Spectral estimates of dynamically-defined and amenable operator families
Extracted source text · arXiv 2110.05763
\pdfoutput=1
\documentclass{amsart}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\makeatletter
%\tagsleft@false
\renewcommand{\labelenumi}{\theenumi.}
\renewcommand{\theenumi}{\@arabic\c@enumi}
\renewcommand{\labelenumii}{\theenumii.}
\renewcommand{\theenumii}{\@alph\c@enumii}
\renewcommand{\p@enumii}{\theenumi.}
\renewcommand{\labelenumiii}{\theenumiii.}
\renewcommand{\theenumiii}{\@roman\c@enumiii}
\renewcommand{\p@enumiii}{\p@enumii\theenumii.}
\renewcommand{\labelenumiv}{\theenumiv.}
\renewcommand{\theenumiv}{\@Alph\c@enumiv}
\renewcommand{\p@enumiv}{\p@enumiii\theenumiii.}
\renewcommand{\labelitemi}{$\vcenter{\hbox{\(\bullet\)}}$}
\renewcommand{\labelitemii}{$\vcenter{\hbox{\scalebox{.85}{\(\bullet\)}}}$}
\renewcommand{\labelitemiii}{$\vcenter{\hbox{\scalebox{.70}{\(\bullet\)}}}$}
\renewcommand{\labelitemiv}{$\vcenter{\hbox{\scalebox{.55}{\(\bullet\)}}}$}
\let\originalitemize\itemize
\let\originalenditemize\enditemize
\renewenvironment{itemize}{%
\NEWleftmargini{10.00002pt}%
\NEWleftmarginii{9.25003pt}%
\NEWleftmarginiii{8.49998pt}%
\NEWleftmarginiv{7.75002pt}%
\originalitemize%
}{%
\originalenditemize%
\NEWleftmargini{12.7778pt}%
\NEWleftmarginii{12.7778pt}%
\NEWleftmarginiii{18.61116pt}%
\NEWleftmarginiv{15.62502pt}%
}
\newenvironment{hlist}{%
\list{}{\leftmargin=\parindent\itemindent=-\parindent\labelsep=0pt\labelwidth=0pt}%
}{%
\endlist%
}
\renewcommand{\thefootnote}{\arabic{footnote}}
\let\fboxT\fbox
\newcommand{\fboxM}[1]{\mathpalette\fboxM@{#1}}
\newcommand{\fboxM@}[2]{\fboxT{$\mathsurround=0pt#1#2$}}
\renewcommand{\fbox}[1]{\ifmmode\fboxM{#1}\else\fboxT{#1}\fi}
\newcommand{\sandwich}[1]{%
\settoheight{\dimen0}{#1}%
\ooalign{%
\clap{\rule[0pt]{200pt}{0.001pt}}\cr%
\clap{\rule[\dimen0-0.001pt]{200pt}{0.001pt}}\cr%
#1}%
}
\newcommand{\swapabovedisplayskip}{\SwapAboveDisplaySkip}
\newcommand{\tsmash}[1]{\smash[t]{#1}}
\newcommand{\bsmash}[1]{\smash[b]{#1}}
\newcommand{\nbaselineskip}{\@ifstar\nbaselineskip@@\nbaselineskip@}
\newcommand{\nbaselineskip@}[1]{\vspace{#1\baselineskip}}
\newcommand{\nbaselineskip@@}[1]{\\*[#1\baselineskip]}
\newcommand{\nvfill}[1]{\vspace{\stretch{#1}}}
\newcommand{\nhfill}[1]{\hspace{\stretch{#1}}}
\makeatother
\newcommand{\NEWOLD}[2]{%
\expandafter\newcommand\csname NEW#1\endcsname[1]{\csname #1\endcsname=##1}%
\expandafter\newcommand\csname OLD#1\endcsname{\csname #1\endcsname=#2}%
}
\NEWOLD{pretolerance}{100}\AtBeginDocument{\NEWpretolerance{-1}}
\NEWOLD{tolerance}{200}\AtBeginDocument{\NEWtolerance{100}}
\NEWOLD{emergencystretch}{0pt}
\NEWOLD{hbadness}{1000}\AtBeginDocument{\NEWhbadness{100}}
\NEWOLD{vbadness}{1000}\AtBeginDocument{\NEWvbadness{100}}
\NEWOLD{clubpenalty}{10000}
\NEWOLD{widowpenalty}{10000}
\NEWOLD{brokenpenalty}{10000}
\NEWOLD{displaywidowpenalty}{50}\AtBeginDocument{\NEWdisplaywidowpenalty{10000}}
\NEWOLD{exhyphenpenalty}{50}\AtBeginDocument{\NEWexhyphenpenalty{10000}}
\NEWOLD{relpenalty}{500}\AtBeginDocument{\NEWrelpenalty{10000}}
\NEWOLD{binoppenalty}{700}\AtBeginDocument{\NEWbinoppenalty{10000}}
\NEWOLD{fboxsep}{3pt}
\NEWOLD{fboxrule}{.4pt}\AtBeginDocument{\NEWfboxrule{0.001pt}\NEWfboxsep{-\fboxrule}}
\NEWOLD{arraycolsep}{5pt}\AtBeginDocument{\NEWarraycolsep{0pt}}
\NEWOLD{labelsep}{5pt}
\NEWOLD{parindent}{12pt}
\NEWOLD{partopsep}{0pt}
\NEWOLD{parsep}{0pt}
\NEWOLD{itemsep}{0pt}
\NEWOLD{listparindent}{0pt}
\NEWOLD{itemindent}{0pt}
\NEWOLD{leftmargini}{2.5em}\AtBeginDocument{\NEWleftmargini{12.7778pt}}
\NEWOLD{leftmarginii}{2.2em}\AtBeginDocument{\NEWleftmarginii{12.7778pt}}
\NEWOLD{leftmarginiii}{1.87em}\AtBeginDocument{\NEWleftmarginiii{18.61116pt}}
\NEWOLD{leftmarginiv}{1.7em}\AtBeginDocument{\NEWleftmarginiv{15.62502pt}}
\usepackage{geometry}\geometry{paper=letterpaper}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\let\bT\b
\let\oT\o
\let\vT\v
\let\ssT\ss
\let\HT\H
\let\LT\L
\let\OT\O
\let\PT\P
\let\ST\S
\let\McT\Mc
\providecommand{\DIFadd}[1]{{\protect\color{blue}\uwave{#1}}}
\providecommand{\DIFdel}[1]{{\protect\color{red}\sout{#1}}}
\providecommand{\DIFaddbegin}{}
\providecommand{\DIFaddend}{}
\providecommand{\DIFdelbegin}{}
\providecommand{\DIFdelend}{}
\providecommand{\DIFaddFL}[1]{\DIFadd{#1}}
\providecommand{\DIFdelFL}[1]{\DIFdel{#1}}
\providecommand{\DIFaddbeginFL}{}
\providecommand{\DIFaddendFL}{}
\providecommand{\DIFdelbeginFL}{}
\providecommand{\DIFdelendFL}{}
\usepackage{comment}
\usepackage{inputenc}
\inputencoding{utf8}
\usepackage{microtype}
\microtypesetup{final,verbose=errors}
\usepackage{mathtools}
\DeclarePairedDelimiter{\sM}{\{}{\}}
\DeclarePairedDelimiter{\pM}{(}{)}
\DeclarePairedDelimiter{\bM}{[}{]}
\DeclarePairedDelimiter{\abM}{\langle}{\rangle}
\DeclarePairedDelimiter{\vM}{\lvert}{\rvert}
\DeclarePairedDelimiter{\vvM}{\lVert}{\rVert}
\DeclareMathOperator{\supp}{supp}
\DeclareMathOperator{\dist}{dist}
\let\lim\undefined
\DeclareMathOperator*{\lim}{lim}
\let\sup\undefined
\DeclareMathOperator*{\sup}{sup}
\let\inf\undefined
\DeclareMathOperator*{\inf}{inf}
\setcounter{MaxMatrixCols}{10}
\renewcommand{\implies}{%
\mspace{5mu plus 0mu minus 5mu}\Longrightarrow\mspace{5mu plus 0mu minus 5mu}%
}
\renewcommand{\impliedby}{%
\mspace{5mu plus 0mu minus 5mu}\Longleftarrow\mspace{5mu plus 0mu minus 5mu}%
}
\renewcommand{\iff}{%
\mspace{5mu plus 0mu minus 5mu}\Longleftrightarrow\mspace{5mu plus 0mu minus 5mu}%
}
\renewcommand{\,}{%
,\mspace{3mu plus 0mu minus 3mu}%
}
\renewcommand{\:}{%
.\sbox0{$.$}\kern-1.819\ht0\rlap{\color{white}\rule{1.006\ht0}{1.006\ht0}}\kern1.819\ht0\llap{.}%
}
\renewcommand{\;}{%
.\sbox0{$.$}\kern-1.819\ht0\rlap{\color{white}\rule{1.006\ht0}{1.006\ht0}}\kern1.819\ht0\llap{,}%
}
\usepackage{mleftright}
\let\originalleft\left
\let\originalright\right
\let\left\mleft
\let\right\mright
\usepackage{tensor}
\renewcommand{\prescript}[3]{{\tensor[^{#1}_{#2}]{#3}{}}}
\renewcommand{\sideset}[3]{{\tensor*[#1]{#3\nolimits}{#2}}}
\usepackage{bm}
\usepackage{xcolor}
\usepackage{amssymb}
\DeclareMathAlphabet{\mathbbm}{U}{bbm}{m}{n}
\SetMathAlphabet\mathbbm{bold}{U}{bbm}{bx}{n}
\DeclareMathAlphabet{\mathbbmss}{U}{bbmss}{m}{n}
\SetMathAlphabet\mathbbmss{bold}{U}{bbmss}{bx}{n}
\DeclareSymbolFont{rsfs}{U}{rsfs}{m}{n}
\DeclareSymbolFontAlphabet{\mathscr}{rsfs}
\usepackage{amsthm}
%\swapnumbers
\newtheorem{prop}{Proposition}[section]
\newtheorem*{prop*}{Proposition}
\newtheorem{thm}[prop]{Theorem}
\newtheorem*{thm*}{Theorem}
\newtheorem{lem}[prop]{Lemma}
\newtheorem*{lem*}{Lemma}
\newtheorem{cor}{Corollary}[prop]
\newtheorem*{cor*}{Corollary}
\theoremstyle{remark}
\newtheorem*{rem}{Remark}
\usepackage{needspace}
\usepackage{enumerate}
\usepackage{array}
\usepackage{booktabs}
\usepackage{ulem}
\normalem
\usepackage{tikz}
\usetikzlibrary{cd}
\usepackage{graphicx}
\usepackage{trimclip}
\usepackage{pdfpages}
\usepackage{pdflscape}
\usepackage{accsupp}
\newcommand{\copypaste}[1]{%
\BeginAccSupp{method=escape,ActualText={\detokenize{#1}}}#1\EndAccSupp{}%
}
\usepackage{url}
\usepackage{bookmark}
\usepackage{hyperref}
\hypersetup{final,hidelinks,bookmarksnumbered=true,bookmarksopen=true}
\usepackage{macros}
\begin{document}
\input{_title_and_abstract}
\section{Introduction}\label{sec: Introduction}
\input{_introduction}
\clearpage
\section{Preliminaries}\label{sec: Preliminaries}
\input{_preliminaries}
\section{Proof of Main Results}\label{sec: Proof of Main Results}
\input{_proof_of_main_results}
\bookmarksetup{startatroot}
\input{_references}
\end{document}
\title[Spectral estimates of dynamically-defined and amenable operator families]{Spectral estimates of dynamically-defined and amenable operator families}
\author[S.~Beckus]{Siegfried Beckus}
\address{Institut f\"ur Mathematik, Universit\"at Potsdam, Potsdam~14476, Germany}
\email{beckus@uni-potsdam.de}
\author[A.~Takase]{Alberto Takase}
\address{Department of Mathematics, University of California at Irvine, CA~92697, USA}
\email{atakase@uci.edu}
\begin{abstract}
For dynamically-defined operator families, the Hausdorff distance of the spectra is estimated by the distance of the underlying dynamical systems while the group is amenable.
We prove that if the group has strict polynomial growth and both the group action and the coefficients are Lipschitz continuous, then the spectral estimate has a square root behavior or, equivalently, the spectrum map is $ \frac{1}{2} $-Hölder continuous.
We prove the behavior can be improved resulting in the spectrum map being Lipschitz continuous if the coefficients are locally-constant.
In 1990, the square root behavior was established for the Almost Mathieu Operator or, more generally, the quasiperiodic operators with Lipschitz continuous potentials.
Our results extend the square root behavior to a bigger class of operators such as (magnetic) discrete Schrödinger operators with finite range and with Lipschitz continuous coefficients.
\end{abstract}
\date{\today}
\thanks{We thank Anton Gorodetski for his guidance and support. We thank Jean Bellissard and Horia Cornean for inspiring discussions on Lipschitz partitions of unity and on amenability. This project was supported by the Deutsche Forschungsgemeinschaft [BE 6789/1-1 to S.B.] and the second author was partially supported by NSF grant DMS-1855541.}
\maketitle
%\clearpage
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%\clearpage
This paper is about estimating the Hausdorff distance of the spectra by the distance of the underlying dynamical systems while the group is amenable.
We start by providing two different illuminating examples that fall in the bigger class of operators under consideration.
The \textit{Almost Mathieu Operator} (AMO) has its origins in solid-state physics and the study of \mbox{electrons}.
The operator also has a connection to the quantum hall effect---a Nobel-prize-worthy \mbox{discovery} by K.~von~Klitzing, G.~Dorda, and M.~Pepper in 1980 \cite{Physics1980QuantizedHallEffect}.
For discussions of discrete Schrödinger operators and the AMO, see \cite{Physics1982QuantizedHallEffect,MP1982BellissardReviewPaper,MP1982SimonReviewPaper,MP2017DamanikReviewPaper,MP2019AMO,MP2019QPSOp}.
In 1990, J.~Avron, P.H.M.~van~Mouche, and B.~Simon \cite{MP1990[Avron][Mouche][Simon]} considered a more general operator family containing the AMO and proved that the Hausdorff distance $ \dist_{\ssm H} $ of the spectra has a square root behavior or, equivalently, the spectrum map is $ \frac{1}{2} $-Hölder continuous:
If $ v\in C_{\ssm b}(\T) $ with $ \T\defeq\R/\Z $ is real-valued and Lipschitz continuous, then there exists $ C>0 $ such that for all sufficiently close $ \alpha_{\ttm 1},\alpha_{\ttm 2}\in \R \;$
\[\ts
\smash{\dist_{\ssm H}(\overline{\bigcup_{\theta_{\ttm 1}\in\T}\sigma(H_{v,\alpha_{\ttm 1},\theta_{\ttm 1}})}
,\overline{\bigcup_{\theta_{\ttm 2}\in\T}\sigma(H_{v,\alpha_{\ttm 2},\theta_{\ttm 2}})})}
\le C\v{\alpha_{\ttm 1}-\alpha_{\ttm 2}}^{1/2}\:
\]
Here the operator family $ H_{v,\alpha,\theta} $ on $ \ell^2(\Z) \;$ for $ v\in C_{\ssm b}(\T)\, \alpha\in \R\, \theta\in \T \;$ is defined by
\[\ts
(H_{v,\alpha,\theta}\psi)(n)=\psi(n+1)+\psi(n-1)+v(\bp{T}{\alpha}{-n}\theta)\psi(n)=\sum_{m\in \Z}k_v(-n\b{\begin{subarray}{1}
\alpha\\
\theta
\end{subarray}},-n+m)\psi(m)
\]
for every $ \psi\in \ell^2(\Z)\, n\in\Z \:$
Here $ T_\alpha:\T\to\T:\theta\mapsto \lfloor-\alpha+\theta\rfloor \;$ $ \Z $~acts~on $ \R\times \T $ via $ n\b{\begin{subarray}{1}
\alpha\\
\theta
\end{subarray}}=\b{\begin{subarray}{1}
\alpha\\
\bp{T}{\alpha}{n}\theta
\end{subarray}} \;$ and
\[\ts
k_v:(\R\times \T)\times \Z\to\C:(\b{\begin{subarray}{1}
\alpha\\
\theta
\end{subarray}},m)\mapsto \de_{1}(m)+\de_{-1}(m)+v(\theta)\de_{0}(m)\:
\]
Note, $ H_{v,\alpha,\theta} $ is a bounded self-adjoint integral operator with kernel $ k_v \:$
We remark that the authors of \cite{MP1990[Avron][Mouche][Simon]} also considered the multifrequency case.
