Fate of dynamical quantum phase transitions from sudden quench to slow quench limit
Extracted source text · arXiv 2607.15649
%\pdfoutput=1
%\documentclass[aps,prl,floatfix,superscriptaddress,reprint]{revtex4-2}
%%\documentclass[article,twocolumn,showpacs,floatfix,longbibliography]{revtex4-2}
\documentclass[twocolumn,superscriptaddress,letter]{revtex4-2}%
\usepackage{eurosym}
\usepackage{amsbsy}
\usepackage{latexsym,epsfig}
\usepackage{graphicx}
\usepackage{dcolumn}
\usepackage{subfigure}
\usepackage{comment}
\usepackage{multirow}
\usepackage{color}
\usepackage{bm}
\usepackage{mathrsfs}
\usepackage{amssymb}
\usepackage{amsfonts}
\usepackage{amsmath}
\usepackage{xspace}
\usepackage{soul}
\usepackage{cancel}
\usepackage{float}
\usepackage{epstopdf}
\usepackage{tabularx}
\usepackage{longtable}
%\usepackage{caption}
%\usepackage{multirow}
\usepackage[colorlinks=true, letterpaper=true, pdfstartview=FitV, linkcolor=red, citecolor=blue, urlcolor=blue]{hyperref}
\usepackage[normalem]{ulem}
\usepackage[version=4]{mhchem}
\setcounter{MaxMatrixCols}{10}
\newcommand{\sgn}{\text{sgn}}
\newcommand{\bhy}{\hat y}
\newcommand{\bhz}{ \hat z}
\newcommand{\bp}{\mathbf p}
\newcommand{\bb}{\mathbf b}
\newcommand{\bc}{\mathbf c}
\newcommand{\bkp}{{\mathbf k}_{\parallel}}
\newcommand{\nn}{\nonumber}
\newcommand{\ft}{\footnote{\tiny}}
\newcommand{\bk}{\mathbf{k}}
\newcommand{\disav}[1]{\left\langle #1\right\rangle}
\newcommand{\bu}{\mathbf{u}}
\newcommand{\bs}{\mathbf{s}}
\newcommand{\ba}{\mathbf{a}}
\newcommand{\bmm}{\mathbf{m}}
\newcommand{\bM}{\mathbf{M}}
\newcommand{\bP}{\mathbf{P}}
\newcommand{\bD}{\mathbf{D}}
\newcommand{\bR}{\mathbf{R}}
\newcommand{\bG}{\mathbf{G}}
\newcommand{\cR}{{\cal R}}
\newcommand{\cS}{{\cal S}}
\newcommand{\cV}{{\cal V}}
\newcommand{\bK} {{\bf{K}}}
\newcommand{\bA} {{\bf{A}}}
\newcommand{\bC} {{\bf C}}
\newcommand{\cD} {{\cal D}}
\newcommand{\bhn}{\bf{\hat n}}
\newcommand{\bhr}{\mathbf{\hat r}}
\newcommand{\bnabla}{\mathbf\nabla}
\newcommand{\br}{\mathbf r}
\newcommand{\bh}{\mathbf h}
\newcommand{\bq}{\mathbf q}
\newcommand{\bx}{\mathbf r}
\newcommand{\by}{\mathbf y}
\newcommand{\bz}{\mathbf z}
\newcommand{\bn}{\mathbf n}
\newcommand{\bj}{\mathbf j}
\newcommand{\bpi}{\bm \pi}
\newcommand{\hn}{\hat n}
\newcommand{\dg}{\dagger}
\newcommand{\bS}{\mathbf{S}}
\newcommand{\bmu} {\bm \mu}
\newcommand{\bchi} {\bm \chi}
\newcommand{\brho} {\bm \rho}
\newcommand{\ve} {\varepsilon}
\newcommand{\be}{\begin{eqnarray}}
\newcommand{\ee}{\end{eqnarray}}
\newcommand{\la}{\langle}
\newcommand{\ra}{\rangle}
\newcommand{\rar}{\rightarrow}
\newcommand{\da}{\downarrow}
\newcommand{\ua}{\uparrow}
\begin{document}
\title{Fate of dynamical quantum phase transitions from sudden quench\\ to slow quench limit}
\author{Xiang Zhang}
\author{Fuxiang Li}
\email[Corresponding author: ]{fuxiangli@hnu.edu.cn}
\affiliation{School of Physics and Electronics, Hunan University, Changsha 410082, China}
\date{\today}
\begin{abstract}
Previous investigations of dynamical quantum phase transition (DQPT) have predominantly focused on sudden quench protocols. In this work, we systematically explore the fate of DQPT from the sudden to the slow quench regime. We establish that DQPT remains robust under slow quenching when the protocol crosses an equilibrium quantum phase transition. Conversely, we show that accidental DQPTs—which occur when the pre- and post-quench Hamiltonians reside in the same equilibrium phase—can be filtered out in the slow quench limit. This demonstrates that slow quenches can reveal the intrinsic connection between DQPT and equilibrium quantum phase transitions, thereby enabling DQPT to serve as a dynamical probe for identifying equilibrium quantum critical points. We illustrate these findings with examples from the XY chain, the Aubry-André model, and the trimer Su-Schrieffer-Heeger model.
\end{abstract}
\maketitle
%{\it Introduction.--}
\section{Introduction}
Driven by the experimental advances in various quantum-simulator platforms such as ultracold atoms \cite{Langen} and trapped ions \cite{Leibfried}, nonequilibrium quantum dynamics \cite{Polkovnikov2011,Liu2024,Zache2019} has thrived as a central research field in physics. Intriguing out-of-equilibrium phenomena, such as many-body localization \cite{Pal,Nandkishore,Abanin,Schreiber,Smith,Choi}, quantum Kibble-Zurek mechanism \cite{Zurek2005, Polkovnikov2005, Sen2008, Barankov2008, Dziarmaga2010, Nowak2021, Kou2023, Liang2024}, time-crystal \cite{Wilczek,Sacha,Zaletel} have been theoretically predicted and experimentally verified.
Among these, the idea of dynamical quantum phase transitions (DQPT) \cite{Heyl2013,Heyl2018dyn} has been proposed as a powerful framework to characterize quantum dynamics of nonequilibrium many-body systems. DQPT has advanced our understanding of nonequilibrium physics in quantum many-body systems, which may lead to potential applications in quantum computing.
Analogous to the nonanalytic behavior observed in equilibrium quantum and classical phase transitions, DQPT has been identified as the nonequilibrium counterpart in the real-time evolution of quantum systems \cite{Heyl2013,Heyl2018dyn}. A paradigmatic way to drive a quantum system far from equilibrium is through a quantum quench \cite{Polkovnikov2011}, where the system is initially prepared in the ground state of one Hamiltonian and subsequently evolved under a different Hamiltonian. In this setting, the evolution time plays the role of control parameter and can be understood as imaginary inverse temperature. A central quantity in the study of DQPT is the Loschmidt echo, defined as $\mathcal{L}(t)=|\langle\psi_0|\psi(t)\rangle|^2$ which measures the overlap between the time evolved state $|\psi(t)\rangle$ and the initial state $|\psi_0\rangle$, and its associated rate function $r(t)=-\lim_{N\to\infty}\ln\mathcal{L}(t)$, where $N$ is the number of degrees of freedom. In non-interacting models, DQPTs are signaled by nonanalyticities in the rate function $r(t)$ at critical times $t_c$, which occur when the time-evolved state becomes orthogonal to the initial state. A substantial body of work has been devoted to DQPT, both theoretically \cite{Heyl2013,Heyl2018dyn,Lang2018dynamical,Vosk2014dynamical,Bhattacharya2017inter,Huang2019dynamical,Link2020dynamical,Huang2016dynamical,Budich2016dynamical} and experimentally \cite{Jurcevic2017direct,Flaschner2018observation,Nie2020experimental,Guo20190bservation,xu2020measuring,Wang2019simulating}. The concept has also been extended to Floquet systems \cite{Yang2019floquet,Zamani2020floquet,Jafari2021floquet,Zamani2022out,Jafari2022floquet}, mixed states \cite{Bhattacharya2017mixed,Heyl2017dynamical}, finite temperatures \cite{Mera2018dynamical,Lang2018concurrence}, open quantum systems \cite{Lang2018dyn,Sedlmayr2018fate,Kyaw2020dynamical,Naji2022dissipative,Kawabata2023dynamical}, high-order topological systems \cite{Xiao2024dynamical,Maslowski2023dynamical}, and the quasiperiodic lattices \cite{Yang2017dynamical,Ye2025disentangling,Ye2024energy}. Moreover, DQPTs have been found to exhibit intriguing connections with various other phenomena, including entanglement dynamics \cite{Nicola2021entanglement,Poyhonen2021entanglement,Jafari2021floquet,Chen2025dynamical}, dynamical transition of the order parameter \cite{Heyl2014dynamical,Heyl2018dynamical,Homrighausen2017}, string-order operator \cite{Bandyopadhyay2021ob}, the work distribution function \cite{Abeling2016quantum}, and the quantum Mpemba effect \cite{Gilles2025reduced}.
