \Bigg)^2 \\
&\geq \frac{\nu}4 \Big( \alpha \ a |\Re(u_{-2})| -\sqrt{1-\alpha^2} \ b \Big)^2
+\frac{\nu}4 \Big( \alpha \ a |\Re(u_2)| -\sqrt{1-\alpha^2} \ b \Big)^2 \ ,
\end{split}
\end{equation}
where
\begin{equation} \label{a:b}
a:=\frac{k_1^2+2}{\sqrt{k_1^2+1}\ \sqrt{k_1^2 +4}}, \qquad
b:=\frac{|k_1|}{\sqrt{k_1^2+1}}.
\end{equation}
Together with~\eqref{est:xRx}, for any $x\in\cD(\VhRzz)$ with $\norm{x}=1$, we obtain
\begin{multline} \label{est:kappa:0}
\sum_{j=0}^1 \normBig{\VhRzz(\hJzz^*)^j x}^2
\\
\geq \nu \alpha^2 +\frac{\nu}4 \Big( \alpha \ a |\Re(u_{-2})| -\sqrt{1-\alpha^2} \ b \Big)^2
+\frac{\nu}4 \Big( \alpha \ a |\Re(u_2)| -\sqrt{1-\alpha^2} \ b \Big)^2 \ .
\end{multline}
\medskip\noindent
\underline{Step 2:}
Finally, we show that there exists $\kappa>0$ such that~\eqref{Op:J-R:unbdd} holds with $m=1$, $\mJ=\hJzz:=\hJkzz$ and $\mR=\hRzz$.
To this end, we estimate the left-hand side of~\eqref{est:kappa:0} uniformly in $\alpha\in[0,1]$, $|\Re(u_2)|\in[0,1]$, and $k_1\in\N$:
We observe that the function
\begin{equation} \label{b}
b^2:\
[1,\infty)\to [0,\infty)\ , \quad
k_1 \mapsto \frac{k_1^2}{k_1^2+1}\ ,
\end{equation}
is monotonically increasing such that $b^2(k_1)\geq b^2(1) =1/2$ for $k_1\in[1,\infty)$.
Then, we estimate the expression on the left-hand side of~\eqref{est:kappa:0} as
\begin{equation} \label{est:kappa}
\begin{split}
& \sum_{j=0}^1 \normBig{\VhRzz(\hJzz^*)^j x}^2
\\
&\geq \nu \alpha^2 +\frac{\nu}4 \Big( \alpha \ a |\Re(u_{-2})| -\sqrt{1-\alpha^2} \ b \Big)^2
+\frac{\nu}4 \Big( \alpha \ a |\Re(u_2)| -\sqrt{1-\alpha^2} \ b \Big)^2
\\
&= \nu \Big(\alpha^2
+\frac{b^2}4 \Big( \alpha \frac{a}{b}\ |\Re(u_{-2})| -\sqrt{1-\alpha^2} \Big)^2
+\frac{b^2}4 \Big( \alpha \frac{a}{b}\ |\Re(u_2)| -\sqrt{1-\alpha^2} \Big)^2 \Big)
\\
&\geq \nu \Big(\alpha^2
+\frac18 \Big( \alpha \frac{a}{b}\ |\Re(u_{-2})| -\sqrt{1-\alpha^2} \Big)^2
+\frac18 \Big( \alpha \frac{a}{b}\ |\Re(u_2)| -\sqrt{1-\alpha^2} \Big)^2 \Big)
\\
&\geq \nu \Big(\alpha^2
+\frac18 \Big( \alpha \frac{a}{b}\ |\Re(u_2)| -\sqrt{1-\alpha^2} \Big)^2 \Big)\ .
\end{split}
\end{equation}
We continue to derive a lower bound for the function
\begin{equation} \label{h:alpha:beta}
h:\
[0,1]\times [0,1] \to [0,\infty)\ , \qquad
(\alpha,\beta) \mapsto \alpha^2
+\frac18 \Big( \alpha \beta c -\sqrt{1-\alpha^2} \Big)^2 \ ,
\end{equation}
where $c:=a/b>0$ and $\beta$ replaces $|\Re(u_2)|$.
The function
\begin{equation} \label{c2}
c^2:\
[1,\infty)\to [0,\infty)\ , \quad
k_1 \mapsto \frac{a^2}{b^2} =\frac{(k_1^2+2)^2}{k_1^2\ (k_1^2 +4)}\ ,
\end{equation}
is monotonically decreasing such that $1\leq c^2(k_1)\leq c^2(1) =9/5$ for $k_1\in[1,\infty)$.
A straightforward computation shows that~$h$ has no extremum in the interior of~$[0,1]^2$.
Therefore, the minimum of~$h=h(\alpha,\beta)$ in~\eqref{h:alpha:beta} is located at the boundary of $(\alpha,\beta)\in[0,1]\times[0,1]$:
For $\beta\in[0,1]$, we derive
\[
h(0,\beta) =\frac18 \ , \qquad
h(1,\beta) =1 +\frac18 c^2 \beta^2 \geq 1 \ .
