Relaxation toward an Ideal Chern Band through Coupling to a Markovian Bath

Extracted source text · arXiv 2511.11394

\documentclass[prb,twocolumn,superscriptaddress,nofootinbib]{revtex4-2}

% Math and physics
\usepackage{amsmath,amssymb,amsthm,amsfonts,mathtools}
\usepackage{physics}
\usepackage{bm,bbm,siunitx}
\usepackage[colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue]{hyperref}

\usepackage{appendix}

\newcommand{\ii}{\mathrm{i}}
\renewcommand{\dd}{\mathrm{d}}
\renewcommand{\tr}{\operatorname{tr}}
\newcommand{\Gr}{\mathrm{Gr}}
\newcommand{\omegaFS}{\omega_{\mathrm{FS}}}
\newcommand{\gFS}{g_{\mathrm{FS}}}
\newcommand{\BZ}{\mathrm{BZ}}
\renewcommand{\bf}[1]{\mathbf{#1}}

% Theorem environments
\newtheorem{theorem}{Theorem}[section]
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{corollary}[theorem]{Corollary}
\theoremstyle{definition}
\newtheorem{definition}[theorem]{Definition}
\theoremstyle{remark}
\newtheorem{remark}[theorem]{Remark}



\begin{document}

\title{Relaxation to an Ideal Chern Band through Coupling to a Markovian Bath}

\author{Bruno Mera}
\affiliation{Instituto de Telecomunica\c{c}\~oes and Departmento de Matem\'{a}tica, Instituto Superior T\'ecnico, Universidade de Lisboa, Avenida Rovisco Pais 1, 1049-001 Lisboa, Portugal}
\affiliation{Advanced Institute for Materials Research (WPI-AIMR), Tohoku University, Sendai 980-8577, Japan}

\author{Tomoki Ozawa}
\affiliation{Advanced Institute for Materials Research (WPI-AIMR), Tohoku University, Sendai 980-8577, Japan}

\begin{abstract}
We propose a microscopic, weak-coupling mechanism by which generic Chern bands relax toward ideal bands. We consider coupling interacting electrons to a Caldeira-Leggett like Ohmic bosonic bath. Using the Born-Markov approximation, Slater determinant states of a Chern band under Hartree-Fock approximation evolve toward Slater determinant states corresponding to an ideal Chern band. We validate our proposal by performing numerical simulation of a massive Dirac model, showing that the Berry curvature and quantum metric indeed co-evolve to saturate the trace condition. Our proposal provides a concrete dissipative route to realize ideal Chern bands, a fundamental building block for the stabilization of fractional Chern insulators.
\end{abstract}

\maketitle
%
\paragraph*{Introduction. ---} The notion of \emph{ideal Chern bands}~\cite{rahul:2014,claassen:2015,ozawa:mera:2021,mera:ozawa:2021:relations,mera:ozawa:2021:engineering,wang:cano:millis:liu:yang:2021,ledwith:vishwanath:parker:2023,estienne:regnault:valentin:2023,mera:ozawa:2023}---flat bands whose Berry curvature and quantum metric saturate a quantum-geometric inequality---has emerged as a central paradigm in the theory of fractional Chern insulators (FCIs)~\cite{bernevig:2011,titus:2011,sheng:2011,sun:2011,wen:2011} and topological flat-band materials~\cite{wang:cano:millis:liu:yang:2021,ledwith:tarnopolsky:khalaf:vishwanath:2020, wang:liu:2022,ledwith:vishwanath:parker:2023,ledwith:vishwanath:khalaf:2022} due to their remarkable similarity to the lowest Landau level. More recently, a generalization of this notion---\emph{generalized Landau levels} or \emph{harmonic bands}~\cite{liu:mera:fujimoto:ozawa:wang:2025,fujimoto:parker:dong:khalaf:vishwanath:ledwith:2025,paiva:wang:ozawa:mera:2025,onishi:avdoshkin:fu:2025}---was formulated, and their suitability for stabilizing non-Abelian FCI phases was numerically confirmed.