Most of the investigations of DQPT have focused on the sudden quench protocols. In this context, the emergence of DQPT is highly sensitive to the specific details of the model under study. In the relatively simple model initially explored by Heyl {\it et al.}\cite{Heyl2013}, a direct connection seemed to exist between DQPT and equilibrium quantum phase transitions—namely, DQPTs appeared to occur when the quench crossed an equilibrium phase boundary. However, subsequent studies soon revealed that such a general one-to-one correspondence does not hold \cite{Vajna2014dis,Vajna2015topo,Ye2025disentangling,Andraschko2014dyn,Karrasch2017dyn}. To date, no clear universal relationship has been established between DQPT and equilibrium criticality, and their connection remains relatively underexplored.
%Recently, our work \cite{zhang2025uni} finds a potential connection between them. We adopt slow quench protocols and present a universal scaling laws between the critical times of DQPTs and the critical exponent of equilibrium quantum phase transitions. In addition, renormalization group transformations have also shown that the DQPTs are critical points associated with unstable fixed points of equilibrium Ising models, indicating that DQPT obeys scaling and universality \cite{Trapin2021,Heyl2015,Wu2020,Halimeh2021,HalimeharXiv}.
In this work, we extend the exploration of DQPT to the slow quench protocols \cite{zhang2025uni, Puskarov2016,Sharma2016}. We establish that the occurrence of DQPT remains robust under the slow quench regime when the quench protocol crosses the equilibrium quantum phase transitions. Conversely, we show that accidental DQPTs—which occur when the pre- and post-quench Hamiltonians reside in the same equilibrium phase—can be filtered out in the slow quench limit. Therefore, the slow quenches can reveal the intrinsic connection between DQPT and equilibrium quantum phase transitions.
We illustrate our finding by taking the XY chain model, the Aubry-Andr{\'e} model and the trimer Su-Schrieffer-Heeger model as examples. For the XY chain, previous study \cite{Vajna2014dis} has reported that DQPTs can occur under sudden quench when the pre-quench and post-quench Hamiltonian belong to the same ferromagnetic phase. Our results reveal that this kind of {\color{blue}DQPT} is accidental and can be filtered out by the slow quench limit. For the Aubry-André model, Ref.~\cite{Ye2025disentangling} reported the emergence of energy-dependent {\color{blue} DQPT} during sudden quenches that begin and end within the same extended phase. We argue that these energy-dependent DQPTs are accidental in nature, as their occurrence explicitly relies on the choice of the initial state. In contrast, as the dynamical counterpart of equilibrium phase transitions, genuine {\color{blue}DQPT} should remain robust and is independent of both the initial state preparation and microscopic details of the Hamiltonian. Importantly, such state-sensitive DQPTs are also eliminated in the slow-quench limit.
% further supporting their classification as non-universal artifacts.
As for the trimer Su-Schrieffer-Heeger model, it hosts two distinct topological phase transitions: one accompanied by bulk gap closing and the other occurring without bulk gap closing. Moreover, we emphasize that the correspondence between slow-quench DQPT and equilibrium phase transitions is fundamentally tied to the presence of bulk gap closure. In its absence, such dynamical signatures fail to reliably capture equilibrium critical behavior. %Furthermore, within the mirror-symmetric regime, we analyze the scaling of the first critical time and extract the critical exponent product $z\nu=1$, consistent with the universal scaling relation reported in Ref.~\cite{zhang2025uni}.
%A fundamental question for the same phase quench protocols concerns the value of the quench rate $\beta$ required to filter out these accidental DQPTs. We further establish a universal criterion in terms of the quench rate $\beta$, and derive an explicit threshold for $\beta$ using the Landau–Zener transition formula.
This paper is organized as follows. In Sec. \ref{s2} we present a general argument of the fate of DQPT in two-band model. Models and quench dynamics are developed in Sec. \ref{s3}. We investigate three distinct models to illustrate our finding: the XY chain, the Aubry-Andr{\'e} model, and the trimer Su-Schrieffer-Heeger model and perform a comprehensive numerical study of DQPT under both sudden and slow quench protocols. We finally conclude in Sec. \ref{s4}.
\section{General Argument: DQPTs from sudden to slow quench protocols}
\label{s2}
We begin with a general one-dimensional two-band model. %Assuming a unit lattice constant, the first Brillouin zone is the circle resulting from the interval $[0,2\pi]$ by identification of its end points.
The Bloch Hamiltonian is of the following general form:
\begin{equation}
H_k(h)=\vec{d}_k(h)\cdot\vec{\sigma},
% =\begin{pmatrix}
% d_{k,3} & d_{k,1}-id_{k,2} \\
% d_{k,1}+id_{k,2} & -d_{k,3}
% \end{pmatrix},
\label{Hk}
\end{equation}
where the Hamiltonian is parametrized by a vector $\vec{d}_k$ and the Pauli matrices $\vec{\sigma}$, and $h$ is an control parameter. For the study of nonequilibrium quantum real-time evolution and DQPTs induced by quantum quench, we prepare the system in the ground state $|\psi_k\rangle$ of the initial Hamiltonian $H_k(h_i)=\vec{d}_k(h_i)\cdot\vec{\sigma}$, then perform a sudden quench or slow quench as illustrated in Fig.~\ref{p1}.
Under sudden quench, a measure of the stability of time reversal is the Loschmidt echo defined by~\cite{Gorin2006dynamics,Quan2006decay,Hasegawa2021irreversibility,Jafari2017loschmidt}
\begin{equation}
\mathcal{L}(t) = \big| \langle \psi(t_i) | e^{i H(h_i) (t-t_i)} e^{-i H(h_f) (t-t_i)} | \psi(t_i) \rangle \big|^2,
\label{eq:loschmidt}
\end{equation}
where $H(h)$ is a time-independent Hamiltonian depending on a microscopic parameter $h$, whose value is assumed to change from $h_f$ during the forward time evolution to $h_i$ during the backward time evolution. $\mathcal{L}(t)$ thus is nothing but the fidelity between the state time-evolved with $H(h_i)$ and the state time-evolved with $H(h_f)$. In this work, we are interested in the recent idea \cite{Heyl2013} to use the Loschmidt echo to study sudden quenches in closed quantum systems. We concentrate on the case where $|\psi(t_i)\rangle$ is the ground state of $H(h_i)$. In this case, the Loschmidt echo reduces to
\begin{equation}
\mathcal{L}(t) =|\mathcal{G}(t)|^2= \big| \langle \psi(t_i) | e^{-i H(h_f) (t-t_i)} | \psi(t_i) \rangle \big|^2,
\label{eq:loschmidt_reduced}
\end{equation}
where $\mathcal{G}(t)$ is the so-called Loschmidt amplitude and quantifies the deviation of the time-evolved state from the initial state \cite{Heyl2018dyn}. The Loschmidt echo \cite{Gorin2006dynamics,Quan2006decay,Hasegawa2021irreversibility,Jafari2017loschmidt} serves as a powerful tool of characterizing the quantum decoherence effect \cite{Zurek2001sub,Zurek2003decoherence,Cucchietti2003decoherence}, quantum chaos \cite{Gutierrez2009long,Jacquod2001golden,Cerruti2002sensitivity,Wisniacki2003short} and quantum criticality \cite{Cucchietti2004universality,tang2022dynamical}.
\begin{figure}[htbp]
\includegraphics[width=0.7\linewidth]{protocol.png}
\caption{Schematic of the quench protocols: Sudden quench protocol \textcircled{1} and slow quench protocol \textcircled{2}. Here, for both quench protocols, the initial $h_i$ and final values $h_f$ of quench parameter $h$ are fixed. For the slow quench protocol \textcircled{2}, $h$ is changed as a function of time $t$ with quench rate $\beta$, i.e., $h_f=h_i+\beta t$.}
\label{p1}
\end{figure}
We first consider the sudden quench protocol, as shown in Fig.~\ref{p1}(\textcircled{1}), the parameter $h$ is switched suddenly from $h_i$ to $h_f$ at time $t=t_i$, resulting in a sudden change $\vec{d}_k(h_i)\to\vec{d}_k(h_f)$. We assume that the system
initially occupies the lower Bloch band of $H_k^i\equiv H_k(h_i)$. The associated lower band Bloch states (ground state) are denoted by $|u_k^{i,-}\rangle$ such that the initial state $|\psi(t_i)\rangle$ is a Slater determinant of all lower band Bloch states. Since lattice translation invariance is maintained at all times, the dynamics of the system can be considered separately for every lattice momentum $k$. Explicitly, we get
\begin{equation}
\begin{split}
|\psi_k(t)\rangle&=e^{-i H_k^f (t-t_i)} |u_k^{i,-}\rangle,\\
&=e^{iE_k^f (t-t_i)}c_{k,1} |u_k^{f,-}\rangle+e^{-iE_k^f (t-t_i)}c_{k,2} |u_k^{f,+}\rangle,
\end{split}
\end{equation}
where $E_k^f=|\vec{d}_k(h_f)|$ denotes the energy eigenvalues of $H_k^f$, i.e., $H_k^f|u_k^{f,\pm}\rangle=\pm E_k^f|u_k^{f,\pm}\rangle$ and $c_{k,1}=\langle u_k^{f,-}|u_k^{i,-}\rangle$, $c_{k,2}=\langle u_k^{f,+}|u_k^{i,-}\rangle$. Because the different momentum sectors are decoupled, ground states exhibit a factorization property. Then the Loschmidt amplitude can be expressed in a compact form:
\begin{equation}
\begin{split}
\mathcal{G}(t)&=\prod_k\mathcal{G}_k(t)=\prod_k\langle u_k^{i,-}|\psi_k(t)\rangle\\
&=\prod_k\left(|c_{k,1}|^2 e^{i E_k^f (t-t_i)}+|c_{k,2}|^2 e^{-i E_k^f (t-t_i)}\right).