\]
For $\alpha\in[0,1]$, we derive
\[
h(\alpha,0)
%=\alpha^2 +\frac18 (1-\alpha^2)
=\frac18 (7\alpha^2 +1)
\geq \frac18 \ ,
\]
and
\begin{equation} \label{h:alpha:1}
h(\alpha,1)
=\alpha^2 +\frac18 \Big( \alpha c -\sqrt{1-\alpha^2} \Big)^2 \ .
\end{equation}
To finish the estimate, we have to derive a lower bound for $h(\alpha,1)$, $\alpha\in[0,1]$:
We observe that $h(0,1)=1/8$ and $h(1,1)=1 +c^2/8\geq 1$.
To find local extrema, we search for $\alpha_*\in(0,1)$ such that
\[
0
=\frac{\partial h}{\partial\alpha} (\alpha_*,1)
=\underbracket{\frac{\alpha_*}4 (7 +c^2)}_{>0}
+\underbracket{\frac{c}4 \frac{2\alpha_*^2 -1}{\sqrt{1-\alpha_*^2}}}_{<0 \text{ on } (0,1/\sqrt{2})} \ , %\label{0=dhdalpha:alpha:1}
\]
or,
\[
\alpha_*^4 -\alpha_*^2 +\frac{c^2}{4c^2 +(7+c^2)^2} =0 \ .
\]
Solving for $\alpha_*^2$, we find the solutions
\[
(\alpha_*^2)_\pm
=\frac12 \Big(1\pm \sqrt{\frac{(7+c^2)^2}{4c^2 +(7+c^2)^2}}\Big)
\in (0,1) \ .
\]
Using the positive root of $(\alpha_*^2)_-$ yields
\[
h\Big(\sqrt{ (\alpha_*^2)_- },1\Big)
\geq (\alpha_*^2)_-
=\frac12 \Big(1 -\sqrt{\frac{(7+c^2)^2}{4c^2 +(7+c^2)^2}}\Big) >0 \ .
\]
\medskip
To derive a lower bound on~$h$, which is uniform w.r.t. $k_1\in[1,\infty)$, we recall that $1\leq c^2(k_1)\leq 2$ for $k_1\in[1,\infty)$ and study the function
\[
g:\
[1,2]\to [0,\infty)\ , \qquad
\gamma \mapsto \frac{(7+\gamma)^2}{4\gamma +(7+\gamma)^2}\ .
\]
The function $g=g(\gamma)$ is monotone decreasing w.r.t. $\gamma\in[1,2]$ such that $81/89=g(2) \leq g(\gamma) \leq g(1)=64/68$.
This implies that
\[
(\alpha_*^2)_- =\frac12 \Big(1 -\sqrt{g(c^2)}\Big)
\geq \frac12 \Big(1 -\sqrt{g(1)}\Big) =0.0149\ldots >1/100 \ .
\]
Altogether, we derive that
\[
h(\alpha,\beta)>1/100 \qquad\text{for } (\alpha,\beta)\in[0,1]\times[0,1] \ ,
\]
which implies the uniform lower bound $\kappa\geq \nu/100$ in~\eqref{est:kappa}.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Proof of Lemma \ref{lemma:Oseen:LMI}}\label{appendix_three}
\begin{proof}
To prove uniform coercivity (w.r.t.\ $k_1\in\Z\setminus\{0\}$) of the self-adjoint operator $\mQ_{k_1} :=-(\mA_{k_1}^H \mX_{k_1} +\mX_{k_1}\mA_{k_1})$, we use $\mX_{k_1} =\mI +\epsilon_{k_1}\mY$ in~\eqref{model:Oseen:X} and $\mA_{k_1} =(\widehat{\mJ}_{k_1})_{2,2} -\widehat{\mR}_{2,2}$ to deduce that
\[ \mQ_{k_1} =2\hRzz -\epsilon_{k_1} (\mA_{k_1}^H \mY +\mY\mA_{k_1}) \ . \]
Hence, $\mQ_{k_1}$ is the sum of the diagonal operator~$2\hRzz$, and the compact operator~$-\epsilon_{k_1} (\mA_{k_1}^H \mY +\mY\mA_{k_1})$ acting on the finite-dimensional subspace $\cH_5:=\spn\{e_{-2},e_{-1},e_0,e_1,e_2\}$.