The \textit{Fibonacci Hamiltonian} (FH) has its origins in solid-state physics and the study of electrons.
The operator also has a connection to quasicrystals---a Nobel-prize-worthy discovery by D.~\mbox{Shechtman}, I.~Blech, D.~Gratias, and J.~W.~Cahn in 1982 \cite{FH1}.
This operator and related models have been extensively studied over the last decades using periodic approximations; see \cite{FH5,FH4,FH3,FH2,MP2017DamanikReviewPaper}.
In 2019, S.~Beckus, J.~Bellissard, and H.~Cornean \cite{MP2019SpectralLipschitzContinuity} considered a more general operator family containing the FH and proved that the Hausdorff distance $ \dist_{\ssm H} $ of the spectra has a linear behavior or, equivalently, the spectrum map is Lipschitz continuous:
If $ \Ac $ is a nonempty finite set and $ \cramped{\Ac^{\Z^d}} $ is endowed with a specific metric and $ v\in C_{\ssm b}(\cramped{\Ac^{\Z^d}}) $ is real-valued and locally-constant, then there exists $ C>0 $ such that for all invariant nonempty closed subsets $ X_{\ttm 1},X_{\ttm 2} $ of $ \cramped{\Ac^{\Z^d}} \;$
\[\ts
\smash{\dist_{\ssm H}(\overline{\bigcup_{x_{\ttm 1}\in X_{\ttm 1}}\sigma(H_{v,x_{\ttm 1}})},\overline{\bigcup_{x_{\ttm 2}\in X_{\ttm 2}}\sigma(H_{v,x_{\ttm 2}})})}
\le Cd_{\ssm H}\p{X_{\ttm 1},X_{\ttm 2}}\:
\]
Here the operator family $ H_{v,x} $ on $ \ell^2(\Z^d) \;$ for $ v\in C_{\ssm b}(\cramped{\Ac^{\Z^d}})\, x\in \cramped{\Ac^{\Z^d}} \;$ is defined by
\[\ts
(H_{v,x}\psi)(n)=\bigp{\sum_{\smash{\vM{j}}=1}\psi(n+j)}+v(-nx)\psi(n)=\sum_{m\in \Z^d}k_v(-n x,-n+m)\psi(m)
\]
for every $ \psi\in \ell^2(\Z^d)\, n\in\Z^{d} \:$
Here $ \Z^d $~acts~on $ \cramped{\Ac^{\Z^d}} $ via $ nx= x\circ-n $ and
\[\ts
k_v:\cramped{\Ac^{\Z^d}}\times \Z^d\to\C:(x,m)\mapsto \bigp{\sum_{\smash{\vM{j}}=1}\de_{j}(m)}+v(x)\de_{0}(m)\:
\]
Note, $ H_{v,x} $ is a bounded self-adjoint integral operator with kernel $ k_v \:$
We remark that the authors of \cite{MP2019SpectralLipschitzContinuity} also considered the infinite range case and a general compact alphabet $ \Ac \:$
The two different illuminating examples above can be unified to have spectral estimates in terms of the underlying dynamics.
Indeed, endow $ \R\times\T $ with a metric via
\[\ts d(\b{\begin{subarray}{1}
\alpha_{\ttm 1}\\
\theta_{\ttm 1}
\end{subarray}},\b{\begin{subarray}{1}
\alpha_{\ttm 2}\\
\theta_{\ttm 2}
\end{subarray}})=\v{\alpha_{\ttm 1}-\alpha_{\ttm 2}}_{\R}\join \v{\theta_{\ttm 1}-\theta_{\ttm 2}}_\T\: \]
Then $ \v{\alpha_{\ttm 1}-\alpha_{\ttm 2}}=d_{\ssm H}(\s{\alpha_{\ttm 1}}\times\T,\s{\alpha_{\ttm 2}}\times\T) $ and $ \s{\alpha}\times\T $ is an invariant nonempty closed subset of $ \R\times \T \:$
Here $ d_{\ssm H} $ is the Hausdorff distance induced by the metric on $ \R\times \T \:$
While investigating the unified setting it became natural to consider a much more general operator family: the \textit{dynamically-defined operator family} $ A_x $ on $ L^2(G,\lambda) $ indexed by $ x\in Z $.
This is the main object of study in this paper.
Specifically, let $ (Z,G) $ be a topological dynamical system, let $ k:Z\times G\to \C $ be a bounded continuous (on $ Z $) map, and let $ A_x $ be the integral operator on $ L^2(G,\lambda) \;$ for $ x\in Z \;$ defined by
\[\ts
(A_x\psi)(g)=\int k(\inv{g}x,\inv{g}h)\psi(h)dh
\]
for every $ \psi\in L^2(G,\lambda)\, g\in G \:$
Define, for $ x\in Z \;$ $ \mathrm{orb}(x)\defeq \s{gx:g\in G} \:$
Then we prove estimates, for $ x_{\ttm 1},x_{\ttm 2}\in Z \;$ of the form
\[\ts
\dist_{\ssm H}(\sigma(A_{x_{\ttm 1}}),\sigma(A_{x_{\ttm 2}}))
\le C d_{\ssm H}(\overline{\mathrm{orb}(x_{\ttm 1})},\overline{\mathrm{orb}(x_{\ttm 2})})^{\alpha}
\]
or, more generally, for invariant nonempty closed subsets $ X_{\ttm 1},X_{\ttm 2} $ of $ Z \;$
\[\ts
\smash{\dist_{\ssm H}(\overline{\bigcup_{x_{\ttm 1}\in X_{\ttm 1}}\sigma(A_{x_{\ttm 1}})},\overline{\bigcup_{x_{\ttm 2}\in X_{\ttm 2}}\sigma(A_{x_{\ttm 2}})})}
\le C d_{\ssm H}(X_{\ttm 1},X_{\ttm 2})^{\alpha}
\]
while $ G $ is amenable, $ A_x $ is normal for every $ x\in Z \;$ and suitable regularity assumptions are placed on both the action $ G\curvearrowright Z $ and the kernel $ k:Z\times G\to\C $ which also satisfies
\[\ts
\begin{cases}
\vv{k_{Z}}_{\infty,1}\defeq \sup_{x\in Z}\int \sup_{g\in G}\v{k(\inv{g}x,h)} dh<+\infty&\iph\text{\(G\) is amenable}\;\\
\vv{k_{Z}}_S\defeq \sup_{x\in Z}\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h} dh<+\infty&\iph\text{\(G\) has strict polynomial growth}\:
\end{cases}
\]
Here $ G $ is a unimodular second-countable locally-compact Hausdorff group endowed with a Haar measure $ \lambda $ and $ Z $ is a metric space.
If $ G $ has strict polynomial growth with respect to a proper continuous left invariant metric, then $ \v{\cdot} $ is the induced norm on $ G \:$
Note, strict polynomial growth implies amenability; see \textsc{subsection~\ref{subsec: amenability}} and Lemma~\ref{lem: (5)} for details.
Also, $ \vv{k_Z}_{S}<+\infty $ implies $ \vv{k_Z}_{\infty,1}<+\infty \:$
Indeed, $ k $ is bounded and $ \s{g\in G:\v{g}\le r} $ is compact for every $ r\ge 0 $.
It routinely follows that dynamically-defined operator families satisfy a covariance relation, are uniformly bounded, and are strongly continuous; see \textsc{subsection~\ref{subsec: DDOF}} for details.
%We remark that $ k $ being continuous on $ Z $ is what guarantees the strong continuity property.
Corollary~\ref{cor: 4.1} along with the covariance relation, the strong continuity property, and assuming that $ A_x $ is normal for every $ x\in Z \;$ implies, by standard arguments, \[\ts \sigma(A_x)=\overline{\bigcup_{y\in X}\sigma(A_y)}\; \] where $ X\defeq \overline{\mathrm{orb}(x)} \:$
We state the main theorem combining the main results in this paper.
\begin{thm}\label{thm: M}
Let $ \Ss $ be the collection of all invariant nonempty closed subsets of $ Z \:$
Let $ \Ks $ be the collection of all nonempty compact subsets of $ \C \:$
Define the spectrum map to be
\[\ts
\bar{\sigma}:(\Ss,d_{\ssm H})\to (\Ks,\dist_{\ssm H}):X\mapsto\overline{\bigcup_{x\in X}\sigma(A_x)}\:
\]
Assume $ G $ is unimodular and $ A_x $ is normal for every $ x\in Z \:$
\begin{enumerate}[(a)]
\item
If $ G $ is amenable, $ A_x $ has a compact range, and $ \vv{k_Z}_{\infty,1}<+\infty \;$ then $ \bar{\sigma} $ is continuous.
\item
If $ G $ has strict polynomial growth, $ \vv{k_Z}_{S}<+\infty \;$ $ G\curvearrowright Z $ is Lipschitz continuous on $ Z \;$ and $ k:Z\times G\to \C $ is Lipschitz continuous on $ Z \;$ then $ \bar{\sigma} $ is $ \frac{1}{2} $-Hölder continuous.
\item
If $ G $ has strict polynomial growth, $ \vv{k_Z}_{S}<+\infty \;$ $ G\curvearrowright Z $ is Lipschitz continuous on $ Z \;$ and $ k:Z\times G\to \C $ is locally-constant on $ Z \;$ then $ \bar{\sigma} $ is Lipschitz continuous.
\end{enumerate}
\begin{comment}
If $ G $ has strict polynomial growth, $ \vv{k_Z}_{S}<+\infty \;$ $ G\curvearrowright Z $ is Lipschitz continuous on $ Z \;$ and $ k:Z\times G\to \C $ is Lipschitz continuous on $ Z \;$ then $ \bar{\sigma} $ is $ \frac{1}{2} $-Hölder continuous.
Furthermore, if $ k:Z\times G\to \C $ is instead locally-constant on $ Z \;$ then $ \bar{\sigma} $ is instead Lipschitz continuous.
\end{comment}
\end{thm}
We elaborate on the claim that the two different illuminating examples above fall in the bigger class of operators under consideration.
For the AMO, the group $ G=\Z $ acts on the metric space $ Z=\R\times \T \:$
For the FH, the group $ G=\Z^d $ acts on the metric space $ Z=\cramped{\Ac^{\Z^d}} \:$
For both the AMO and the FH, the Haar measure is the counting measure.
Let $ \Rc_x \;$ for $ x\in Z \;$ be the support of the map $ h\mapsto \sup_{g\in G}\v{k(\inv{g}x,h)} \:$
Say the operator family has a \textit{compact range} (or a \textit{finite range} if $ G $ is discrete) if $ \overline{\bigcup_{x\in Z}\Rc_x} $ is a compact subset of $ G \:$
For both the AMO and the FH, the operator family has a finite range.
Indeed, the range for both examples is $ \s{g\in G:\v{g}\le 1} \:$
%It should be noted that $ A_x $ having a compact range helps with making sure both $ \vv{k_Z}_{\infty,1} $ and $ \vv{k_Z}_S $ are finite which is the case in both examples.
We explain the regularity assumptions that are placed on the group action and the kernel.
Specifically, $ G\curvearrowright Z $ is \textit{Lipschitz continuous} on $ Z $ if there exists $ m>0 $ such that for each $ g\in G $ and for all $ x,y\in Z \;$
\[\ts d(gx,gy)\le (m\v{g}+1)d(x,y)\:\]
Moving onto the kernel, $ k:Z\times G\to \C $ is \textit{Lipschitz continuous} on $ Z $ if there exists $ M\in L^1(G) $ such that for $ \lambda $-a.e.\ $ h\in G \;$ $ M(h)\ge 0 $ and for all $ x,y\in Z \;$
\[\ts\v{k(x,h)-k(y,h)}\le M(h)d(x,y)\:\]
Moreover, $ k:Z\times G\to \C $ is \textit{locally-constant} on $ Z $ if there exists $ M\in L^\infty(G) $ such that for $ \lambda $-a.e.\ $ h\in G \;$ $ M(h)\ge 1 $ and for all $ x,y\in Z \;$
\[\ts d(x,y)\le \frac{1}{M(h)}\lRightarrow k(x,h)=k(y,h)\:\]
The first example above involves a group action that is Lipschitz continuous on $ Z \:$
Indeed, for the AMO, the group $ \Z $ acts on the metric space $ Z=\R\times \T \:$
Furthermore, the metric $ d $ on $ Z $ is defined via
$ d(\b{\begin{subarray}{1}
\alpha_{\ttm 1}\\
\theta_{\ttm 1}
\end{subarray}},\b{\begin{subarray}{1}
\alpha_{\ttm 2}\\
\theta_{\ttm 2}
\end{subarray}})=\v{\alpha_{\ttm 1}-\alpha_{\ttm 2}}_{\R}\join \v{\theta_{\ttm 1}-\theta_{\ttm 2}}_\T\: $
Then $ \Z $ acts Lipschitz continuously on $ Z $ since for each $ n\in \Z $ and for all $ \b{\begin{subarray}{1}
\alpha_{\ttm 1}\\
\theta_{\ttm 1}
\end{subarray}},\b{\begin{subarray}{1}
\alpha_{\ttm 2}\\
\theta_{\ttm 2}
\end{subarray}}\in Z \;$ $ d(n\b{\begin{subarray}{1}
\alpha_{\ttm 1}\\
\theta_{\ttm 1}
\end{subarray}},n\b{\begin{subarray}{1}
\alpha_{\ttm 2}\\
\theta_{\ttm 2}
\end{subarray}})\le (\v{n}+1)d(\b{\begin{subarray}{1}
\alpha_{\ttm 1}\\
\theta_{\ttm 1}
\end{subarray}},\b{\begin{subarray}{1}
\alpha_{\ttm 2}\\
\theta_{\ttm 2}
\end{subarray}}) \:$
The second example above involves a group action that is Lipschitz continuous on $ Z \:$
Indeed, for the FH, the group $ \Z^d $ acts on the metric space $ Z=\cramped{\Ac^{\Z^d} }\:$
Furthermore, the metric $ d $ on $ Z $ is defined via
$ d(x_{\ttm 1},x_{\ttm 2})= \inf\s{\smash[b]{\frac{1}{r+1}}:r\in[0,+\infty)\, \smash[b]{x_{\ttm 1}|_{B(r)}}=\smash[b]{x_{\ttm 2}|_{B(r)}}}\:$
Then $ \Z^d $ acts Lipschitz continuously on $ Z $ since for each $ n\in \Z^{d} $ and for all $ x_{\ttm 1},x_{\ttm2}\in Z \;$ $ d(nx_{\ttm 1},n x_{\ttm 2})\le (\v{n}+1)d(x_{\ttm 1},x_{\ttm 2}) \:$
We conclude this section with a few words on Theorem~\ref{thm: M}, related mathematical results, and prospects for the future.