\label{sudden}
\end{split}
\end{equation}
{\color{blue}DQPT} occurs when the Loschmidt amplitude $\mathcal{G}(t)$ vanishes at critical times $\tau_c$ for a critical momentum $k_c$ \cite{Heyl2018dyn}. Here this can only happen if
\begin{equation}
|c_{k_c,1}|^2=|c_{k_c,2}|^2 \quad {\rm and} \quad \tau_{c,n}=\frac{(2n-1)\pi}{2E_{k_c}^f}-t_i,\quad n\in \mathbb{N}.
\label{tc}
\end{equation}
{\color{blue}DQPT} hence occurs at critical momenta $k_c$ where the initial lower Bloch state is an equal weight superposition of the final Bloch states and the critical time $\tau_c$ is determined by the spectrum $E_{k_c}^f$ of the final Hamiltonian $H_{k_c}^f$. From the perspective of Eq.~(\ref{tc}), it is apparent that the survival conditions for DQPTs are not inherently linked to equilibrium quantum phase transitions. As demonstrated in Refs.~\cite{Vajna2014dis,Ye2025disentangling}, the conditions of Eq.~(\ref{tc}) can also be satisfied under sudden quench protocols when the quenching process originates and terminates within the same phases, the so-called accidental DQPTs. Therefore, the connection between DQPT and equilibrium quantum phase transitions is ambiguous when considering sudden quench protocols.
% Recently, this study \cite{Vajna2014} has found that DQPTs can occur even though the sudden quenches don't cross equilibrium quantum phase transition.
% In this paper, we try to build a bridge between DQPTs and equilibrium quantum phase transition such that we can use DQPTs to characterize some properties of equilibrium quantum phase transition.
In the following, we will generalize the sudden quench protocols \textcircled{1} to slow quench protocols \textcircled{2} (Fig.~\ref{p1}) to investigate DQPTs \cite{Sharma2016,zhang2025uni,Puskarov2016}. As shown in Fig.~\ref{p1}, the quench parameter $h$ is slowly changed from $h_i$ to $h_f$ with a quench rate $\beta$, i.e., $h_f=h_i+\beta t$ during a finite time interval $[t_i,t_f]$. After the slow quenching, the subsequent temporal evolution ($t>t_f$) will be governed by the final time-independent Hamiltonian $H^f$. In the slow quench protocols, two different approaches can be adopted to quantify the Loschmidt amplitude \cite{Heyl2018dyn}. In the first approach, the Loschmidt amplitude $\mathcal{G}(t)$ and the corresponding rate function $r(t)$ are defined as
\be
\mathcal{G}(t) = \langle \psi(t_i)|\psi(t)\rangle, ~~r\left(t\right)=-\lim_{N\rightarrow\infty}\frac{1}{N}\ln |\mathcal{G}(t)|^2,
\label{slow}
\ee
where $N$ is the number of degrees of freedom. When the quench rate $\beta$ satisfies $\beta\to \infty$, Eq.~(\ref{slow}) will return to the expression Eq.~(\ref{sudden}) of sudden quench protocols. The nonanalyticity at critical times $\tau_c$ in the rate function corresponds to DQPTs.
Here, $|\psi(t)\rangle$ {\color{blue} is} the time evolved state in the slow quench duration and $|\psi(t_i)\rangle$ {\color{blue}is} the initial state at initial time $t_i$. This approach has been adopted in our recent work \cite{zhang2025uni} with sufficiently slow quenching rate so that the first DQPT occurs during the slow quench duration.
In this work, all discussions will be based on the first approach. In the two-band model, the state vector $|\psi(t)\rangle$ during the slow quench process can always be formally written as a linear superposition of the instantaneous eigenstates of the time-dependent Hamiltonian $H_k(t)$:
\begin{equation}
|\psi_k(t)\rangle=p_{k,1}(t) |u_k^{-}(t)\rangle+ p_{k,2}(t) |u_k^{+}(t)\rangle.
\end{equation}
Here, $p_{k,1}(t)$ and $p_{k,2}(t)$ are the occupation amplitudes of state $|\psi_k(t) \ra$ on the two instantaneous eigenstates $|u_k^{\pm}(t)\rangle$ of time-dependent Hamiltonian $H_k(t)$, and satisfy $|p_{k,1}(t)|^2 + |p_{k,2}(t)|^2=1$. If we start from the initial ground state $|\psi(t_i)\rangle= \prod_k|u_k^{i,-}\rangle$, the Loschmidt amplitude
$\mathcal{G}(t)$ in Eq.~(\ref{slow}) reads
\begin{equation}
\mathcal{G}(t)=\prod_k\left(p_{k,1}(t)\langle u_k^{i,-}|u_k^-(t)\rangle+p_{k,2}(t)\langle u_k^{i,-}|u_k^+(t)\rangle\right).
\label{le}
\end{equation}
For simplicity, the occupation amplitudes can be rewritten as $p_{k,1}(t)=\langle u_{k}^{-}(t)|\psi_k(t)\rangle=|p_{k,1}(t)|e^{-i\theta_{k,1}(t)}$ and $p_{k,2}(t)=\langle u_{k}^{+}(t)|\psi_k(t)\rangle=|p_{k,2}(t)|e^{i\theta_{k,2}(t)}$. Meanwhile, the transition amplitudes between initial state $|u_{k}^{i,-}\rangle$ and instantaneous eigenstates $|u_k^{\pm}(t)\rangle$ can also be rewritten as
$\langle u_k^{i,-}|u_k^-(t)\rangle=|c_{k,1}(t)|e^{i\phi_{k,1}(t)}$ and $\langle u_k^{i,-}|u_k^+(t)\rangle=|c_{k,2}(t)|e^{-i\phi_{k,2}(t)}$. Therefore, the Loschmidt amplitude is
\begin{equation}
\mathcal{G}_k(t)=\prod_k \left(|p_{k,1}||c_{k,1}|e^{i\Phi_{k,1}(t)}+|p_{k,2}||c_{k,2}|e^{-i\Phi_{k,2}(t)}\right).
\label{le1}
\end{equation}
Here, $\Phi_{k,i}(t)=\phi_{k,i}(t)-\theta_{k,i}(t)$, $i=1,2$ and we have ignored the temporal symbols for coefficients. For sudden quench protocol \textcircled{1}, Eq.~(\ref{le1}) return to Eq.~(\ref{sudden}). Note that the magnitude of the coefficients $|c_{k,i}|$ is not related to the quench rate $\beta$, which can be seen as a constant and $|p_{k,i}|$ is closely related to the quench rate $\beta$. The existence of critical momentum $k_c$ of DQPTs requires that the magnitudes of the coefficients satisfy
\begin{equation}
|p_{k_c,1}||c_{k_c,1}|=|p_{k_c,2}||c_{k_c,2}|.
\label{cond}
\end{equation}
The above condition is different from the condition (\ref{tc}) (sudden quench) and the concrete values ($|p_{k_c,i}|$ and $|c_{k_c,i}|$) of the magnitudes for coefficients depend on models.
Before we discuss the specific models in next section, we want to first understand the fate of DQPTs under slow quench by making a qualitative discussion of Eq.~(\ref{cond}). One can obtain the values of $p_{k, 1}(t)$ and $p_{k, 2}(t)$ by numerically solving the Schrodinger equation with time-dependent Hamiltonian:
\begin{equation}
i\frac{d|\psi_k(t)\rangle}{dt}=H_k(t)|\psi_k(t)\rangle.
\label{se}
\end{equation}
If one linearly quenches the parameter $h$ in the two-band model (\ref{Hk}), the time-dependent Hamiltonian can be written in the following form
\begin{equation}
H_k(t)=\begin{pmatrix}
h_k(t) & \Delta_k \\
\Delta^*_k & -h_k(t)
\end{pmatrix},
\label{lz}
\end{equation}
after, if necessary, making a transformation of representation. A specific example can be found in Sec.~\ref{sec:XY} for the XY chain model. Here, $h_k(t)=h_i+\beta t$, and $\Delta_k$ is the coupling between the two energy levels.
This is the famous Landau-Zener-St${\rm \ddot{u}}$ckelberg-Majorana problem \cite{Landau1932theorie,Zener1932non,Stueckelberg1932theorie,Majorana1932atomi}, or simply called Landau-Zener problem. Landau-Zener problem provides an analytical solution for the transition probability from the ground state at initial time $t\rightarrow -\infty$ to the excited state at final time $t\rightarrow \infty$
\begin{equation}
P=e^{-\frac{\pi\Delta_k^2}{\beta}}.