Since $\ip{x}{\mQ_{k_1} x}=\ip{x}{2\hRzz x}\geq 18\|x\|^2$ for all $x\in(\cH_5)^\perp$, we are left to prove uniform coercivity of $\mQ_{k_1}\big|_{\cH_5}$, or equivalently, of the following $5\times 5$-matrix:
For $\alpha :=\epsilon_{k_1} k_1$, the matrices~$\mQ$ representing $\mQ_{k_1}\big|_{\cH_5}$ read
\[
\mQ
=
\begin{bmatrix}
8\nu & 0 & -\alpha\beta_2 & 0 & 0 \\
0 & -2\alpha\beta_1 +2\nu & -\alpha\nu/k_1 & 2\alpha\beta_1 & 0 \\
-\alpha\beta_2 & -\alpha\nu/k_1 & 4\alpha\beta_1 & \alpha\nu/k_1 & -\alpha\beta_2 \\
0 & 2\alpha\beta_1 & \alpha\nu/k_1 & -2\alpha\beta_1 +2\nu & 0 \\
0 & 0 & -\alpha\beta_2 & 0 & 8\nu
\end{bmatrix}
\]
where $\beta_1, \beta_2$ are functions given as
\[
\beta_1:
\Z\to\R\ , \quad
k_1 \mapsto \frac{k_1^2}{2\sqrt{k_1^2 +1}\ |k_1|} \ ,
\qquad \text{and }
%
\beta_2:
\Z\to\R\ , \quad
k_1 \mapsto \frac{k_1^2+2}{2\sqrt{k_1^2 +4}\ \sqrt{k_1^2 +1}} \ ,
\]
which satisfy
\begin{equation}\label{bounds:beta1}
\beta_{1,\min} := \frac1{2\sqrt{2}} \leq \beta_1(k_1) < \frac12
\quad
\text{for all } k_1\in\Zo \ ,
\end{equation}
and
\begin{equation}\label{bounds:beta2}
%\frac25
\beta_{2,\min} := \frac{3}{\sqrt{40}} \leq \beta_2(k_1) \leq \frac12
\quad
\text{for all } k_1\in\Zo \ .
\end{equation}
A Hermitian matrix is positive definite if and only if all of its leading principal minors are positive definite \cite{HoJo13}.
Using permutations of rows and columns, it is evident that all principal minors have to be positive definite.
Indeed, we consider other (not only the leading) principal minors to highlight restrictions on $\alpha$.
The $1\times 1$ minors are the diagonal elements of~$\mQ$.
The leading principal $1\times 1$ minor $\mQ_{1,1}=8\nu>0$ is positive, since the diffusion coefficient $\nu>0$ is positive.
The coefficient $\mQ_{2,2} =-2\alpha\beta_1 +2\nu$ is positive if and only if $\alpha \beta_1 < \nu$ for all $k_1\in\Zo$ which holds if
\begin{equation}\label{minor:Q2}
\alpha \le 2\nu
\quad
\text{for all $k_1\in\Zo$} ,
\end{equation}
due to~\eqref{bounds:beta1}.
The coefficient $\mQ_{3,3} =4\alpha\beta_1$ is positive if and only if
$0<\alpha \beta_1$ for all $k_1\in\Zo$, which holds if
\begin{equation}\label{minor:Q3}
0<\alpha \quad \text{for all $k_1\in\Zo$} ,
\end{equation}
due to~\eqref{bounds:beta1}.
The leading principal $2\times 2$ minor $\det\mQ_{\{1,2\}\times\{1,2\}}=8\nu\ (-2\alpha\beta_1 +2\nu)$ is positive, since it is the product of two diagonal elements.
The principal minor $\det\mQ_{\{2,4\}\times\{2,4\}}=(-2\alpha\beta_1 +2\nu)^2 -(2\alpha\beta_1)^2 =2\nu\ (-4\alpha\beta_1 +2\nu)>0$, if and only if $-4\alpha\beta_1 +2\nu>0$ for all $k_1\in\Zo$, which is satisfied if
\begin{equation}\label{minor:Q24}
\alpha \le\nu
\quad
\text{for all $k_1\in\Zo$} ,
\end{equation}
due to~\eqref{bounds:beta1}.
The principal minor $\det\mQ_{\{2,3\}\times\{2,3\}}=\alpha\big(8\beta_1 \nu -\alpha(8\beta_1^2 +\nu^2/k_1^2)\big)$ is positive if and only if $\alpha< 8\beta_1 \nu /(8\beta_1^2 +\nu^2/k_1^2)$ for all $k_1\in\Zo$.
This holds if
\begin{equation}\label{minor:Q23}
\alpha <\nu \frac{2\sqrt{2}}{2+\nu^2}
\quad
\text{for all $k_1\in\Zo$} ,
\end{equation}
since ${2\sqrt{2}}/(2+\nu^2) \leq 8\beta_1 /(8\beta_1^2 +\nu^2/k_1^2)$ for all $k_1\in\Zo$ due to~\eqref{bounds:beta1}.
The leading principal $3\times 3$ minor $\det\mQ_{\{1,2,3\}\times\{1,2,3\}}=2\beta_1 \beta_2^2 \alpha^3 -2(32\beta_1^2 \nu +\beta_2^2 \nu +4\nu^3 /{k_1^2})\ \alpha^2 +64\beta_1 \nu^2 \alpha$ is positive if
\[
\underbracket{\beta_1 \beta_2^2}_{=:a_3} \alpha^2 \underbracket{-\nu (32\beta_1^2 +\beta_2^2 +4\nu^2)}_{=:b_3} \alpha +\underbracket{32\beta_1 \nu^2}_{=:c_3}
>
0
\]
for all $k_1\in\Zo$.
This quadratic polynomial in~$\alpha$ has two positive roots, since $b_3<0$ and $c_3>0$.
