A Characteristic free approach to skew-gentle algebras
Extracted source text ยท arXiv 2212.06467
\documentclass[11pt]{article}
\usepackage{amsthm,amssymb,amsmath,amscd}
\usepackage[all]{xy}
\usepackage{mathrsfs}
\usepackage{multirow}
\usepackage{color}
\usepackage{mathptm}
\usepackage{url}
%\usepackage[notcite,notref]{showkeys}
\usepackage{caption}
\usepackage{subfigure}
\usepackage{graphicx}
\usepackage{tikz-cd}\usepackage{enumitem}
\DeclareMathAlphabet{\mathcal}{OMS}{cmsy}{m}{n}
\DeclareSymbolFont{largesymbols}{OMX}{cmex}{m}{n}
%\pagestyle{empty}
\oddsidemargin 8pt \voffset=-1.5cm
\addtolength{\textheight}{4cm}\addtolength{\textwidth}{3cm}
\newtheorem{Def}{Definition}[section]
\newtheorem{Prop}[Def]{Proposition}
\newtheorem{Theo}[Def]{Theorem}
\newtheorem{Lem}[Def]{Lemma}
\newtheorem{Koro}[Def]{Corollary}
\newtheorem{Claim}[Def]{Claim}
\newtheorem{Remark}[Def]{Remark}
\newtheorem{Example}[Def]{Example}
%\newenvironment{proof}{{\it proof: }}{$\square$\hfill}
%%%%%%%%%%%%%%%%%%%% definition of commands %%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newcommand{\add}{\operatorname{add}\,}
\newcommand{\con}{\operatorname{con}}
\newcommand{\soc}{{\rm soc}}
\newcommand{\rad}{\operatorname{rad}\,}
\renewcommand{\top}{\operatorname{top}}
\newcommand{\Tr}{{\rm Tr}}
\newcommand{\Coker}{{\rm Coker}}
\newcommand{\Ker}{{\rm Ker\,}}
\newcommand{\opp}{^{\rm op}}
\newcommand{\Imf}{{\rm Im}}
\newcommand{\domdim}{\operatorname{dom.dim}\,}
\newcommand{\vdomdim}{\operatorname{\nu-dom.dim}\,}
\newcommand{\repdim}{\operatorname{rep.dim}\,}
\newcommand{\gldim}{\operatorname{gl.dim}\,}
\newcommand{\findim}{\operatorname{fin.dim}}
\newcommand{\pd}{{\rm pd }}
\newcommand{\id}{{\rm id}}
\newcommand{\per}{{\rm per}}
\newcommand{\ann}{{\rm Ann}}
\newcommand{\Tria}{{\rm Tria}}
\newcommand{\Hom}{\operatorname{Hom}}
\newcommand{\rHom}{{\rm\bf R}{\rm Hom}}
\newcommand{\HomP}{{\rm Hom}^{\bullet}}
\newcommand{\StHom}{{\rm \underline{Hom}}}
\newcommand{\otimesL}{\otimes^{\rm\bf L}}
\newcommand{\otimesP}{\otimes^{\bullet}}
\newcommand{\End}{\operatorname{End}}
\newcommand{\Ext}{\operatorname{Ext}}
\newcommand{\Tor}{\operatorname{Tor}}
\newcommand{\w}{\widetilde}
%\newcommand{\E}[4]{{\rm E}_{#1}^{#2}(#3, #4)}
%\newcommand{\Ex}[3]{{\rm E}_{#1}^{#2}(#3)}
%\newcommand{\EP}[4]{{\rm E}_{#1}^{#2\bullet}(#3, #4)}
\newcommand{\E}{\textsf{E}}
\newcommand{\Oc}[3]{{#1}^{#2,#3}}
\newcommand{\cpx}[1]{#1^{\bullet}}
\newcommand{\D}[1]{\mathscr{D}(#1)}
\newcommand{\Dz}[1]{{\mathscr{D}}^+(#1)}
\newcommand{\Df}[1]{{\mathscr{D}}^-(#1)}
\newcommand{\Db}[1]{ \mathscr{D}^{\rm b}(#1)}
\newcommand{\Dsg}[1]{ \mathscr{D}_{\rm sg}(#1)}
\newcommand{\Ddef}[1]{ \mathscr{D}_{\rm def}(#1)}
\newcommand{\Dif}{\rm{Dif}}
\newcommand{\Lt}{\rm{L}}
\newcommand{\Rt}{\rm{R}}
\newcommand{\WDb}[1]{\widetilde{\mathscr{D}^{\rm b}(#1)}}
\newcommand{\C}[1]{\mathscr{C}(#1)}
\newcommand{\Cz}[1]{{\mathscr{C}}^+(#1)}
\newcommand{\Cf}[1]{{\mathscr{C}}^-(#1)}
\newcommand{\Cb}[1]{{\mathscr{C}^b}(#1)}
\newcommand{\K}[1]{\mathscr{K}(#1)}
\newcommand{\Kz}[1]{\mathscr{K}^+(#1)}
\newcommand{\Kzb}[1]{\mathscr{K}^{+, b}(#1)}
\newcommand{\Kf}[1]{\mathscr{K}^-(#1)}
\newcommand{\Kfb}[1]{\mathscr{K}^{-, b}(#1)}
\newcommand{\Kb}[1]{ \mathscr{K}^{\rm b}(#1)}
\newcommand{\modcat}[1]{#1\mbox{{\rm -mod}}}
\newcommand{\percat}[1]{#1\mbox{{\rm -perfect}}}
\newcommand{\Modcat}[1]{#1\mbox{{\rm -Mod}}}
\newcommand{\stmodcat}[1]{#1\mbox{{\rm -{\underline{mod}}}}}
\newcommand{\stpmodcat}[1]{#1\mbox{{\rm -{\underline{Gproj}}}}}
\newcommand{\istmodcat}[1]{#1\mbox{{\rm -{\overline{mod}}}}}
\newcommand{\pmodcat}[1]{#1\mbox{{\rm -proj}}}
\newcommand{\gmodcat}[1]{#1\mbox{{\rm -Gproj}}}
\newcommand{\fmodcat}[1]{#1\mbox{{\rm -flat}}}
\newcommand{\imodcat}[1]{#1\mbox{{\rm -inj}}}
\newcommand{\HH}{\rm HH}
\newcommand{\Imodcat}[1]{#1\mbox{{\rm -Inj}}}
\newcommand{\Pmodcat}[1]{#1\mbox{{\rm -Proj}}}
\newcommand{\lra}{\longrightarrow}
\newcommand{\rra}{\longleftarrow}
\newcommand{\ra}{\rightarrow}
\newcommand{\lraf}[1]{\stackrel{#1}{\lra}}
\newcommand{\wi}{\widetilde}
\def\scr#1{\mathscr{#1}}
\def\mcal#1{\mathcal{#1}}
\def\rann{{\mathcal{R}}}
\def\lann{{\mathcal{L}}}
\def\fac{{\mathcal{F}}}
\def\calI{\mathcal{I}}
\def\calJ{\mathcal{J}}
%%%%%%%%%%%%%%%%%%% End of commands %%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\title{A characteristic free approach for skew-gentle algebras}
\author{Yiping Chen }
\date{}
\begin{document}
\maketitle
\renewcommand{\thefootnote}{\alph{footnote}}
\setcounter{footnote}{-1} \footnote{2010 Mathematics Subject
Classification: 16E35, 16G10; 16P10, 16P20.}
\renewcommand{\thefootnote}{\alph{footnote}}
\setcounter{footnote}{-1} \footnote{Keywords:Derived equivalences; Gentle algebras; Skew-gentle algebras; Recollements}
\begin{abstract} In this article, we write an explicit recollement for the derived category of every skew-gentle algebra. As an application, we show that gentle algebras and skew-gentle algebras share some common properties independent of the characteristic of the field. For example, all skew-gentle algebras are Gorenstein. An indecomposable skew-gentle algebra is selfinjective if and only if the set of special loops is empty and the corresponding gentle algebra is selfinjective. Finally, we show that
the finitistic dimension conjecture and the Auslander-Reiten conjecture hold for all skew-gentle algebras.
\end{abstract}
%\tableofcontents
\section{Introduction}
Gentle algebras are first introduced in \cite{AS1987} to classify the iterated tilted algebras of type $\widetilde{\mathbb{A}}$. Gentle algebras are Gorenstein \cite{Geis2005}, and are stable under derived equivalences \cite{SZ2003}. And the derived categories and the singularity categories can be described explicitly \cite{Bekkert2003, Kalck2015}.
Recently, gentle algebras have became increasingly important in many branches of mathematics. Examples are ranging from the Jacobian algebra of a triangulation of an unpunctured surface \cite{ABC2010, Fragoso2009} to the closed connection between graded gentle algebras and Fukaya categories \cite{HKK2017} and geometric descriptions of the module categories, derived categories and derived equivalences \cite{Baur2021, Opper2018, Amiot2019}.
Skew-gentle algebras, a generalization of gentle algebras, are first introduced in \cite{GP1999} to study a class of tame matrix problems. And skew-gentle algebras also attract more attention for the connection with cluster theory \cite{Geiss2016} and geometric models of the module categories and the derived categories \cite{Amiot2022, He2020, Fragoso2022}.
Let $k$ be a field, and $A$ be a finite dimensional $k$-algebra. We write $char(k)$ for the characteristic of $k$. If $char(k)\neq 2$, then a skew-gentle algebra $A$ is Morita equivalent to a skew group algebra $\Gamma G$ with $\Gamma$ a gentle algebra and $G=\langle e, g\mid g^2=e\rangle$ a finite group \cite{GP1999}.
For a skew group algebra $\Gamma G$, when the order of $G$ is invertible in $k$, skew group constructions preserve many invariants such as Gorensteiness, selfinjectivity, finitistic dimensions (\cite[Theorem 1.3]{Reiten1985b}, \cite[Proposition 3.6]{huang2012}). So, gentle algebras and skew-gentle algebras share some common properties in this situation. A natural question is:
{\em If or not these properties are preserved when $char(k)=2$? More generally, is it possible to supply a characteristic free approach to investigate skew-gentle algebras via gentle algebras? }
In this paper, we will give an affirmative answer to this question.
Let $(Q, I)$ be a gentle pair with $A(Q, I):=kQ/\langle I\rangle$ the corresponding gentle algebra. Let $Sp$ be a set of special loops. Then, the triple $(Q, I, Sp)$ is called a {\em skew-gentle triple} if $(Q', I\cup \{\alpha^2\mid \alpha\in Sp\})$ is a gentle pair where $Q'$ is the quiver by adding the special loops to $Q$. A skew-gentle algebra associated with a skew-gentle triple $(Q, I, Sp)$ is defined to be the algebra $kQ'/\langle I\cup \{\alpha^2-\alpha\mid \alpha\in Sp\}\rangle$. Obviously, if $Sp=\emptyset$, then the corresponding skew-gentle algebra is just the gentle algebra $A(Q, I)$.
For the precise definitions of skew-gentle triples and skew-gentle algebras, we refer the readers to Section $3$.
In this article, we assume $Sp\neq\emptyset$. Let $Q^{Sp}_0$ be the set of special vertices where special loops are attached, and $Q^{or}_0$ be the set of the remaining ordinary vertices. The quiver of a skew-gentle algebra is presented by splitting every special vertex into two. More precisely, the set of vertices is a disjoint union of $Q^{or}_0, Q^{Sp+}_0$ and $Q^{Sp-}_0$ with $Q^{Sp+}_0:=\{i^+\mid i\in Q^{Sp}_0\}, Q^{Sp-}_0:=\{i^-\mid i\in Q^{Sp}_0\}$.
Let $e$ be the sum of idempotents corresponding to the vertices in $Q^{Sp-}_0$. Then, the Pierce decomposition of $A(Q, I, Sp)$ can be written as
$$A=\begin{pmatrix}(1-e)A(1-e)&(1-e)Ae\\eA(1-e)&eAe\end{pmatrix}.
$$
where we write $A$ for $A(Q, I, Sp)$ for short. We write $E_{22}$ for the idempotent matrix whose $(2, 2)$-entry is $e$, and other entries are zero. In this paper, we prove that $A E_{22}A$ is a stratifying ideal in the sense of Cline, Parshall and Scott \cite{Cline1996}. This yields a stratifying structure for the derived category of $A(Q, I, Sp)$.
Throughout this paper, we assume that $k$ is an arbitrary field. The main result is stated as follows.
\begin{Theo}\label{main result} Let $A(Q, I, Sp)$ be a skew-gentle algebra with $Sp\neq\emptyset$. Then,
\begin{enumerate}[label=(\alph*)]
\item
We have a recollement
\begin{equation}\label{recollement}\xymatrix{
\D{\Modcat{A/A E_{22}A}}\ar[rr]^{i_*=i_!}&&\D{\Modcat{A}}\ar@/^1pc/[ll]^{i^!}\ar@/_1pc/[ll]_{i^*}\ar[rr]^{j^!=j^*}&&\D{\Modcat{E_{22}A E_{22}}}\ar@/_1pc/[ll]_{j_!}\ar@/^1pc/[ll]^{j_*}},\end{equation}
where $$i^*=A/A E_{22}A\otimesL_A-, \; i_*=A/A E_{22}A\otimesL_{A/A E_{22}A}-, i^!=\rHom_A(A/A E_{22}A, -),$$
$$j_!=A E_{22}\otimesL_{E_{22}A E_{22}}-, \;j^!=E_{22}A\otimesL_A-,\; j_*=\rHom_{E_{22}A E_{22}}(E_{22}A, -).$$
\item
The algebra $A/A E_{22}A$ is isomorphic to $A(Q, I)$ as algebras.
\item
The indecomposable direct factor of $E_{22}A E_{22}$ is $k$ or a hereditary algebra of the linearly oriented quiver of type $\mathbb{A}_n$ for $n\geqslant 2$.
\end{enumerate}
\end{Theo}
As applications, we consider Gorensteiness, selfinjectivity, singularity categories, the finitistic dimension conjecture and the Auslander-Reiten conjecture for skew-gentle algebras.