\end{equation}
For general cases of time evolution, the above Landau-Zener formula can serve as a good approximation.
When the system is quenched across the quantum phase transition, there always exists a $k_0$ point such that $\Delta_{k_0}=0$, so that two energy levels simply cross with each other with no transition between the two levels. At this $k_0$ point, if starting from the ground state, the system will finally go to the upper level, and the transition amplitude $|p_{k_0,2}|=1$. For other $k$ points away from the critical point $k_0$, $\Delta_k$ is finite, and the transition probability $|p_{k_0, 2}|$ decreases towards zero with decreasing quench rate $\beta$. In the thermodynamic limit ($N\to\infty$), momentum is continuous and thus we can always find a special point $k_c$ such that the condition (\ref{cond}) is satisfied, which constitutes a necessary condition for the DQPTs.
One the other hand, if we do not quench system across the quantum phase transition, but within the same phase, the system is always gapped, and $\Delta_k$ is always nonzero. As long as the quench rate is sufficiently slow, i.e., very small $\beta$, the transition amplitude $p_{k, 2}$ would be always very small for each $k$, so that the condition (\ref{cond}) for DQPTs can not be satisfied, and no DQPTs can occur.
% {\color{blue} For intuitive understanding, we can consider a limiting case $\beta\to0$, namely, adiabatic evolution, then the coefficient $p_{k,2}$ equals to $0$ at all times such that the condition ($\ref{cond}$) can not be satisfied. This is a simplest case, we should note that the invalidation of condition ($\ref{cond}$) is not only limited to the case $\beta\to0$. For a slow but not adiabatic quench rate $\beta$, the condition ($\ref{cond}$) can also be invalid, which may have a potential application in the experiment due to the difficulty of adiabaticity.}
Therefore, slow quench dynamics can rule out accidental DQPTs that appear within the same phase under sudden quench.
There is an alternative approach for the Loschmidt amplitude $\mathcal{G}(t)$, which is defined as
\begin{equation}
\mathcal{G}(t)=\langle\psi(t_f)|\psi(t)\rangle,
\end{equation}
in which $|\psi(t)\rangle$ is the time evolved state governed by the final time-independent Hamiltonian $H^f$ after the slow quench duration, i.e., $|\psi(t)\rangle=e^{-iH^f(t-t_f)}|\psi(t_f)\rangle$ and $|\psi(t_f)\rangle$ is the state vector at the end of the slow quench duration at time $t_f$ \cite{Puskarov2016,Sharma2016}. Note that, in the slow quench protocols, the second approach is equivalent to preparing a different initial state, which renders the comparison with its sudden qunech counterpart unreasonable. Therefore, we will restrict our discussions within the first approach.
%To qualitatively understand the fate of DQPTs under slow quench, we consider a sufficiently small quench rate but the system do not stay in the adiabatic regime, namely, the excitation probability $p_{k,2}$ is small and can be neglected compared to probability $p_{k,1}$. In this regime, DQPTs occur when the Loschmidt amplitude satisfies \(\mathcal{G}(t)=0\). From Eq.~(\ref{le}), this condition can be fulfilled only if \(p_{k_c,1}(\tau_c)\langle u_{k_c}^{i,-}|u_{k_c}^-(\tau_c)\rangle = 0\) at a critical time \(\tau_c\) and critical momentum \(k_c\). For a two-band model, the initial and time-evolved states can be represented as points on the Bloch sphere. By choosing the initial state to lie at the north pole at initial time \(t_i\), the condition \(p_{k_c,1}(\tau_c)\langle u_{k_c}^{i,-}|u_{k_c}^-(\tau_c)\rangle = 0\) implies that the time-evolved state reaches the south pole at the critical time \(\tau_c\). Under slow quench, this can be satisfied only if the slow quench crosses the quantum critical points. Therefore, slow quench protocols reveal an intrinsic connection between DQPTs and equilibrium quantum phase transitions: DQPTs occur only when the slow quench traverses a quantum phase transition point, whereas they are completely absent for slow quenches confined to a single phase. Therefore, DQPTs under slow quench protocol can serve as a faithful dynamical signature to identify equilibrium quantum critical points.
% {\color{blue} The time-evolved wave function $|\psi(t)\rangle$ can be obtained based on the adiabatic-impulse approximation in the Kibble-Zurek mechanism \cite{Kibble1976topology,Kibble1980some,Zurek1985cosmological,Zurek1996cosmological,Del2014universality,Damski2005}. According to this approximation, the slowly quenched process across a quantum transition point can be divided into two regimes, the adiabatic and impulse regimes, depending on the relative magnitude of quench rate of control parameter and the relaxation rate of the many-body system. At initial time (large negative time $t_i$), the system can adiabatically follow the instantaneous ground state $|\Psi_0(t)\rangle$ until at the transition time $-\hat{t}$ when the timescale of change of the control parameter is equal to the relaxation time of the system. Thereafter, the state is effectively frozen to the state at the transition time $|\Psi_0(-\hat{t} ) \rangle$.
% The frozen regime ends at time $\hat{t}$, and the system then enters into the adiabatic regime, in which the probability remains unchanged on the adiabatic eigenstates, and only the dynamical phase $\eta$ is accumulated.
% Therefore, at time $t>\hat{t}$, one can write down the time-evolved wave function as (see details in our recent work \cite{zhang2025uni})
% \begin{equation}
% |\psi(t)\rangle = \sum_{n} \sqrt{P_n} e^{-i \eta_n (t)} |\Psi_n(t) \ra,
% \label{wf}
% \end{equation}
% where $|\Psi_n(t) \ra$ being the $n$-th instantaneous eigenstate. Note that in the adiabatic regime, the probability $P_n=|\la \Psi_0(-\hat{t} )|\Psi_n(\hat{t})\ra|^2$ remains unchanged. Then the Loschmidt amplitude $\mathcal{L}(t)$ is
% \begin{equation}
% \mathcal{L}(t)=\sum_{n} \sqrt{P_n} e^{-i \eta_n (t)} \langle\psi(t_i)|\Psi_n(t) \ra.
% \end{equation}
% Taking the two-band model as an example, i.e., $n=0,1$, the dynamical phase $\eta$ is just opposite. DQPTs occur when $|\mathcal{L}(\tau_c)|^2=0$, which can be satisfied at the critical time $\tau_c$ if $\sqrt{P_n} \langle\psi(t_i)|\Psi_n(\tau_c) \ra=1/2$. However, for slow quench within the same phase (without gap closing), the adiabatic-impulse approximation is invalid and the time-evolved wave function can be always in the instantaneous ground state $|\Psi_0(t)\rangle$ at all times under the slow quench limit with very small quench rate $\beta$. In this case, the condition $|\mathcal{L}(\tau_c)|^2=0$ is impossible.}
% {\color{blue} In this work, we aim to study the fate of DQPTs from sudden quench to slow quench. We have shown that the survival conditions for DQPTs are not inherently linked to equilibrium quantum phase transitions from the perspective of Eq.~(\ref{tc}). In stark contrast, through the numerical calculation, slow quench protocols reveal an intrinsic connection between DQPTs and equilibrium quantum phase transitions: DQPTs occur only when the slow quench traverses a quantum phase transition point, whereas they are completely absent for slow quenches confined to a single phase. Therefore, DQPTs under slow quench protocol can serve as a faithful dynamical signature to identify equilibrium quantum critical points.}
% In the first approach, the Loschmidt amplitude is defined as $\mathcal{L}(t)=\langle\psi(t_i)|\psi(t)\rangle$ with $|\psi(t)\rangle$ being the time evolved state in the slow quench duration and the subsequent temporal evolution duration, and $|\psi(t_i)\rangle$ being the initial state at initial time $t_i$. This approach has been adopted in our recent work \cite{zhang2025uni} with sufficiently slow quenching rate so that the first DQPT occurs during the slow quench duration. In the second approach,
% {\color{blue} Note that, in the slow quench protocols, the other approach \cite{Sharma2016,Puskarov2016} is adopted to quantify the Loschmidt amplitude. The Loschmidt amplitude is defined as $\mathcal{L}(t)=\langle\psi(t_f)|\psi(t)\rangle$, in which $|\psi(t)\rangle$ is the time evolved state governed by the final time-independent Hamiltonian $H^f$ after the slow quench,i.e., $|\psi(t_f)\rangle=e^{-iH^f(t-t_f)}|\psi(t_f)\rangle$, and $|\psi(t_f)\rangle$ is selected as the initial state at the end of the duration of slow quench at time $t_f$ \cite{Puskarov2016,Sharma2016}.
% Crucially, the slow quench in this case is equivalent to preparing a different initial state, which renders a meaningful comparison with its sudden qunech counterpart unreasonable.}
% When adopting the second approach, the quenching rate $\beta$ is usually very large (corresponding to short quench time), and the first DQPT occurs usually after the slow quench duration. Therefore, this approach is more suitable for the comparison between slow quench protocol and sudden quench protocol. In this work, most of the discussions will be based on the second approach.