With regard to Theorem~\ref{thm: M}~(a), dynamical-and-spectral continuity has been known in the larger realm of amenable groupoids \cite{MP2018SpecrtalContinuityI}.
Our contribution is a new proof not using $C^*$-algebraic techniques.
To explain why the continuous behavior of the spectra has been studied with various models, there are cases where one has a sequence of approximations and thus a sequence of convergent spectra.
This has been useful to analyze the spectral theory of operators---one only needs suitable approximations.
For example, rational approximations provide a deeper understanding of the AMO and related quasiperiodic models \cite{MP1990[Avron][Mouche][Simon],almostLipschitz,JM12,MP2017DamanikReviewPaper}.
Similarly, periodic approximations provide a deeper understanding of the FH and related Sturmian models \cite{FH5,FH4,FH3,FH2,MP2017DamanikReviewPaper}.
Essentially, approximations are a practical tool to compute the spectrum \cite{Hof76,OK85,BS91}; see \cite{MP2018SpecrtalContinuityI} for a detailed discussion.
We remark that the continuous behavior of the spectra has applications beyond periodic approximation \cite{KP19}.
This explains the interest in dynamical-and-spectral continuity for almost-periodic Schrödinger operators \cite{Ell82} and for magnetic Schrödinger operators \cite{MP05,CP12,CP15,PR16}.
We mentioned that Theorem~\ref{thm: M}~(a) has been known in the larger realm of amenable groupoids \cite{MP2018SpecrtalContinuityI}.
The notion of `continuous fields of $C^*$-algebras' \cite{Kap51,DD63,Dix69} proved to be successful to obtain dynamical-and-spectral continuity.
It has been used in various instances to prove convergence of the spectra, but the systematic study of the spectrum map with respect to the dynamics is recent \cite{BB16,Bec16,MP2018SpecrtalContinuityI,MP2019SpectralLipschitzContinuity}.
%By imposing regularity assumptions on the operator coefficients, or more specficially, the kernel, we obtain insight about the regularity of the spectrum map, and we obtain insight about the convergence rate of the spectra.
With regard to Theorem~\ref{thm: M}~(b)~and~(c), quantitative dynamical-and-spectral continuity has been known in certain models, e.g., the AMO and the FH \cite{HS87,MP1990[Avron][Mouche][Simon],Bel94,CP12,CP15,MP2019SpectralLipschitzContinuity}.
Our contribution is an extension and unification of the results within \cite{MP1990[Avron][Mouche][Simon],MP2019SpectralLipschitzContinuity} to a bigger class of dynamically-defined operators, and it is done with spectral theory techniques, e.g., using approximate eigenfunctions---the same technique can be found within \cite{MP1990[Avron][Mouche][Simon]}.
Another method using the resolvent instead of approximate eigenfunctions can be found within \cite{CP12,CP15,MP2019SpectralLipschitzContinuity}.
A third method using exponential localization for Diophantine frequencies along with analysis of generalized eigenfunctions can be found within \cite{almostLipschitz} for the AMO.
Interestingly, a $C^*$-algebraic technique which uses commuting unitaries implied that the spectrum map for the AMO is $ \frac{1}{2} $-Hölder continuous \cite{GMS20}.
Another contribution of this paper is the connection between amenability and the regularity of the spectrum map such as the square root behavior when the group has strict polynomial growth and both the group action and the kernel is Lipschitz continuous.
Remarkably, the behavior is independent of the dimension although the constant in front is dependent.
Furthermore, the behavior can be improved resulting in the spectrum map being Lipschitz continuous if the kernel is locally-constant.
Proving regularity of the spectrum map has led to spectral properties being preserved in the limit such as openness of spectral gaps and the Lebesgue measure \cite{onethird,MP1990[Avron][Mouche][Simon],almostLipschitz}.
With regard to the future, a thorough study on the existence of periodic approximations for general dynamical systems is the natural next step.
Indeed, periodic approximations turn out to not only be practical, as in \cite{Hof76,OK85,BS91}, but also efficient since there are cases where they converge exponentially fast \cite{expfast}.
Papers in this direction include \cite{BBdN20} for one-dimensional systems and \cite{Bec16} for certain substitution multi-dimensional systems.
In the larger realm of amenable groupoids, where dynamical systems can be described via groupoids but not vice-versa, the continuity of the spectrum map has been shown in \cite{MP2018SpecrtalContinuityI}.
Systems in this larger realm include discrete Schrödinger operators on graphs or Delone sets such as the Penrose tiling or the Octogonal tiling.
They are of particular interest in the area of aperiodic order and so the question arises: how exactly does one find suitable approximations?
Can one obtain quantitative continuity of the spectrum map in this larger realm?
One of the obstacles is that the operators are not defined on a common Hilbert space.
In \textsc{section~\ref{sec: Preliminaries}}, we state the preliminaries.
In \textsc{section~\ref{sec: Proof of Main Results}}, we prove the main results.
\begin{comment}
\begin{itemize}
\item Counter-example of spectral convergence as $ y\to x $
\item three epsilon compact range
\item We remark that the square root behavior is known to be optimal since the spectrum map for the critical AMO.
Conjecture
\begin{itemize}
\item groupoids?
\item Penrose Tiling
\item Octogonal Tiling
\item Maryland Model?
\item Extended Harper Model?
\item Magnetic Model?
\item Frame Model? (see Baake and Grimm)
\item Odometer?
\item Almost periodic
\begin{itemize}
\item periodic
\item q-periodic
\item l-periodic
\end{itemize}
\item coupling and perturbation
\item Countable Groups and Schrödinger-type operators with respect to Cayley graphs
\item Fractional Laplacian?
\item Nondiscrete Groups and Delone sets (see Solomak)
\end{itemize}
\end{itemize}
\end{comment}
%\clearpage
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%\clearpage
\subsection{Unimodularity}\label{subsec: unimodularity}
In this subsection, we provide a short exposition of unimodular groups;~see~\cite{GR2015[Folland]} for a textbook exposition.
Let $ \ab{G,\lambda} $ be a second-countable locally-compact Hausdorff group endowed with a left Haar measure.
Let $ \Bs $ be the collection of all Borel subsets of $ G \:$
Define, for $ g\in G\, \Sc\in\Bs \;$ $\lambda^g(\Sc)\defeq\lambda(\Sc g)\: $
Note, for $ g\in G \;$ $ \lambda^g $ is a left Haar measure and there exists $ c(g)=c>0 $ such that $ \lambda^g=c\lambda \:$
Define
\[\ts D:G\to (0,+\infty):g\mapsto c(g) \: \]
Here $ D $ is called the \textit{modular function} of $ G \;$ and it can be shown that $ D $ is a continuous \mbox{homomorphism}.
Say $ \ab{G,\lambda} $ is \textit{unimodular} if $ D\equiv 1 \;$ or, equivalently, $ \lambda^g=\lambda $ for every $ g\in G \:$
Define, for $ \Sc\in\Bs \;$ $ \lambda^*\p{\Sc}\defeq \lambda(\inv{\Sc})\: $
It can be shown that if $ \ab{G,\lambda} $ is unimodular, then $ \lambda^*=\lambda \:$
Note, if $ G $ is discrete or abelian, then $ \ab{G,\lambda} $ is unimodular.
Observe if $ G/[G,G] $ is compact, then $ \ab{G,\lambda} $ is unimodular.
Indeed, the image of $ [G,G]\defeq \s{[x,y]:x,y\in G} $ under $ D $ is trivial since $ D $ is a homomorphism and $ D(G)=D(G/[G,G]) $ is compact since $ D $ is continuous and the only compact subgroup of $ (0,+\infty ) $ is $ \s{1} \:$
As a result, if $ G $ is compact, then $ \ab{G,\lambda} $ is unimodular.
Other examples can be found within \cite{GR2015[Folland]} such as connected nilpotent Lie groups, e.g., the Heisenberg group.
Henceforth, if $ \ab{G,\lambda} $ is unimodular, then we write $ dh=d\lambda(h) $ for every $ h\in G \:$
\subsection{Amenability}\label{subsec: amenability}
In this subsection, we provide a short exposition of amenable groups; see \cite{AG} for a textbook exposition.
Let $ \ab{G,\lambda} $ be a second-countable locally-compact Hausdorff group endowed with a left Haar measure.
Say $ \ab{G,\lambda} $ is \textit{amenable} if for each $ \eta>0 $ and for each compact subset $\Fc$ of $ G \;$ there exists a compactly-supported continuous nonzero $ \chi \in L^2(G) $ such that for each $ h\in \Fc\;$
\[\ts \vv{\cramped{\chi^h-\chi}}\le \eta\vv{\chi}\: \]
Here $ \cramped{\chi^h}:G\to \C:g\mapsto\chi(gh) \:$
Note, if $ G $ is compact, then $ \ab{G,\lambda} $ is amenable.
It can be shown that if $ G $ is solvable, then $ \ab{G,\lambda} $ is amenable.
As a result, if $ G $ is abelian, then $ \ab{G,\lambda} $ is amenable.
The free group $ \F_2 $ with two generators is an example of a group that is not amenable.
%The class of all unimodular groups contains the compact groups, the abelian groups, and the discrete groups.
%The class of all amenable groups contains the compact groups and the abelian groups.
%A metric is said to be compatible to a topological space $ Z $ if it induces the same topology on $ Z \:$
Because $G$ is second-countable, locally-compact, and Hausdorff, there exists a proper continuous metric $d$ on $G$ that is left invariant, i.e., for all $j,g,h\in G\;$ $ d(jg,jh) = d(g,h)\: $
Say $ \ab{G,\lambda} $ has \textit{strict polynomial growth} if there exist $ b\ge 1\, c_1>c_0>0 $ such that for each $ r\ge 1 \;$
\[\ts c_0r^b\le\lambda(B(r))\le c_1r^b \:\]
Here $ B(r) $ is the open ball centered at the identity element of $ G $ with radius $ r \:$
We remark that this notion of growth can be found within \cite{DS2006[Nevo]}, and it can be shown that it implies $ \ab{G,\lambda} $ is amenable.
Henceforth, if $ G $ has a proper continuous left invariant metric, then we write $ \v{\inv{g}h}=d(g,h) $ for every $ g,h\in G \:$
\subsection{Hausdorff Distance}\label{subsec: hausdorff distance}
Let $ Z $ be a metric space with metric $ d \:$
Let $ S_{\ttm 1} $ and $ S_{\ttm 2} $ be subsets of $ Z \:$
The \textit{Hausdorff distance} between $ S_{\ttm 1} $ and $ S_{\ttm 2} $ induced by $ d $ is
\[\ts \max\s{\sup_{s_{\ttm 1}\in S_{\ttm 1}}d(s_{\ttm 1},S_{\ttm 2}),\sup_{\smash{s_{\ttm 2}}\in \smash{S_{\ttm 2}}}d(s_{\ttm 2},S_{\ttm 1})}\eqdef d_{\mathsf{H}}(S_{\ttm 1},S_{\ttm 2}) \:\]
If $ (Z,G) $ is a topological dynamical system and $ Z $ is a compact metric space, then it can be shown that the collection $ \Ss $ of all invariant nonempty closed subsets of $ Z $ endowed with the Hausdorff metric $ d_{\ssm H} $ is a compact metric space; this result can be found within \cite{MP2018SpecrtalContinuityI}.
%\clearpage
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%\clearpage
\subsection{Dynamically-Defined Operator Families}\label{subsec: DDOF}
In this subsection, basic facts are recovered.
Specifically, we prove a covariance relation, a uniform boundedness, and a strong continuity property for dynamically-defined operator families and their adjoints.
Let $ \ab{G,\lambda} $ be a second-countable locally-compact Hausdorff group endowed with a left Haar measure, where $ \ab{G,\lambda} $ is unimodular.
Let $ \ab{Z,G} $ be a topological space endowed with a left \mbox{continuous} action.
Let $ k $ be a bounded continuous (on $ Z $) complex-valued function on $ Z\times G \;$ where $ \sup_{x\in Z}\int \sup_{g\in G}\v{k(\inv{g}x,h)} dh<+\infty \:$
Define $ A_x:L^2(G)\to L^2(G) \;$ for $ x\in Z \;$ by
\[\ts
(A_x\psi)(g)=\int k(\inv{g}x,\inv{g}h)\psi(h) dh
\]
for every $ \psi\in L^2(G)\, g\in G \:$
\begin{lem}\label{lem: (1)}
Define $ \smash{U_j}:L^2(G)\to L^2(G) \;$ for $ j\in G \;$ by
\[\ts (\smash{U_j}\psi)(g)=\psi(\inv{j}g) \]
for every $ \psi\in L^2(G)\, g\in G \:$
Fix $ x\in Z \:$
Then $ A_x $ is linear and for each $ j\in G \;$
\[\ts \smash{U_{j}}A_x\smash{U_{\inv{j}}}=A_{jx} \]
and for each $ \vp\in L^2(G) \;$
\[\ts \vv{A_x\vp}\le \p{\int \sup_{g\in \supp\p{\vp}\inv{h}}\v{k(\inv{g}x,h)} dh}\vv{\vp} \]
and for each $ y\in Z \;$
\[\ts \vv{(A_y-A_x)\vp}\le \p{\int \sup_{g\in \supp\p{\vp}\inv{h}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh}\vv{\vp}\: \]
\end{lem}
\begin{proof}
\allowdisplaybreaks%
Observe $ A_x $ is linear.