Moreover, since $4 a_3 c_3$ is uniformly bounded away from 0, there exists a constant $\alpha_{3,\min}>0$, independent of $k_1$, which is a lower bound (uniformly in $k_1\in\Zo$) for the smaller root~$\alpha^{(3)}_{-} :=(-b_3 -\sqrt{b_3^2 -4 a_3 c_3})/(2a_3)$.
Consequently, for $\alpha\in(0,\alpha_{3,\min})$, the leading principal $3\times 3$ minor $\det\mQ_{\{1,2,3\}\times\{1,2,3\}}$ is positive for all $k_1\in\Zo$.
% \begin{equation}\label{minor:Q123}
% \alpha
% <
% \alpha_{-}^{(3)} := \frac{-b_3 -\sqrt{b_3^2 -4 a_3 c_3}}{2 a_3}
% \quad
% \text{for all $k_1\in\Zo$} .
% \end{equation}
The leading principal $4\times 4$ minor $\det\mQ_{\{1,2,3,4\}\times\{1,2,3,4\}}=8\beta_1 \beta_2^2 \nu \alpha^3 -4(64\beta_1^2 \nu^2 +\beta_2^2 \nu^2 +8\nu^4 /{k_1^2})\ \alpha^2 +128\beta_1 \nu^3 \alpha$ is positive if
\[
\underbracket{2\beta_1 \beta_2^2}_{=:a_4} \alpha^2 \underbracket{-\nu (64\beta_1^2 +\beta_2^2 +8\nu^2)}_{=:b_4} \alpha +\underbracket{32\beta_1 \nu^2}_{=:c_4}
>
0
\]
for all $k_1\in\Zo$.
This quadratic polynomial in~$\alpha$ has two positive roots, since $b_4<0$ and $c_4>0$.
Moreover, since $4 a_4 c_4$ is uniformly bounded away from 0, there exists a constant $\alpha_{4,\min}>0$, independent of $k_1$, which is a lower bound (uniformly in $k_1\in\Zo$) for the smaller root~$\alpha^{(4)}_{-} :=(-b_4 -\sqrt{b_4^2 -4 a_4 c_4})/(2a_4)$.
Consequently, for $\alpha\in(0,\alpha_{4,\min})$, the leading principal $4\times 4$ minor $\det\mQ_{\{1,2,3,4\}\times\{1,2,3,4\}}$ is positive for all $k_1\in\Zo$.
The (leading) principal $5\times 5$ minor $\det\mQ=128\beta_1 \beta_2^2 \nu^2 \alpha^3 -64(32\beta_1^2 \nu^3 +\beta_2^2 \nu^3 +4\nu^5 /{k_1^2})\ \alpha^2 +1024\beta_1 \nu^4 \alpha$ is positive if
\[
\underbracket{2\beta_1 \beta_2^2}_{=:a_5} \alpha^2 \underbracket{-\nu (32\beta_1^2 +\beta_2^2 +4\nu^2)}_{=:b_5} \alpha +\underbracket{16\beta_1 \nu^2}_{=:c_5}
>
0
\]
for all $k_1\in\Zo$.
This quadratic polynomial in~$\alpha$ has two positive roots, since $b_5<0$ and $c_5>0$.
Moreover, since $4 a_5 c_5$ is uniformly bounded away from 0, there exists a constant $\alpha_{5,\min}>0$, independent of $k_1$, which is a lower bound (uniformly in $k_1\in\Zo$) for the smaller root~$\alpha^{(5)}_{-} :=(-b_5 -\sqrt{b_5^2 -4 a_5 c_5})/(2a_5)$.
Consequently, for $\alpha\in(0,\alpha_{5,\min})$, the leading principal $5\times 5$ minor $\det\mQ$ is positive for all $k_1\in\Zo$.
Thus, we choose
\begin{equation}\label{cond:Oseen:alpha}
0 <\alpha <\min\big\{ 1/{\sqrt2},\ \nu,\ \nu\frac{2\sqrt{2}}{2+\nu^2},\, \alpha_{3,\min},\ \alpha_{4,\min},\ \alpha_{5,\min} \big\} =: \alpha_{\min},
\end{equation}
due to the restriction $|\epsilon_{k_1}|<1/{\sqrt2}$ to ensure that $\mX_{k_1}$ is positive definite and $\epsilon_{k_1}= \alpha/k_1$, and those conditions to ensure that $\mQ =\mQ_{k_1}$ is positive definite.\\
Finally we determine $\mu_{k_1}>0$ (bounded below, uniformly in $k_1\in\Zo$) such that the LMIs~\eqref{DAE:Oseen:LMI} hold.
Let $\{\lambda_1,\ \lambda_2,\ \ldots,\ \lambda_5\}$ be the eigenvalues of the positive definite Hermitian matrix~$\mQ$ arranged in increasing order.
We seek a lower bound on $\lambda_1$.
Note that the arithmetic-geometric mean inequality yields
\[
\lambda_1
=
\frac{\det\mQ}{\lambda_2 \lambda_3 \lambda_4 \lambda_5}
\geq
\big(\frac{\lambda_2 +\lambda_3 +\lambda_4 +\lambda_5}4\big)^{-4} \det{\mQ}
\geq
256 \frac{\det\mQ}{(\tr\mQ)^4} .