% In the two band models, the state vector $|\psi(t_f)\rangle$ at the end of the slow quench time $t_f$ can always be represented as a linear superposition of the eigenstate of the final Hamiltonian $H_k^f$:
% \begin{equation}
% |\psi_k(t_f)\rangle=p_{k,1} |u_k^{f,-}\rangle+ p_{k,2} |u_k^{f,+}\rangle.
% \end{equation}
% Here, $p_{k, 1}$ and $p_{k, 2}$ are the occupation amplitudes of state $|\psi(t_f) \ra$ on the two eigenstates of final Hamiltonian $H_k^f$.
% Then the Loschmidt amplitude is reduced as
% \begin{equation}
% \mathcal{L}(t)=\prod_k\left(|p_{k,1}|^2 e^{i E_k^f t}+|p_{k,2}|^2 e^{-i E_k^f t}\right).
% \end{equation}
% For DQPTs to occur, the condition is
% \begin{equation}
% |p_{k_c,1}|^2=|p_{k_c,2}|^2 \quad {\rm and} \quad t_{c,n}=\frac{(2n-1)\pi}{2E_{k_c}^f},\quad n\in \mathbb{N}.
% \label{t}
% \end{equation}
% Consequently, for DQPTs to occur under a slow quench protocol across the interval $[t_i,t_f]$ with an rate $\beta$, the state $|\psi_k(t_f)\rangle$ must be represented as an equal weight superposition of the eigenstate of the final Hamiltonian $H_k^f$:
% \begin{equation}
% |\psi_k(t_f)\rangle=\sqrt\frac{1}{2}|u_k^{f,-}\rangle+\sqrt\frac{1}{2} |u_k^{f,+}\rangle.
% \label{phi0}
% \end{equation}
% This criterion (\ref{phi0}) reveals an intrinsic connection between the emergence of DQPTs in slow quenches and the equilibrium quantum phase transitions of the system. When the pre-quench and post-quench Hamiltonians reside within the same equilibrium phase, a quench rate $\beta$ can always be found such that Eq.~(\ref{phi0}) is not satisfied. Conversely, if they belong to distinct phases, crossing the equilibrium quantum critical point universally induces a gap closure at a momentum $k_0$, which inevitably satisfies the condition in Eq.~(\ref{phi0}). At this gap closure momentum $k_0$, the excitation probability $|p_{k_0,2}|^2$ into the higher eigenenergy level of the final Hamiltonian approaches unity, i.e.,
% $|p_{k_0,2}|^2\to 1$, regardless of the quench rate $\beta$. In regions of momentum space where the energy gap remains open, one can always select a value of $\beta$ such that $|p_{k,1}|^2\to 1$ for the lower level. Due to the continuity of momentum $k$ in the thermodynamic limit, there must exist at least one critical momentum $k_c$ at which the excitation probabilities satisfy $|p_{k_c,1}|^2=|p_{k_c,2}|^2=\frac{1}{2}$, which constitutes the necessary condition for the occurrence of the dynamical quantum phase transitions. Therefore, slow quench protocols can filter out the accidental DQPTs that appear in the sudden quench protocol with the pre- and post-quench Hamiltonian being in the same equilibrium phase, and thus establish a fundamental link between dynamical and equilibrium quantum phase transitions.
% A fundamental question is the value of the slow quench rate $\beta$ required to filter out these accidental DQPTs. To address this, we turn to the Landau-Zener model. The Landau-Zener (LZ) model, or the Landau-Zener-St${\rm \ddot{u}}$ckelberg-Majorana model, is perhaps the most well-known time-dependent quantum model which is exactly solvable. Since its introduction in the four pioneering works in 1932
% by Landau \cite{Landau1932theorie}, Zener \cite{Zener1932non}, St${\rm \ddot{u}}$ckelberg \cite{Stueckelberg1932theorie} and Majorana \cite{Majorana1932atomi}, this model has become a prototype to study nonadiabatic transitions, and it has found realizations and applications in a variety of physical systems \cite{Shevchenko2010landau,Ivakhnenko2023nonadiabatic}. The LZ model describes a two-level system evolving under the Schr${\rm \ddot{o}}$dinger equation with linear driving:
% \begin{equation}
% i\frac{d|\psi_k(t)\rangle}{dt}=H_k(t)|\psi_k(t)\rangle, \quad \\ H_k(t)=\begin{pmatrix}
% h_k(t) & \Delta_k \\
% \Delta_k & -h_k(t)
% \end{pmatrix},
% \end{equation}
% where $h_k(t)=h_i+\beta t$ and $\Delta_k$ is the coupling between the two energy levels. The transition probability from the ground state at initial time $t\rightarrow -\infty$ to the excited state at final time $t\rightarrow \infty$ has an exact analytic formula, namely, the LZ formula:
% \begin{equation}
% P=e^{-\frac{\pi\Delta_k^2}{\beta}}.
% \end{equation}
% The condition for DQPTs under slow quench protocol to occur, as specified by Eq.~(\ref{phi0}), requires the transition probability to equal 1/2. This leads to the threshold quench rate:
% \begin{equation}
% \beta^{th}=\frac{\pi\Delta_k^2}{\ln 2}.
% \end{equation}
% Here, $k$ denotes the momentum at which the energy gap is minimum, thereby determining the dominant contribution to the threshold.
\section{Model and Quench dynamics}
\label{s3}
In this section, we investigate three different models: the XY chain, the Aubry-André model, and the trimer Su-Schrieffer-Heeger model, and numerically study the DQPT under both sudden and slow quench protocols, within the same phase and across the phase transitions. We will explore the fate of DQPT from the sudden to the slow quench regime. We will demonstrate that DQPTs remain robust under slow quenching when the protocol crosses an equilibrium quantum phase transition. In contrast, accidental DQPTs, which occur when the pre- and post-quench Hamiltonians reside in the same equilibrium phase, will disappear in the slow quench limit.
% {\color{blue} Note that the XY chain can be mapped onto a two-band model in momentum space via a Jordan-Wigner transformation followed by a Fourier transform, thereby falling within the scope of our general argument in Sec.~\ref{s2}. For the remaining two models, we provide numerical evidence demonstrating that our argument extends beyond two-band systems to encompass multi-band models as well.
% }
\subsection{XY Chain \label{sec:XY}}
We begin with the spin-1/2 quantum XY chain under a uniform transverse field (homogeneous for each spin). The Hamiltonian of the transverse-field XY chain with nearest-neighbor interaction is given by
\begin{equation}
H=-\sum_{j=1}^N\left[\frac{1+\gamma}{2}\sigma_j^x\sigma_{j+1}^x+\frac{1-\gamma}{2}\sigma_j^y\sigma_{j+1}^y+h\sigma_j^z\right],
\end{equation}
where $N$ counts the number of spins, $\sigma_j^i$ $(i=x,y,z)$ are the Pauli matrices acting on the $j$th spin, $\gamma$ represents the strength of anisotropy, and $h$ measures the strength of the transverse field. The system reduces to the isotropic XY chain for $\gamma\to 0$ and {\color{blue}to} the Ising chain for $\gamma\to 1$.
This model is exactly solvable by the Jordan-Wigner transformation, which maps a system of spin 1/2 to a system of spinless free fermions \cite{Jordan1928paulische,Dziarmaga2005}. After the Jordan-Wigner transformation, the XY chain Hamiltonian can be rewritten as
\begin{equation}
\begin{split}
H&=-\sum_{j=1}^N\left(c_j^\dagger c_{j+1}+\gamma c_j^\dagger c_{j+1}^\dagger+{\rm H.c.}\right)\\
&+h\sum_{j=1}^N\left(2c_j^\dagger c_j-1\right),
\end{split}
\end{equation}
where H.c. denotes the Hermitian conjugate. After the Fourier
transformation of the fermionic operators, we obtain the momentum-space representation of the Hamiltonian,
\begin{equation}
H=\sum_k\psi_k^\dagger H_k\psi_k
\label{xyha}
\end{equation}
where $\psi_k=(c_k,c_{-k}^\dagger)^T$ are Nambu spinors and $H_k=\vec{d_k}\cdot\vec{\sigma}=\gamma \sin k\sigma_y+(h-\cos k)\sigma_z$. The energy is $E(k)=\pm\sqrt{\gamma^2 \sin^2k+(h-\cos k)^2}$. The system has quantum critical points at $h=\pm1$. For the $h=1$, the gap closing occurs at $k=0$, while for $h=-1$, it occurs at $k=\pm\pi$. This model exhibits two phases, $|h|<1$ for ferromagnetic phase and $|h|>1$ for paramagnetic phase. %Within the ferromagnetic phase, the order parameter is $\langle\sigma_x\rangle$ for $\gamma>0$ and the order parameter is $\langle\sigma_y\rangle$ for $\gamma<0$.
Previous studies of DQPTs under sudden quench in the XY model have shown that, even when the pre- and post-quench Hamiltonians are in the same ferromagnetic phases,
DQPT can still occur \cite{Vajna2014dis}. Here, we extend the investigation of DQPT from sudden quench regime to the slow quench regime, and show that under slow quench regime, accidental DQPTs which occur when the pre- and post-quench Hamiltonians reside in the same equilibrium phase, will disappear. Explicitly, we implement two distinct quench protocols within the same ferromagnetic phase with fixed $\gamma$: (i) sudden quench protocol, where the field $h$ is abruptly switched from $0$ to $0.9$ at time $t=0$; (ii) slow linear quench protocol, with $h(t)=h_i+ \beta t$, where the initial and final values $h_i=0$ and $h_f=0.9$ which are the same as those of the sudden quench protocol.