Also, for $ j\in G\, \psi\in L^2(G)\, g\in G \;$
\begin{align*}
&\ts({{U}_{j}A_x{U}_{\inv{j}}}\psi)(g)\\
&\ts=\int k(\inv{(\inv{j}g)}x,\inv{(\inv{j}g)}h)\psi(jh) dh\\
&\ts=\int k(\inv{g}jx,\inv{g}h)\psi(h) dh\\
&\ts=(A_{jx}\psi)(g)\:
\end{align*}
Also, for $ \vp\in L^2(G) \;$
\begin{align*}
&\ts\vv{A_x\vp}^2\\
&\ts=\int\v{\int k(\inv{g}x,h)\vp(gh) dh}^2dg\\
&\ts\le\int\p{\int \v{k(\inv{g}x,h)}\v{\vp(gh)} dh}^2dg\\
&\ts\le\int\p{\int \v{k(\inv{g}x,h)}\1_{\supp\p{\vp}}(gh) dh}\p{\int \v{k(\inv{g}x,h)}\v{\vp(gh)}^2 dh}dg
\tag*{\scalebox{1}{\(\mathllap{\smash{\p{\lsubstack{\textrm{Hölder's}\\\textrm{inequality}}}}}\)}}\\
&\ts\le\int \p{\int \sup_{\gch\in \supp\p{\vp}\inv{h}}\v{k(\inv{\gch}x,h)} dh}\p{\int \sup_{\check{g}\in \supp\p{\vp}\inv{h}}\v{k(\inv{\check{g}}x,h)}\v{\vp(gh)}^2 dh}dg\\
&\ts= \p{\int \sup_{g\in \supp\p{\vp}\inv{h}}\v{k(\inv{g}x,h)} dh}^2\vv{\vp}^2\:
\end{align*}
Also, for $ \vp\in L^2(G)\, y\in Z \;$
\begin{align*}
&\ts\vv{(A_{y}-A_{x})\vp}^2\\
&\ts=\int\v{\int (k(\inv{g}y,h)-k(\inv{g}x,h))\vp(gh) dh}^2dg\\
&\ts\le\int\p{\int \v{k(\inv{g}y,h)-k(\inv{g}x,h)}\v{\vp(gh)} dh}^2dg\\
&\ts\le\int\p{\int \v{k(\inv{g}y,h)-k(\inv{g}x,h)}\1_{\supp\p{\vp}}(gh) dh}\p{\int \v{k(\inv{g}y,h)-k(\inv{g}x,h)}\v{\vp(gh)}^2 dh}dg\\
%\tag*{\scalebox{1}{\(\mathllap{\smash{\p{\lsubstack{\textrm{Hölder's}\\\textrm{inequality}}}}}\)}}\\
&\ts\le\int
\!\begin{array}[t]{l}
\p{\int \sup_{\check{g}\in \supp\p{\vp}\cramped{\inv{h}}}\v{k(\inv{\check{g}}y,h)-k(\inv{\check{g}}x,h)} dh}\\
{}\cdot\p{\int \sup_{\check{g}\in\supp\p{\vp}\cramped{\inv{h}}}\v{k(\inv{\check{g}}y,h)-k(\inv{\check{g}}x,h)}\v{\vp(gh)}^2 dh}dg
\end{array}\\
&\ts=\p{\int \sup_{g\in \supp\p{\vp}\inv{h}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh}^2\vv{\vp}^2\:\qedhere
\end{align*}
\end{proof}
\begin{lem}
Define $ a_x:G\times G\to \C \;$ for $ x\in Z \;$ by
\[\ts a_x(g,h)=k(\inv{g}x,\inv{g}h) \]
for every $ g,h\in G \:$
Assume $ k $ is continuous on $ Z\times G \:$
Fix $ x\in Z \:$
Then $ a_x $ is continuous and
\begin{align*}
\ts A_x=A_x\mspace{-3mu}{}^* &\ts\iff \p{\f g,h\in G}\b{a_x(g,h)=\overline{a_x(h,g)}\mspace{1.5mu}}\;\\
\ts A_xA_x\mspace{-3mu}{}^*=A_x\mspace{-3mu}{}^*A_x &\ts\iff \p{\f g,h\in G}\b{\int a_x(g,j)\overline{a_x(h,j)} dj=\int \overline{a_x(j,g)}a_x(j,h) dj}\:
\end{align*}
\end{lem}
\begin{proof}
\allowdisplaybreaks%
Observe $ a_x $ is continuous.
Also, for $ \vp,\psi\in L^2(G) \;$
\begin{align*}
\ab{\vp,A_x\psi}&\ts=\iint\overline{\vp(g)}a_x(g,h)\psi(h)dhdg\;\\
\ab{A_x\vp,\psi}&\ts=\iint\overline{a_x(h,g)}\mspace{1.5mu}\overline{\vp(g)}\psi(h)dgdh\:
\end{align*}
Also, for $ \psi\in L^2(G)\, g\in G \;$
\begin{align*}\swapabovedisplayskip
(A_xA_x\mspace{-3mu}{}^{*}\psi)(g)&\ts=\iint a_x(g,j)\overline{a_x(h,j)}\psi(h)dhdj\;\\
(A_x\mspace{-3mu}{}^{*}A_x\psi)(g)&\ts=\iint \overline{a_x(j,g)}a_x(j,h)\psi(h)dhdj\:\qedhere
\end{align*}
\end{proof}
\begin{lem}\label{lem: (3)}
Fix $ x\in Z \:$
Then for each $ \vp\in L^2(G) \;$
\[\ts \vv{A_x\mspace{-3mu}{}^{*}\vp}\le \p{\int \sup_{g\in \supp\p{\vp}}\v{k(\inv{g}x,h)} dh}\vv{\vp} \]
and for each $ y\in Z \;$
\[\ts \vv{(A_y-A_x)^*\vp}\le \p{\int \sup_{g\in \supp\p{\vp}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh}\vv{\vp}\: \]
\end{lem}
\begin{proof}
\allowdisplaybreaks%
Observe, for $ \vp\in L^2(G) \;$
\begin{align*}
&\ts\vv{A_x\mspace{-3mu}{}^{*}\vp}^2\\
&\ts=\int\v{\int \overline{k(\inv{h}\inv{g}x,\inv{h})}\vp(gh) dh}^2dg\\
&\ts\le\int\p{\int \v{\overline{k(\inv{h}\inv{g}x,\inv{h})}}\v{\vp(gh)} dh}^2dg\\
&\ts\le\int\p{\int \v{k(\inv{(gh)}x,\inv{h})}\1_{\supp\p{\vp}}(gh) dh}\p{\int \v{k(\inv{(gh)}x,\inv{h})}\v{\vp(gh)}^2 dh}dg\\
&\ts\le\int \p{\int \sup_{\gch\in \supp\p{\vp}\inv{h}}\v{k(\inv{(\gch h)}x,\inv{h})} dh}\p{\int \sup_{\check{g}\in \supp\p{\vp}\inv{h}}\v{k(\inv{(\check{g}h)}x,\inv{h})}\v{\vp(gh)}^2 dh}dg\\
&\ts= \p{\int \sup_{g\in \supp\p{\vp}}\v{k(\inv{g}x,\inv{h})} dh}^2\vv{\vp}^2\:
\end{align*}
Also, for $ \vp\in L^2(G)\, y\in Z \;$
\begin{align*}
&\ts\vv{(A_{y}-A_{x})^*\vp}^2\\
&\ts=\int\v{\int (\overline{k(\inv{h}\inv{g}y,\inv{h})}-\overline{k(\inv{h}\inv{g}x,\inv{h})})\vp(gh) dh}^2dg\\
&\ts\le\int\p{\int \v{\overline{k(\inv{h}\inv{g}y,\inv{h})}-\overline{k(\inv{h}\inv{g}x,\inv{h})}}\v{\vp(gh)} dh}^2dg\\
&\ts\le\int
\!\begin{array}[t]{l}
\p{\int \v{{k((gh)^{-1}y,\inv{h})}-{k((gh)^{-1}x,\inv{h})}}\1_{\supp\p{\vp}}(gh) dh}\\
{}\cdot\p{\int \v{{k((gh)^{-1}y,\inv{h})}-{k((gh)^{-1}x,\inv{h})}}\v{\vp(gh)}^2 dh}dg
\end{array}\\
&\ts\le\int
\!\begin{array}[t]{l}
\p{\int \sup_{\check{g}\in \supp\p{\vp}\cramped{\inv{h}}}\v{{k((\check{g}h)^{-1}y,\inv{h})}-{k((\check{g}h)^{-1}x,\inv{h})}} dh}\\
{}\cdot\p{\int \sup_{\check{g}\in\supp\p{\vp}\cramped{\inv{h}}}\v{{k((\check{g}h)^{-1}y,\inv{h})}-{k((\check{g}h)^{-1}x,\inv{h})}}\v{\vp(gh)}^2 dh}dg
\end{array}\\
&\ts=\p{\int \sup_{g\in \supp\p{\vp}}\v{{k({g}^{-1}y,\inv{h})}-{k({g}^{-1}x,\inv{h})}} dh}^2\vv{\vp}^2\:\qedhere
\end{align*}
\end{proof}
\subsection{Continuity of Spectrum Maps}\label{subsec: CSM}
In this subsection, we establish sufficient conditions, e.g., amenability, implying semi-continuity of the spectrum $ \sigma(A_x) $ and implying continuity of the spectrum map $ X\mapsto \overline{\bigcup_{x\in X}\sigma(A_x)} $ with respect to a dynamically-defined operator family.
Let $ \ab{G,\lambda} $ be a second-countable locally-compact Hausdorff group endowed with a left Haar measure, where $ \ab{G,\lambda} $ is unimodular.
Let $ e $ be the identity element of $ G \:$
Let $ \ab{Z,G} $ be a topological space endowed with a left continuous action.
Let $ k $ be a bounded continuous (on $ Z $) complex-valued function on $ Z\times G \;$ where $ \sup_{x\in Z}\vv{k_x}_{\infty,1}<+\infty \:$
Here
\[\ts \vv{k_x}_{\infty,1}\defeq\int\sup_{g\in G}\v{k(\inv{g}x,h)} dh \:\]
Define $ A_x:L^2(G)\to L^2(G) \;$ for $ x\in Z \;$ by
\[\ts
(A_x\psi)(g)=\int k(\inv{g}x,\inv{g}h)\psi(h) dh
\]
for every $ \psi\in L^2(G)\, g\in G \:$
Let $ \Rc_x \;$ for $ x\in Z \;$ be the support of the map $ h\mapsto \sup_{g\in G}\v{k(\inv{g}x,h)} \:$
\begin{thm}\label{thm: (4)}
Fix $ x\in Z $ with $ A_xA_x\mspace{-3mu}{}^*=A_x\mspace{-3mu}{}^*A_x \:$
Fix $ E\in\sigma(A_x) \:$
Fix $ \ep>0 \:$
Let $ \vp:G\to\C $ be a nonzero $ L^2 $-vector satisfying $ \vv{(A_{x}-E)\vp}\le \ep\vv{\vp} \:$
Let $ \chi:G\to\C $ be a compactly-supported continuous nonzero $ L^2 $-vector.
Define, for $ j\in G \;$
\begin{align*}
\ts\chi_{j}&\ts:G\to\C:g\mapsto\chi(\inv{j}g)\;\\
\ts\cramped{\chi^{j}}&\ts:G\to\C:g\mapsto\chi(gj)\:
\end{align*}
Then there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\sup_{h\in\Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}}
+2\ep
}\vv{\vp_j}\;
\]
where $ \vp_j:G\to \C:g\mapsto \chi_j(g)\vp(g) \:$
\end{thm}
\begin{proof}
\allowdisplaybreaks%
Define
\[\ts
\de\defeq
\begin{cases}
\ep\p{\vv{k_x}_{\infty,1}\sup_{h\in\Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}}}^{-1}&\iph\p{\vv{k_x}_{\infty,1}\sup_{h\in\Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}}}\ne 0\;\\
3^{-1}&\elss\:
\end{cases}
\]
Let $ \Sc $ be the support of the map $ j\mapsto \vv{\vp_{j}}^2 \:$
Observe, for $ j\in G \;$
\begin{align*}
&\ts\v{\vv{\vp_j}^2-\vv{\vp_e}^2}\\
&\ts\le \int\v{\v{\vp_j(g)}^2-\v{\vp_e\p{g}}^2}dg\\
&\ts\le 2\vv{\chi}_{\infty}\vv{\chi_j-\chi}_{\infty}\vv{\vp}^2\:
\end{align*}
Because $ \chi\in C_{\ssm c}(G) \;$ the map $ j\mapsto \vv{\vp_{j}}^2 $ is continuous and $ 0<\lambda(\Sc) \:$
Define $ \bigb{A,B}:L^2(G)\to L^2(G) \;$ for $ A,B\in \Ls\p{L^2(G)} \;$ by
\[\ts
(\bigb{A,B}\psi)(g)=(AB\psi)(g)-(BA\psi)(g)
\]
for every $ \psi\in L^2(G)\, g\in G \:$
Define $ \bigb{\chi_j(\cdot)}:L^2(G)\to L^2(G) \;$ for $ j\in G \;$ by
\[\ts
(\bigb{\chi_j(\cdot)}\psi)(g)=\chi_j(g)\psi(g)
\]
for every $ \psi\in L^2(G)\, g\in G \:$
Define, for $ j\in G \;$
\begin{align*}
\vp_{\ep}&\defeq (A_{x}-E)\vp\;\\
\vp_{j,\ep}&\defeq (A_{x}-E)\bigb{\chi_{j}(\cdot)}\vp\;\\
\vp_{\ep,j}&\defeq \bigb{\chi_{j}(\cdot)}(A_{x}-E)\vp\:
\end{align*}
Observe
\begin{align*}\swapabovedisplayskip
&\ts\int \vv{\vp_{\ep,j}}^2dj\\
&\ts=\iint\v{\chi_j(g)}^2\v{\vp_{\ep}(g)}^2dgdj\\
&\ts=\iint\v{\chi_j(g)}^2\v{\vp_{\ep}(g)}^2djdg\\
&\ts=\vv{\vp_{\ep}}^2\vv{\chi}^2\\
&\ts\le\ep^2\vv{\vp}^2\vv{\chi}^2\\
&\ts=\iint\ep^2\v{\chi_j(g)}^2\v{\vp(g)}^2djdg\\
&\ts=\iint\ep^2\v{\chi_j(g)}^2\v{\vp(g)}^2dgdj\\
&\ts=\int \ep^2\vv{\vp_{j}}^2dj
\end{align*}
and
\begin{align*}\swapabovedisplayskip
&\ts\int\vv{\bigb{A_x-E,\bigb{\chi_j\p{\cdot}}}\vp}^2dj\\
&\ts=\iint\v{\int k(\inv{g}x,h)(\chi_j(gh)-\chi_j(g))\vp(gh)dh}^2dgdj\\
&\ts\le\iint\p{\int \v{k(\inv{g}x,h)}\v{\chi_j(gh)-\chi_j(g)}\v{\vp(gh)}dh}^2dgdj\\
&\ts\le\iint\p{\int \v{k(\inv{g}x,h)}dh}\p{\int \v{k(\inv{g}x,h)}\v{\chi_j(gh)-\chi_j(g)}^2\v{\vp(gh)}^2dh}dgdj\\
&\ts\le\iint
\!\begin{array}[t]{l}
\p{\int \sup_{\gch\in G}\v{k(\inv{\gch}x,h)}dh}\\
{}\cdot\p{\int \sup_{\gch\in G}\v{k(\inv{\gch}x,h)}\v{\chi_j(gh)-\chi_j(g)}^2\v{\vp(gh)}^2dh}djdg
\end{array}\\
&\ts=\p{\int \sup_{g \in G}\v{k(\inv{g}x,h)}dh}\p{\int \sup_{g \in G}\v{k(\inv{g}x,h)}\vv{\cramped{\chi^h}-\chi}^2\vv{\vp}^2dh}\\
&\ts=\p{\int \sup_{g \in G}\v{k(\inv{g}x,h)}dh}\p{\int \sup_{g \in G}\v{k(\inv{g}x,h)}\vv{\cramped{\chi^h}-\chi}^2\frac{1}{\vv{\chi}^2}dh}\vv{\chi}^2\vv{\vp}^2
\\
&\ts=\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}dh}\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\smash{\frac{\vv{\cramped{\chi^h-\chi}}^2}{\vv{\chi}^2}}dh}\int\vv{\vp_j}^2dj\\
&\ts\le\int(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})^2\vv{\vp_j}^2dj\:
\end{align*}
Therefore
\begin{align*}\swapabovedisplayskip
&\ts\int\vv{\vp_{j,\ep}}^2dj\\
&\ts=\int\vv{\bigb{A_{x}-E,\bigb{\chi_j\p{\cdot}}}\vp+\vp_{\ep,j}}^2dj\\
&\le\ts \int (1+\de)\vv{\bigb{A_{x}-E,\bigb{\chi_j\p{\cdot}}}\vp}^2+(1+\inv{\de})\vv{\vp_{\ep,j}}^2dj
\tag*{\scalebox{1}{\(\mathllap{\smash{\p{\lsubstack{\textrm{Peter--Paul's}\\\textrm{inequality}}}}}\)}}\\
&\le\ts \int (1+\de)(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})^2\vv{\vp_{j}}^2+(1+\inv{\de})\ep^2\vv{\vp_{j}}^2dj\:
%&=\ts \int \bigp{(1+\de)(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})^2+(1+\inv{\de})\ep^2}\vv{\vp_{j}}^2dj\:
\end{align*}
Because $ 0<\lambda(\Sc) \;$ there exists $ j\in \Sc $ such that $ \vv{\vp_j}\ne 0 $ and
\[\ts \vv{\vp_{j,\ep}}^2\le\bigp{(1+\de)(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})^2+(1+\inv{\de})\ep^2}\vv{\vp_{j}}^2\: \]
Therefore
\[\ts
\begin{cases}
\vv{\vp_{j,\ep}}\le \bigp{(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})+\ep}\vv{\vp_j}&\iph\p{\vv{k_x}_{\infty,1}\sup_{h\in\Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}}}\ne 0\;\\
\vv{\vp_{j,\ep}}\le 2\ep\vv{\vp_j}&\elss\:
\end{cases}
\]
Also, for $ y\in Z \;$
\begin{align*}
&\ts\vv{(A_{y}-E)\vp_{j}}\\
&\ts=\vv{(A_{y}-A_{x}+A_{x}-E)\vp_{j}}\\
&\ts\le\vv{(A_{y}-A_{x})\vp_{j}}+\vv{\vp_{j,\ep}}\\
&\ts\le\vv{(A_{y}-A_{x})\vp_{j}}+\bigp{(\vv{k_x}_{\infty,1}\sup_{h\in \Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}})+2\ep}\vv{\vp_{j}}\:\qedhere
%&\ts=\Bigp{
%\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
%+\vv{k_x}_{\infty,1}\sup_{h\in\Rc_x\cup\s{e}}\smash{\frac{\vv{\cramped{\chi^h-\chi}}}{\vv{\chi}}}
%+2\ep
%}\vv{\vp_j}
\end{align*}
\end{proof}
\begin{cor}\label{cor: 4.1}
Fix $ x\in Z \:$
Assume $ \ab{G,\lambda} $ is amenable and $ \Rc_x $ is compact and $ A_yA_y\mspace{-3mu}{}^*=A_y\mspace{-3mu}{}^*A_y $ for every $ y $ in some neighborhood of $ x \:$
Then
\[\ts \lim_{y\xrightarrow[]{\smash{Z}}x}\sup_{E\in\sigma(A_x)}\dist(E,\sigma(A_y))=0\: \]
\end{cor}
\begin{proof}
\allowdisplaybreaks%
Fix $ \eta>0 \:$
Because $ \ab{G,\lambda} $ is amenable and $ \Rc_x $ is compact, there exists a compactly-supported continuous nonzero $ L^2 $-vector $ \chi:G\to\C $ satisfying
\[\ts \sup_{h\in\Rc_x\cup\s{e}}\vv{\cramped{\chi^h-\chi}}\le \eta\vv{\chi}\: \]
Here $ \cramped{\chi^h}:G\to \C:g\mapsto\chi(gh) \:$
Define, for $ j\in G \;$ $ \chi_j:G\to \C:g\mapsto\chi(\inv{j}g) \:$
Fix $ \ep>0 \:$
Because $ A_x $ is a bounded operator, there exist elements $ E_1,\ldots,E_N $ of $ \sigma(A_x) $ satisfying
\[\ts \sigma(A_x)\subseteq \bigcup_{n=1}^N B(E_n,\ep) \:\]
Here $ B(E_n,\ep) $ is the open ball centered at $ E_n $ with radius $ \ep \:$
Fix $ n\in\s{1,\ldots,N} \:$
Because $ A_x $ is a normal operator, there exists a nonzero $ L^2 $-vector $ \vp:G\to\C $ satisfying
\[\ts \vv{(A_{x}-E_n)\vp}\le \ep\vv{\vp} \:\]
Here $ \vp $ is an approximate eigenfunction.