The famous finitistic dimension conjecture and the Auslander-Reiten conjecture, which are still open now, play important roles in representation theory of algebra and homological algebra. They are stated as follows.
{\bf The finitistic dimension conjecture}: The finitistic dimension of $A$ (the supremum of the projective dimensions of finitely generated $A$-modules with finite projective dimension) is finite.
{\bf The Auslander-Reiten Conjecture}: A finitely generated $A$-module $X$ satisfying $\Ext_A^i(X, X\oplus A)=0$ for all $i>0$ is necessarily projective.
There is a close connection between the finitistic dimension conjecture and the Auslander-Reiten conjecture. More precisely, the finitistic dimension conjecture holds for all algebras if and only if so does the Auslander-Reiten conjecture. And it is still unknown whether the Auslander-Reiten conjecture holds automatically for an algebra satisfying the finitistic dimension conjecture \cite[Theorem 3.4.3]{Yamagata1996}.
The finitistic dimension conjecture holds for some classes of algebras such as monomial algebras, algebras with radical cube zero, and algebras whose representation dimension no more than $3$ \cite{Green1991, Green1991a, Igusa2005}. For the history of this conjecture, we refer the readers to \cite{Zimmermann1995} for details.
The Auslander-Reiten conjecture holds for algebras of finite representation type, torsionless-finite algebras, symmetric biserial algebras, algebras with radical square zero, local algebras with radical cube zero, monomial algebras \cite{Auslander1975c, Xu2013a, Xu2015, Marczinzik2019}. It also holds for certain commutative rings that are not necessarily finite dimensional algebras, such as
Gorenstein commutative local rings with radical cube zero, complete intersection local rings and some commutative CM rings \cite{Auslander1993, Hoshino1984, Huneke2004a}.
\begin{Koro}\label{coro}
\begin{enumerate}[label=(\alph*)]
\item
Skew-gentle algebras are Gorenstein.
\item
The singularity categories $\Dsg{A(Q, I)}$ and $\Dsg{A(Q, I, Sp)}$ are equivalent as triangulated categories.
\item
An indecomposable skew-gentle algebra $A(Q, I, Sp)$ is selfinjective if and only if $Sp=\emptyset$ and the gentle algebra $A(Q, I)$ is selfinjective. Equivalently, $Sp=\emptyset$ and the gentle $A(Q, I)$ is simple or a selfinjective Nakayama algebra with radical square zero.
\item
The finitistic dimension conjecture holds for all skew-gentle algebras.
\item
The Auslander-Reiten conjecture holds for all skew-gentle algebras.
\end{enumerate}
\end{Koro}
This paper is organized as follows: In Section 2, we recall some notations and definitions. In Section 3, some basic facts about gentle algebras and skew-gentle algebras will be given. And in the last section, we will prove the main results.
\medskip
\section{Notations and definitions}
In this section, we will fix some notations and recall some basic definitions needed in our proofs.
Throughout this article, $k$ denotes a fixed field with arbitrary characteristic. Let $A$ be a finite dimensional $k$-algebra. We write $A^e$ for the enveloping algebra
$A\otimes_kA^{op}$. We denote by $\Modcat{A}$ the category
of left $A$-modules, and by $\modcat{A}$ the category of all finitely
generated left $A$-modules. The notation Mod-$A$ means the category of right
$A$-modules. $\Pmodcat{A}$ (resp. $\pmodcat{A}$) is the full subcategory of $\Modcat{A}$ (resp. $\modcat{A}$) with objects the (resp. finitely generated) projective left $A$-modules. Dually, $\Imodcat{A}$ (resp. $\imodcat{A}$) is the full subcategory of $\Modcat{A}$ (resp. $\modcat{A}$) with objects the (resp. finitely generated) injective left $A$-modules.
By $\add(M)$, we shall mean the full subcategory of
$A$-Mod, whose objects are direct summands of finite direct sums of
copies of $M$.
For two morphisms $f:{}X \ra Y$ and $g:{}Y \ra Z$ in
$\modcat{A}$, the composition of $f$ and $g$ is written as $fg$, which is
a morphism from $X$ to $Z$. However, the composition of functors is from right to left. Namely, for two functors $F:{}\mathcal{C}\ra \mathcal{D}$ and $G:{}\mathcal{D}\ra \mathcal{E}$, we write $GF$ for their composite instead of $FG$, where $\mathcal{C}, \mathcal{D}$ and $\mathcal{E}$ are categories.
Let $A$ be a finite dimensional $k$-algebra.
We denote by $\C{A}$ the category of complexes of finitely generated left $A$-modules, and by $\D{\Modcat{A}}$ the derived category of $\Modcat{A}$. The notions $\Kb{A}$ and $\Db{A}$ are the bounded homotopy and the bounded derived category of $\modcat{A}$, respectively.
\medskip
In \cite{Beuilinson1982}, recollements were first introduced by Beilinson, Bernstein and Deligne to study perverse sheaves. A {\em recollement} of triangulated $k$-categories is a diagram
$$\xymatrix{
\mathcal{T}'\ar[rr]^{i_*=i_!}&&\mathcal{T}\ar@/^1pc/[ll]^{i^!}\ar@/_1pc/[ll]_{i^*}\ar[rr]^{j^!=j^*}&&\mathcal{T}''\ar@/_1pc/[ll]_{j_!}\ar@/^1pc/[ll]^{j_*}}$$
of triangulated categories and triangle functors such that
$(1)$ $(i^*, i_*), (i_!, i^!), (j_!, j^!), (j^*, j_*)$ are adjoint pairs;
$(2)$ $i_*, j_*, j_!$ are fully and faithful;
$(3)$ $i^!j_*=0, j^!i_!=0, i^*j_!=0$;
$(4)$ Let $M$ be an object in $\mathcal{T}$. Then there are canonical triangles
$$i_!i^!(M)\lra M\lra j_*j^*(M)\lra i_!i^!(M)[1]$$
$$j_!j^!(M)\lra M\lra i_*i^*(M)\lra j_!j^!(M)[1]$$
where the morphisms are given by adjunctions.
There exists a close connection between recollements and homological epimorphisms.
Recall that
a ring homomorphism $\varphi:{}A\lra B$ is a {\em homological epimorphism} if $\varphi$ is a ring epimorphism and $\Tor^A_i(B, B)=0$ for all $i>0$.
By \cite[Theorem 4.4]{GL1991} and \cite[Section 4]{NicolasSaorin2009a}, $\varphi$ is a homological epimorphism if and only if the induced map $\Ext^i_B(M, N)\lra \Ext^i_A(\varphi^*(M), \varphi^*(N))$ is bijective for all $i\geqslant 0$, equivalently,
the induced functor $\varphi_*:{}\D{\Modcat{B}}\lra \D{\Modcat{A}}$ is fully faithful. Let $q:{} \D{\Modcat{A}}\lra \D{\Modcat{A}}/\Imf(\varphi_*)$ be the Verdier quotient of $\D{\Modcat{A}}$ via the essential image of $\varphi_*$. Note that $\varphi_*$ has a left adjoin $B\otimesL_A-$ and a right adjoint $\rHom_A(B, -)$. Then, by \cite[Proposition 9.1.18]{Neeman2001}, the Bousfield (co)localisation functor exists. Namely, the Verdier quotient $q$ has a left adjoint and a right adjoint.
In this way, we get
a recollement
$$\xymatrix{
\D{\Modcat{B}}\ar[rr]^{\varphi_*}&&\D{\Modcat{A}}\ar@/^1pc/[ll]^{\rHom_A(B, -)}\ar@/_1pc/[ll]_{B\otimesL_A-}\ar[rr]^{}&&\chi\ar@/_1pc/[ll]\ar@/^1pc/[ll]}$$
where $\chi$ is a triangulated category.
Let $\varphi:{}A\lra B$ be a surjective homomorphism. Assume that $\varphi$ is a homological epimorphism with the kernel $AeA$ where $e$ is an idempotent of $A$. By \cite[Remark 2.1.2 and Definition 2.1.1]{Cline1996}, the kernel $AeA$ is a {\em stratifying ideal}, i.e., $Ae\otimesL_{eAe}eA\lra AeA$ is an isomorphism. By \cite[Example 4.5]{AKL2011}, this gives a recollement
$$\xymatrix{
\D{\Modcat{A/AeA}}\ar[rr]^{i_*}&&\D{\Modcat{A}}\ar@/^1pc/[ll]^{i^!}\ar@/_1pc/[ll]_{i^*}\ar[rr]^{j^!}&&\D{\Modcat{eAe}}\ar@/_1pc/[ll]_{j_!}\ar@/^1pc/[ll]^{j_*}}$$
where
$$i^*=A/AeA\otimesL_A-, \; i_*=A/AeA\otimesL_{A/AeA}-, i^!=\rHom_A(A/AeA, -),$$
$$j_!=Ae\otimesL_{eAe}-, \;j^!=eA\otimesL_A-,\; j_*=\rHom_{eAe}(eA, -).$$
\medskip
Let $A$ be an algebra, and $M$ be a finitely generated left $A$-module. The finitely generated module $M$ is called {\em Gorenstein projective} if there is an exact sequence $P^{\bullet}\in \C{A}:$
$$\cdots\lra P^{-2}\lra P^{-1}\stackrel{d^{-1}}\lra P^0\stackrel{d^0}\lra P^1\lra P^2\lra \cdots$$
such that $\Hom_A(P^{\bullet}, A)$ is exact and $X\cong \Imf(d^0)$. Clearly, all finitely generated projective left $A$-modules are finitely generated Gorenstein projective. For any $M\in\modcat{A}$, we write $\id_A(M)$ for the injective dimension of $M$.
An algebra $A$ is called {\em Gorenstein} if $\id_A(A)<\infty$ and $\id_{A^{\opp}}(A)<\infty$. We write $d$ for the injective dimension of the left regular module $A$. Then, the algebra $A$ is Gorenstein if and only if $\Omega^d(M)$ is Gorenstein-projective for all $M\in \modcat{A}$, where $\Omega$ is the syzygy functor \cite[Proposition 3.10]{Beligiannis2005}.
It is also equivalent to saying $\Kb{\pmodcat{A}}=\Kb{\imodcat{A}}$ \cite[Section 1]{Happel1991f}.
For convenience, we write $(\D{\Modcat{B}}, \D{\Modcat{A}}, \D{\Modcat{C}})$ for the recollement
\begin{equation}\label{R}
\xymatrix{
\D{\Modcat{B}}\ar[rr]^{i_*}&&\D{\Modcat{A}}\ar@/^1pc/[ll]^{i^!}\ar@/_1pc/[ll]_{i^*}\ar[rr]^{j^!}&&\D{\Modcat{C}}.\ar@/_1pc/[ll]_{j_!}\ar@/^1pc/[ll]^{j_*}}
\end{equation}
We say that the recollement (\ref{R}) can be restricted to $\Db{\rm mod}$ and $\Kb{\rm proj}$ if $(\Db{B}, \Db{A}, \Db{C})$ and $(\Kb{B}, \Kb{A}, \Kb{C})$ are recollements with the same triangle functors in (\ref{R}).
The following Propositions, taken from \cite{AKLY2017}, \cite{CK2017} and \cite{QinHan2016}, are about the connections between Gorensteiness and recollements.
\begin{Prop}\label{1} (\cite[Corollary 4.7]{AKLY2017}) If $\gldim{C}<\infty$, then the recollement (\ref{R}) can be restricted to $\Db{\rm mod}$.
\end{Prop}
\begin{Prop}\label{2} (\cite[Proposition 3.4]{CK2017}, \cite{QinHan2016})
Assume the recollement (\ref{R}) can be restricted to $\Db{\rm mod}$. Then, the following statements are equivalent:
\begin{enumerate}[label=(\alph*)]
\item
The algebra $A$ is Gorenstein.
\item
The algebras $A$, $B$ and $C$ are Gorenstein.
\item
The algebras $B$ and $C$ are Gorenstein, and the recollement (\ref{R}) can be restricted to $\Kb{{\rm proj}}$.
\item
The algebras $B$ and $C$ are Gorenstein, and $i^!(A)\in\Kb{\pmodcat{B}}$.
\end{enumerate}
\end{Prop}
\medskip
\section{Skew-gentle algebras}
Recall that a {\em quiver} $Q=(Q_0, Q_1, s, t)$ is an oriented graph, where $Q_0$ is the set of vertices and $Q_1$ is the set of arrows. The maps $s:{}Q_1\lra Q_0$ and $t:{}Q_1\lra Q_0$ are defined by $s(\alpha)=i$ and $t(\alpha)=j$ for any arrow $\alpha:{}i\lra j$. Namely, $s(\alpha)$ is the source of $\alpha$, and $t(\alpha)$ is the terminal of $\alpha$.
A sub-quiver of $Q$ is a quiver $(Q_0', Q_1', s', t')$ satisfying that $Q'_0\subseteq Q_0, Q'_1\subseteq Q_1$, and $s', t'$ are the restrictions of $s$ and $t$ to $Q'_1$, respectively. If both $Q_0$ and $Q_1$ are finite sets, then $Q$ is called a {\em finite quiver}. Throughout this article, all quivers are supposed to be finite.
Let $Q=(Q_0, Q_1, s, t)$ be a quiver. For any $l\in\mathbb{N}$, a path of length $l$ is an oriented sequences of arrows $p=\alpha_1\alpha_2\cdots\alpha_l$ such that $t(\alpha_i)=s(\alpha_{i+1})$ for all $i=1, \cdots, l-1$. When $l=0$, the path is called a {\em trivial path}, and is denoted by $e_i$ for $i\in Q_0$. A nontrivial path $p$ is called to be an {\em oriented cycle} if $s(p)=e(p)$. And an oriented cycle of length $1$ is also said to be a {\em loop}.
A path algebra $kQ$ is a $k$-algebra whose underlying $k$-vector space has its basis the set of all paths in $Q$. Let $p$ and $p'$ be two paths in $Q$. The product of $p$ and $p'$ is defined by $\delta_{t(p)s(p')}pp'$, where $\delta$ is the Kronecker delta.