Under the slow quench, the time-dependent Hamiltonian $H_k(t)$ is
\begin{equation}
H_k(t)=\begin{pmatrix}
\beta t-\cos k & -i \gamma\sin k \\
i \gamma\sin k & -\beta t+\cos k
\end{pmatrix}.
\end{equation}
Compared to Eq.~(\ref{lz}), $h_k(t)=\beta t-\cos k $ and $\Delta_k=-i\gamma \sin k$. At $k_0=0$ or $\pi$, the off-diagonal element cancels $\Delta_k=0$ so that the two energy levels cross with each other. Therefore, if the system is quenched across the quantum phase transition, i.e., $t$ is taken from $0$ to a final time $t_f>1/\beta$, the transition amplitude reaches $|p_{k_0,2}|=1$. For other $k$ points away from the critical point, $\Delta_k$ is finite and the transition probability decreases towards zero with decreasing $\beta$. Therefore, we can always find a special point $k_c$ such that condition (\ref{cond}) is satisfied, which constitutes a necessary condition for the DQPTs.
% At critical time $t_c$, we can plot the occupation probabilities $|c_{k, j}|^2$ for the two bands $j=1$ and $2$ in Fig.~\ref{xy1} (a2). It clearly shows that at the critical momenta $k_c$, the condition Eq.~(\ref{tc}) for DQPTs is satisfied.
% This absence is further reflected in Fig.~\ref{xy1} (b2), where the level occupation probabilities $p_k$ do not cross with each other and therefore, the condition Eq.~(\ref{tc}) for DQPTs is not satisfied.
% {\color{blue}Before we perform the numerical calculation, we can understand the fate of DQPTs under slow quench by making a qualitative discussion of Eq.~(\ref{le}).
% One can obtain the values of $p_{k, 1}(t)$ and $p_{k, 2}(t)$ by numerically solving the Schrodinger equation with time-dependent Hamiltonian:
% \begin{equation}
% \begin{split}
% &i\frac{d|\psi_k(t)\rangle}{dt}=h_k(t)|\psi_k(t)\rangle
% \end{split}
% \label{se}
% \end{equation}
% If one linearly quenches the parameter $h$ in Eq.~(\ref{xyha}), the time-dependent Hamiltonian can be written in the following form after a transformation
% \begin{equation}
% h_k(t)=\begin{pmatrix}
% h_k(t) & \Delta_k \\
% \Delta^*_k & -h_k(t)
% \end{pmatrix}.
% \end{equation}
% Here, $h_k(t)=\beta t-\cos k $ and $\Delta_k=\gamma \sin k$. This is the famous Landau-Zener-St${\rm \ddot{u}}$ckelberg-Majorana problem \cite{Landau1932theorie,Zener1932non,Stueckelberg1932theorie,Majorana1932atomi}, or simply called Landau-Zener problem. It provides an analytical solution for the transition probability from the ground state at initial time $t\rightarrow -\infty$ to the excited state at final time $t\rightarrow \infty$
% \begin{equation}
% P=e^{-\frac{\pi\Delta_k^2}{\beta}}.
% \end{equation}
% For general cases of time evolution, the above Landau-Zener formula can serve as a good approximation.
% When the system is quenched across the quantum phase transition, there exists always a $k_0=0$ or $\pi$ point that $\Delta_{k_0}=0$, so that the two energy levels simply cross with each other with no transition. At this $k_0$ point, if starting from the ground state, the system will finally go to the upper level, and the transition amplitude $|p_{k_0,2}|=1$. For other $k$ points away from the critical point $k_0$, $\Delta_k$ is finite, and the transition probability $|p_{k_0, 2}|$ decreases towards zero with decreasing quench rate $\beta$. Therefore, we can always find a special point $k_c$ such that $|p_{k_0, 1}|=|p_{k_0, 2}|=1/\sqrt{2}$, which constitutes a necessary condition for the DQPTs.
% One the other hand, if we do not quench system across the quantum phase transition, but within the same phase, the system is always gapped, and $\Delta_k$ is always nonzero. As long as the quench rate is sufficiently slow, i.e., very small $\beta$, the transition amplitude $p_{k, 2}$ would be always very small for each $k$, so that the condition for DQPTs can not be satisfied, and no DQPTs can occur. Therefore, slow quench dynamics can rule out the accidental DQPTs that appear within the same phase under sudden quench. }
We now begin to study DQPT under slow quench by numerically solving Eq.~(\ref{se}). As shown in Fig.~\ref{xy1}(a), the nonanalyticities in the rate function $r(t)$ appear that correspond to the occurrence of DQPTs under a sudden quench even when the pre-quench and post-quench Hamiltonians belong to the same equilibrium phase, which is consistent with the results from Ref.~\cite{Vajna2014dis}. In stark contrast, under the slow quench protocol with the same initial and final value of control parameters, DQPTs disappear that are manifested as the absence of the nonanalyticity, as shown in Fig.~\ref{xy1} (b). Here, the quench rate is $\beta=0.05$. These results indicate that the DQPTs observed in the corresponding sudden quench are accidental in nature and can be filtered out by adopting a sufficiently slow quench protocol.
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{xy.pdf}
\caption{DQPTs in the XY chain under sudden and slow quench protocols when the system is quenched within the same ferromagnetic phase and across the phase transition from ferromagnetic phase to paramagnetic phase, respectively. (a) and (b) The rate function $r(t)$ of the Loschmidt amplitude under sudden and slow quench protocol within the same phase. (a) Parameter $h$ is suddenly changed from $h_i=0$ to $h_f=0.9$ at time $t=0$ while $\gamma=0.3$. (b) The quench parameter $h$ is slowly changed as $h(t)=h_i+\beta t$ with the quench rate $\beta=0.05$. (c) and (d) The rate function $r(t)$ of the Loschmidt amplitude under sudden and slow quench protocol across the equilibrium phase transition. (c) $h$ is suddenly changed from $h_i=0$ to $h_f=2$ at time $t=0$. (d) The quench parameter $h$ is slowly changed as $h(t)=h_i+\beta t$ with same quench rate $\beta=0.05$}.
\label{xy1}
\end{figure}
For comparison, we also investigate DQPT under quench protocols across the equilibrium quantum phase transitions. Specifically, we also consider two different quench protocols with fixed $\gamma$: (i) sudden quench protocol, where the field $h$ is abruptly switched from $0$ to $2$ at time $t=0$; (ii) slow linear quench protocol, with $h(t)=h_i+ \beta t$, where the initial and final values $h_i=0$ and $h_f=2$ which are the same as those of the sudden quench protocol. Here, we choose the same quench rate $\beta=0.05$. As shown in Fig.~\ref{xy1} (c) and (d), we reveal that DQPT is robust even under slow quench protocols, as shown in Fig.~\ref{xy1} (d). Therefore, we can establish a intrinsic connection between DQPT and equilibrium quantum phase transitions under the slow quench protocol, thereby enabling DQPT to serve as a dynamical probe for identifying equilibrium quantum phase transition.
We now clarify that, to filter out the accidental DQPTs, the slow quench rate does not have to be infinitely small to make the quench dynamics adiabatic. As shown in Fig.~\ref{condition}, we present three-dimensional plots of the quantities \(|p_{k,i}||c_{k,i}|\) and the occupation amplitudes \(|p_{k,i}|^2\) for quench rates \(\beta = 0.08\) (upper panels a1–a2) and \(\beta = 0.1\) (bottom panels b1–b2). In the vicinity of the threshold quench rate \(\beta_{\text{th}} = 0.09\), condition (\ref{cond}) for DQPTs ceases to be valid. This threshold \(\beta_{\text{th}} = 0.09\) characterizes a “sufficiently slow quench.” For quench rates above \(\beta_{\text{th}}\) (b1), condition (\ref{cond}) is satisfied, ensuring the occurrence of DQPTs. Conversely, below \(\beta_{\text{th}}\) (a1), condition (\ref{cond}) is violated, and no DQPTs appear. Meanwhile, we note that in (a2), the system does not enter the adiabatic regime, as the excitation amplitude \(|p_{k,2}|^2\) (blue region) remains finite.
\begin{figure*}[htbp]
\includegraphics[width=0.9\linewidth]{condition.pdf}
\centering
\caption{The condition (\ref{cond}) for quench rate $\beta=0.08$ (a1) and $\beta=0.1$ (b1) in the XY chain within the same phase. Occupation amplitudes $p_{k,1}$ and $p_{k,2}$ for quench rate $\beta=0.08$ (a2) and $\beta=0.1$ (b2). $t_e$ represents the ending time of slow quench protocols. After $t_e$, the subsequent temporal evolution will be governed by the final time-independent Hamiltonian. In (b1) and (b2), the red dotted lines represent the time evolution for critical momentum $k_c$.}
\label{condition}
\end{figure*}
Next, we further study the intrinsic connection between DQPT and equilibrium quantum phase transitions under the slow quench protocol and illustrate that the critical time $\tau_c$ of DQPTs exhibits a universal scaling law with respect to the quench rate $\beta$ \cite{zhang2025uni}. Remarkably, the exponent of this scaling law is governed by the critical exponent of the underlying equilibrium phase transitions, as shown in Fig.~\ref{xy2}. The first critical time $\tau_c$ exhibits a universal scaling with respect to the quench rate $\beta$, $\tau_c\sim \beta^{-\sigma}$ with $\sigma=1/2$. This result is consistent with the general argument \cite{zhang2025uni} that $\sigma=z\nu/(1+z\nu)$ with $z\nu=1$ in the XY chain model. Here, $z$ and $\nu$ are the critical exponents of the equilibrium quantum phase transition.