\]
Since $\tr\mQ =20\nu$ is independent of $k_1$, we finally obtain the bound
\begin{equation}\label{model:Oseen:lambda1}
\begin{split}
\lambda_1
&\geq
256 \frac{\det\mQ}{(\tr\mQ)^4}
\\
&=
\big( 128\beta_1 \beta_2^2 \nu^2 \alpha^3 -64(32\beta_1^2 \nu^3 +\beta_2^2 \nu^3 +4\nu^5 /{k_1^2})\ \alpha^2 +1024\beta_1 \nu^4 \alpha\big) \frac{256}{(20\nu)^4}
\\
&=
\big( 2\beta_1 \beta_2^2 \alpha^2 -(32\beta_1^2 +\beta_2^2 +4\nu^2 /{k_1^2})\nu \alpha +16\beta_1 \nu^2 \big) 64\nu^2 \alpha \frac{256}{(20\nu)^4}
%&=
% \big( 2\beta_1 \beta_2^2 \alpha^2 -(32\beta_1^2 +\beta_2^2 +4\nu^2 /{k_1^2})\nu \alpha +16\beta_1 \nu^2\big) \alpha \frac{64}{5^4 \nu^2}
\\
&>
\big( 2\beta_{1,\min} \beta_{2,\min}^2 \alpha^2 -({33}/{4} +4\nu^2)\nu \alpha +16\beta_{1,\min} \nu^2\big) \alpha \frac{64}{5^4 \nu^2}
=:\lambda_{1,\min}(\alpha) \ .
\end{split}
\end{equation}
Note that $\lambda_{1,\min}>0$ for $\alpha>0$ small enough, and it is a lower bound on $\lambda_1$, uniform in $k_1\in\Zo$.
Then, for each admissible $\alpha$ from~\eqref{cond:Oseen:alpha} such that $\lambda_{1,\min}(\alpha)>0$, the uniform estimates $\mQ\geq \lambda_{1,\min}(\alpha)\mI$ and $\mX_{k_1} \leq (1+\sqrt{2}|\epsilon_{k_1}|)\ \mI \leq 2\mI$ using $|\epsilon_{k_1}|\leq 1/\sqrt{2}$ imply that $\mQ \geq \lambda_{1,\min}\mI \geq (\lambda_{1,\min}/2) \mX_{k_1}$.
Hence, the inequalities~\eqref{DAE:Oseen:LMI} hold for some constants $\mu_{k_1}>0$ which are uniformly (w.r.t.~$k_1$) bounded from below.
One may choose, e.g., $\mu_{k_1} =\lambda_{1,\min}/4$, $k_1\ne0$.
\end{proof}
%------------arXiv version-----------------------------
\begin{comment}
\section*{Acknowledgments}
The first author (FA) was supported by the FWF-funded SFB \# F65.
The second author (AA) was partially supported by the FWF-doctoral school "Dissipation and dispersion in non-linear partial differential equations'' and the FWF-funded SFB \# F65.
The third author (VM) was supported by DFG SFB \# 910.
%-----------------------------------------------------------
\fontfamily{\rmdefault}
\fontsize{8bp}{10}
\selectfont\baselineskip=10pt
%\bibliographystyle{amsplain}
\bibliographystyle{abbrv}
\end{comment}
%------------arXiv version-----------------------------
\section*{Declarations}
\textbf{Ethical Approval}
%(applicable for both human and/ or animal studies. Ethical committees, Internal Review Boards and guidelines followed must be named. When applicable, additional headings with statements on consent to participate and consent to publish are also required)
is not applicable.
\medskip\noindent
\textbf{Authors' contributions}
%(applicable for submissions with multiple authors)
All authors wrote and reviewed the manuscript in a collaborative process.
\medskip\noindent
\textbf{Funding}
%(details of any funding received)
The first author (FA) was supported by the Austrian Science Fund (FWF) via the FWF-funded SFB \# F65.
The second author (AA) was supported by the Austrian Science Fund (FWF), partially via the FWF-doctoral school "Dissipation and dispersion in non-linear partial differential equations'' (\# W1245) and the FWF-funded SFB \# F65.
The third author (VM) was supported by Deutsche Forschungsgemeinschaft (DFG) via the DFG-funded SFB \# 910.
\medskip\noindent
\textbf{Availability of data and materials}
% (a statement on how any datasets used can be accessed)
In this research no datasets were generated or analysed.
\medskip\noindent
\textbf{Conflict of interest}
% Authors must disclose all relationships or interests that could have direct or potential influence or impart bias on the work:
The authors declare that they have no conflict of interest.
\medskip\noindent
\textbf{Acknowledgements}
For the purpose of open access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.