Although the importance of the ideal Chern bands and generalized Landau levels have been increasingly appreciated, how one can obtain such bands, either theoretically or experimentally, remains to be nontrivial. Theoretically, one needs special models which fulfill the ideal conditions. Experimentally, there is no known simple method to obtain ideal Chern bands or generalized Landau levels.
In this paper, we propose a generic mechanism under which ideal Chern bands or generalized Landau levels can be obtained. We show that, by coupling weakly interacting electrons to Caldeira-Leggett type Markovian bath, mean-field (Slater determinant) states evolve into states which minimize the Dirichlet energy functional determined from the quantum geometry. When the Dirichlet energy is dominant over the kinetic energy, the resulting state one obtains in a long-time limit is nothing but the ideal Chern band. Numerical simulations for the massive Dirac model confirm this evolution, with the Dirichlet energy approaching the topological lower bound $\pi|\mathcal{C}|$, where $\mathcal{C}$ is the Chern number of the occupied states, as the Berry curvature and quantum metric co-evolve toward the ideal-band condition. In this picture, ideal Chern bands correspond to the global minima of the Dirichlet energy, while generalized Landau levels are stationary points.
We note that recently, in Refs.~\cite{yu:lian:ryu:2025,fukui:2025}, gradient flow with respect to the Dirichlet energy of the Bloch projector was proposed to achieve the ideal-band conditions. Compared to these recent works, our work provides a concrete dissipative mechanism through which ideal bands can be achieved. 

%
\paragraph*{Model and projection to Slater dynamics. ---}  We first consider, for demonstration, a single-site model and later extend it to a lattice model. The model consists of a system of interacting electrons coupled to a Caldeira-Legget like bosonic bath,
$\widehat H=\widehat H_S+\widehat H_B+\widehat H_I$,
where
\begin{align}
\widehat H_S&=\sum_{i,j}t_{ij}\,c_i^\dagger c_j
+\tfrac12\sum_{i,j,k,l}v_{ijkl}\,c_i^\dagger c_j^\dagger c_l c_k, \nonumber\\
\widehat H_B&=\sum_\alpha \Omega_\alpha\, b_\alpha^\dagger b_\alpha,\quad
\widehat H_I=\lambda\sum_{i,j,\alpha}
\left(b_\alpha^\dagger \;c^\dagger_{i}\left(V_\alpha\right)_{ij}c_j +\text{H.c.}\right).
\label{eq: microscopic model}
\end{align}
Above, $c_i^\dagger$ and $c_i$ create and annihilate an electron in orbital \(i\),
while \(b_\alpha^\dagger\) and \(b_\alpha\) create and annihilate a bosonic bath mode \(\alpha\) of frequency \(\Omega_\alpha\).
The $t_{ij}$ encode single-particle hoppings, and
$v_{ijkl}$ denote the two-body interaction matrix elements.
The system–bath coupling is parameterized by a real constant $\lambda$ and by the matrices $V_\alpha$,
whose elements $(V_\alpha)_{ij}$ specify the interaction vertex between orbital $i$ and $j$
mediated by bath mode $\alpha$. 

Assuming (i) an initial Slater-determinant state and Gaussian fermionic density matrices at all times---so that Wick’s theorem holds---(ii) a weak-coupling Born factorization
$\widehat\rho(t)\simeq \widehat\rho_S(t)\otimes\widehat\rho_B(t)$,
and (iii) a Markov approximation,
the system’s dynamics remains confined to the manifold of Slater determinants,
which is parametrized by the Grassmannian manifold of rank equal to the particle number.
We will denote this manifold simply as $\Gr$. Each point of $\Gr$ corresponds to a Slater determinant state uniquely described by the single-particle density matrix
$P=[P_{ij}]=[\langle c_j^\dagger c_i\rangle]$,
an orthogonal projector of rank equal to the particle number.