A relation $\sigma$ is a $k$-linear combination of paths. More precisely, $\sigma=k_1p_1+k_2p_2+\cdots +k_np_n$ with $k_i\in k$ and $s(p_i)=s(p_j), e(p_i)=e(p_j)$ for all $1\leqslant i, j\leqslant n$. Let $\rho$ be the set of relations. The pair $(Q, \rho)$ is called {\em a quiver with relations}. We write $kQ/\langle \rho\rangle$ for the quotient algebra of $kQ$ with respect to the ideal generated by the set $\rho$ in $kQ$. Let $J$ be the ideal generated by all arrows in $kQ$. The ideal $\langle \rho\rangle$ is called {\em admissible} if $J^t\subseteq \langle \rho\rangle \subseteq J^2$ for some positive integer $t$. Let $Q$ be a finite quiver, and $\langle \rho\rangle$ be an admissible ideal of $kQ$. Then $kQ/\langle \rho\rangle$ is a finite dimensional algebra. Throughout this paper, we write $A(Q, \rho)$ for the path algebra corresponding to the pair $(Q, \rho)$.
For more facts about path algebras, we refer the readers to \cite[Chapter III]{Auslander1995} and \cite[Chapter II]{Assem2006}.
\medskip
A quiver with relations $(Q, I)$ is called a {\em gentle pair}, if the following hold.
\begin{itemize}
\item
Each vertex of $Q$ is start point of at most two arrows, and end point of at most two arrows.
\item
For each arrow $\alpha$ in $Q$, there is at most one arrow $\beta$ with $t(\alpha)=s(\beta)$ such that $\alpha\beta\notin I$, and at most one arrow $\gamma$ with $t(\gamma)=s(\alpha)$ such that $\gamma\alpha\notin I$.
\item
For each arrow $\alpha$ in $Q$, there is at most one arrow $\beta$ with $t(\alpha)=s(\beta)$ such that $\alpha\beta\in I$, and there is at most one arrow $\gamma$ with $t(\gamma)=s(\alpha)$ such that $\gamma\alpha\in I$.
\item
The algebra $A(Q, I)$ is finite dimensional.
\end{itemize}
A finite dimensional algebra $A(Q, I)$ is called a {\em gentle algebra} if $(Q, I)$ is a gentle pair. Let $(Q, I)$ be a gentle pair. We add some special loops in $Q$, and denote the set of special loops by Sp. We call $(Q, I, Sp)$ the {\em skew-gentle triple} if $(Q', I\cup\{\alpha^2\mid \alpha\in Sp\})$ is a gentle pair where $Q'$ is the quiver by adding the special loops to $Q$. The finite dimensional algebra $A(Q, I, Sp)=kQ/\langle I\cup \{\alpha^2-\alpha\mid \alpha\in Sp\}\rangle$ is called a {\em skew-gentle algebra} corresponding to the skew-gentle triple $(Q, I, Sp)$. Obviously, a gentle algebra is a skew-gentle algebra when Sp is empty. It is also equivalent to saying that $(Q, I)$ is a gentle pair for any skew-gentle triple $(Q, I, Sp)$.
Note that the ideal $\langle I\cup \{\alpha^2-\alpha\mid \alpha\in Sp\}\rangle$ is not admissible. Then, we can not read off the quiver of a skew-gentle algebra by the corresponding skew-gentle triple $(Q, I, Sp)$. For a skew-gentle triple $(Q, I, Sp)$, the vertices of $Q$ consists of the special vertices where a special loop is attached and the remaining ordinary vertices. We write the sets of special vertices and ordinary vertices by $Q_0^{Sp}$ and $Q_0^{or}$, respectively. Now, we split every special vertex into two. More precisely, for a special vertex $i\in Q_0^{Sp}$, we get two vertices $i^-$ and $i^+$. This yields two sets of vertices:
$$Q_0^{Sp-}:=\{i^-\mid i\in Q_0^{Sp}\}; \quad Q_0^{Sp+}:=\{i^+\mid i\in Q_0^{Sp}\}.$$
The quiver of the skew-gentle algebra $A(Q, I, Sp)$ is described as follows:
\begin{itemize}
\item
The set of vertices $Q^A_0$: $Q_0^{or}\cup Q_0^{Sp+}\cup Q_0^{Sp-}$.
\item
The set of arrows $Q^A_1$: For an arrow $\alpha:{}i\lra j$ in $Q$,
\begin{itemize}
\item
If both $i$ and $j$ are ordinary, then $\alpha$ is an arrow in $Q_1^A$.
\item
If $i\in Q_0^{Sp}$ and $j\in Q_0^{or}$, then there are two arrows
$$\xymatrix@=5mm{
i^+\ar[rd]^{{}^+\alpha}&\\&j\\i^-\ar[ur]_{{}^-\alpha}&
}$$ in $Q_1^A$.
\item
Dually, when $i\in Q_0^{or}$ and $j\in Q_0^{Sp}$, we also get two arrows $\alpha^+:{}i\lra j^+$ and $\alpha^-:{}i\lra j^-$ in $Q_1^A$.
\item
If both $i$ and $j$ are in $Q_0^{Sp}$, we have four arrows in $Q_1^A$.
$$\xymatrix@=15mm{
i^+\ar[r]^{{}^{+}\alpha^+}\ar[rd]^(.3){{}^+\alpha^-}&j^+\\
i^-\ar[r]^{{}^-\alpha^-}\ar[ur]^(.3){{}^-\alpha^+}&j^-
}$$
\end{itemize}
\end{itemize}
Note that $I$ consists of some paths of length $2$. The relations $I^A$ is presented as follows. For a path
$$j\stackrel{\alpha}\lra i\stackrel{\beta}\lra k$$
in $I$,
\begin{itemize}
\item
The zero relations:{}
If $i$ is an ordinary vertex, then the paths ${}^{\pm}\alpha\beta^{\pm}, {}^\pm \alpha\beta$, $\alpha\beta^{\pm}$ and $\alpha\beta$ lie in $I^A$. The existence of ${}^{\pm}\alpha$ and $\beta^{\pm}$ depends on if or not $j$ and $k$ are special vertices.
\item
The commutative relations:{}
If $i$ is a special vertex, then there are three potential cases where all squares are anti-commutative.
$$\xymatrix@=5mm{
j^+\ar[r]^{{}^+\alpha^+}\ar[rdd]_(.3){{}^+\alpha^-}&i^+\ar[rd]^{{}^+\beta}&\\&&k\\
j^-\ar[r]^{{}^-\alpha^-}\ar[uur]_(.7){{}^-\alpha^+}&i^-\ar[ur]_{{}^-\beta}&
}\quad\quad\quad\xymatrix@=5mm{
&i^+\ar[r]^{{}^+\beta^+}\ar[rdd]^{{}^+\beta^-}&k^+\\
j\ar[ur]^{\alpha^+}\ar[rd]^{\alpha^-}&&\\
&i^-\ar[r]^{{}^-\beta^-}\ar[uur]^{{}^-\beta^+}&k^-
}\quad\quad\quad \xymatrix{
j^+\ar[r]^{{}^+\alpha^+}\ar[rd]^{{}^+\alpha^-}&i^+\ar[r]^{{}^+\beta^+}\ar[rd]^{{}^+\beta^-}&k^+\\
j^-\ar[r]^{{}^-\alpha^-}\ar[ur]^{{}^-\alpha^+}&i^-\ar[r]^{{}^-\beta^-}\ar[ur]^{{}^-\beta^+}&k^-
}$$
\end{itemize}
Let $G=\langle e, g\mid g^2=e\rangle$ be a finite group. In \cite[Proposition 4.6]{GP1999}, Geiss and de la Pe\~{n}a prove that a skew-gentle algebra $A(Q, I, Sp)$ is Morita equivalent to a skew-group algebra $\Gamma G$ in the case of $char(k)\neq 2$. Here, the algebra $\Gamma$ is a gentle algebra, and is defined by the quiver with relations $(Q^{\Gamma}, I^{\Gamma})$. The quiver $Q^{\Gamma}$ has two vertices for every ordinary vertex, and the set of arrows are split into two parts. More precisely, the quiver with relations $(Q^{\Gamma}, I^{\Gamma})$ is given by
\begin{itemize}
\item
The set of vertices:{} $Q^{\Gamma}_0=Q^{or}_0\times\{-1, 1\}\cup \{i\mid i\in Q^{Sp}_0\};$
\item
The set of arrows:{} $Q^{\Gamma}_1=Q_1\times\{-1, 1\}=\{\alpha^+, \alpha^-\mid \forall \alpha\in Q_1\}.$
$$s(\alpha^\pm):=\left\{
\begin{array}{ll}
s(\alpha)^{\pm}, &\mbox{ if } s(\alpha)\in Q^{or}_0;\\
s(\alpha), &\mbox{ if } s(\alpha)\in Q^{Sp}_0,
\end{array}
\right.\quad
t(\alpha^\pm):=\left\{
\begin{array}{ll}
t(\alpha)^{\pm}, &\mbox{ if } t(\alpha)\in Q^{or}_0;\\
t(\alpha), &\mbox{ if } t(\alpha)\in Q^{Sp}_0.
\end{array}\right.
$$
\item
The relations $I^{\Gamma}=\{\alpha^+\beta^+, \alpha^-\beta^-\mid \alpha\beta\in I, t(\alpha)\in Q^{or}_0\}\cup\{\alpha^+\beta^-, \alpha^-\beta^+\mid \alpha\beta\in I, t(\alpha)\in Q^{Sp}_0\}.$
\end{itemize}
The operation of $G$ on $\Gamma$ is a function $G\times\Gamma\lra \Gamma$ satisfying that
$$\left\{
\begin{array}{ll}
g(i)=i, &\mbox{ if } i\in Q^{Sp}_0;\\
g(i^+)=i^-, &\mbox{ if } i\in Q^{or}_0;\\
g(\alpha^+)=\alpha^-& \mbox{ if } \alpha\in Q_1.
\end{array}
\right.$$
For more details for skew-gentle algebras, we refer the readers to \cite{GP1999}, \cite[Section 4]{BH2007} and \cite[Section 2.4]{BMM2003}.
\begin{Example} Let $A(Q, I)$ be a gentle algebra which is given by the quiver
$$1\stackrel{\alpha}\lra 2\stackrel{\beta}\lra 3\stackrel{\gamma}\lra 4$$
with relation $\{\alpha\beta\}$. The potential special loops are $\delta_i$ which are attached with the vertices $i$ for $i=1, 2, 4$. Note that $\delta_3$ is not a special loop. Because $(Q',I\cup\{\delta^2_3\}\rangle$ is not a gentle pair.
Assume that the set of the special vertices is $\{1, 2\}$. More precisely, the skew-gentle triple is
\[
\begin{tikzcd}
1 \arrow[r,left,"\alpha"] \arrow[out=180,in=90,loop, "\delta_1"] & 2\arrow[in=140,out=70,loop, "\delta_2"]\arrow[r,"\beta"]&3\arrow[r, "\gamma"]&4
\end{tikzcd}
\]
with $I=\{\alpha\beta\}$ and $Sp=\{\delta_1, \delta_2\}$. Then, the quiver of $A(Q, I, Sp)$ is displayed as
$$\xymatrix@=15mm{
1^+\ar[r]^{{}^+\alpha^+}\ar[rd]_(.3){{}^+\alpha^-}&2^+\ar[r]^{{}^+\beta}&3\ar[r]^{\gamma}&4\\
1^-\ar[r]^{{}^-\alpha^-}\ar[ur]_(.7){{}^-\alpha^+}&2^-\ar[ur]^{{}^-\beta}
}
$$
with relations $\{{}^+\alpha^+{}^+\beta+{}^+\alpha^-{}^-\beta, {}^-\alpha^+{}^+\beta+{}^-\alpha^-{}^-\beta\}$. The skew-gentle algebra is Morita equivalent to the skew-group algebra $\Gamma G$, where $\Gamma$ is given by the quiver
$$\xymatrix@=3mm{
&&&&3^-\ar[rr]^{\gamma^-}&&4^-\\
1\ar@/_1pc/[rr]_{\alpha^+}\ar@/^1pc/[rr]^{\alpha^-}&&2\ar[urr]^{\beta^-}\ar[rrd]_{\beta^+}&&&\\
&&&&3^+\ar[rr]^{\gamma^+}&&4^+}$$
with relations $\{\alpha^+\beta^-, \alpha^-\beta^+\}$.
\end{Example}
\medskip
\section{Main results}
In this section, we will turn to the proof of the main results.
Let $B$ and $C$ be rings with identity, and ${}_BM_C, {}_CN_B$ be bimodules. Recall that $(B, C, {}_BM_C, {}_CN_B, f, g)$ is a Morita context if $f:{}M\otimes_CN\lra B$ and $g:{}N\otimes_BM\lra C$ are morphisms of bimodules satisfying that
$$f(m\otimes n)m'=mg(n\otimes m')\mbox{ and }nf(m\otimes n')=g(n\otimes m)n'$$
for all $m, m'\in M$ and $n, n'\in N$. With such Morita context, we associate the ring
$$A=\small{\begin{pmatrix}B&M\\N&C
\end{pmatrix}}
$$
with the operations the formal operations of $2\times 2$-matrices. More precisely,
$$\small{\begin{pmatrix}b&m\\n&c
\end{pmatrix}\begin{pmatrix}b'&m'\\n'&c'
\end{pmatrix}:=\begin{pmatrix}bb'+f(m\otimes n')&bm'+mc'\\nc'+cn'&g(n\otimes m')+cc'
\end{pmatrix}}
$$
with $b, b'\in B, c, c'\in C, m, m'\in M, n, n'\in N$. The ring $A$ is called the {\em Morita context ring} associated with the Morita context $(B, C, {}_BM_C, {}_CN_B, f, g)$.
For a Morita context ring $A$, we suppose $g:{}N\otimes_AM\lra \rad(C)$. Then $f(M\otimes_C N)$ is a nilpotent ideal of $B$. Because
$$f(m\otimes n)f(m'\otimes n')=f(m\otimes nf(m'\otimes n'))=f(m\otimes g(n\otimes m')\otimes n')$$
for all $m, m'\in M$ and $n, n'\in N$. Let$$\small{E_{22}=\begin{pmatrix}0&0\\0&e
\end{pmatrix}.}
$$ be an idempotent of $A$. The following Lemma has been proved in \cite{Koenig2009}.
\begin{Lem}\cite[Lemma 4.1]{Koenig2009}\label{KN09} If $\Tor^C_n(M, N)=0$ for all $n\geqslant 1$, then $A E_{22}A$ is a stratifying ideal.