\end{appendices}
%% BibTeX users please use one of
%%\bibliographystyle{spbasic} % basic style, author-year citations
%\bibliographystyle{spmpsci} % mathematics and physical sciences
%%\bibliographystyle{spphys} % APS-like style for physics
%%\bibliography{} % name your BibTeX data base
\bibliography{AAM-2022-HCIndex}
\end{document}
\begin{thebibliography}{51}
\providecommand{\natexlab}[1]{#1}
\providecommand{\url}[1]{{#1}}
\providecommand{\urlprefix}{URL }
\providecommand{\doi}[1]{\url{https://doi.org/#1}}
\providecommand{\eprint}[2][]{\url{#2}}
\bibcommenthead
\bibitem[{Achleitner et~al(2015)Achleitner, Arnold, and St\"{u}rzer}]{AASt15}
Achleitner F, Arnold A, St\"{u}rzer D (2015) Large-time behavior in
non-symmetric {F}okker--{P}lanck equations. Riv Math Univ Parma (NS)
6(1):1--68
\bibitem[{Achleitner et~al(2016)Achleitner, Arnold, and Carlen}]{AAC16}
Achleitner F, Arnold A, Carlen EA (2016) On linear hypocoercive {BGK} models.
In: From particle systems to partial differential equations. {III}, Springer
Proc. Math. Stat., vol 162. Springer, Cham, p 1--37,
\doi{10.1007/978-3-319-32144-8_1}
\bibitem[{Achleitner et~al(2018)Achleitner, Arnold, and Carlen}]{AAC18}
Achleitner F, Arnold A, Carlen EA (2018) On multi-dimensional hypocoercive
{BGK} models. Kinet Relat Models 11(4):953--1009. \doi{10.3934/krm.2018038}
\bibitem[{Achleitner et~al(2019)Achleitner, Arnold, and Signorello}]{AAS19}
Achleitner F, Arnold A, Signorello B (2019) On optimal decay estimates for
{ODE}s and {PDE}s with modal decomposition. In: Stochastic dynamics out of
equilibrium, Springer Proc. Math. Stat., vol 282. Springer, Cham, p 241--264,
\doi{10.1007/978-3-030-15096-9_6}
\bibitem[{Achleitner et~al(2022)Achleitner, Arnold, and Carlen}]{AAC22b}
Achleitner F, Arnold A, Carlen EA (2022) {Construction of strict Lyapunov
functionals for semi-dissipative ODEs. The hypocoercivity index for the
large-time behavior of {ODE}s}. work in progress
\bibitem[{Achleitner et~al(2023{\natexlab{a}})Achleitner, Arnold, and
Carlen}]{AAC22}
Achleitner F, Arnold A, Carlen EA (2023{\natexlab{a}}) The hypocoercivity index
for the short time behavior of linear time-invariant {ODE} systems. J
Differential Equations 371:83--115. \doi{10.1016/j.jde.2023.06.027}
\bibitem[{Achleitner et~al(2023{\natexlab{b}})Achleitner, Arnold, and
Mehrmann}]{AAM21}
Achleitner F, Arnold A, Mehrmann V (2023{\natexlab{b}}) Hypocoercivity and
controllability in linear semi-dissipative {H}amiltonian ordinary
differential equations and differential-algebraic equations. ZAMM Z Angew
Math Mech 103(7):Paper No. e202100171, 31. \doi{10.1002/zamm.202100171}
\bibitem[{Achleitner et~al(2023{\natexlab{c}})Achleitner, Arnold, and
Mehrmann}]{AAM22b}
Achleitner F, Arnold A, Mehrmann V (2023{\natexlab{c}}) Hypocoercivity and
hypocontractivity concepts for linear dynamical systems. Electron J Linear
Algebra 39:33--61. \doi{10.13001/ela.2023.7531}
\bibitem[{Amrouche and Consiglieri(2011)}]{AmrCon11}
Amrouche C, Consiglieri L (2011) On the stationary {O}seen equations in {$\Bbb
R^3$}. Commun Math Anal 10(1):5--29
\bibitem[{Arnold and Erb(2014)}]{ArEr14}
Arnold A, Erb J (2014) Sharp entropy decay for hypocoercive and non-symmetric
{F}okker--{P}lanck equations with linear drift. arXiv preprint arXiv:14095425
\bibitem[{Arnold et~al(2021)Arnold, Dolbeault, Schmeiser, and
W\"{o}hrer}]{ArDoScWo21}
Arnold A, Dolbeault J, Schmeiser C, et~al (2021) Sharpening of decay rates in
{F}ourier based hypocoercivity methods. In: Recent advances in kinetic
equations and applications, Springer INdAM Ser., vol~48. Springer, Cham, p
1--50, \doi{10.1007/978-3-030-82946-9\_1}
\bibitem[{Arnold et~al(2022)Arnold, Schmeiser, and Signorello}]{ASS20}
Arnold A, Schmeiser C, Signorello B (2022) Propagator norm and sharp decay
estimates for {F}okker--{P}lanck equations with linear drift. Comm Math Sc
20(4):1047--1080
\bibitem[{Bae and Jin(2008)}]{BaJi08}
Bae HO, Jin BJ (2008) Estimates of the wake for the 3{D} {O}seen equations.