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{scaling_xy.pdf}
\caption{The scaling behavior of critical time of DQPT when the system is slowly quenched across the phase transition in the XY chain. (a) Rate function $r(t)$ calculated by the initial state $|\psi(t_i)\rangle$ with different quench rate. (b) The first critical time $\tau_c$ as a function of the quench rate $\beta$.}
\label{xy2}
\end{figure}
\subsection{Aubry-André model}
%Dynamical quantum phase transitions can not only appear in quantum spin chain models, but also in the quasiperiodic lattices.
We now investigate the quasiperiodic Aubry-André (AA) model, which is a toy model for the understanding of localization-delocalization transition. It is a chain of single-particle fermions with nearest-neighbor hopping $J$ subjected to on-site quasiperiodic potential $V_n$. The Hamiltonian is given by
\begin{equation}
H=-J\sum_{n=1}^{N}\left(c_n^\dagger c_{n+1}+c_{n+1}^\dagger c_{n}\right)+\sum_{n=1}^{N}V_nc_n^\dagger c_{n},
\end{equation}
where $c_n^\dagger$ and $c_n$ are the free fermionic creation and annihilation operators at lattice site $n$, respectively, and $N$ is the lattice size. For simplicity, all energy scales are measured in units of $J$, which hereafter is set to unity. The on-site energy $V_n$ of the AA model is a quasiperiodic potential given by
\begin{equation}
V_n=V\cos{(2\pi\alpha n+\phi)},
\end{equation}
where $V$ is the strength of the incommensurate potential, $\alpha=(\sqrt{5}-1)/2$ being an irrational number which is a deterministic quasiperiodic one, and $\phi$ is the phase parameter, which is set to zero without loss of any generality. The incommensurate potential can be viewed as a kind of quasi-random disorder, which drives the system undergoing a delocalization-localization transition at $V=2J$. When $V<2J$, all the eigenstates are extended, whereas all the eigenstates are localized, when $V>2J$.
It has been demonstrated in Ref.~\cite{Yang2017dynamical} that DQPTs can occur in the Aubry–André model when the system is suddenly quenched across the delocalization–localization phase boundary. Recently, Ref.~\cite{Ye2025disentangling} reported the emergence of energy-dependent DQPTs after sudden quenches that begin and end within the same extended phase. Here, the term ``energy-dependent" signifies that the manifestation of DQPTs is sensitive to the choice of the initial state.
Here, we make a comprehensive study of DQPT in the AA model under both sudden and slow quenches of the potential strength $V$. We first consider the quench dynamics within the same delocalized phase. For sudden quench protocol within the same phase, one can observe that DQPTs occur at some critical times, while for slow quench protocols, DQPTs disappear, as shown in Fig.~\ref{aah}.
For the sudden quench, see Fig.~\ref{aah}(a), the Loschmidt echo $\mathcal{L}(t)$ becomes zero at some times (touching with the red dashed line), signaling energy-dependent DQPTs. In contrast, the Loschmidt echo $\mathcal{L}(t)$ remains finite at all times and can not touch with the zero axis under slow quench protocol, indicating that such energy-dependent DQPTs are filtered out when the quench rate is sufficiently slow, see Fig.~\ref{aah}(b). Here, we choose the initial state to be the eigenstate corresponding to the middle eigenvalue of the Hamiltonian with system size $N=400$.
We then study DQPT when the system is driven across the the delocalization–localization phase boundary. Here the potential strength $V$ is either suddenly or slowly quenched from $V_i=0$ in the delocalized phase to $V_f=3$ in the localized regime. Fig.~\ref{aah}(c) displays the time evolution of the Loschmidt echo $\mathcal{L}(t)$ following a sudden quench. The recurrence of $\mathcal{L}(t)=0$ at multiple times signals the occurrence of DQPTs. However, some of these may be identified as accidental, as they are eliminated under slow quench protocol [Fig.~\ref{aah}(d)]. Only those DQPTs, intrinsically linked to the delocalization–localization transition, remain robust under slow quench protocols.
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{aah_1.pdf}
\caption{Loschmidt echo $\mathcal{L}(t)$ (blue lines) with sudden ((a) and (c)) and slow quench ((b) and (d)) in the Aubry-André model. (a) The commensurate potential $V$ is suddenly changed from $V_i=0$ to $V_f=1.8$ at time $t=0$. (b) Slow quench protocol: $V(t)=V_i+\beta t$, quench rate $\beta=0.1$. The pre-quench and post-quench Hamtonians are both are in the delocalization phase. (c) and (d) Quench protocol crosses the delocalization–localization phase boundary ($V_i=0 \to V_f=3$). Here, the red dashed lines represent $\mathcal{L}(t)=0$.}
\label{aah}
\end{figure}
\subsection{Trimer Su-Schrieffer-Heeger model}
%DQPTs can appear not only in the two band model but also in the multiband model \cite{Huang2016dynamical}.
In this subsection, we extend the discussion of DQPT from two band models to multiband models by considering a three-band model, trimer Su-Schrieffer-Heeger model (SSH3) \cite{Jin2017top,Martinez2019edge,Anastasiadis2022bulk}. SSH3 is an extended SSH model with a unit cell that consists of three sites. The hopping between the sites is controlled by the couplings $t_1$ and $t_2$, while the different unit cells are coupled with the intercell coupling $t_3$. The system is governed by the Hamiltonian
\begin{equation}
\begin{split}
H=\sum_{j=1}^{N} \left(t_1 c_{j,B}^\dagger c_{j,A}+t_2 c_{j,C}^\dagger c_{j,B}+t_3 c_{j+1,A}^\dagger c_{j,C}\right)+{\rm H.c.}
\end{split}
\label{ssh3}
\end{equation}
Here, $c_{j,\alpha}^\dagger$ ($c_{j,\alpha}$) denotes the creation (annihilation) operator at site $\alpha$ (which can be either A, B or C type) of the $j$th unit cell. Without loss of generality, we will henceforth assume that the hopping parameters are real and non-negative. Considering periodic boundary conditions (PBC) along the length of the chain and performing Fourier transform of creation and annihilation operators, $c_{j,\alpha}^\dagger=(1/\sqrt{N})\sum_k e^{-ikj}c_{k,\alpha}^\dagger$, we can write the Hamiltonian in momentum space as $H=\sum_k\psi_k^\dagger H_k\psi_k$ with $\psi_k^\dagger=(c_{k,a}^\dagger,c_{k,b}^\dagger,c_{k,c}^\dagger)$, where
\begin{equation}
H_k=\begin{pmatrix}
0 & t_1 & t_3 e^{-ik} \\
t_1 & 0 & t_2\\
t_3 e^{ik} & t_2 & 0
\end{pmatrix}.
\end{equation}
For PBC, the spectrum of the above Hamiltonian consists of three dispersive bands are shown in Fig.~\ref{ssh3-energy}(a-b). The three bands touch each other at $k=0$ and $\pi$, only when $t_1=t_2=t_3$, i.e., in the absence of trimerization. As the parameters are shifted from that condition ($t_1\neq t_2$), two band gaps appear in the band structure, as shown in Fig.~\ref{ssh3-energy}(b). Interestingly, these band gaps may host very peculiar edge states as those shown in Fig.~\ref{ssh3-energy}(d), where we have plotted the energy spectrum with open boundary conditions (OBC) of a finite trimer lattice ($N=50$).