\end{Lem}
In the following, we will show that every skew-gentle algebra with ${Sp}\neq\emptyset$ is isomorphic to a Morita context ring with respect to the Morita context $(B, C, M, N, f, g)$ such that $g:{}N\otimes_CM\lra \rad(B)$, and $M, N$ are projective right $C$-module and projective left $C$-module, respectively.
We write $e$ for the sum of idempotents (or trivial paths) corresponding to the vertices in $Q_0^{Sp-}$. Then, $1=e+(1-e)$ is a decomposition of the identity of $A(Q, I, Sp)$. The Peirce decomposition of $A(Q, I, Sp)$ will be written in the form
$$A(Q, I, Sp)=\begin{pmatrix}(1-e)A(Q, I, Sp)(1-e)&(1-e)A(Q, I, Sp)e\\eA(Q, I, Sp)(1-e)&eA(Q, I, Sp)e
\end{pmatrix}.
$$
In the following, we write $A$ for the skew-gentle algebra $A(Q, I, Sp)$, and write
$$A=\begin{pmatrix}B&M\\N&C
\end{pmatrix}
$$
for the Peirce decomposition of $A$ for simplicity.
Here, the sets $M$ and $N$ are left $B$-right $C$-bimodule and left $C$-right $B$-bimodule, respectively. As a $k$-vector space, the basis of $M$ (resp. $N$) consists of
$$\{p+\langle I^A\rangle\mid p \mbox{ is a path in } Q^A \mbox{ with } s(p)\in Q^{Sp+}_0\cup Q^{or}_0, e(p)\in Q^{Sp-}_0\}$$$$ (resp. \{p+\langle I^A\rangle\mid p \mbox{ is a path in } Q^A \mbox{ with } s(p)\in Q^{Sp-}, e(p)\in Q^{Sp+}_0\cup Q_0^{or}\}).$$
The left $B$-right $C$-bimodule homomorphism
$f:{}M\otimes_{C}N\lra B$
and left $C$-right $B$-bimodule homomorphism
$g:{}N\otimes_{B}M\lra C$ are given by the multiplications in $A$. Due to the associativity of the multiplication in $A$, the bimodule homomorphisms are compatible with each other. In this way, $A$ is a Morita context ring associated with the Morita context $(B, C, M, N, f, g)$.
Now, we want to construct a stratifying ideal of $A$.
First, we will explore some properties of skew-gentle triples.
\begin{Lem} \label{algebra C}Let $(Q, I, Sp)$ be a skew-gentle triple with $Sp\neq\emptyset$. Then, for any positive integer $l$ greater than $1$, given $i^-, j^-, k^-\in Q_0^{Sp-}$,
\begin{enumerate}[label=(\alph*)]
\item
There is at most one vertex $j^-\in Q_0^{Sp-}$ such that $i^-\lra j^-$ is an arrow in $Q^A$. Dually, there is at most one vertex $k^-\in Q_0^{Sp-}$ such that $k^-\lra i^-$ is an arrow in $Q^A$.
\item
There is at most one vertex $k\in Q^{or}_0\cup Q^{Sp+}_0$ such that $k\lra j^-$ is an arrow in $Q^A$. Dually, there is at most one vertex $i\in Q^{or}_0\cup Q^{Sp+}_0$ such that $j^-\lra i$ is an arrow in $Q^A$.
\item
There is not an oriented cycle $p:{}i^-\lra \cdots\lra i^-$ in $Q^A$ such that $p\notin \langle I^A\rangle$.
\item
Let $p:{}i^-\stackrel{\alpha_1}\lra \cdots \stackrel{\alpha_l}\lra j^-$ be a path in $Q^A$ satisfying $p\notin \langle I^A\rangle$ and $t(\alpha_i)\in Q^{or}_0\cup Q^{Sp+}_0$ for all $i=1, \cdots, l-1$.
\begin{enumerate}[label=(d\arabic*)]
\item
If there exists some $t(\alpha_{i_0})\in Q^{Sp+}_0$ with $1\leqslant t_0\leqslant l-1$, then there is a path $p'\in Q^A$ through some vertex in $Q^{Sp-}_0$ such that $p+p'\in \langle I^A\rangle$.
\item
Assume that $t(\alpha_i)\in Q^{Sp+}_0$ for all $i=1, \cdots, l-1$. Then, there is a path $p':{}i^-\stackrel{\beta_1}\lra \cdots\stackrel{\beta_l}\lra j^-$ in $Q^A$ such that $p-(-1)^lp'\in\langle I^A\rangle$ and $t(\beta_i)\in Q^{Sp-}_0$ for all $i=1, \cdots, l-1$.
\item
Suppose that all of $\alpha_{i}$ satisfies that $t(\alpha_{i})\in Q^{or}_0$ for all $i=1, \cdots, l-1$. Then, the path $p$ satisfying $p\notin\langle I^A\rangle$ is unique in $Q^A$. Namely, if $p':{}i^-\stackrel{\beta_1}\lra \cdots\stackrel{\beta_l}\lra j^-$ is another path with $t(\beta_{j})\in Q^{or}_0$ for all $j=1, \cdots, l-1$ in $Q^A$, then $p'\in \langle I^A\rangle$.
\end{enumerate}
\item
Dually, let $p:{}i\stackrel{\alpha_1}\lra \cdots\stackrel{\alpha_l}\lra j$ be a path in $Q^A$ with $i, j\in Q^{or}_0\cup Q^{Sp+}_0$ and $t(\alpha_i)\in Q_0^{Sp-}$ for all $l=1, \cdots, l-1$. Then, there exists a path $p':{}i\stackrel{\beta_1}\lra \cdots\stackrel{\beta_l}\lra j$ in $Q^A$ with $t(\beta_i)\in Q^{Sp+}_0$ satisfying $p-(-1)^lp'\in\langle I^A\rangle$ for all $i=1, \cdots, l-1$.
\item
Let $p_1:{}i^-\stackrel{\alpha_1}\lra \cdots\stackrel{\alpha_l}\lra j^-$ and $p_2:{}j^-\stackrel{\beta_1}\lra \cdots\stackrel{\beta_q}\lra k^-$ be paths in $Q^A$ such that $p_1, p_2\notin \langle I^A\rangle$
with $t(\alpha_i), t(\beta_j)\in Q^{or}_0\cup Q^{Sp+}_0\cup \emptyset$ for all $i=1, \cdots, l-1, j=1, \cdots, q-1$. Then, we have $p_1p_2\notin \langle I^A\rangle$. Here, the notion $t(\alpha_{i_0})=\emptyset$ for some $i_0$ implies that the path $p_1$ is an arrow $i^-\lra j^-$.
\item
Let $\alpha_1:{}j\lra i^-, \alpha_2:{}j\lra k^-, \beta_1:{}j^-\lra i$ and $\beta_2:{}k^-\lra i$ be arrows in $Q^A$ with $i, j\in Q^{or}_0\cup Q^{Sp+}_0$.
\begin{enumerate}[label=(g\arabic*)]
\item
For all paths $p_1\alpha_1p_2$ and $p_3\alpha_2p_4$ in $Q^A$ satisfying $p_1\alpha_1 p_2, p_3\alpha_2p_4\notin I^A$, we have
$(p_1\alpha_1p_2+\langle I^A\rangle)+(p_3\alpha_2p_4+\langle I^A\rangle)=$
$$\begin{cases}(p_1+p_3)(\alpha_1+\alpha_2)(p_2+p_4)+\langle I^A\rangle&l(p_1)=l(p_3)=0 \mbox{ or }l(p_1)>0, l(p_3)>0;\\
(p_1+p_3)(\alpha_1+\alpha_2)(p_2+p_4)-\alpha_1p_2+\langle I^A\rangle& l(p_1)>0 \mbox{ and }l(p_3)=0;\\
(p_1+p_3)(\alpha_1+\alpha_2)(p_2+p_4)-\alpha_2p_4+\langle I^A\rangle& l(p_3)>0 \mbox{ and }l(p_1)=0.\end{cases}$$
\item
Dually, for all paths $p_1'\beta_1p_2'$ and $p_3'\beta_2p_4'$ in $Q^A$ satisfying $p_1'\beta_1p_2', p_3'\beta_2p_4'\notin I^A$, we have
$(p_1'\beta_1p_2'+\langle I^A\rangle)+(p_3'\beta_2p_4'+\langle I^A\rangle)=$
$$\begin{cases}(p_1'+p_3')(\beta_1+\beta_2')(p_2'+p_4')+\langle I^A\rangle&l(p_2')=l(p_4')=0 \mbox{ or } l(p_2')>0, l(p_4')>0;\\
(p_1'+p_3')(\beta_1+\beta_2')(p_2'+p_4')-p_1'\beta_1+\langle I^A\rangle& l(p_2')>0 \mbox{ and }l(p_4')=0;\\
(p_1'+p_3')(\beta_1+\beta_2')(p_2'+p_4')-p_3'\beta_2+\langle I^A\rangle& l(p_4')>0 \mbox{ and }l(p_2')=0.
\end{cases}$$
\end{enumerate}
\end{enumerate}
\end{Lem}
\begin{proof} $(a)$ Suppose that $i^-\lra s^-$ is another arrow in $Q^A$ with $s^-\in Q^{Sp-}_0$. Then,
$$\xymatrix@=3mm{
&j\\
i\ar[ur]^{\alpha}\ar[dr]^{\gamma}&\\
&s
}$$
is a sub-quiver of $Q$ with $i$ a special vertex. Since $(Q, I, Sp)$ is a skew-gentle triple, the vertex $i$ can not be a special vertex. This is a contradiction. Dually, there is at most one arrow $k^-\lra i^-$ in $Q^A$.
$(b)$ The proof of $(b)$ is similar with $(a)$.
$(c)$ Suppose that $p:{}i^-\stackrel{\alpha_1}\lra \cdots \stackrel{\alpha_l}\lra i^-$ is an oriented cycle in $Q^A$ such that $p\notin \langle I^A\rangle$. Note that the zero relations in $I^A$ are paths of length of $2$. Since $i$ is a special vertex, we have $\alpha_l\alpha_1$ is not a zero relation.
So, we have $p^n\notin \langle I^A\rangle$ for all positive integer $n$. Hence $A$ is infinite dimensional. This is a contradiction.
$(d1), (d2)$ and $(e)$ follow from the commutative relations in $I^A$. We assume that $p$ is given by
$$i^-\lra k_1^+\lra k_2^+\lra \cdots\lra k_s^+\lra j^-$$
with all $k_i^{+}\in Q^{Sp+}_0$ for all $i=1, \cdots, s$. Thus, the quiver $Q^A$ contains the following sub-quiver with all squares anti-commutative
$$\xymatrix{
&k_1^+\ar[r]\ar[rd]&k_2^+\ar[r]\ar[rd]&k_3^+\ar[r]\ar[rd]&\cdots\ar[r]\ar[rd]&k_s^+\ar[rd]&\\
i^-\ar[r]\ar[ur]&k_1^-\ar[r]\ar[ur]&k_2^-\ar[r]\ar[ur]&k_3^-\ar[r]\ar[ur]&\cdots\ar[r]\ar[ur]&k_s^-\ar[r]&j^-
}$$
Thus, we have
$(-1)^sp=i^-\lra k^-_1\lra \cdots \lra k^-_s\lra j^-\mbox{ modulo }\langle I^A\rangle.$
$(d3)$ For the vertex $i^-$, there is at most one arrow in $Q^A$ with the terminal in $Q^{or}_0\cup Q^{Sp+}_0$. Otherwise, there are two arrows out from the special vertex $i$. It follows that $(Q, I, Sp)$ is not a skew-gentle triple.
We assume that $t(\alpha_1)$ is an ordinary vertex. Since $(Q, I)$ is a gentle pair, there are at most two arrows out from the vertex $t(\alpha_1)$. Let $\alpha'_2$ and $\alpha''_2$ be arrows starting in $t(\alpha_1)$. It also follows from $(Q, I)$ be a gentle pair that $\alpha_1\alpha_2'\in I^A$ or $\alpha_1\alpha''_2\in I^A$. The proof is completed.
$(f)$ The assumption that $p_1p_2$ lies in $\langle I^A\rangle$ is equivalent to saying that $\alpha_l\beta_1\in I^A$ and $j$ is an ordinary vertex. The proof is completed.
$(g)$ We just prove $(g1)$, and $(g2)$ is dual.
Since $i$ and $k$ are special vertices, we get $i^-\neq k^-$. Otherwise, $i$ can not be a special vertex. For the same reason above, the vertex $j$ can not lie in $Q^{Sp+}_0$. Now, we assume that $j$ is an ordinary vertex. Then, we get two distinct paths $p_1\alpha_1p_2$ and $p_3\alpha_2p_4$ satisfying $s(\alpha_1)=s(\alpha_2)$ and $t(\alpha_1)\neq t(\alpha_2)$.
If $l(p_1)=0, l(p_3)=0$, then we have $\alpha_1p_2+\alpha_2p_4=(\alpha_1+\alpha_2)(p_3+p_4)$. The conclusion is true. Assume $l(p_1)>0$ and $l(p_3)>0$. Note that the arrows $\alpha_1$ and $\alpha_2$ correspond the arrows in the gentle pair $(Q, I)$.
We have $p_1\alpha_1\in \langle I^A\rangle, p_3\alpha_2\in \langle I^A\rangle$ or $p_1\alpha_2\in\langle I^A\rangle, p_3\alpha_1\in\langle I^A\rangle$. Since $p_1\alpha_1p_2, p_3\alpha_2p_4\notin I^A$, we get $p_1\alpha_2, p_3\alpha_1\in \langle I^A\rangle$.