Discrete Contin Dyn Syst Ser B 10(1):1--18. \doi{10.3934/dcdsb.2008.10.1}
\bibitem[{Batchelor(2000)}]{Ba2000}
Batchelor G (2000) An introduction to fluid dynamics. Cambridge University
Press, Cambridge
\bibitem[{Bernstein(2018)}]{Be18}
Bernstein DS (2018) Scalar, vector, and matrix mathematics. Princeton
University Press, Princeton, NJ
\bibitem[{Bouin et~al(2020)Bouin, Dolbeault, Mischler, Mouhot, and
Schmeiser}]{BoDoMiMoSc20}
Bouin E, Dolbeault J, Mischler S, et~al (2020) Hypocoercivity without
confinement. Pure Appl Anal 2(2):203--232. \doi{10.2140/paa.2020.2.203}
\bibitem[{Brezis(2011)}]{Br11}
Brezis H (2011) Functional analysis, {S}obolev spaces and partial differential
equations. Universitext, Springer, New York
\bibitem[{Chemin et~al(2000)Chemin, Desjardins, Gallagher, and Grenier}]{CDGG}
Chemin JY, Desjardins B, Gallagher I, et~al (2000) Fluids with anisotropic
viscosity. M2AN Math Model Numer Anal 34(2):315--335.
\doi{10.1051/m2an:2000143}
\bibitem[{Chemin et~al(2006)Chemin, Desjardins, Gallagher, and
Grenier}]{CheDesGalGre06}
Chemin JY, Desjardins B, Gallagher I, et~al (2006) Mathematical geophysics: An
introduction to rotating fluids and the Navier-Stokes equations, Oxford
Lecture Series in Mathematics and its Applications, vol~32. The Clarendon
Press, Oxford University Press, Oxford
\bibitem[{Chen(1999)}]{Ch99}
Chen CT (1999) Linear system theory and design, 3rd edn. The Oxford series in
electrical and computer engineering, Oxford University Press, New York, NY
\bibitem[{Curtain and Pritchard(1978)}]{CuPr78}
Curtain R, Pritchard A (1978) Infinite dimensional linear systems theory,
Lecture Notes in Control and Information Sciences, vol~8. Springer-Verlag,
Berlin-New York
\bibitem[{Curtain and Zwart(2020)}]{CuZw20}
Curtain R, Zwart H (2020) Introduction to infinite-dimensional systems theory,
Texts in Applied Mathematics, vol~71. Springer, New York,
\doi{10.1007/978-1-0716-0590-5}
\bibitem[{Datko(1970)}]{Da70}
Datko R (1970) Extending a theorem of {A}. {M}. {L}iapunov to {H}ilbert space.
J Math Anal Appl 32:610--616. \doi{10.1016/0022-247X(70)90283-0}
\bibitem[{Dautray and Lions(2000)}]{DL3}
Dautray R, Lions J (2000) Mathematical Analysis and Numerical Methods for
Science and Technology, Vol. 3: {S}pectral Theory and Applications. Springer
\bibitem[{Dolbeault et~al(2009)Dolbeault, Mouhot, and Schmeiser}]{DoMoSc09}
Dolbeault J, Mouhot C, Schmeiser C (2009) Hypocoercivity for kinetic equations
with linear relaxation terms. C R Math Acad Sci Paris 347(9-10):511--516.
\doi{10.1016/j.crma.2009.02.025}
\bibitem[{Dolbeault et~al(2015)Dolbeault, Mouhot, and Schmeiser}]{DoMoSc15}
Dolbeault J, Mouhot C, Schmeiser C (2015) Hypocoercivity for linear kinetic
equations conserving mass. Trans Amer Math Soc 367(6):3807--3828.
\doi{10.1090/S0002-9947-2015-06012-7}
\bibitem[{Emmrich and Mehrmann(2013)}]{EM13}
Emmrich E, Mehrmann V (2013) Operator differential-algebraic equations arising
in fluid dynamics. Computational Methods in Applied Mathematics
13(4):443--470
\bibitem[{Engel and Nagel(2000)}]{EnNa00}
Engel KJ, Nagel R (2000) One-parameter semigroups for linear evolution
equations, Graduate Texts in Mathematics, vol 194. Springer-Verlag, New York
\bibitem[{Evans(2010)}]{Ev10}
Evans LC (2010) Partial differential equations, Graduate Studies in
Mathematics, vol~19, 2nd edn. American Mathematical Society, Providence, RI,
\doi{10.1090/gsm/019}
\bibitem[{Foias et~al(2001)Foias, Manley, Rosa, and Temam}]{FMRT01}
Foias C, Manley O, Rosa R, et~al (2001) Navier-{S}tokes equations and
turbulence, Encyclopedia of Mathematics and its Applications, vol~83.