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{ssh3.png}
\caption{The spectrum of the bulk Hamiltonian of SSH3 in the first Brillouin zone under periodic boundary conditions for $t_1=t_2$ (a) and $t_1\neq t_2$ (b). (c) and (d) The real space energy spectrum with respect to the intercell hopping coupling $t_3$ for $t_1=t_2$ and $t_1\neq t_2$. Here, $N$ = 50. The red and blue dotted lines represent the topological edge states. }
\label{ssh3-energy}
\end{figure}
For OBC, we show the energy spectrum of a finite trimer lattice with $N = 50$ unit cells in Fig.~\ref{ssh3-energy}(c) and (d), as a function of the intercell hopping amplitude $t_3$ for two different cases with $t_1=t_2$ and $t_1\neq t_2$, respectively. We observe a different number of in-gap edge states emerging as $t_3$ changes. In Fig.~\ref{ssh3-energy}(c), edge states (red dotted lines) can appear in the case of a mirror-symmetric SSH3 (i.e., when intracell coupling $t_1=t_2$) and they emerge at the point where the gap closes. However, edge states can also appear in the case of an SSH3 (red dotted lines and blue dotted lines) that does not possess mirror symmetry (when intracell coupling $t_1\neq t_2$) and their emergence is not related to a gap closing (Fig.~\ref{ssh3-energy}(d)). %In the mirror-symmetric case it is obvious that the appearance of edge states is accompanied by the gap closing where $t_1=t_2=t_3$ is unavoidable, while in a chain without mirror symmetry this gap closing can be avoided. However, in the latter case, e
In this case ($t_1\neq t_2$), the edge states appear after the passing of mirror-symmetric points, i.e., when $t_3=t_1$ and $t_3=t_2$. Note that two-edge states (blue dotted lines) are localized on the right boundary of the system when intercell hopping $t_3$ satisfies $t_1<t_3<t_2$, and two new edge states (red dotted lines) can appear at the left boundary of the system for when $t_1<t_2<t_3$. The bulk-edge correspondence that can predict the emergence of all these edge states has been made via the sublattice Zak phase, which also remains valid in the absence of mirror symmetry \cite{Anastasiadis2022bulk}.
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{ssh3.pdf}
\caption{Sudden and slow quench dynamics in the SSH3 model for mirror-symmetric case $(t_1=t_2)$ (a-b) and no mirror-symmetric case $(t_1\neq t_2)$ (c-d).
(a) The rate function $r(t)$ of the Loschmidt amplitude under sudden quench protocol. $t_3$ is suddenly changed from 0.5 to 2.5 at time $t=0$. (b) The rate function $r(t)$ of the Loschmidt amplitude under slow quench protocol. $t_3(t)=t_{3,i}+\beta t$ with quench rate $\beta=0.1$. (c) The sudden quench protocol is same as (a). (d) The slow quench protocol is also same as (b) with quench rate $\beta=0.1$.}
\label{ssh3-symmetry}
\end{figure}
% \begin{figure}[htbp]
% \includegraphics[width=1\linewidth]{ssh3-symmetry.pdf}
% \caption{Sudden and slow quench dynamics in the SSH3 model.
% (a1) The rate function $r(t)$ of the Loschmidt amplitude under sudden quench protocol. $t_3$ is suddenly changed from 0.5 to 2.5 at time $t=0$. (b1) The rate function $r(t)$ of the Loschmidt amplitude under slow quench protocol. $t_3(t)=t_{3,i}+\beta t$. (a2) and (b2) are the level occupation probabilities. Here, the intracell coupling $t_1=t_2=1$.}
% \label{ssh3-symmetry}
% \end{figure}
We now examine both sudden and slow quench protocols for the mirror-symmetric case ($t_1= t_2$). The parameter $t_3$ is quenched across the topological phase boundary, a process accompanied by the closure of the energy gap at the critical point $t_3=t_1=t_2$. Fig.~\ref{ssh3-symmetry} (a) depictes the rate function $r(t)$ for the sudden quench protocol, while (b) presentes the corresponding results for the slow quench protocol.
In this mirror-symmetric case, DQPT is protected by the symmetry of the Hamiltonian and the critical times of DQPTs present a distinct periodicity.
% The critical momentum $k_c$ is identified when the ground-state survival probability equals the transition probability to the second energy level. Meanwhile, the excitation probability to the third level remains small under sudden quench but vanishes entirely under slow quench.
We next investigate sudden and slow quench protocols in the absence of mirror symmetry ($t_1\neq t_2$). Although edge states may emerge in real space, their appearance is not associated with a bulk gap closing. This leads to markedly distinct dynamical behaviors between the two quench protocols. As shown in Figs.~\ref{ssh3-symmetry}(c), DQPTs emerge under sudden quench. In contrast, DQPTs are not observed under slow quench, as evidenced in Figs.~\ref{ssh3-symmetry}(d). We emphasize that the connection between slow-quench DQPT and equilibrium phase transitions fundamentally relies on the occurrence of a bulk gap closure; with its absence, such dynamical signatures do not faithfully reflect the underlying equilibrium criticality.
% Next, we consider sudden quench protocol and slow quench protocol when mirror symmetry of the Hamiltonian is broken ($t_1\neq t_2$). Although there can appear edge states in real space, their emergence is not related to a gap closing. This case leads to a completely different results for sudden quench and slow quench. In Fig.~\ref{ssh3-nosymmetry}(a1) and (a2), DQPTs can appear and the critical momentum $k_c$ exist under sudden quench. To the contrast, DQPTs and critical momentum $k_c$ disappear under slow quench in Fig.~\ref{ssh3-nosymmetry}(b1) and (b2).
% \begin{figure}[htbp]
% \includegraphics[width=1\linewidth]{ssh3-nosymmetry.pdf}
% \caption{Sudden and slow quench in the SSH3 model.
% (a1) The rate function $r(t)$ of the Loschmidt amplitude under sudden quench protocol. $t_3$ is suddenly changed from 0.5 to 2.5 at time $t=0$. (b1) The rate function $r(t)$ of the Loschmidt amplitude under slow quench protocol. $t_3(t)=t_{3,i}+\beta t$. (a2) and (b2) are the level occupation probabilities. Here, the intracell coupling $t_1=1$ and $t_2=2$.}
% \label{ssh3-nosymmetry}
% \end{figure}
Beyond the previously discussed advantages of slow quench protocols, the critical time $\tau_c$ of dynamical quantum phase transitions can be leveraged to extract the critical exponent of the corresponding equilibrium quantum phase transition \cite{zhang2025uni}. We pay our attention to the mirror-symmetric regime and use the ground state of the pre-quench Hamiltonian as the initial wavefunction. The rate function $r(t)$ with different quench rate $\beta$ is shown in Fig.~\ref{ssh3-scaling}(a). By extracting the first critical time
we observe a universal scaling relation:
\begin{equation}
\tau_c\sim\beta^{-1/2}+\tau_0.
\end{equation}
as illustrated in Fig.~\ref{ssh3-scaling} (b). Here, $\tau_0$ is an constant shift. This scaling behavior yields the critical exponent product $z\nu=1$, consistent with the universality class reported in \cite{zhang2025uni}. This observation further indicates that this connection between DQPT and equilibrium phase transitions is not only valid in two-band models but also in multiband models.
\begin{figure}[htbp]
\includegraphics[width=1\linewidth]{ssh3_scaling.pdf}
\caption{ Rate function $r(t)$ (a) and the scaling behavior of the critical time $\tau_c$ (b) of Loschmidt amplitude $\mathcal{L}(t)$ in the trimer Su-Schrieffer-Heeger model in the mirror-symmetric case ($t_1 = t_2$) under slow quench protocol. (a) Rate function $r(t)$ for different quench rates $\beta$. The vertical black dashed lines correspond to the first critical times $\tau_c$.}
\label{ssh3-scaling}
\end{figure}
\section{Conclusion}
\label{s4}
In summary, we investigate the fate of dynamical quantum phase transitions when the quench protocol is switched from sudden to slow quench limit. We have made a general argument that accidental DQPTs can be filtered out under the slow quench protocol. We have investigated three distinct models to illustrate our finding: the XY chain, the Aubry-André model, and the trimer Su-Schrieffer-Heeger model by performing a comprehensive numerical study of DQPT under both sudden and slow quench protocols. For the XY chain \cite{Vajna2014dis}, previous studies have reported that DQPT can occur under sudden quench when the pre-quench and post-quench Hamiltonian belong to the same ferromagnetic phase.
For the Aubry-André model, Ref.~\cite{Ye2025disentangling} reported the emergence of energy-dependent dynamical quantum phase transitions during sudden quenches that begin and end within the same extended phase. Under slow quench dynamics, we show that both above cases of DQPTs are accidental and can be filtered out.
For the trimer Su-Schrieffer-Heeger model, there exist two different kinds of topological phase transitions. One is accompanied by a bulk gap closure and the other one is not. We emphasize that the connection between slow-quench DQPT and equilibrium phase transitions fundamentally relies on the occurrence of a bulk gap closure; in its absence, such dynamical signatures do not faithfully reflect the underlying equilibrium criticality. Therefore, we have demonstrated that slow quenche protocol can be used to reveal the intrinsic connection between DQPT and equilibrium quantum phase transitions, thereby enabling DQPT to serve as a dynamical probe for identifying the equilibrium quantum phase transition and the corresponding critical exponents.
%We further study the scaling of the first critical times within the mirror-symmetric regime and obtain the critical exponent product $z\nu=1$, in agreement with the prediction in \cite{zhang2025uni}.
\section*{Acknowledgements}
This work was supported by the National Natural Science Foundation of China (No. 12275075), the National Key Research and Development Program of Ministry of Science and Technology (No. 2021YFA1200700), the China Postdoctoral Science Foundation (Grant No. 2021M690970) and the Fundamental Research Funds for the Central Universities from China. \\
\section*{DATAAVAILABILITY}
All the data that support the plots and other findings of this study are available from the corresponding author upon reasonable request.
%\bibliographystyle{apsrev4-2}
\bibliography{ref}
\vspace{0.5cm}
\end{document}