Thus, we have
$$(p_1\alpha_1p_2+\langle I^A\rangle)+(p_3\alpha_2p_4+\langle I^A\rangle)=(p_1+p_2)(\alpha_1+\alpha_2)(p_3+p_4)+\langle I^A\rangle.$$
The other cases are similar and straightforward.\end{proof}
For a skew-gentle triple $(Q, I, Sp)$, we divide the set of relations $I$ into two parts. Namely, $I$ is the disjoint union of
$$I^{or}:{}=\{\cdot\stackrel{\alpha}\lra i\stackrel{\beta}\lra \cdot\in I\mid i\in Q^{or}_0 \} \mbox{ and }I^{Sp}:{}=\{\cdot\stackrel{\alpha}\lra i\stackrel{\beta}\lra \cdot\in I\mid i\in Q^{Sp}_0\}.$$
The next lemma is about the relation between $A(Q, I)$ and $A(Q, I^{or})$.
\begin{Lem}Let $(Q, I, Sp)$ be a skew-gentle triple. Then, $A(Q, I^{or})$ is a gentle algebra, and $A(Q, I)$ is isomorphic to the quotient of $A(Q, I^{or})$ via $\langle I\rangle/\langle I^{or}\rangle$.
\end{Lem}
\begin{proof} In fact, the potential special vertices are
$$\xymatrix{
\cdots\ar[r]&i
},
\quad\quad\quad \xymatrix{i&\cdots\ar[l]},\quad\quad\quad
\xymatrix{\cdots\ar[r]^{\alpha}&i\ar[r]^{\beta}&\cdots
}$$
with $\alpha\beta\in I$. Obviously, $(Q, I^{or})$ is still a gentle pair if we can prove that $A(Q, I^{or})$ is finite dimensional. For convenience, we suppose $Q^{Sp}_0=\{i\}$.
Assume that $A(Q, I^{or})$ is infinite dimensional, and $p:{}\cdots\stackrel{\alpha}\lra i\stackrel{\beta}\lra\cdots$ is a path with infinite length satisfying $p\notin\langle I^{or}\rangle$. Note that $A(Q, I)$ is finite dimensional. Then, we have $p\in \langle I\rangle$. It follows $p\in\langle I^{Sp}\rangle$. So, there is some arrow $\alpha$ in $p$ such that $t(\alpha)=i$ is a special vertex.
Now, we add a special loop at the vertex $i$, and consider the path $p':{}\cdots\stackrel{\alpha}\lra i\stackrel{\delta
_i}\lra i\stackrel{\beta}\lra \cdots$. Obviously, $p'$ is a path with infinite length, and $p'\notin \langle I\cup \{\delta_i^2\}\rangle$. This is a contraction. Because $A(Q, I\cup\{\delta_i^2\})$ is a finite dimensional algebra. It is immediate to check the isomorphism.
\end{proof}
\medskip
Now, we want to describe the quivers of $B$ and $C$.
Recall that, for an idempotent $f$ of $A$, the radical of $fAf$ is $f\rad(A) f$ (see \cite[Proposition 3.2.4]{Drozd1994}). And there is a $k$-linear isomorphism $fAf/f\rad(A)f\lra f(A/\rad(A))f$ which is given by $fxf+f(\rad(A))f\mapsto f(x+\rad(A))f$ with $x\in A$. Thus, the sets of vertices in $Q^B$ (resp. $Q^C$) consists of the ones in $Q^{or}_0\cup Q^{Sp+}_0$ (resp. $Q^{Sp-}_0$).
Note that the arrows of $fAf$ are given by $\rad(fAf)/\rad^2(fAf) (=f\rad(A)f/f\rad(A)f\rad(A)f)$, and there exists a $k$-linear isomorphism
$$f\rad(A)f/f\rad(A)f\rad(A)f\lra f(\rad(A)/\rad(A)f\rad(A))f,$$
which sends $fxf+f\rad(A)f\rad(A)f$ to $f(x+\rad(A)f\rad(A))f$ for all $x\in \rad(A)$. Then, the arrows of $Q^{fAf}$ are determined by the paths in $f\rad(A)f$ satisfying that they do not pass through the vertices corresponding the idempotent $f$.
Consider the path $p:{}i^-\stackrel{\alpha_1}\lra \cdots\stackrel{\alpha_l}\lra j^-\notin \langle I^A\rangle$ in $Q^A$ with $i^-, j^-\in Q^{Sp-}_0$. If there is some $t(\alpha_{i_0})\in Q^{Sp+}_0\cup Q^{Sp-}_0$, then we have $p+\langle I^A\rangle\in \rad(A)e\rad(A)$ by Lemma \ref{algebra C} $(d1)$. By Lemma \ref{algebra C} $(d3)$, there is a unique $p\notin \langle I^A\rangle$ satisfying $t(\alpha_i)\in Q^{or}_0$ for all $i=1, \cdots, l-1$.
To sum up, the set of arrows $Q^C_1$ consists of two parts:
\begin{itemize}
\item
If ${}^-\alpha^-:{}i^-\lra j^-$ with $i^-, j^-\in Q_0^{Sp-}$ is an arrow in $Q^A_1$, then ${}^-\alpha^-$ lies in $Q^C_1$.
\item
Let $l$ be a positive integer greater than $1$, and $i^-\stackrel{\alpha_1}\lra \cdots\stackrel{\alpha_l}\lra j^-\notin \langle I^A\rangle$ be a path in $Q^A$ with $i^-, j^-\in Q^{Sp-}_0, t(\alpha_i)\in Q^{or}_0$ for all $i=1,\cdots, l-1$. Then there is an arrow $i^-\lra j^-$ in $Q^C_1$.
\end{itemize}
\begin{Koro}\label{hereditary} Let $(Q, I, Sp)$ with $Sp\neq \emptyset$ be a skew-gentle triple. Then, there is an algebra isomorphism $C\lra kQ^C$.
Any indecomposable direct factor of $C$ is $k$ or a hereditary algebra of the linearly oriented quiver of type $\mathbb{A}_n$ with $n\geqslant 2$.
\end{Koro}
\begin{proof} Let $f:{} kQ^C\lra C$ be a $k$-linear map which sends $i^-\in Q^C_0$ to $i^-+\langle I^A\rangle$, sends an arrow ${}^-\alpha^-:{}i^-\lra j^-\in Q^C_1$ to ${}^-\alpha^-+\langle I^A\rangle$ or $p+\langle I^A\rangle$ where $p:{}i^-\stackrel{\alpha_1}\lra \cdots\stackrel{\alpha_l}\lra j^-\notin \langle I^A\rangle$ is a path in $Q^A$ with $t(\alpha_i)\in Q^{or}_0$ for all $i=1, \cdots, l-1$. It is easy to check that $f$ is surjective and an algebra homomorphism.
Note that the paths of the form
$$i^-\stackrel{{}^-\alpha^-}\lra j^-\stackrel{{}^-\beta^-}\lra k^-, \;\; i^-\stackrel{{}^-\alpha^-}\lra j^-\stackrel{{}^-\beta}\lra k, \;\;i\stackrel{\alpha^-}\lra j^-\stackrel{{}^-\beta^-}\lra k^- $$
are not zero relations in $I^A$. And it follows from Lemma \ref{algebra C} $(a)$ and $(d3)$ that the image of any combination of paths in $Q^C$ under $f$ does not lie in $I^A$. Then, the map $f$ is an algebra isomorphism.
Thanks to Lemma \ref{algebra C} $(a), (d3)$, we have that any indecomposable direct factor of $C$ is a Nakayama algebra. And under the help of Lemma \ref{algebra C} $(c)$ that it is $k$ or a Nakayama algebra of linearly ordered quiver of type $\mathbb{A}_n$ with $n\geqslant 2$.\end{proof}
Next, we turn to the quiver with relations of $B$. Thanks to Lemma \ref{algebra C} $(e)$, the set of arrows $Q^B_1$ is a full subset of $Q^A_1$. Namely, an arrow $\alpha:{}i\lra j$ with $i, j\in Q^B_0$ is in $Q^B_1$ if and only if $\alpha$ lies in $Q^A_1$. For a relation $j\stackrel{\alpha}\lra i\stackrel{\beta}\lra k\in I^{or}$,
\begin{itemize}
\item
If $j\in Q^{or}_0$ and $k\in Q^{Sp}_0$, then $\alpha\beta^+\in I^B$.
\item
If $j\in Q^{Sp}_0$ and $k\in Q^{or}_0$, then ${}^+\alpha\beta\in I^B$.
\item
If $j, k\in Q^{or}_0$, then $\alpha\beta\in I^B$.
\end{itemize}
Clearly, the algebra $A(Q^B, I^B)$ is isomorphic to the gentle algebra $A(Q, I^{or})$. The following lemma tells us that $(Q^B, I^B)$ is actually the quiver with relations of $B$.
\begin{Lem} \label{algebra iso}Notations as above. There exists an algebra isomorphism $A(Q^B, I^B)\lra B $, which is given by $p+\langle I^B\rangle\mapsto p+\langle I^A\rangle$ with a path $p\in kQ^B$.
\end{Lem}
\begin{proof} Consider the canonical algebra homomorphism $\pi:{}kQ^B\lra B$ which is given by $p\mapsto p+\langle I^A\rangle$ for all $p\in kQ^B$. The kernel of $\pi$ is indeed the ideal $\langle I^B\rangle$. \end{proof}
At last, we want to describe the left $B$-right $C$-bimodule $M$ and the left $C$-right $B$-bimodule $N$. Set
$$Q^M_1:=\{i\stackrel{\alpha}\lra j\in Q^A_1\mid i\in Q^{or}_0\cup Q^{Sp
+}_0, j\in Q^{Sp-}_{0}\},\; Q^N_1:=\{i\stackrel{\alpha}\lra j\in Q^A_1\mid i\in Q^{Sp-}_0, j\in Q^{or}_0\cup Q^{Sp+}_0\}.$$
Obviously, the set $\{\alpha+\langle I^A\rangle\mid \alpha\in Q^M_1\}$ (resp. $\{\alpha+\langle I^A\rangle\mid \alpha\in Q^N_1\}$) is a generating set of $M$ (resp. $N$) as a left $B$-right $C$-bimodule (resp. a left $C$-right $B$-bimodule).
Now,
we suppose that $C=C_1\times C_2\times\cdots \times C_l$ is a decomposition of the algebra $C$ into a direct product of indecomposable algebras $C_1, \cdots, C_l$.
By Corollary \ref{hereditary}, the algebra $C_p$ is $k$ or a hereditary algebra of the linearly oriented quiver for $p=1, \cdots, l$. In this way, the set $Q^{Sp-}_0$ is a disjoint union of $Q^{Sp-}_{01}, \cdots, Q^{Sp-}_{0l}$ where the simple modules of $C_p$ correspond to the vertices in $Q^{Sp-}_{0p}$ for all $p=1, \cdots, l$. Hence, we have $e=e_1+\cdots+e_l$ with $e_p=\sum_{q\in Q^{Sp-}_{0p}}e_q$ for all $p=1, \cdots, l$. Meanwhile, we
set
$$Q^{M}_{1p}:=\{i\stackrel{\alpha}\lra j\in Q^A_1\mid i\in Q^{or}_0\cup Q^{Sp
+}_0, j\in Q^{Sp-}_{0p}\},\;Q^N_{1q}:{}=\{i\stackrel{\alpha}\lra j\in Q^A_1\mid i\in Q^{Sp-}_{0q} , j\in Q^{or}_0\cup Q^{Sp
+}_0\}$$
for all $p, q=1, \cdots, l$. It follows that $Q^M_1$ (resp. $Q^N_1$) is a disjoint union of $Q^M_{11}, \cdots, Q^M_{1l}$ (resp. $Q^N_{11}, \cdots, Q^N_{1l}$).
Let $p_1\alpha_1p_2$ and $p_3\alpha_2 p_4$ be two paths in $Q^A$ with $\alpha_1, \alpha_2\in Q^M_1$. If $s(\alpha_1)\neq s(\alpha_2)$ and $t(\alpha_1)\neq t(\alpha_2)$, then we have $p_1\alpha_1p_2+p_3\alpha_2p_4=(p_1+p_3)(\alpha_1+\alpha_2)(p_2+p_4)$. And it is impossible that $s(\alpha_1)\neq s(\alpha_2)$ and $t(\alpha_1)=t(\alpha_2)$ for $t(\alpha_1)$ being a special vertex. Then, by Lemma \ref{algebra C} $(g)$, we have $$\sum_{\alpha\in Q^M_{1p}}(1-e)Ae_{s(\alpha)}\alpha e_{t(\alpha)}Ae=B(\sum_{\alpha\in Q^M_{1p}}\alpha) C.$$
Dually, we get
$$\sum_{\alpha\in Q^N_{1q}}eAe_{s(\alpha)}\alpha e_{t(\alpha)}A(1-e)=C(\sum_{\alpha\in Q^N_{1q}}\alpha)B.$$
Consequently, we have
$$M=(1-e)Ae=\sum_{\alpha\in Q^M_1}(1-e)Ae_{s(\alpha)}\alpha e_{t(\alpha)}Ae=\sum_{p=1, \cdots, l}\sum_{\alpha\in Q^M_{1p}}(1-e)Ae_{s(\alpha)}\alpha e_{t(\alpha)}Ae=\sum_{p=1, \cdots, l}B(\sum_{\alpha\in Q^M_{1p}}\alpha) C,$$
$$N=eA(1-e)=\sum_{\alpha\in Q^N_1}eAe_{s(\alpha)}\alpha e_{t(\alpha)}A(1-e)=\sum_{q=1, \cdots, l}\sum_{\alpha\in Q^N_{1q}}eAe_{s(\alpha)}\alpha e_{t(\alpha)}A(1-e)=\sum_{q=1, \cdots, l}C(\sum_{\alpha\in Q^N_{1q}}\alpha)B.$$
By Corollary \ref{algebra C} and Lemma \ref{algebra iso}, the algebras $B$ and $C$ are isomorphic to the path algebras $A(Q^B, I^B)$ and $A(Q^C, \emptyset)$, respectively. Then, when we mention a path $p_1\alpha p_2$ with $p_1\in Q^B, p_2\in Q^C$ and $\alpha\in Q^M_1$, it is actually a path $p_1'\alpha p_2'$ in $Q^A$ satisfying $s(p_1'), t(p_2')\in Q^{or}_0\cup Q^{Sp+}_0$, where $p_1'$ and $p_2'$ are the images of $p_1$ and $p_2$ under the algebra isomorphisms, respectively.
\begin{Lem} \label{bimodules}For all arrows in $Q^M_{1p}$ and $Q^N_{1q}$ with $p, q=1, \cdots, l$, we have
\begin{enumerate}[label=(\alph*)]
\item
Let $\alpha$ and $\beta$ be arrows in $Q^M_{1p}$ and $Q^N_{1q}$, respectively.