By Theorem~\ref{thm: (4)}, there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E_n)\vp_j}\le
\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta
+2\ep
}\vv{\vp_j}\;
\]
where $ \vp_j:G\to \C:g\mapsto \chi_{j}(g)\vp(g) \:$
Let $ y $ be an element of $ Z $ satisfying
\[\ts A_yA_y\mspace{-3mu}{}^*=A_y\mspace{-3mu}{}^*A_y\: \]
By properties of normal operators,
\begin{align*}
&\ts \dist(E_n,\sigma(A_y))\\
&\ts =\vv{(A_y-E_n)^{-1}}^{-1}\\
&\ts\le {
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta
+2\ep
}\\
&\ts =
\vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta
+2\ep\:
\end{align*}
Observe
\[\ts
\supp\p{\vp_j}\subseteq \supp\p{\chi_j}=j\supp\p{\chi}\:
\]
By Lemma~\ref{lem: (3)},
\begin{align*}\swapabovedisplayskip
&\ts \vv{(A_y-A_x)^*\smash[b]{\frac{\vp_j}{\vv{\vp_j}}}}\\
&\ts \le \int \sup_{g\in \supp\p{\vp_j}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh\\
&\ts \le \int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\inv{j}y,h)-k(\inv{g}\inv{j}x,h)} dh\:
\end{align*}
Therefore
\begin{align*}\swapabovedisplayskip
&\ts\sup_{E\in \sigma(A_x)\cap B(E_n,\ep)}\dist(E,\sigma(A_y))\\
&\ts\le \int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\inv{j}y,h)-k(\inv{g}\inv{j}x,h)} dh
+\vv{k_x}_{\infty,1}\eta
+3\ep\:
\intertext{Because $ k $ is continuous on $ Z $ and because $ \supp\p{\chi} $ is compact,}
&\ts \lim_{y\xrightarrow[]{\smash{Z}}x}\sup_{E\in\sigma(A_x)}\dist(E,\sigma(A_y))=0\:\qedhere
\end{align*}
\end{proof}
\begin{cor}\label{cor: 4.2}
Let $ d $ be a continuous metric on $ Z \:$
Let $ \Ss $ be the collection of all invariant nonempty closed subsets of $ Z \:$
Let $ \Ks $ be the collection of all nonempty compact subsets of $ \C \:$
Define
\[\ts
\bar{\sigma}:(\Ss,d_{\ssm H})\to (\Ks,\dist_{\ssm H}):X\mapsto\overline{\bigcup_{x\in X}\sigma(A_x)}\:
\]
Assume $ \ab{G,\lambda} $ is amenable and $ \overline{\bigcup_{x\in Z}\Rc_x} $ is compact and $ A_yA_y\mspace{-3mu}{}^*=A_y\mspace{-3mu}{}^*A_y $ for every $ y\in Z \:$
Then
\[\ts
\lim_{\de\to 0^+}\sup_{X,Y\in\Ss,d_{\ssm{H}}(X,Y)\le \de}\dist_{\ssm{H}}(\bar{\sigma}(X),\bar{\sigma}(Y))=0\:
\]
Here $ d_{\ssm H}(X,Y) $ and $ \dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y)) $ are Hausdorff distances between subsets of metric spaces.
\end{cor}
\begin{proof}
\allowdisplaybreaks%
Fix $ \eta>0 \:$
Define $ \Rc\defeq \overline{\bigcup_{x\in Z}\Rc_x} \:$
Because $ \ab{G,\lambda} $ is amenable and $ \Rc $ is compact, there exists a compactly-supported continuous nonzero $ L^2 $-vector $ \chi:G\to\C $ satisfying
\[\ts \sup_{h\in\Rc\cup\s{e}}\vv{\cramped{\chi^h-\chi}}\le \eta\vv{\chi}\: \]
Here $ \cramped{\chi^h}:G\to \C:g\mapsto\chi(gh) \:$
Define, for $ j\in G \;$ $ \chi_j:G\to \C:g\mapsto\chi(\inv{j}g) \:$
Fix $ \ep>0 \:$
Fix $ X,Y\in\Ss \:$
Define $ \de_{\ssm H}\defeq d_{\ssm H}(X,Y) \:$
Fix $ E\in\bigcup_{x\in X}\sigma(A_x) \:$
Let $ x $ be an element of $ X $ satisfying
\[\ts E\in\sigma(A_x) \:\]
Because $ A_x $ is a normal operator, there exists a nonzero $ L^2 $-vector $ \vp:G\to \C $ satisfying
\[\ts \vv{(A_x-E)\vp}\le \ep\vv{\vp} \:\]
Here $ \vp $ is an approximate eigenfunction.
By Theorem~\ref{thm: (4)}, there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta+2\ep
}\vv{\vp_j}\;
\]
where $ {\vp}_{j}:G\to \C:g\mapsto \chi_{j}(g){\vp}(g) \:$
Let $ y $ be an element of $ Y $ satisfying
\[\ts d(\inv{j}{x},\inv{j}{y})= d(\inv{j}{x},Y)\le \de_{\ssm H} \:\]
By properties of normal operators,
\begin{align*}
&\ts \dist(E,\sigma(A_y))\\
&\ts =\vv{(A_y-E)^{-1}}^{-1}\\
&\ts\le {
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta
+2\ep
}\\
&\ts ={
\vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{\infty,1}\eta
+2\ep
}\:
\end{align*}
Observe
\[\ts
\supp\p{\vp_j}\subseteq \supp\p{\chi_j}=j\supp\p{\chi}\:
\]
By Lemma~\ref{lem: (3)},
\begin{align*}\swapabovedisplayskip
&\ts \vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}\\
&\ts \le \int \sup_{g\in \supp\p{\vp_j}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh\\
&\ts \le \int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\inv{j}y,h)-k(\inv{g}\inv{j}x,h)} dh\:
\end{align*}
Therefore
\begin{align*}
&\ts\sup_{E\in \bigcup_{x\in X}\sigma(A_x)}\dist(E,\bigcup_{y\in Y}\sigma(A_y))\\
&\ts\le\sup_{\lsubstack{\check{x}\in X,\check{y}\in Y\\\mathrlap{d(\check{x},\check{y})\le \de_{\ssm H}}}}\int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\check{y},h)-k(\inv{g}\check{x},h)} dh
+\p{\sup_{\check{z}\in Z}\vv{k_{\check{z}}}_{\infty,1}}\eta+2\ep\:
\intertext{Because $ k $ is continuous on $ Z $ and because $ \supp\p{\chi} $ is compact,}
&\ts\lim_{\de\to 0^+}\sup_{X,Y\in\Ss,d_{\ssm{H}}(X,Y)\le \de}\dist_{\ssm{H}}(\bar{\sigma}(X),\bar{\sigma}(Y))=0\:\qedhere
\end{align*}
\end{proof}
\subsection{Quantitative Continuity of Spectrum Maps}\label{subsec: QCSM}
In this subsection, we establish sufficient conditions, e.g., strict polynomial growth and Lipschitz continuity on $ Z $ of both the action $ G\curvearrowright Z $ and the kernel $ k:Z\times G\to\C \;$ implying $ \frac{1}{2} $-Hölder continuity of the spectrum map $ X\mapsto \overline{\bigcup_{x\in X}\sigma(A_x)} $ with respect to a dynamically-defined operator family.
Furthermore, if the kernel is instead locally-constant on $ Z $, then the spectrum map is instead Lipschitz continuous.
Let $ \ab{G,\lambda} $ be a second-countable locally-compact Hausdorff group endowed with a left Haar measure, where $ \ab{G,\lambda} $ is unimodular.
Let $ e $ be the identity element of $ G \:$
Let $ d $ be a proper continuous left invariant metric on $ G \;$ where $ \ab{G,\lambda} $ has strict polynomial growth, i.e., there exist $ b\ge 1\, c_1>c_0>0 $ such that for each $ r\ge 1 \;$
\[\ts c_0r^b\le\lambda(B(r))\le c_1r^b \:\]
Here $ B(r) $ is the open ball centered at $ e $ with radius $ r \:$
Define, for $ j\in G \;$ $ \v{j}\defeq d(e,j) $ so that $ d(g,h) \;$ for $ g,h\in G \;$ is instead now $ \v{\inv{g}h} \:$
Let $ \ab{Z,G} $ be a topological space endowed with a left continuous action.
Let $ k $ be a bounded continuous (on $ Z $) complex-valued function on $ Z\times G \;$ where $ \sup_{x\in Z}\vv{k_x}_{S}<+\infty \:$
Here
\[\ts \vv{k_x}_{S}\defeq\int\sup_{g\in G}\v{k(\inv{g}x,h)}\v{h} dh \:\]
Define $ A_x:L^2(G)\to L^2(G) \;$ for $ x\in Z \;$ by
\[\ts
(A_x\psi)(g)=\int k(\inv{g}x,\inv{g}h)\psi(h) dh
\]
for every $ \psi\in L^2(G)\, g\in G \:$
\begin{lem}\label{lem: (5)}
Define $ t\defeq\p{{\frac{(2+b)^{2+b}c_1}{2b^bc_0}}}^{1/2} \:$
Fix $ r\ge 1 \:$
Define $ \chi:G\to \C:g\mapsto \1_{B(r)}(g)\p{\smash{\frac{r-\v{g}}{r}}} \:$
Define, for $ h\in G \;$ $ \cramped{\chi^h}:G\to\C:g\mapsto\chi(gh) \:$
Then $ \chi $ is a compactly-supported continuous nonzero $ L^2 $-vector and for each $ h\in G \;$
\[\ts
\vv{\cramped{\chi^h-\chi}}\le\smash{\frac{t}{r}}\v{h}\vv{\chi}\:
\]
\end{lem}
\begin{rem}
Lemma~\ref{lem: (5)} establishes the implication that strict polynomial growth implies amenability.
\end{rem}
\begin{proof}
\allowdisplaybreaks%
Define $ \smash{\de\defeq \p{\frac{2}{2+b}}r} \:$
Observe $ \chi $ is a compactly-supported continuous nonzero $ L^2 $-vector.
Also, for $ h\in G \;$
\begin{align*}
&\ts\vv{\cramped{\chi^h-\chi}}^2\\
&\ts=\int_{B(r)\inv{h}\setminus B(r)}(\frac{r-\v{gh}}{r})^2dg
+\int_{B(r)\setminus B(r)\inv{h}}(\frac{r-\v{g}}{r})^2dg
+\int_{B(r)\cap B(r)\inv{h}}(\frac{\v{g}-\v{gh}}{r})^2dg
\\
&\ts\le\int_{B(r)\inv{h}\setminus B(r)}(\frac{\v{g}-\v{gh}}{r})^2dg
+\int_{B(r)\setminus B(r)\inv{h}}(\frac{\v{gh}-\v{g}}{r})^2dg
+\int_{B(r)\cap B(r)\inv{h}}(\frac{\v{g}-\v{gh}}{r})^2dg
\\
&\ts\le\int_{B(r)\inv{h}\setminus B(r)}r^{-2}\v{h}^2dg
+\int_{B(r)\setminus B(r)\inv{h}}r^{-2}\v{h}^2dg
+\int_{B(r)\cap B(r)\inv{h}}r^{-2}\v{h}^2dg
\\
&\ts\le r^{-2}\v{h}^22(c_1r^b)
\shortintertext{and}
&\ts\vv{\chi}^2\\
&\ts=\int_{B(r)\setminus B(r-\de)}(\frac{r-\v{g}}{r})^2dg
+\int_{B(r-\de)}(\frac{r-\v{g}}{r})^2dg
\\
&\ts\ge\int_{B(r-\de)}(\frac{r-\v{g}}{r})^2dg\\
&\ts\ge\int_{B(r-\de)}r^{-2}\de^2dg\\
&\ts\ge r^{-2}\de^2(c_0(r-\de)^b)\:\qedhere
\end{align*}
\end{proof}
\begin{thm}\label{thm: (6)}
Fix $ x\in Z $ with $ A_xA_x\mspace{-3mu}{}^*=A_x\mspace{-3mu}{}^*A_x \:$
Fix $ E\in\sigma(A_x) \:$
Fix $ \ep>0 \:$
Let $ \vp:G\to\C $ be a nonzero $ L^2 $-vector satisfying $ \vv{(A_{x}-E)\vp}\le \ep\vv{\vp} \:$
Let $ \chi:G\to\C $ be a compactly-supported continuous nonzero $ L^2 $-vector.