Cambridge University Press, Cambridge, \doi{10.1017/CBO9780511546754}
\bibitem[{Galdi(2011)}]{Gal11}
Galdi GP (2011) An introduction to the mathematical theory of the
{N}avier-{S}tokes equations, 2nd edn. Springer Monographs in Mathematics,
Springer, New York, \doi{10.1007/978-0-387-09620-9}
\bibitem[{Grabowski(1990)}]{Gr90}
Grabowski P (1990) On the spectral-{L}yapunov approach to parametric
optimization of distributed-parameter systems. IMA J Math Control Inform
7(4):317--338. \doi{10.1093/imamci/7.4.317}
\bibitem[{Hansen and Weiss(1997)}]{HaWe97}
Hansen S, Weiss G (1997) New results on the operator {C}arleson measure
criterion. In: Townley S (ed) Distributed parameter systems: analysis,
synthesis and applications, Part 1, IMA J. Math. Control Inform., vol~14.
Oxford University Press, p 3--32, \doi{10.1093/imamci/14.1.3}
\bibitem[{Horn and Johnson(2013)}]{HoJo13}
Horn RA, Johnson CR (2013) Matrix analysis, 2nd edn. Cambridge University
Press, Cambridge
\bibitem[{John(2016)}]{Jo16}
John V (2016) Finite element methods for incompressible flow problems, Springer
Series in Computational Mathematics, vol~51. Springer, Cham,
\doi{10.1007/978-3-319-45750-5}
\bibitem[{Kailath(1980)}]{Ka80}
Kailath T (1980) Linear systems. Prentice-Hall, Inc., Englewood Cliffs, N.J.
\bibitem[{Kato(1995)}]{Kato}
Kato T (1995) Perturbation theory for linear operators. Classics in
Mathematics, Springer-Verlag, Berlin, reprint of the 1980 edition
\bibitem[{Lagerstrom(1996)}]{La1996}
Lagerstrom P (1996) Laminar flow theory. Princeton University Press, Princeton,
NJ
\bibitem[{Layton(2008)}]{Lay08}
Layton W (2008) Introduction to the Numerical Analysis of Incompressible
Viscous Flows. {SIAM} Publications, Philadelphia, PA
\bibitem[{Mehl et~al(2018)Mehl, Mehrmann, and Wojtylak}]{MehMW18}
Mehl C, Mehrmann V, Wojtylak M (2018) Linear algebra properties of dissipative
{H}amiltonian descriptor systems. SIAM J Matrix Anal Appl 39(3):1489--1519.
\doi{10.1137/18M1164275}
\bibitem[{Mikhailov(2023)}]{Mik23}
Mikhailov SE (2023) Stationary anisotropic {S}tokes, {O}seen and
{N}avier-{S}tokes systems: periodic solutions in {$\Bbb R^n$}. Math Methods
Appl Sci 46(9):10903--10928
\bibitem[{Paicu(2005{\natexlab{a}})}]{Pa}
Paicu M (2005{\natexlab{a}}) \'{E}quation anisotrope de {N}avier-{S}tokes dans
des espaces critiques. Rev Mat Iberoamericana 21(1):179--235.
\doi{10.4171/RMI/420}
\bibitem[{Paicu(2005{\natexlab{b}})}]{Pa05periodic}
Paicu M (2005{\natexlab{b}}) \'{E}quation periodique de {N}avier-{S}tokes sans
viscosit\'{e} dans une direction. Comm Partial Differential Equations
30(7-9):1107--1140. \doi{10.1080/036053005002575529}
\bibitem[{Pazy(1983)}]{Pazy}
Pazy A (1983) Semigroups of linear operators and applications to partial
differential equations. Springer
\bibitem[{Rannacher(2000)}]{Ran00}
Rannacher R (2000) Finite element methods for the incompressible
{N}avier-{S}tokes equations. In: Fundamental directions in mathematical fluid
mechanics. Adv. Math. Fluid Mech., Birkh\"{a}user, Basel, p 191--293
\bibitem[{Shinbrot(1973)}]{Shi73}
Shinbrot M (1973) Lectures on fluid mechanics. Gordon and Breach, Science
Publ., New York
\bibitem[{Temam(1995)}]{T95}
Temam R (1995) Navier-{S}tokes equations and nonlinear functional analysis,
CBMS-NSF Regional Conference Series in Applied Mathematics, vol~66, 2nd edn.
Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA,
\doi{10.1137/1.9781611970050}
\bibitem[{Temam(1997)}]{T97}
Temam R (1997) Infinite-dimensional dynamical systems in mechanics and physics,
Applied Mathematical Sciences, vol~68, 2nd edn. Springer-Verlag, New York,
\doi{10.1007/978-1-4612-0645-3}
\bibitem[{Temam and Wang(1996)}]{TW96}
Temam R, Wang X (1996) Asymptotic analysis of {O}seen type equations in a
channel at small viscosity. Indiana Univ Math J 45(3):863--916
\bibitem[{Villani(2009)}]{Vi09}
Villani C (2009) Hypocoercivity. Mem Amer Math Soc 202(950):iv+141.
\doi{10.1090/S0065-9266-09-00567-5}
\bibitem[{Wonham(1985)}]{Wo85}
Wonham WM (1985) Linear multivariable control, Applications of Mathematics (New
York), vol~10, 3rd edn. Springer-Verlag, New York,
\doi{10.1007/978-1-4612-1082-5}
\end{thebibliography}