Assume that $p_1\alpha p_2$ is a path with $p_1\in Q^B, p_2\in Q^C$ satisfying $p_1\alpha p_2\notin \langle I^A\rangle$. Then, there exists an arrow $\alpha'\in Q^M_{1p}$ and an element $\widetilde{p}_1\in kQ^B$ such that $p_1\alpha p_2-\widetilde{p}_1\alpha'\in\langle I^A\rangle$.
Dually, suppose that $p_3\beta p_4$ is a path such that $p_3\beta p_4\notin \langle I^A\rangle$ with $p_3\in Q^C, p_4\in Q^B$. Then, there is an arrow $\beta'\in Q^N_{1q}$ and an element $\widetilde{p}_4\in kQ^B$ satisfying $p_3\beta p_4-\beta'\widetilde{p}_4'\in\langle I^A\rangle$.
\item
For all $1\leqslant p\neq q\leqslant l$, we have
$$B(\sum_{\alpha\in Q^M_{1p}}\alpha)C\bigcap B(\sum_{\alpha\in Q^M_{1q}}\alpha)C=\{0\}, \;C(\sum_{\alpha\in Q^N_{1p}}\alpha)B\bigcap C(\sum_{\alpha\in Q^N_{1q}}\alpha)B=\{0\}.$$
\end{enumerate}
\end{Lem}
\begin{proof} $(a)$ We just prove the first statement. The second one is dual. Assume that $\alpha:{}i\lra j^-$ is an arrow with $i\in Q^{or}_0\cup Q^{Sp+}_0$ and $j^-\in Q^{Sp-}_0$. Note that $C_p$ is the indecomposable algebra corresponding to the set $Q^{M}_{1p}$. By Corollary \ref{hereditary}, the algebra $C_p$ is $k$ or a hereditary algebra of linearly oriented quiver of type $\mathbb{A}_n$ for $n\geqslant 2$. If there is not arrow $j^-\lra k^-$ in $eAe$ with $k^-\in Q^{Sp-}_0$, then the path $p_1\alpha p_2$ lies in $\langle I^A\rangle$, where the length of $p_2$ is greater than $0$. Because the bimodule $B\alpha C$ is a simple right $C$-module.
Since $p_1\alpha p_2\notin \langle I^A\rangle$, the path $p_1\alpha p_2$ is actually $p_1\alpha e_{t(\alpha)}$.
Now, we assume that there is an arrow $\alpha_1:{}j^-\lra k^-$ in $eAe$ with $k^-\in Q^{Sp-}_0$, and the length of $p_2$ is greater than $0$. Then, it follows from the commutative relations in $I^A$ that there exist elements $p'_1\in Q^B$ and $p_2'\in Q^C$ such that $p_1\alpha p_2+p_1'\alpha_1 p_2'\in \langle I^A\rangle$. Note that the length of $p_2'$ is indeed $l(p_2)-1$. Then, by induction on $l(p_2)$, we finally get an arrow $\alpha'\in Q^M_{1q}$ and an element $\widetilde{p}_1\in kQ^B$ such that $p_1\alpha p_2-\widetilde{p}_1\alpha'\in\langle I^A\rangle$.
$(b)$ Let $x\in B(\sum_{\alpha\in Q^M_{1p}}\alpha)C\bigcap B(\sum_{\alpha\in Q^M_{1q}}\alpha)C$. Note that the set $B(\sum_{\alpha\in Q^M_{1p}}\alpha )C$ is actually $\sum_{\alpha\in Q^M_{1p}}B\alpha C$ for all $p=1, \cdots, l$. By $(a)$, we set $x=\sum_{\alpha\in Q^{M}_{1p}}k_\alpha p_{\alpha}\alpha+ \langle I^A\rangle$ and $x=\sum_{\beta\in Q^{M}_{1q}}k_\beta p_{\beta}\beta+ \langle I^A\rangle$ with $k_{\alpha}, k_\beta\in k$ and $p_\alpha, p_\beta\in Q^B$ for all $\alpha\in Q^M_{1p}$ and $\beta\in Q^M_{1q}$. Here, we assume that both the sets $\{p_{\alpha}\alpha+\langle I^A\rangle\mid \alpha\in Q^M_{1p}\}$ and $\{p_{\beta}\beta+\langle I^A\rangle\mid \beta\in Q^M_{1q}\}$ are $k$-linearly independent.
Since $p$ and $q$ are different positive integers, the set $\{p_{\alpha}\alpha+\langle I^A\rangle, p_\beta\beta+\langle I^A\rangle\mid \alpha\in Q^M_{1p}, \beta\in Q^M_{1q}\}$ is $k$-linearly independent. Because the paths $p_{\alpha}\alpha$ and $p_{\beta}\beta$ have different terminals.
It follows from $\sum_{\alpha\in Q^{M}_{1p}}k_\alpha p_{\alpha}\alpha-\sum_{\beta\in Q^{M}_{1q}}k_\beta p_{\beta}\beta\in\langle I^A\rangle$ that $k_{\alpha}=0=k_\beta$ for all $\alpha\in Q^M_{1p}$ and $\beta\in Q^M_{1q}$. The second statement is dual. The proof is finished.
\end{proof}
Thanks to Lemma \ref{bimodules} $(b)$, we get $M=\bigoplus^l_{p=1}B(\sum_{\alpha\in Q^M_{1p}}\alpha)C$ and $N=\bigoplus^l_{q=1}C(\sum_{\alpha\in Q^N_{1q}}\alpha)B$ as a left $B$-right $C$-bimodule and a left $C$-right $B$-bimodule, respectively.
In the following, we want to describe the bimodules $M$ and $N$ explicitly.
For all $n\geqslant 1$, let
$$1^-\stackrel{{}^-\alpha_1^-}\lra 2^-\stackrel{{}^-\alpha^-_2}\lra 3^-\stackrel{{}^-\alpha_3^-}\lra\cdots \stackrel{{}^-\alpha_{n-1}^-}\lra n^-$$
be some direct factor $C_{p}$. Note that the arrows in the quiver of $C_p$ correspond to arrows or paths in $Q^A$. Then, we should consider the following two cases.
The first case is that there is no arrows from the ordinary vertex to the special vertices $1^+$ and $1^-$, then the following is a sub-quiver of $Q^A$ with all squares anti-commutative.
\begin{equation}
\xymatrix@=5mm{1^+\ar[r]^{}\ar[rdd]_(.3){}&2^+\ar[r]^{}\ar[rdd]_(.3){}&3^+\ar[r]^{}\ar[rdd]_(.3){}&\cdots\ar[r]^{}\ar[rdd]_(.3){}&s^+\ar[rd]&&&&(n-t)^+\ar[r]\ar[rdd]&\cdots\ar[r]\ar[rdd]&n^+\ar[rd]&\\
&&&&&\cdot\ar[r]&\cdots\ar[r]&\cdot\ar[ur]\ar[rd]&&&&\cdot\\
1^-\ar[r]^{{}^-\alpha_1^-}\ar[uur]^(.3){}&2^-\ar[r]^{{}^-\alpha_2^-}\ar[uur]^(.3){}&3^-\ar[r]^{{}^-\alpha_3^-}\ar[uur]&\cdots\ar[r]^{{}^-\alpha_{s-1}^-}\ar[uur]&s^-\ar[ur]&&&&(n-t)^-\ar[r]\ar[uur]&\cdots\ar[r]\ar[ruu]&n^-\ar[ur]&}\end{equation}
In this situation, the left $B$-right $C$-bimodule $B(\sum_{\alpha\in Q^M_{1p}}\alpha)C$ is given by $(\sum_{j=1}^{n-1}e_{j^+})A(\sum_{\alpha\in Q^M_{1p}}\alpha)Ae$ which is actually $(\sum_{\alpha\in Q^M_{1p}}\alpha)Ae$ for the commutative relations in $I^A$. Thus, the $k$-vector space $(\sum_{\alpha\in Q^M_{1p}}\alpha)Ae$ is a left $B$-right $C$-bimodule.
Note that $C_p$ is a hereditary algebra of the linearly oriented quiver of type $\mathbb{A}_n$, and the set of relations $I^A$ consists of zero relations and commutative relations. Then, the map
$$f_p:{}(\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)})Ae\lra (\sum_{\alpha\in Q^M_{1p}}\alpha)Ae$$ which sends $\sum_{\alpha\in Q^M_{1p}}k_{\alpha}e_{t(\alpha)}p_\alpha+\langle I^A\rangle$ to $\sum_{\alpha\in Q^M_{1p}}k_{\alpha}\alpha p_{\alpha}+\langle I^A\rangle$ is a $k$-linear isomorphism with $k_{\alpha}\in k$ and paths $p_{\alpha}\in Q^C$ for all $\alpha\in Q^M_{1p}$.
In fact, the $k$-vector space $(\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)})Ae$ is a left $B$-right $C$-bimodule. The right $C$-module structure is the multiplication in $A$, and the left $B$-module structure is given by multiplication by $(\sum_{\alpha\in Q^M_{1p}}\alpha)$.
Set $b\in B$ and $x\in (\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)})Ae$.
For the multiplication $b(\sum_{\alpha\in Q^M_{1p}}\alpha)x$, there exists $x'\in kQ^C$ such that $b(\sum_{\alpha\in Q^M_{1p}}\alpha)x-(\sum_{\alpha\in Q^M_{1p}}\alpha)x'x\in \langle I^A\rangle$ by the commutative relations in $I^A$.
Then, the left $B$-module structure $b\cdot x$ is indeed given by $f^{-1}_p((\sum_{\alpha\in Q^M_{1p}}\alpha)x'x)$. And it is straightforward that $f_p$ is a left $B$-right $C$ isomorphism.
The second case is that there are arrows $i\lra 1^{\pm}$ with $i$ an ordinary vertex. The corresponding sub-quiver of $Q^A$ is presented as follows.
\begin{equation}
\xymatrix@=5mm{&1^+\ar[r]^{}\ar[rdd]_(.3){}&2^+\ar[r]^{}\ar[rdd]_(.3){}&3^+\ar[r]^{}\ar[rdd]_(.3){}&\cdots\ar[r]^{}\ar[rdd]_(.3){}&s^+\ar[rd]&&&&(n-t)^+\ar[r]\ar[rdd]&\cdots\ar[r]\ar[rdd]&n^+\ar[rd]&\\
i\ar[ur]\ar[rd]_{\alpha^-}&&&&&&\cdot\ar[r]&\cdots\ar[r]&\cdot\ar[ur]\ar[rd]&&&&\cdot\\
&1^-\ar[r]^{{}^-\alpha^-_1}\ar[uur]^(.3){}&2^-\ar[r]^{{}^-\alpha_2^-}\ar[uur]^(.3){}&3^-\ar[r]^{{}^-\alpha_3^-}\ar[uur]&\cdots\ar[r]^{{}^-\alpha_{s-1}^-}\ar[uur]&s^-\ar[ur]&&&&(n-t)^-\ar[r]\ar[uur]&\cdots\ar[r]\ar[ruu]&n^-\ar[ur]&}\end{equation}
Let $p_1(\sum_{\alpha\in Q^M_{1p}}\alpha)p'_1+\langle I^A\rangle$ be an element in $B(\sum_{\alpha\in Q^M_{1p}}\alpha)C$ with $p_1\in Q^B$ and $p_1'\in Q^C$. If the path $p_1$ passes through the vertex $i$, then there exists a subpath of $p_1$, which is denoted by $p_2$, and a path $p_2'\in Q^C$ such that $p_1(\sum_{\alpha\in Q^M_{1p}}\alpha)p_1'-p_2\alpha^-p_2'\in\langle I^A\rangle$. If the path $p_1$ does not pass through the vertex $i$, then the discussion is similar with the first case. In this way, the $k$-vector space $B(\sum_{\alpha\in Q^M_{1p}}\alpha)C$ can be written as $(1-e)Ae_i\otimes_ke_{1^-}Ae\oplus (\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)})Ae$ which is actually a $\mathbb{N}$-graded left $B$-right $C$-bimodule. And the grading depends on the length of $p_2$. More precisely, the $0$-th degree is the $k$-vector space $\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)}Ae$, and the $m$-th degree is a $k$-vector space generated by $p_2\alpha^-p_2'+\langle I^A\rangle$ with $p_2\alpha^-p_2'\notin \langle I^A\rangle$ and $l(p_2)=m$ for all $p_2\in Q^B, p_2'\in Q^C$ and $m\geqslant 1$.
Dually, the similar discussion still works for $C(\sum_{\alpha\in Q^N_{1q}}\alpha)B$. Thus, we get the following corollary.
\begin{Koro}\label{projective} Let $(Q, I, Sp)$ be a skew-gentle triple with $Sp\neq \emptyset$. Then,
\begin{enumerate}[label=(\alph*)]
\item
The left $B$-right $C$-bimodule $M$ is a projective right $C$-module.
\item
The left $C$-right $B$-bimodule $N$ is a projective left $C$-module.
\item
The $B$-$B$-bimodule homomorphism
$f:{} M\otimes_{C}N\lra B$ is given by the multiplication in $A$.
The image is an ideal of $B$ is
$$\sum_{p=1, \cdots, l}B(\sum_{\alpha\in Q^M_{1p}}\alpha)C(\sum_{\beta\in Q^N_{1p}}\beta)B.$$
which
is an ideal of $B$, and the generating set is given as follows:
$$ \alpha^+{}^+\beta, \alpha^+{}^+\beta^+, {}^+\alpha^+{}^+\beta, {}^+\alpha^+{}^+\beta^+$$
for a relation $j\stackrel{\alpha}\lra i\stackrel{\beta}\lra k\in I$ with $i$ a special vertex. Namely, the quotient algebra $B/f(M\otimes_CN)$ is isomorphic to the path algebra $A(Q, I)$.