Define, for $ j\in G \;$
\begin{align*}
\ts\chi_{j}&\ts:G\to\C:g\mapsto\chi(\inv{j}g)\;\\
\ts\cramped{\chi^{j}}&\ts:G\to\C:g\mapsto\chi(gj)\:
\end{align*}
Assume there exists $ \eta>0 $ such that for each $ h\in G \;$ $ \vv{\cramped{\chi^h-\chi}}\le \eta\v{h}\vv{\chi} \:$
Then there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\eta
+2\ep
}\vv{\vp_j}\;
\]
where $ \vp_j:G\to \C:g\mapsto \chi_j(g)\vp(g) \:$
\end{thm}
\begin{rem}
Theorem~\ref{thm: (6)} does not need that $ \ab{G,\lambda} $ has strict polynomial growth.
\end{rem}
\begin{proof}
\allowdisplaybreaks%
Define
\[\ts
\de\defeq
\begin{cases}
\ep\p{\vv{k_x}_{S}\eta}^{-1}&\iph\vv{k_x}_{S}\ne 0\;\\
3^{-1}&\elss\:
\end{cases}
\]
Let $ \Sc $ be the support of the map $ j\mapsto \vv{\vp_{j}}^2 \:$
Observe, for $ j\in G \;$
\begin{align*}
&\ts\v{\vv{\vp_j}^2-\vv{\vp_e}^2}\\
&\ts\le \int\v{\v{\vp_j(g)}^2-\v{\vp_e\p{g}}^2}dg\\
&\ts\le 2\vv{\chi}_{\infty}\vv{\chi_j-\chi}_{\infty}\vv{\vp}^2\:
\end{align*}
Because $ \chi\in C_{\ssm c}(G) \;$ the map $ j\mapsto \vv{\vp_{j}}^2 $ is continuous and $ 0<\lambda(\Sc) \:$
Define $ \bigb{A,B}:L^2(G)\to L^2(G) \;$ for $ A,B\in \Ls\p{L^2(G)} \;$ by
\[\ts
(\bigb{A,B}\psi)(g)=(AB\psi)(g)-(BA\psi)(g)
\]
for every $ \psi\in L^2(G)\, g\in G \:$
Define $ \bigb{\chi_j(\cdot)}:L^2(G)\to L^2(G) \;$ for $ j\in G \;$ by
\[\ts
(\bigb{\chi_j(\cdot)}\psi)(g)=\chi_j(g)\psi(g)
\]
for every $ \psi\in L^2(G)\, g\in G \:$
Define, for $ j\in G \;$
\begin{align*}
\vp_{\ep}&\defeq (A_{x}-E)\vp\;\\
\vp_{j,\ep}&\defeq (A_{x}-E)\bigb{\chi_{j}(\cdot)}\vp\;\\
\vp_{\ep,j}&\defeq \bigb{\chi_{j}(\cdot)}(A_{x}-E)\vp\:
\end{align*}
Observe
\begin{align*}\swapabovedisplayskip
&\ts\int \vv{\vp_{\ep,j}}^2dj\\
%&\ts=\iint\v{\chi_j(g)}^2\v{\vp_{\ep}(g)}^2dgdj\\
%&\ts=\iint\v{\chi_j(g)}^2\v{\vp_{\ep}(g)}^2djdg\\
&\ts=\vv{\vp_{\ep}}^2\vv{\chi}^2\\
&\ts\le\ep^2\vv{\vp}^2\vv{\chi}^2\\
%&\ts=\iint\ep^2\v{\chi_j(g)}^2\v{\vp(g)}^2djdg\\
%&\ts=\iint\ep^2\v{\chi_j(g)}^2\v{\vp(g)}^2dgdj\\
&\ts=\int \ep^2\vv{\vp_{j}}^2dj
\end{align*}
and
\begin{align*}\swapabovedisplayskip
&\ts\int\vv{\bigb{A_x-E,\bigb{\chi_j\p{\cdot}}}\vp}^2dj\\
&\ts=\iint\v{\int k(\inv{g}x,h)(\chi_j(gh)-\chi_j(g))\vp(gh)dh}^2dgdj\\
&\ts\le\iint\p{\int \v{k(\inv{g}x,h)}\v{\chi_j(gh)-\chi_j(g)}\v{\vp(gh)}dh}^2dgdj\\
&\ts\le\iint\p{\int \v{k(\inv{g}x,h)}\v{h}dh}\p{\int \v{k(\inv{g}x,h)}\v{h}^{-1}\v{\chi_j(gh)-\chi_j(g)}^2\v{\vp(gh)}^2dh}dgdj\\
&\ts\le\iint
\!\begin{array}[t]{l}
\p{\int \sup_{\gch\in G}\v{k(\inv{\gch}x,h)}\v{h}dh}\\
{}\cdot\p{\int \sup_{\gch\in G}\v{k(\inv{\gch}x,h)}\v{h}^{-1}\v{\chi_j(gh)-\chi_j(g)}^2\v{\vp(gh)}^2dh}djdg
\end{array}\\
&\ts=\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}dh}\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}^{-1}\vv{\cramped{\chi^h}-\chi}^2\vv{\vp}^2dh}\\
&\ts=\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}dh}\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}^{-1}\vv{\cramped{\chi^h}-\chi}^2\frac{1}{\vv{\chi}^2}dh}\vv{\chi}^2\vv{\vp}^2\\
&\ts=\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}dh}\p{\int \sup_{g\in G}\v{k(\inv{g}x,h)}\v{h}^{-1}\smash{\frac{\cramped{\vv{\chi^h-\chi}^2}}{\vv{\chi}^2}}dh}\int\vv{\vp_j}^2dj\\
&\ts\le\int(\vv{k_x}_{S}\eta)^2\vv{\vp_j}^2dj\:
\end{align*}
Therefore
\begin{align*}\swapabovedisplayskip
&\ts\int\vv{\vp_{j,\ep}}^2dj\\
&\ts=\int\vv{\bigb{A_{x}-E,\bigb{\chi_j\p{\cdot}}}\vp+\vp_{\ep,j}}^2dj\\
&\le\ts \int (1+\de)\vv{\bigb{A_{x}-E,\bigb{\chi_j\p{\cdot}}}\vp}^2+(1+\inv{\de})\vv{\vp_{\ep,j}}^2dj\\
&\le\ts \int (1+\de)(\vv{k_x}_{S}\eta)^2\vv{\vp_{j}}^2+(1+\inv{\de})\ep^2\vv{\vp_{j}}^2dj\:
%&=\ts \int \bigp{(1+\de)(\vv{k_x}_{S}\eta)^2+(1+\inv{\de})\ep^2}\vv{\vp_{j}}^2dj\:
\end{align*}
Because $ 0<\lambda(\Sc) \;$ there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and
\[\ts \vv{\vp_{j,\ep}}^2\le\bigp{(1+\de)(\vv{k_x}_{S}\eta)^2+(1+\inv{\de})\ep^2}\vv{\vp_{j}}^2\: \]
Therefore
\[\ts
\begin{cases}
\vv{\vp_{j,\ep}}\le \bigp{(\vv{k_x}_{S}\eta)+\ep}\vv{\vp_j}&\iph\vv{k_x}_{S}\ne 0\;\\
\vv{\vp_{j,\ep}}\le 2\ep\vv{\vp_j}&\elss\:
\end{cases}
\]
Also, for $ y\in Z \;$
\begin{align*}\swapabovedisplayskip
&\ts\vv{(A_{y}-E)\vp_{j}}\\
&\ts=\vv{(A_{y}-A_{x}+A_{x}-E)\vp_{j}}\\
&\ts\le\vv{(A_{y}-A_{x})\vp_{j}}+\vv{\vp_{j,\ep}}\\
&\ts\le\vv{(A_{y}-A_{x})\vp_{j}}+\bigp{(\vv{k_x}_{S}\eta)+2\ep}\vv{\vp_{j}}\:\qedhere
%&\ts=\Bigp{\vv{(A_{y}-A_{x})\smash[b]{\frac{\vp_{j}}{\vv{\vp_j}}}}+\vv{k_x}_{S}\eta+2\ep}\vv{\vp_{j}}\:\qedhere
\end{align*}
\end{proof}
\begin{cor}\label{cor: 6.1}
Let $ d $ be a continuous metric on $ Z \;$ where the group action is Lipschitz continuous on $ Z $ and the kernel is Lipschitz continuous on $ Z \:$
Specifically, there exists $ m>0 $ such that for each $ g\in G $ and for all $ x,y\in Z \;$
\[\ts d(gx,gy)\le (m\v{g}+1)d(x,y)\]
and there exists $ M\in L^1(G) $ such that for $ \lambda $-a.e.\ $ h\in G \;$ $ M(h)\ge 0 $ and for all $ x,y\in Z \;$
\[\ts\v{k(x,h)-k(y,h)}\le M(h)d(x,y)\:\]
Define $ t\defeq \smash[t]{\p{\frac{(2+b)^{2+b}c_1}{2b^bc_0}}^{1/2}} \:$
Define $ \vv{k_Z}_S\defeq \smash[t]{\sup_{x\in Z}\vv{k_x}_{S}} \:$
Let $ \Ss $ be the collection of all invariant nonempty closed subsets of $ Z \:$
Let $ \Ks $ be the collection of all nonempty compact subsets of $ \C \:$
Define
\[\ts
\bar{\sigma}:(\Ss,d_{\ssm H})\to (\Ks,\dist_{\ssm H}):X\mapsto\overline{\bigcup_{x\in X}\sigma(A_x)}\:
\]
Assume $ \vv{k_Z}_S>0 $ and $ \vv{M}_1>0 $ and $ A_yA_y\mspace{-3mu}{}^*=A_y\mspace{-3mu}{}^*A_y $ for every $ y\in Z \:$
Fix $ X,Y\in\Ss \:$
Then
\[\ts
d_{\ssm H}(X,Y)\le{\min\s{\frac{t \vv{k_Z}_S}{m\vv{M}_1},1}}
\lRightarrow
\dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y))\le {\bigp{2(t\vv{k_Z}_Sm\vv{M}_1)^{1/2}+\vv{M}_1}} d_{\ssm H}(X,Y)^{1/2}\:
\]
Here $ d_{\ssm H}(X,Y) $ and $ \dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y)) $ are Hausdorff distances between subsets of metric spaces.
\end{cor}
\begin{proof}
\allowdisplaybreaks%
Define $ \de_{\ssm H}\defeq d_{\ssm H}(X,Y) \;$ $ r\defeq (\frac{t\vv{k_Z}_S}{\de_{\ssm H}m\vv{M}_1})^{1/2} \;$ $ \chi:G\to \C:g\mapsto\1_{B(r)}(g)(\frac{r-\v{g}}{r}) \:$
Fix $ \ep>0 \:$
Fix $ E\in \bigcup_{x\in X}\sigma(A_x) \:$
Let $ x $ be an element of $ X $ satisfying $ E\in \sigma(A_x)\:$
Because $ A_x $ is a normal operator, there exists a nonzero $ L^2 $-vector $ \vp:G\to \C $ satisfying
\[\ts \vv{(A_x-E)\vp}\le \ep\vv{\vp} \:\]
Here $ \vp $ is an approximate eigenfunction.
By Lemma~\ref{lem: (5)} and Theorem~\ref{thm: (6)}, there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
{\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}}\vv{\vp_j}\;
\]
where $ \vp_j:G\to \C:g\mapsto \chi_{j}(g)\vp(g) $ and $ \chi_{j}:G\to \C:g\mapsto\chi(\inv{j}g) \:$
Indeed, $ r\ge 1 \:$
Let $ {y} $ be an element of $ Y $ satisfying $ d(\inv{j}{x},\inv{j}{y})= d(\inv{j}{x},Y)\le\de_{\ssm H} \:$
By properties of normal operators,
\begin{align*}
&\ts \dist(E,\sigma(A_y))\\
&\ts =\vv{(A_y-E)^{-1}}^{-1}\\
&\ts\le {
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}\\
&\ts ={
\vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}\:
\end{align*}
Observe
\[\ts
\supp\p{\vp_j}\subseteq \supp\p{\chi_j}=j\supp\p{\chi}\:
\]
By Lemma~\ref{lem: (3)},
\begin{align*}\swapabovedisplayskip
&\ts \vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}\\
&\ts \le \int \sup_{g\in \supp\p{\vp_j}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh\\
&\ts \le \int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\inv{j}y,h)-k(\inv{g}\inv{j}x,h)} dh\:
\end{align*}
Therefore
\begin{align*}
&\ts\sup_{E\in \bigcup_{x\in X}\sigma(A_x)}\dist(E,\bigcup_{y\in Y}\sigma(A_y))\\
&\ts\le\sup_{\lsubstack{\check{x}\in X,\check{y}\in Y\\\mathrlap{d(\check{x},\check{y})\le \de_{\ssm H}}}}\int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\check{y},h)-k(\inv{g}\check{x},h)} dh
+\p{\sup_{\check{z}\in Z}\vv{k_{\check{z}}}_{S}}\frac{t}{r}+2\ep\\
&\ts\le \vv{M}_1(m r+1)\de_{\ssm H}
+\vv{k_Z}_{S}\smash[b]{\frac{t}{r}}+2\ep\:\qedhere
\end{align*}
\end{proof}
\begin{cor}\label{cor: 6.2}
Let $ d $ be a continuous metric on $ Z \;$ where the group action is Lipschitz continuous on $ Z $ and the kernel is locally-constant on $ Z \:$
Specifically, there exists $ m>0 $ such that for each $ g\in G $ and for all $ x,y\in Z \;$
\[\ts d(gx,gy)\le (m\v{g}+1)d(x,y)\]
and there exists $ M\in L^\infty(G) $ such that for $ \lambda $-a.e.\ $ h\in G \;$ $ M(h)\ge 1 $ and for all $ x,y\in Z \;$
\[\ts d(x,y)\le \frac{1}{M(h)}\lRightarrow k(x,h)=k(y,h)\:\]
Define $ t\defeq \smash[t]{\p{\frac{(2+b)^{2+b}c_1}{2b^bc_0}}^{1/2}} \:$
Define $ \vv{k_Z}_S\defeq \smash[t]{\sup_{x\in Z}\vv{k_x}_{S}} \:$
Let $ \Ss $ be the collection of all invariant nonempty closed subsets of $ Z \:$
Let $ \Ks $ be the collection of all nonempty compact subsets of $ \C \:$
Define
\[\ts
\bar{\sigma}:(\Ss,d_{\ssm H})\to (\Ks,\dist_{\ssm H}):X\mapsto\overline{\bigcup_{x\in X}\sigma(A_x)}\:
\]
Assume $ \vv{k_Z}_S>0 $ and $ A_yA_y\mspace{-3mu}{}^*=A_y\mspace{-3mu}{}^*A_y $ for every $ y\in Z \:$
Fix $ X,Y\in\Ss \:$
Then
\[\ts
d_{\ssm H}(X,Y)\le{\frac{1}{(m+1)\vv{M}_{\infty}}}
\lRightarrow
\dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y))\le {\bigp{t\vv{k_Z}_S(m+1)\vv{M}_{\infty}}} d_{\ssm H}(X,Y)\:
\]
Here $ d_{\ssm H}(X,Y) $ and $ \dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y)) $ are Hausdorff distances between subsets of metric spaces.