\end{enumerate}
\end{Koro}
\begin{proof} $(a)$ By Lemma \ref{bimodules} and the discussion above, the direct summand of $M$ is isomorphic to $(\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)})Ae$ or $(1-e)Ae_i\otimes_ke_{1^-}Ae\oplus (\sum_{\alpha\in Q^M_{1p}}e_{t(\alpha)}Ae)$ as left $B$-right $C$-bimodules. Obviously, they are projective right $C$-modules for $\{1^-, t(\alpha)\mid \alpha\in Q^M_{1p}\}$ being a subset of $Q^{Sp-}_0$. Similarly, we get $(b)$.
$(c)$ Note that, for the quiver of $C$, there is not path between the vertices in $Q^{Sp-}_{1p}$ and $Q^{Sp-}_{1q}$ for all $1\leqslant p\neq q\leqslant l$. Then, we have
$$B(\sum_{\alpha\in Q^M_{1p}}\alpha)C(\sum_{\beta\in Q^N_{1q}}\beta)B=0$$ for all
$1\leqslant p\neq q\leqslant l$.
By Lemma \ref{bimodules} $(a)$, the left $B$-right $C$-bimodule $B(\sum_{\alpha\in Q^M_{1p}}\alpha)C$ can be written as $B(\sum_{\alpha\in Q^M_{1p}}\alpha)$. Dually, the left $C$-right $B$-bimodule $C(\sum_{\beta\in Q^N_{1p}}\beta)B$ is actually $(\sum_{\beta\in Q^N_{1p}}\beta)B$. Then, the generating set of $f(M\otimes_CN)$ as a $B$-$B$-bimodule is
$$\{ \alpha^+{}^+\beta, \alpha^+{}^+\beta^+, {}^+\alpha^+{}^+\beta, {}^+\alpha^+{}^+\beta^+\mid j\stackrel{\alpha}\lra i\stackrel{\beta}\lra k\in I^{Sp}\}.$$\end{proof}
\begin{Prop}\label{skew-gentle sd} Let $(Q, I, Sp)$ be a skew-gentle triple with $Sp\neq\emptyset$. Then
\begin{enumerate}[label=(\alph*)]
\item
The set
$$A E_{22}A=\begin{pmatrix}f(M\otimes_CN)&M\\N&C
\end{pmatrix}
$$ is a stratifying ideal of $A$.
\item
$\pd_{A^e}(A E_{22}A)<\infty$.
\item
The two algebra $A/A E_{22}A$ and $A(Q, I)$ are isomorphic.
\end{enumerate}
\end{Prop}
\begin{proof} $(a)$ It is immediate from Corollary \ref{projective} and Lemma \ref{KN09}.
$(b)$ It is essentially from the proof of \cite[Proposition 3.3]{Koenig2009}. For convenience for the readers, we include the proof. By $(a)$, $A E_{22}A$ is a stratifying ideal of $A$. Thus, $A E_{22}A$ is isomorphic to $A E_{22}\otimes_{C}E_{22}A$ as $A$-$A$-bimodules. Let $X$ be a left $A^e$-module. We have
$$\begin{array}{ll}
\Ext^n_{A^e}(A E_{22}A, X)&\cong \Ext^n_{A^e}(A E_{22}\otimes_{C}E_{22}A, X)\\
&\cong \Ext^n_{A\otimes_kC^{\opp}}(A E_{22}, \Hom_{A^{\opp}}(E_{22}A, X))\\
&\cong \Ext^n_{A\otimes_kC^{\opp}}(A E_{22}\otimes_{C}C, XE_{22})\\
&\cong \Ext^n_{C^{e}}(C, E_{22}XE_{22}).
\end{array}
$$
By \cite[Section 1.5]{Happel1989d}, we have $\gldim (C)=\pd_{C^e}C$. So, it follows from Corollary \ref{hereditary} that the projective dimension of $A E_{22}A$ as a left $A^e$-module is not greater than $1$.
$(c)$ By Corollary \ref{projective} $(c)$.
\end{proof}
{\em Proof of Theorem \ref{main result}:} It is immediate from Proposition \ref{skew-gentle sd} and Corollary \ref{hereditary}. $\square$
{\em Proof of Corollary \ref{coro}} If $Sp=\emptyset$, then the skew-gentle algebra is just a gentle algebra, which is quadratic monomial. Then, we have $A$ is Gorenstein \cite{Geis2005}, and satisfies the finitistic dimension conjecture \cite{Green1991} and the Auslander-Reiten conjecture \cite[Proposition 1.5]{Marczinzik2019}. Now, we assume $Sp\neq\emptyset$.
By Theorem \ref{main result}, we get a recollement (\ref{recollement}) with $\gldim(C)\leqslant 1$ and $A/A E_{22}A\cong A(Q, I)$. It follows from Proposition \ref{1} that (\ref{recollement}) can be restricted to $\Db{\rm mod}$.
$(a)$
By Proposition \ref{skew-gentle sd} $(b)$, we have $i^!(A)\in\Kb{\pmodcat{A/A E_{22}A}}$. By Proposition \ref{1}, the recollement can be restricted to $\Kb{\rm proj}$. Due to Proposition \ref{2}, the skew-gentle algebra $A(Q, I, Sp)$ is Gorenstein.
$(b)$ Recall that the singularity category $\Dsg{A}$ is the Verdier quotient of $\Db{A}$ with respect to $\Kb{\pmodcat{A}}$.
Since (\ref{recollement}) can be restricted to $\Db{\rm mod}$ and $\Kb{\rm proj}$, by \cite[Lemma 1.2]{Orlov2004}, the adjoint pair $(i^*, i_*)$ admits an adjoint pair $(\overline{i^*}, \overline{i_*})$ between the singularity categories $\Dsg{A/A E_{22}A}$ and $\Dsg{A}$. It follows from Theorem \ref{main result} $(c)$ that $(\overline{i^*}, \overline{i_*})$ is a triangle equivalence.
$(c)$ Assume that $A(Q, I, Sp)$ is indecomposable self-injective. Then, by \cite[Theorem 3.6]{CK2017}, the skew-gentle algebra is derived equivalent to $A(Q, I)$ or $C$. This is a contradiction.
Because derived equivalences preserves the number of isomorphism classes of simple modules. So, we get $Sp=\emptyset$, and $A(Q, I)$ is an indecomposable selfinjective gentle algebra.
Assume that $A(Q, I)$ is an indecomposable selfinjective gentle algebra. It follows
$\gldim(A(Q, I))=0$ or $\infty$. If $\gldim(A(Q, I))=0$, then $A(Q, I)$ is just $k$ for $A(Q, I)$ being indecomposable. Now, we suppose that
$\gldim(A(Q, I))=\infty$. Note that the singularity category $\Dsg{A(Q, I)}$ is not trivial, and is given by the Gorenstein projective (not projective) modules. By \cite[Theorem 2.5]{Kalck2015}, $A(Q, I)$ has a full cycle of relations, that is there exists a oriented cycle $\alpha_0\alpha_1\cdots\alpha_n$ such that $s(\alpha_0)=t(\alpha_n)$ and $\alpha_i\alpha_{i+1}, \alpha_n\alpha_0\in I$ for all $i=0, \cdots, n-1$. Since $A(Q, I)$ is selfinjective, any indecomposable projective module over $A(Q, I)$ is injective. So the socle of any indecomposable projective module is simple. Thus, there are not other arrows starting (or ending) at $s(\alpha_i)$ for all $i=0, \cdots, n$. Hence, $A(Q, I)$ is a selfinjective Nakayama algebra with radical square zero.
$(d)$ It follows from \cite[Theorem 3.3]{Happel1993} and \cite{Green1991}.
$(e)$ Thanks to \cite[Proposition 1.5]{Marczinzik2019} and \cite[Theorem 1.2]{Chen2020}.$\square$
\medskip
{\em Remark.} Corollary \ref{coro} $(b)$ has been proved in \cite{ChenLu2015} and \cite[Theorem 7.1]{Fragoso2022}. In \cite[Section 3.1]{Burban2017}, Burban and Drozd have considered recollements of some special skew-gentle algebras with global dimension $2$, and proved the Rouquier dimension of the bounded derived category is at most $1$.
\providecommand{\bysame}{\leavevmode\hbox to3em{\hrulefill}\thinspace}
\providecommand{\noopsort}[1]{}
\providecommand{\mr}[1]{\href{http://www.ams.org/mathscinet-getitem?mr=#1}{MR~#1}}
\providecommand{\zbl}[1]{\href{http://www.zentralblatt-math.org/zmath/en/search/?q=an:#1}{Zbl~#1}}
\providecommand{\jfm}[1]{\href{http://www.emis.de/cgi-bin/JFM-item?#1}{JFM~#1}}
\providecommand{\arxiv}[1]{\href{http://www.arxiv.org/abs/#1}{arXiv~#1}}
\providecommand{\doi}[1]{\url{https://doi.org/#1}}
\providecommand{\MR}{\relax\ifhmode\unskip\space\fi MR }
% \MRhref is called by the amsart/book/proc definition of \MR.
\providecommand{\MRhref}[2]{%
\href{http://www.ams.org/mathscinet-getitem?mr=#1}{#2}
}
\providecommand{\href}[2]{#2}
\begin{thebibliography}{AHKLY17}
\bibitem[AB22]{Amiot2022}
\bgroup\scshape{}C.~Amiot\egroup{} and
\bgroup\scshape{}T.~Br\"{u}stle\egroup{}, Derived equivalences between
skew-gentle algebras using orbifolds, \emph{Doc. Math.} \textbf{27} (2022),
933--982.
\bibitem[APS19]{Amiot2019}
\bgroup\scshape{}C.~Amiot\egroup{}, \bgroup\scshape{}P.-G. Plamondon\egroup{},
and \bgroup\scshape{}S.~Schroll\egroup{}, A complete derived invariant for
gentle algebras via winding numbers and arf invariants, (2019). Available at
\url{https://arxiv.org/pdf/1904.02555.pdf}.
\bibitem[AHKL11]{AKL2011}
\bgroup\scshape{}L.~Angeleri~H{{\"u}}gel\egroup{},
\bgroup\scshape{}S.~Koenig\egroup{}, and \bgroup\scshape{}Q.~Liu\egroup{},
Recollements and tilting objects, \emph{J. Pure Appl. Algebra} \textbf{215}
no.~4 (2011), 420--438.
\bibitem[AHKLY17]{AKLY2017}
\bgroup\scshape{}L.~Angeleri~H{{\"u}}gel\egroup{},
\bgroup\scshape{}S.~Koenig\egroup{}, \bgroup\scshape{}Q.~Liu\egroup{}, and
\bgroup\scshape{}D.~Yang\egroup{}, Ladders and simplicity of derived module
categories, \emph{J. Algebra} \textbf{472} (2017), 15--66.
\bibitem[ABCJP10]{ABC2010}
\bgroup\scshape{}I.~Assem\egroup{}, \bgroup\scshape{}T.~Br\"{u}stle\egroup{},
\bgroup\scshape{}G.~Charbonneau-Jodoin\egroup{}, and \bgroup\scshape{}P.-G.
Plamondon\egroup{}, Gentle algebras arising from surface triangulations,
\emph{Algebra Number Theory} \textbf{4} no.~2 (2010), 201--229.
\bibitem[ASS06]{Assem2006}
\bgroup\scshape{}I.~Assem\egroup{}, \bgroup\scshape{}D.~Simson\egroup{}, and
\bgroup\scshape{}A.~Skowro{\'n}ski\egroup{}, \emph{Elements of the
representation theory of associative algebras. {V}ol. 1}, \emph{London
Mathematical Society Student Texts} \textbf{65}, Cambridge University Press,
Cambridge, 2006.
\bibitem[AS87]{AS1987}
\bgroup\scshape{}I.~Assem\egroup{} and
\bgroup\scshape{}A.~Skowro\'{n}ski\egroup{}, Iterated tilted algebras of type
{$\tilde{\bf A}_n$}, \emph{Math. Z.} \textbf{195} no.~2 (1987), 269--290.
\bibitem[ADS93]{Auslander1993}
\bgroup\scshape{}M.~Auslander\egroup{}, \bgroup\scshape{}S.~Ding\egroup{}, and
\bgroup\scshape{}{\O}.~Solberg\egroup{}, Liftings and weak liftings of
modules, \emph{J. Algebra} \textbf{156} no.~2 (1993), 273--317.
\bibitem[AR75]{Auslander1975c}
\bgroup\scshape{}M.~Auslander\egroup{} and \bgroup\scshape{}I.~Reiten\egroup{},
On a generalized version of the {N}akayama conjecture, \emph{Proc. Amer.
Math. Soc.} \textbf{52} (1975), 69--74.
\bibitem[ARO95]{Auslander1995}
\bgroup\scshape{}M.~Auslander\egroup{}, \bgroup\scshape{}I.~Reiten\egroup{},
and \bgroup\scshape{}S.~O.\egroup{}, \emph{Representation theory of {A}rtin
algebras}, \emph{Cambridge Studies in Advanced Mathematics} \textbf{36},
Cambridge University Press, Cambridge, 1995.
\bibitem[BCSo21]{Baur2021}
\bgroup\scshape{}K.~Baur\egroup{} and \bgroup\scshape{}R.~Coelho
Sim\~{o}es\egroup{}, A geometric model for the module category of a gentle
algebra, \emph{Int. Math. Res. Not. IMRN} no.~15 (2021), 11357--11392.
\bibitem[BBD82]{Beuilinson1982}
\bgroup\scshape{}A.~A. Be{\u \i}linson\egroup{},
\bgroup\scshape{}J.~Bernstein\egroup{}, and
\bgroup\scshape{}P.~Deligne\egroup{}, Faisceaux pervers, in \emph{Analysis
and topology on singular spaces, {I} ({L}uminy, 1981)}, \emph{Ast{\'e}risque}
\textbf{100}, Soc. Math. France, Paris, 1982, pp.~5--171.
\bibitem[BMM03]{BMM2003}
\bgroup\scshape{}V.~Bekkert\egroup{}, \bgroup\scshape{}E.~N. Marcos\egroup{},
and \bgroup\scshape{}H.~Merklen\egroup{}, Indecomposables in derived
categories of skewed-gentle algebras, \emph{Comm. Algebra} \textbf{31} no.~6
(2003), 2615--2654.
\bibitem[BM03]{Bekkert2003}
\bgroup\scshape{}V.~Bekkert\egroup{} and \bgroup\scshape{}H.~A.
Merklen\egroup{}, Indecomposables in derived categories of gentle algebras,
\emph{Algebr. Represent. Theory} \textbf{6} no.~3 (2003), 285--302.
\bibitem[Bel05]{Beligiannis2005}
\bgroup\scshape{}A.~Beligiannis\egroup{}, Cohen-{M}acaulay modules, (co)torsion
pairs and virtually {G}orenstein algebras, \emph{J. Algebra} \textbf{288}
no.~1 (2005), 137--211.
\bibitem[BH07]{BH2007}
\bgroup\scshape{}C.~Bessenrodt\egroup{} and \bgroup\scshape{}T.~Holm\egroup{},
q-cartan matrices and combinatorial invariants of derived categories for
skewed-gentle algebras, \emph{Pacific J. Math.} \textbf{229} no.~1 (2007),
25--47.
\bibitem[BD17]{Burban2017}
\bgroup\scshape{}I.~Burban\egroup{} and \bgroup\scshape{}Y.~Drozd\egroup{}, On
the derived categories of gentle and skew-gentle algebras: homological
algebra and matrix problems, (2017). Available at
\url{https://arxiv.org/pdf/1706.08358v1.pdf}.
\bibitem[CL15]{ChenLu2015}
\bgroup\scshape{}X.~Chen\egroup{} and \bgroup\scshape{}M.~Lu\egroup{},
Singularity categories of skewed-gentle algebras, \emph{Colloq. Math.}
\textbf{141} no.~2 (2015), 183--198.
\bibitem[CHQW20]{Chen2020}
\bgroup\scshape{}Y.~Chen\egroup{}, \bgroup\scshape{}W.~Hu\egroup{},
\bgroup\scshape{}Y.~Qin\egroup{}, and \bgroup\scshape{}R.~Wang\egroup{},
Singular equivalences and {A}uslander-{R}eiten conjecture, (2020). Available
at \url{https://arxiv.org/pdf/2011.02729.pdf}.
\bibitem[CK17]{CK2017}
\bgroup\scshape{}Y.~Chen\egroup{} and \bgroup\scshape{}S.~Koenig\egroup{},
Recollements of self-injective algebras, and classification of self-injective
diagram algebras, \emph{Math. Z.} \textbf{287} no.~3-4 (2017), 1009--1027.
\bibitem[CPS96]{Cline1996}
\bgroup\scshape{}E.~Cline\egroup{}, \bgroup\scshape{}B.~Parshall\egroup{}, and
\bgroup\scshape{}L.~Scott\egroup{}, Stratifying endomorphism algebras,
\emph{Mem. Amer. Math. Soc.} \textbf{124} no.~591 (1996), viii+119.
\bibitem[DK94]{Drozd1994}
\bgroup\scshape{}Y.~A. Drozd\egroup{} and \bgroup\scshape{}V.~V.
Kirichenko\egroup{}, \emph{Finite-dimensional algebras}, Springer-Verlag,
Berlin, 1994.
\bibitem[GL91]{GL1991}
\bgroup\scshape{}W.~Geigle\egroup{} and \bgroup\scshape{}H.~Lenzing\egroup{},
Perpendicular categories with applications to representations and sheaves,
\emph{J. Algebra} \textbf{144} no.~2 (1991), 273--343.
\bibitem[GR05]{Geis2005}
\bgroup\scshape{}C.~Gei{\ss}\egroup{} and \bgroup\scshape{}I.~Reiten\egroup{},
Gentle algebras are {G}orenstein, in \emph{Representations of algebras and
related topics}, \emph{Fields Inst. Commun.} \textbf{45}, Amer. Math. Soc.,
Providence, RI, 2005, pp.~129--133.
\bibitem[GLFS16]{Geiss2016}
\bgroup\scshape{}C.~Gei\ss\egroup{},
\bgroup\scshape{}D.~Labardini-Fragoso\egroup{}, and
\bgroup\scshape{}J.~Schr\"{o}er\egroup{}, The representation type of
{J}acobian algebras, \emph{Adv. Math.} \textbf{290} (2016), 364--452.
\bibitem[GdlP99]{GP1999}
\bgroup\scshape{}C.~Geiss\egroup{} and \bgroup\scshape{}J.~A. de~la
Pe{\~n}a\egroup{}, Auslander-{R}eiten components for clans, \emph{Bol. Soc.
Mat. Mexicana} \textbf{5} no.~2 (1999), 307--326.
\bibitem[GKK91]{Green1991}
\bgroup\scshape{}E.~L. Green\egroup{}, \bgroup\scshape{}E.~Kirkman\egroup{},
and \bgroup\scshape{}J.~Kuzmanovich\egroup{}, Finitistic dimensions of
finite-dimensional monomial algebras, \emph{J. Algebra} \textbf{136} no.~1
(1991), 37--50.
\bibitem[GZH91]{Green1991a}
\bgroup\scshape{}E.~L. Green\egroup{} and
\bgroup\scshape{}B.~Zimmermann-Huisgen\egroup{}, Finitistic dimension of
{A}rtinian rings with vanishing radical cube, \emph{Math. Z.} \textbf{206}
no.~4 (1991), 505--526.
\bibitem[HKK17]{HKK2017}
\bgroup\scshape{}F.~Haiden\egroup{}, \bgroup\scshape{}L.~Katzarkov\egroup{},
and \bgroup\scshape{}M.~Kontsevich\egroup{}, Flat surfaces and stability
structures, \emph{Publ. Math. Inst. Hautes \'{E}tudes Sci.} \textbf{126}
(2017), 247--318.
\bibitem[Hap89]{Happel1989d}
\bgroup\scshape{}D.~Happel\egroup{}, Hochschild cohomology of
finite-dimensional algebras, in \emph{S\'{e}minaire d'{A}lg{\`e}bre {P}aul
{D}ubreil et {M}arie-{P}aul {M}alliavin, 39{\`e}me {A}nn\'{e}e ({P}aris,
1987/1988)}, \emph{Lecture Notes in Math.} \textbf{1404}, Springer, Berlin,
1989, pp.~108--126.
\bibitem[Hap91]{Happel1991f}
\bgroup\scshape{}D.~Happel\egroup{}, On {G}orenstein algebras, in
\emph{Representation theory of finite groups and finite-dimensional algebras
({B}ielefeld, 1991)}, \emph{Progr. Math.} \textbf{95}, Birkh\"{a}user, Basel,
1991, pp.~389--404.
\bibitem[Hap93]{Happel1993}
\bgroup\scshape{}D.~Happel\egroup{}, Reduction techniques for homological
conjectures, \emph{Tsukuba J. Math.} \textbf{17} no.~1 (1993), 115--130.
\bibitem[HZZ20]{He2020}
\bgroup\scshape{}P.~He\egroup{}, \bgroup\scshape{}Y.~Zhou\egroup{}, and
\bgroup\scshape{}B.~Zhu\egroup{}, A geometric model for the module category
of a skew-gentle algebra, (2020). Available at
\url{https://arxiv.org/pdf/2004.11136.pdf}.
\bibitem[Hos84]{Hoshino1984}
\bgroup\scshape{}M.~Hoshino\egroup{}, Modules without self-extensions and
{N}akayama's conjecture, \emph{Arch. Math. (Basel)} \textbf{43} no.~6
(1984), 493--500.
\bibitem[HS12]{huang2012}
\bgroup\scshape{}Z.~Huang\egroup{} and \bgroup\scshape{}J.~Sun\egroup{},
Invariant properties of representations under excellent extensions, \emph{J.
Algebra} \textbf{358} (2012), 87--101.
\bibitem[HcV04]{Huneke2004a}
\bgroup\scshape{}C.~Huneke\egroup{}, \bgroup\scshape{}L.~M. \c{S}ega\egroup{},
and \bgroup\scshape{}A.~N. Vraciu\egroup{}, Vanishing of {E}xt and {T}or over
some {C}ohen-{M}acaulay local rings, \emph{Illinois J. Math.} \textbf{48}
no.~1 (2004), 295--317.
\bibitem[IT05]{Igusa2005}
\bgroup\scshape{}K.~Igusa\egroup{} and \bgroup\scshape{}G.~Todorov\egroup{}, On
the finitistic global dimension conjecture for artin algebras,
\emph{Representations of algebras and related topics} \textbf{45} (2005),
201--204.
\bibitem[Kal15]{Kalck2015}
\bgroup\scshape{}M.~Kalck\egroup{}, Singularity categories of gentle algebras,
\emph{Bull. Lond. Math. Soc.} \textbf{47} no.~1 (2015), 65--74.
\bibitem[KN09]{Koenig2009}
\bgroup\scshape{}S.~Koenig\egroup{} and \bgroup\scshape{}H.~Nagase\egroup{},
Hochschild cohomology and stratifying ideals, \emph{J. Pure Appl. Algebra}
\textbf{213} no.~5 (2009), 886--891.
\bibitem[LF09]{Fragoso2009}
\bgroup\scshape{}D.~Labardini-Fragoso\egroup{}, Quivers with potentials
associated to triangulated surfaces, \emph{Proc. Lond. Math. Soc. (3)}
\textbf{98} no.~3 (2009), 797--839.
\bibitem[LFSV22]{Fragoso2022}
\bgroup\scshape{}D.~Labardini-Fragoso\egroup{},
\bgroup\scshape{}S.~Schroll\egroup{}, and
\bgroup\scshape{}Y.~Valdivieso\egroup{}, Derived categories of skew-gentle
algebras and orbifolds, \emph{Glasg. Math. J.} \textbf{64} no.~3 (2022),
649--674.
\bibitem[Mar19]{Marczinzik2019}
\bgroup\scshape{}R.~Marczinzik\egroup{}, On weakly gorenstein algebras,
(2019). Available at \url{https://arxiv.org/pdf/1908.04738.pdf}.
\bibitem[Nee01]{Neeman2001}
\bgroup\scshape{}A.~Neeman\egroup{}, \emph{Triangulated categories}, no. 148,
Princeton University Press, 2001.
\bibitem[NS09]{NicolasSaorin2009a}
\bgroup\scshape{}P.~Nicol{{\'a}}s\egroup{} and
\bgroup\scshape{}M.~Saor{\'{\i}}n\egroup{}, Parametrizing recollement data
for triangulated categories, \emph{J. Algebra} \textbf{322} no.~4 (2009),
1220--1250.
\bibitem[OPS18]{Opper2018}
\bgroup\scshape{}S.~Opper\egroup{}, \bgroup\scshape{}P.-G. Plamondon\egroup{},
and \bgroup\scshape{}S.~Schroll\egroup{}, A geometric model for the derived
category of gentle algebras, (2018). Available at
\url{https://arxiv.org/pdf/1801.09659.pdf}.
\bibitem[Orl04]{Orlov2004}
\bgroup\scshape{}D.~O. Orlov\egroup{}, Triangulated categories of singularities
and {D}-branes in {L}andau-{G}inzburg models, \emph{Tr. Mat. Inst. Steklova}
\textbf{246} no.~Algebr. Geom. Metody, Svyazi i Prilozh. (2004), 240--262.
\bibitem[QH16]{QinHan2016}
\bgroup\scshape{}Y.~Qin\egroup{} and \bgroup\scshape{}Y.~Han\egroup{}, Reducing
{H}omological {C}onjectures by n-{R}ecollements, \emph{Algebr. Represent.
Theory} \textbf{19} no.~2 (2016), 377--395.
\bibitem[RR85]{Reiten1985b}
\bgroup\scshape{}I.~Reiten\egroup{} and \bgroup\scshape{}C.~Riedtmann\egroup{},
Skew group algebras in the representation theory of {A}rtin algebras,
\emph{J. Algebra} \textbf{92} no.~1 (1985), 224--282.
\bibitem[SZ03]{SZ2003}
\bgroup\scshape{}J.~Schr{{\"o}}er\egroup{} and
\bgroup\scshape{}A.~Zimmermann\egroup{}, Stable endomorphism algebras of
modules over special biserial algebras, \emph{Math. Z.} \textbf{244} no.~3
(2003), 515--530.
\bibitem[Xu13]{Xu2013a}
\bgroup\scshape{}D.~M. Xu\egroup{}, A note on the {A}uslander-{R}eiten
conjecture, \emph{Acta Math. Sin. (Engl. Ser.)} \textbf{29} no.~10 (2013),
1993--1996.
\bibitem[Xu15]{Xu2015}
\bgroup\scshape{}D.~Xu\egroup{}, Auslander-{R}eiten conjecture and special
biserial algebras, \emph{Arch. Math. (Basel)} \textbf{105} no.~1 (2015),
13--22.
\bibitem[Yam96]{Yamagata1996}
\bgroup\scshape{}K.~Yamagata\egroup{}, Frobenius algebras, in \emph{Handbook
of algebra, {V}ol.\ 1}, North-Holland, Amsterdam, 1996, pp.~841--887.
\bibitem[ZH95]{Zimmermann1995}
\bgroup\scshape{}B.~Zimmermann~Huisgen\egroup{}, The finitistic dimension
conjectures---a tale of {$3.5$} decades, in \emph{Abelian groups and modules
({P}adova, 1994)}, \emph{Math. Appl.} \textbf{343}, Kluwer Acad. Publ.,
Dordrecht, 1995, pp.~501--517.
\end{thebibliography}
%\bibliographystyle{aomalpha}
%\bibliography{../../refData}
\bigskip
Yiping Chen, School of Mathematics and Statistics, Wuhan University, Wuhan 430072, China.
{\tt Email: ypchen@whu.edu.cn}
\end{document}