\end{cor}
\begin{proof}
\allowdisplaybreaks%
Define $ \de_{\ssm H}\defeq d_{\ssm H}(X,Y) \;$ $ r\defeq \frac{1-\de_{\ssm H}\vv{M}_\infty}{\de_{\ssm H}m\vv{M}_\infty} \;$ $ \chi:G\to \C:g\mapsto\1_{B(r)}(g)(\frac{r-\v{g}}{r}) \:$
Fix $ \ep>0 \:$
Fix $ E\in \bigcup_{x\in X}\sigma(A_x) \:$
Let $ x $ be an element of $ X $ satisfying $ E\in \sigma(A_x)\:$
Because $ A_x $ is a normal operator, there exists a nonzero $ L^2 $-vector $ \vp:G\to \C $ satisfying
\[\ts \vv{(A_x-E)\vp}\le \ep\vv{\vp} \:\]
Here $ \vp $ is an approximate eigenfunction.
By Lemma~\ref{lem: (5)} and Theorem~\ref{thm: (6)}, there exists $ j\in G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
{\Bigp{
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}}\vv{\vp_j}\;
\]
where $ \vp_j:G\to \C:g\mapsto \chi_{j}(g)\vp(g) $ and $ \chi_{j}:G\to \C:g\mapsto\chi(\inv{j}g) \:$
Indeed, $ r\ge 1 \:$
Let $ {y} $ be an element of $ Y $ satisfying $ d(\inv{j}{x},\inv{j}{y})= d(\inv{j}{x},Y)\le\de_{\ssm H} \:$
By properties of normal operators,
\begin{align*}
&\ts \dist(E,\sigma(A_y))\\
&\ts =\vv{(A_y-E)^{-1}}^{-1}\\
&\ts\le {
\vv{(A_y-A_x)\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}\\
&\ts ={
\vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\vv{k_x}_{S}\frac{t}{r}
+2\ep
}\:
\end{align*}
Observe
\[\ts
\supp\p{\vp_j}\subseteq \supp\p{\chi_j}=j\supp\p{\chi}\:
\]
By Lemma~\ref{lem: (3)},
\begin{align*}\swapabovedisplayskip
&\ts \vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}\\
&\ts \le \int \sup_{g\in \supp\p{\vp_j}}\v{k(\inv{g}y,h)-k(\inv{g}x,h)} dh\\
&\ts \le \int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\inv{j}y,h)-k(\inv{g}\inv{j}x,h)} dh\:
\end{align*}
Therefore
\begin{align*}
&\ts\sup_{E\in \bigcup_{x\in X}\sigma(A_x)}\dist(E,\bigcup_{y\in Y}\sigma(A_y))\\
&\ts\le\sup_{\lsubstack{\check{x}\in X,\check{y}\in Y\\\mathrlap{d(\check{x},\check{y})\le \de_{\ssm H}}}}\int \sup_{g\in \supp\p{\chi}}\v{k(\inv{g}\check{y},h)-k(\inv{g}\check{x},h)} dh
+\p{\sup_{\check{z}\in Z}\vv{k_{\check{z}}}_{S}}\frac{t}{r}+2\ep\\
&\ts\le 0
+\vv{k_Z}_{S}\smash[b]{\frac{t}{r}}+2\ep\:
\end{align*}
Indeed, for each $ \xch\in X $ and for each $ \ych\in Y \;$ if $ d(\xch,\ych)\le \de_{\ssm H} \;$ then for each $ g\in\supp\p{\chi} \;$
\[\ts
d(\inv{g}\ych,\inv{g}\xch)
\le(mr+1)\de_{\ssm H}
\le \frac{1}{\vv{M}_{\infty}}\:\qedhere
\]
\end{proof}
\begin{comment}
\begin{lem}
Let $ \ab{(\G,d),\#} $ be a countable discrete left-metric group endowed~with the counting measure, where for each $ m\ge 0 \;$
\[\ts \#(B(m))<+\infty\:\]
Let $ \beta:[0,+\infty)\to [1,+\infty) $ be an order-preserving embedding, where for all $ m_{\ttm 1},m_{\ttm 2}\in [0,+\infty) \;$ $\beta(m_{\ttm 1}+m_{\ttm 2})\le \beta(m_{\ttm 1})\beta(m_{\ttm 2})$ and $ \beta(0)=1 $ and $ \sup_{m\ge 0} \beta(m) = +\infty \:$
Let $ l $ be an integer, where $ l\ge 2 \:$
Define $ \Ac\defeq \s{0,\ldots,l-1} \:$
Define $ Z\defeq \Ac^\G \:$
Define, for $ x,y\in{Z} \;$
\[\ts d(x,y)\defeq 1\meet \inf\s{\smash[b]{\frac{1}{\beta(m)}}:m\in[0,+\infty)\, \smash[b]{x|_{B(m)}}=\smash[b]{y|_{B(m)}}}\:\]
Define, for $ g\in \G\, x\in{Z} \;$
\[\ts g x\defeq x\circ\inv{g}\:\]
Let $ e $ be the identity element of $ \G \:$
Then
\begin{enumerate}[(i)]
\item
$ d $ is an ultrametric on $ {Z} \:$
\item
$ \p{{Z},d} $ is compact and totally-disconnected.
\item
$ \p{{Z},d} $ has no isolated points $ \impliedby $ $ \G $ is infinite.
\item
$ ex=x $ for every $ x\in {Z} \:$
\item
$ h(gx)=(hg)x $ for every $ h,g\in\G\, x\in {Z} \:$
\item
$ d(g x,g y)\le \beta(\v{g})d(x,y) $ for every $ g\in \G\, x,y\in {Z} \:$
\end{enumerate}
\end{lem}
\begin{proof}
\allowdisplaybreaks%
\textit{(i)}
Fix $ x,y,z\in {Z} \:$
Observe $ d(x,y)=0 $ if and only if $ x=y \:$
Observe $ d(x,y)=d(y,x) \:$
Suppose $ d(x,y)\join d(y,z)<d(x,z) \:$
Observe $ 0<d(x,y)\meet d(y,z)\meet d(x,z) $ and $ x(e)=z(e) $ and $ d(x,y)\join d(y,z)\join d(x,z)<1 \:$
Let $ m_{xy},m_{yz},m_{xz} $ be elements of $ [0,+\infty) $ satisfying $ d(x,y)=\frac{1}{\beta(m_{xy})} $ and $ d(y,z)=\frac{1}{\beta(m_{yz})} $ and $ d(x,z)=\frac{1}{\beta(m_{xz})} \:$
Observe $ \beta(m_{xz})<\beta(m_{xy})\meet \beta(m_{yz}) $ and $ m_{xz}<m_{xy}\meet m_{yz} $ and $ x|_{B(m_{xy}\meet m_{yz})}=y|_{B(m_{xy}\meet m_{yz})}=z|_{B(m_{xy}\meet m_{yz})} $ and $ m_{xy}\meet m_{yz}\le m_{xz} $---a contradiction, and $ d(x,z)\le d(x,y)\join d(y,z) \:$
\smallskip\\
\textit{(ii)}
Fix $ \s{x^{(n)}}_{n\in\N}\subseteq {Z} \:$
Suppose for each $ \ep>0 \;$ there exists $ n_\ep\in\N $ such that for all $ i,j\ge n_\ep \;$ $ d(x^{(i)},x^{(j)})<\ep \:$
Define $ x^{(\infty)}\defeq (g\mapsto x^{(n_{(1/\beta(\v{g}))})}(g)) \:$
Observe $ \lim_{n\to +\infty}x^{(n)}=x^{(\infty)} \;$ and $ ({Z},d) $ is complete.
Observe for each $ \ep>0 \;$ there exists $ m_\ep\in [0,+\infty) $ such that $ 1/\beta(m_\ep)<\ep $ and $ \Ac^{B(m_\ep)} $ is finite, and $ ({Z},d) $ is totally-bounded.
Fix $ y,z\in {Z} \:$
Fix $ r>0 \:$
Suppose $ d(x^{(n)},y)<r $ for every $ n\in\N $ and $ \lim_{n\to+\infty}x^{(n)}=z \:$
Observe $ d(z,y)\le d(z,x^{(n)})\join d(x^{(n)},y) $ for every $ n\in\N $ and $ d(z,y)<r $ and $ \s{x\in {Z}:d(x,y)<r} $ is clopen, and $ \p{{Z},d} $ is totally-disconnected.
\smallskip\\
\textit{(iii)}
Fix $ x\in {Z} \:$
Fix $ u,v\in\Ac $ with $ u\ne v \:$
Define, for $ n\in\N \;$
\[\ts
x^{(n)}:G\to\Ac:g\mapsto
\begin{cases}
x(g)&\iph g\in B(n)\\
u&\iph g\notin B(n)\, x(g) \ne u\\
v&\iph g\notin B(n)\, x(g) = u\:
\end{cases}
\]
Observe $ x^{(n)}\ne x $ for every $ n\in\N $ and $ \lim_{n\to +\infty}x^{(n)}=x \:$
\smallskip\\
\textit{(iv)}
Fix $ x\in {Z} \:$
Observe $ ex=x\circ\inv{e}=x\circ e=x\: $
\smallskip\\
\textit{(v)}
Fix $ h,g\in\G \:$
Fix $ x\in {Z} \:$
Observe $ h(gx)=h(x\circ \inv{g})=x\circ \inv{g}\circ \inv{h}=x\circ (hg)^{-1}=(hg)x\:$
\smallskip\\
\textit{(vi)}
Fix $ g\in \G \:$
Fix $ x,y\in{Z} \:$
If $ 0=d(gx,gy)\meet d(x,y) $ or $ x(e)\ne y(e) $ or $ (gx)(e)\ne (gy)(e) \;$ then $ d(g x,g y)\le \beta(\v{g})d(x,y)\:$
Suppose $ 0<d(gx,gy)\meet d(x,y) $ and $ x(e)=y(e) $ and $ (gx)(e)= (gy)(e) \:$
Let $ m_g,m_e $ be elements of $ [0,+\infty) $ satisfying $ d(gx,gy)=\frac{1}{\beta(m_g)} $ and $ d(x,y)=\frac{1}{\beta(m_e)} \:$
Observe $ m_e\le \v{g}+m_g $ and $ \beta\p{m_e}\le \beta\p{\v{g}+m_g} $ and $d(g x,g y)\le \beta(\v{g})d(x,y)\:$
\end{proof}
\begin{thm}
Let $ \ab{(\G,d),\#} $ be a countable discrete left-metric group endowed~with the counting measure, where there exist $ b\ge 1\, a_1>a_0>0 $ such that for each $ r\ge 1 \;$
\[\ts a_0r^b\le\#(B(r))\le a_1r^b \:\]
Let $ \beta:[0,+\infty)\to [1,+\infty) $ be an order-preserving embedding, where for all $ m_{\ttm 1},m_{\ttm 2}\in [0,+\infty) \;$ $\beta(m_{\ttm 1}+m_{\ttm 2})\le \beta(m_{\ttm 1})\beta(m_{\ttm 2})$ and $ \beta(0)=1 $ and $ \sup_{m\ge 0} \beta(m) = +\infty \:$
Let $ l $ be an integer, where $ l\ge 2 \:$
Define $ \Ac\defeq \s{0,\ldots,l-1} \:$
Define $ Z\defeq \Ac^\G \:$
Define, for $ x,y\in Z \;$
\[\ts d(x,y)\defeq 1\meet \inf\s{\smash[b]{\frac{1}{\beta(m)}}:m\in[0,+\infty)\, \smash[b]{x|_{B(m)}}=\smash[b]{y|_{B(m)}}}\:\]
Define, for $ g\in \G\, x\in Z \;$
\[\ts g x\defeq x\circ\inv{g}\:\]
Let $ k:Z\times \G\to \C $ be a bounded continuous map, where $ \sup_{x\in Z}\sum_{h\in \G} \sup_{g\in \G}\v{k(\inv{g}x,h)}\v{h} <+\infty $ and there exists $ D\in L^1(\G) $ such that $ \vv{D}_1>0 $ and for each $ h\in \G \;$ $ D(h)\ge 0 $ and for all $ x,y\in Z \;$
\[\ts\v{k(x,h)-k(y,h)}\le D(h)\p{\inv{\beta}\p{d(x,y)^{-1}}}^{-1}\:\]
Define, for $ x\in Z \;$
\[\ts \smash[b]{\vv{k_x}_{\infty,1_+}}\defeq\sum_{h\in\G}\sup_{g\in \G}\v{k(\inv{g}x,h)}\v{h}\: \]
Define $ A_x:L^2(\G)\to L^2(\G) \;$ for $ x\in Z \;$ by
\[\ts
(A_x\psi)(g)=\sum_{h\in\G} k(\inv{g}x,\inv{g}h)\psi(h)
\]
for every $ \psi\in L^2(\G)\, g\in \G \:$
Assume for each $ x\in Z \;$ $ \smash[b]{\vv{k_x}_{\infty,1_+}}\ne 0 $ and $ A_xA_x\mspace{-3mu}{}^*=A_x\mspace{-3mu}{}^*A_x \:$
Let $ \Ss $ be the collection of all invariant nonempty closed subsets of $ Z \:$
Define, for $ X\in\Ss \;$ $\bar{\sigma}(X)\defeq \overline{{\bigcup_{x\in X}\sigma(A_{x})}}\:$
Fix $ X,Y\in\Ss \:$
Then
\[\ts
d_{\ssm H}(X,Y)\le\smash{\frac{1}{\beta(\frac{c+1}{c})}}
\lRightarrow
\dist_{\ssm H}(\bar{\sigma}(X),\bar{\sigma}(Y))\le \smash{\bigp{\p{\sqrt{\vv{D}_1}+\sqrt{st}}^2}} \p{\inv{\beta}\p{d_{\ssm H}(X,Y)^{-1}}}^{-1}\;
\]
where $ s\defeq\sup_{x\in Z}\smash[b]{\vv{k_x}_{\infty,1_+}} $ and $ t\defeq \smash[b]{\p{\frac{(2+b)^{2+b}a_1}{2b^ba_0}}^{1/2}} $ and $ c\defeq\smash[b]{(\frac{st}{\vv{D}_1})^{1/2}} \:$
\end{thm}
\begin{proof}
\allowdisplaybreaks%
Define $ \de_{\ssm H}\defeq d_{\ssm H}(X,Y) \;$ $ r\defeq \frac{\p{\inv{\beta}\p{\bp{\de}{\ssm H}{\smash{-1}}}}c}{c+1} \;$ $ \chi:\G\to \C:g\mapsto\1_{B(r)}(g)(\frac{r-\v{g}}{r}) \:$
Fix $ \ep>0 \:$
Fix~$ E\in \bigcup_{x\in X}\sigma(A_x) \:$
Let $ x $ be an element of $ X $ satisfying $E\in \sigma(A_x)\:$
Let $ \vp:\G\to\C $ be a nonzero $ L^2 $-vector satisfying $ \vv{(A_{x}-E)\vp}\le \ep\vv{\vp} \:$
By Lemma~\ref{lem: (5)} and Theorem~\ref{thm: (6)}, there exists $ j\in \G $ such that $ \vv{\vp_j}\ne 0 $ and for each $ y\in Z \;$
\[\ts
\vv{(A_y-E)\vp_j}\le
{\Bigp{
\vv{(A_y-A_x)^*\smash{\frac{\vp_j}{\vv{\vp_j}}}}
+\smash[b]{\vv{k_x}_{\infty,1_+}}\frac{t}{r}
+\ep
}}\vv{\vp_j}\;
\]
where $ \vp_j:\G\to \C:g\mapsto \chi_{j}(g)\vp(g) $ and $ \chi_{j}:\G\to \C:g\mapsto\chi(\inv{j}g) \:$
Let $ {y} $ be an element of $ Y $ satisfying $ d(\inv{j}{x},\inv{j}{y})= d(\inv{j}{x},Y)\le\de_{\ssm H} \:$
Observe $\supp\p{\vp_j}\subseteq \supp\p{\chi_j}=j\supp\p{\chi}=jB(r)$ and
$ \smash[b]{\sup_{g\in B(r)}}d(\inv{g}\inv{j}y,\inv{g}\inv{j}x)\le\smash[b]{\frac{1}{\vphantom{\de^{-1}}\beta(\inv{\beta}\p{\bp{\de}{\ssm H}{\smash{-1}}}-r)}} $ and
\begin{align*}
&\ts\dist(E,\sigma(A_y))\\
&\ts=\vv{(A_y-E)^{-1}}^{-1}\\
&\ts\le {\Bigp{
\p{\sum_{h\in \G} \sup_{g\in B(r)}\v{k(\inv{g}\inv{j}y,\inv{h})-k(\inv{g}\inv{j}x,\inv{h})}}
+\smash[b]{\vv{k_x}_{\infty,1_+}}\frac{t}{r}
+\ep
}}
\shortintertext{and}
&\ts\sup_{E\in\bigcup_{x\in X}\sigma(A_x)}\dist(E,\bigcup_{y\in Y}\sigma(A_y))\\
&\ts\le \sup_{\lsubstack{\check{x}\in X,\check{y}\in Y\\\mathrlap{d(\check{x},\check{y})\le \de_{\ssm H}}}}\p{\sum_{h\in \G} \sup_{g\in B(r)}\v{k(\inv{g}\check{y},\inv{h})-k(\inv{g}\check{x},\inv{h})}}
+\p{\sup_{\check{z}\in Z}\smash[b]{\vv{k_{\check{z}}}_{\infty,1_+}}}\frac{t}{r}
+\ep\\
&\ts\le \vv{D}_1\smash[b]{\frac{1}{\vphantom{\de^{-1}}\inv{\beta}\p{\bp{\de}{\ssm H}{\smash{-1}}}-r}}
+s\smash[b]{\frac{t}{r}}+\ep\:\qedhere
\end{align*}
\end{proof}
\end{comment}
%\clearpage
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%\clearpage
%\subsection*{Acknowledgements}
\NEWvbadness{10000}
\let\originalurl\url
\renewcommand{\url}[1]{\nobreak\hfil\allowbreak\hfilneg\originalurl{#1}}
\renewcommand{\refname}{References}
\renewcommand{\bibname}{\refname}
\begin{thebibliography}{00}%\addcontentsline{toc}{chapter}{\bibname}
%\bibitem{PDF}
%NAME,
%\textit{TITLE},
%JOURNAL \textbf{VOL},
%(YEAR),
%000--999.
%AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
\bibitem{MP1990[Avron][Mouche][Simon]}
J.~Avron and P.H.M.~van~Mouche and B.~Simon,
\textit{On the measure of the spectrum for the almost Mathieu operator},
Springer, Communications in Mathematical Physics \textbf{132}, no.\ 1,
(1990),
103--118.
%BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
\bibitem{BB16}
S.~Beckus and J.~Bellissard,
\textit{Continuity of the spectrum of a field of self-adjoint operators},
Springer, Annales Henri Poincar{\'e} \textbf{17}, no.\ 12,
(2016),
3425--3442.
\bibitem{Bec16}
S.~Beckus,
\textit{Spectral approximation of aperiodic Schrödinger operators},
Ph.D.\ thesis, Friedrich-Schiller-University Jena, Germany,
(2016),
\url{https://arxiv.org/abs/1610.05894}.
\bibitem{MP2018SpecrtalContinuityI}
S.~Beckus and J.~Bellissard and G.~De~Nittis,
\textit{Spectral continuity for aperiodic quantum systems I. General theory},
Elsevier, Journal of Functional Analysis \textbf{275}, no.\ 11,
(2018),
2917--2977.
\bibitem{MP2019SpectralLipschitzContinuity}
S.~Beckus and J.~Bellissard and H.~Cornean,
\textit{Hölder continuity of the spectra for aperiodic Hamiltonians},
Springer, Annales Henri Poincar{\'e} \textbf{20}, no.\ 11,
(2019),
3603--3631.
\bibitem{BBdN20}
S.~Beckus and J.~Bellissard and G.~De~Nittis,
\textit{Spectral continuity for aperiodic quantum systems: Applications of a folklore theorem},
AIP Publishing LLC, Journal of Mathematical Physics \textbf{61}, no.\ 12,
(2020),
123505.
\bibitem{MP1982BellissardReviewPaper}
J.~Bellissard,
\textit{Schrödinger operators with almost periodic potemntial: An overview},
Springer, Mathematical Problems in Theoretical Physics,
(1982),
356--363.
\bibitem{FH4}
J.~Bellissard and B.~Iochum and E.~Scoppola and D.~Testard,
\textit{Spectral properties of one-dimensional quasi-crystals},
Springer, Communications in Mathematical Physics \textbf{125}, no.\ 3,
(1989),
527--543.
\bibitem{FH3}
J.~Bellissard and B.~Iochum and D.~Testard,
\textit{Continuity properties of the electronic spectrum of 1 D quasicrystals},
Springer, Communications in Mathematical Physics \textbf{141}, no.\ 2,
(1991),
353--380.
\bibitem{Bel94}
J.~Bellissard,
\textit{Lipshitz continuity of gap boundaries for Hofstadter-like spectra},
Springer, Communications in Mathematical Physics \textbf{160}, no.\ 3,
(1994),
599--613.
\bibitem{BS91}
V.~G.~Benza and C.~Sire,
\textit{Band spectrum of the octagonal quasicrystal: Finite measure, gaps, and chaos},
APS, Physical Review B \textbf{44}, no.\ 18,
(1991),
10343.
%CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
\bibitem{onethird}
M.-D.~Choi and G.~A.~Elliott and N.~Yui,
\textit{Gauss polynomials and the rotation algebra},
Springer, Inventiones Mathematicae \textbf{99}, no.\ 1,
(1990),
225--246.
\bibitem{CP12}
H.~D.~Cornean and R.~Purice,
\textit{On the regularity of the Hausdorff distance between spectra of perturbed magnetic Hamiltonians},
Springer, Spectral analysis of quantum Hamiltonians,
(2012),
55--66.
\bibitem{CP15}
H.~D.~Cornean and R.~Purice,
\textit{Spectral edge regularity of magnetic Hamiltonians},
Oxford University Press, Journal of the London Mathematical Society \textbf{92}, no.\ 1,
(2015),
89--104.
%DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD
\bibitem{FH2}
D.~Damanik and A.~Gorodetski and W.~Yessen,
\textit{The Fibonacci Hamiltonian},
Springer, Inventiones Mathematicae \textbf{206}, no.\ 3,
(2016),
629--692.
\bibitem{MP2017DamanikReviewPaper}
D.~Damanik,
\textit{Schrödinger operators with dynamically defined potentials},
Cambridge University Press, Ergodic Theory and Dynamical Systems \textbf{37}, no.\ 6,
(2017),
1681--1764.
\bibitem{DD63}
J.~Dixmier and A.~Douady,
\textit{Champs continus d’espaces hilbertiens et de C*-algebras},
Bull.\ Soc.\ Math.\ France \textbf{91},
(1963),
227--284.
\bibitem{Dix69}
J.~Dixmier and C.~Les,
\textit{algebras et leurs repr{\'e}sentations, Deuxiemeedition},
Cahiers Scientifiques, Fasc. XXIX, Gauthier-Villars {\'E}diteur, Paris,
(1969),
11.
%EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE
\bibitem{Ell82}
G.~A.~Elliott,
\textit{Gaps in the spectrum of an almost periodic Schrödinger operator},
K{\o}benhavns Universitet.\ Matematisk Institut,
(1983).
%FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
\bibitem{GR2015[Folland]}
G.~B.~Folland,
\textit{A course in abstract harmonic analysis},
CRC Press \textbf{29},
(2015),
51--53.
%GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG
\bibitem{GMS20}
M.~Gerhold and O.~Moshe Shalit,
\textit{Dilations of $ q $-commuting unitaries},
International Mathematics Research Notices,
(2019).
%HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
\bibitem{HS87}
B.~Helffer and J.~Sjöstrand,
\textit{Effet tunnel pour l'{\'e}quation de Schr{\"o}dinger avec champ magn{\'e}tique},
Annali della Scuola Normale Superiore di Pisa-Classe di Scienze \textbf{14}, no.\ 4,
(1987),
625--657.
\bibitem{Hof76}
D.~R.~Hofstadter,
\textit{Energy levels and wave functions of Bloch electrons in rational and irrational magnetic fields},
APS, Physical Review B \textbf{14}, no.\ 6,
(1976),
2239.
%IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
%JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ
\bibitem{almostLipschitz}
S.~Jitomirskaya and Y.~Last,
\textit{Anderson localization for the almost Mathieu equation, III. Semi-uniform localization, continuity of gaps, and measure of the spectrum},
Springer, Communications in Mathematical Physics \textbf{195}, no.\ 1,
(1998),
1--14.
%\bibitem{MP2007JitomirskayaReviewPaper}
%S.~Jitomirskaya,
%\textit{Ergodic Schrodinger Operators (on one foot)},
%Providence, RI; American Mathematical Society; 1998, Proceedings of Symposia in Pure Mathematics \textbf{76}, no.\ 2,
%(2007),
%613.
\bibitem{JM12}
S.~Jitomirskaya and C.~A.~Marx,
\textit{Analytic quasi-periodic Schr{\"o}dinger operators and rational frequency approximants},
Springer, Geometric and Functional Analysis \textbf{22}, no.\ 5,
(2012),
1407--1443.
\bibitem{MP2019AMO}
S.~Jitomirskaya,
\textit{Critical phenomena, arithmetic phase transitions, and universality: some recent results on the almost Mathieu operator},
Current Developments in Mathematics,
(2019).
\bibitem{MP2019QPSOp}
S.~Jitomirskaya and W.~Liu and S.~Zhang,
\textit{Arithmetic spectral transitions: a competition between hyperbolicity and the arithmetics of small denominators},
IAS/Park City Mathematics Series \textbf{27},
(2020),
35--72.
%KKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKKK
\bibitem{Kap51}
I.~Kaplansky,
\textit{The structure of certain operator algebras},
JSTOR, Transactions of the American Mathematical Society \textbf{70}, no.\ 2,
(1951),
219--255.
\bibitem{KP19}
J.~Kellendonk and E.~Prodan,
\textit{Bulk--boundary correspondence for Sturmian Kohmoto-like models},
Springer, Annales Henri Poincar{\'e} \textbf{20}, no.\ 6,
(2019),
2039--2070.
\bibitem{Physics1980QuantizedHallEffect}
K.~von~Klitzing and G.~Dorda and M.~Pepper,
\textit{New method for high-accuracy determination of the fine-structure constant based on quantized Hall resistance},
APS, Physical Review Letters \textbf{45}, no.\ 6,
(1980),
494.
%LLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLLL
%\bibitem{MP2005LastReviewPaper}
%Y.~Last,
%\textit{Spectral theory of Sturm-Liouville operators on infinite intervals: a review of recent developments},
%Springer, Sturm-Liouville Theory,
%(2005),
%99--120.
%MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
\bibitem{MP05}
M.~M{\u{a}}ntoiu and R.~Purice,
\textit{Strict deformation quantization for a particle in a magnetic field},
American Institute of Physics, Journal of Mathematical Physics \textbf{46}, no.\ 5,
(2005),
052105.
%\bibitem{MP2017MarxJitomirskayaReviewPaper}
%C.~A.~Marx and S.~Jitomirskaya,
%\textit{Dynamics and spectral theory of quasi-periodic Schrödinger-type operators},
%Cambridge University Press, Ergodic Theory and Dynamical Systems \textbf{37}, no.\ 8,
%(2017),
%2353--2393.
%NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
\bibitem{DS2006[Nevo]}
A.~Nevo,
\textit{Pointwise Ergodic Theorems for Actions of Groups},
Elsevier B.V., Handbook of Dynamical Systems, vol.\ 1B,
(2006),
871--982.
%OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
\bibitem{OK85}
S.~Ostlund and S.-H.~Kim,
\textit{Renormalization of quasiperiodic mappings},
IOP Publishing, Physica Scripta \textbf{1985}, no.\ T9,
(1985),
193.
%PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPP
\bibitem{PR16}
D.~Parra and S.~Richard,
\textit{Continuity of the spectra for families of magnetic operators on $ \Z^d $},
Springer, Analysis and Mathematical Physics \textbf{6}, no.\ 4,
(2016),
327--343.
\bibitem{AG}
J.-P.~Pier,
\textit{Amenable locally compact groups},
Wiley-Interscience,
(1984).
\bibitem{expfast}
E.~Prodan,
\textit{Quantum transport in disordered systems under magnetic fields: A study based on operator algebras},
Oxford University Press, Applied Mathematics Research eXpress \textbf{2013}, no.\ 2,
(2013),
176--265.
%QQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQQ
%RRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR
%SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
\bibitem{FH1}
D.~Shechtman and I.~Blech and D.~Gratias and J.~W.~Cahn,
\textit{Metallic phase with long-range orientational order and no translational symmetry},
APS, Physical Review Letters \textbf{53}, no.\ 20,
(1984),
1951.
\bibitem{MP1982SimonReviewPaper}
B.~Simon,
\textit{Almost periodic Schrödinger operators: a review},
Elsevier, Advances in Applied Mathematics \textbf{3}, no.\ 4,
(1982),
463--490.
\bibitem{FH5}
A.~S{\"u}t{\HT{o}},
\textit{Singular continuous spectrum on a Cantor set of zero Lebesgue measure for the Fibonacci Hamiltonian},
Springer, Journal of statistical physics \textbf{56}, no.\ 3,
(1989),
525--531.
%TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT
\bibitem{Physics1982QuantizedHallEffect}
D.~J.~Thouless; M.~Kohmoto; M.~P.~Nightingale; M.~den~Nijs,
\textit{Quantized Hall conductance in a two-dimensional periodic potential},
APS, Physical Review Letters \textbf{49}, no.\ 6,
(1982),
405.
%UUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUU
%VVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVV
%WWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWW
%XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
%YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY
%ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ
\end{thebibliography}
%\clearpage
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%\clearpage