In the absence of system--bath coupling ($\widehat H_I=0$),
the dynamics reduces to the time-dependent Hartree-Fock equation,
\begin{equation}
\ii\frac{\dd P}{\dd t }=[T+V(P),P],\qquad
V_{ik}(P)=\sum_{jl}\!\big(v_{ijkl}-v_{ijlk}\big)P_{lj},
\label{eq:tdHF}
\end{equation}
where $T=[t_{ij}]$ is the hopping matrix and $V(P)=[V_{ij}(P)]$,
with $V_{ij}(P)=\sum_{kl}\!\big(v_{ikjl}-v_{iklj}\big)P_{lk}$ the mean-field interaction potential.
This evolution can be written geometrically as a Hamiltonian flow on the Grassmannian,
\begin{equation}
\frac{\dd P}{\dd t}=\ii\,[P,\frac{\partial H_S}{\partial P}],
\end{equation}
where the Hamiltonian function $H_S=H_S(P)=\tr\left[\left(T+\frac{1}{2}V(P)\right)P\right]$ is the expectation value of the system Hamiltonian $\widehat H_S$
evaluated on a Slater determinant, and $\frac{\partial H_S}{\partial P}=[\frac{\partial H_S}{\partial P_{ij}}]$ denotes the derivative with respect to $P$,
which coincides with the single-particle Hartree-Fock Hamiltonian $\frac{\partial H_S}{\partial P}=T+V(P)$.
Here, ``Hamiltonian flow'' means Hamilton’s equations generated by $H_S$
with respect to the Fubini-Study symplectic form on the Grassmannian,
$\omegaFS=-\ii\,\mathrm{tr}\!\left(P\,\dd P\wedge\dd P\right)$.
%
\paragraph*{Ohmic bath leads to metriplectic flow. ---}
Solving the bath equations formally using retarded Green’s functions, inserting the result into the system’s equation of motion, and performing a standard Markov limit with an Ohmic spectral weight
$J(\omega)=\eta\,\omega e^{-\omega/\omega_c}$
yields (see the appendix for a detailed derivation)
\begin{equation}
\frac{\dd P}{\dd t}
= \ii\,\Bigl[P,\frac{\partial H_S}{\partial P}\Bigr]
- \gamma\,\Bigl[P,\Bigl[P,\frac{\partial H_S}{\partial P}\Bigr]\Bigr],
\qquad
\gamma = \lambda^2 \eta > 0.
\label{eq: metriplectic}
\end{equation}
Here $\eta$ sets the overall dissipation rate and $\omega_c$ is a high-frequency cutoff,
assumed to be large so that memory effects are negligible and the dynamics becomes local in time.
Equation~\eqref{eq: metriplectic} defines a \emph{Kähler metriplectic} dynamics on the Grassmannian:
the first term generates Hamiltonian motion with respect to the Fubini-Study symplectic form,
while the second term corresponds to gradient flow in the compatible Fubini-Study metric $\gFS=\mathrm{tr}\!\left(\dd P\odot\dd P\right)$, where $\odot$ denotes the symmetrized tensor product. 

Eq.~\eqref{eq: metriplectic} defines a metriplectic flow in the sense introduced by Morrison in Ref.~\cite{morrison:1986}, combining Hamiltonian with gradient flow (see the appendix for more details). Here the metriplectic structure is very natural because the Grassmannian is K\"ahler, and hence we have both symplectic and Riemannian structures, compatible through the complex structure.
%
\paragraph*{Monotonically decreasing energy along the flow. ---}
Consider the system's Hamiltonian function $H_S(P)$.
Using the identity $\Pi_P(X)=[P,[P,X]]=(1-P)XP+PX(1-P)$,
which gives the orthogonal projection of an operator $X$ onto the tangent space
$T_P\mathrm{Gr}$ with respect to the Hilbert-Schmidt inner product,
one finds
\begin{equation}
\frac{\dd H_S}{\dd t}
=\mathrm{tr}\!\left[\frac{\partial H_S}{\partial P}
\frac{\dd P}{\dd t} \right]
=-\gamma\,\mathrm{tr}\!\left[\Pi_P\Bigl(\frac{\partial H_S}{\partial P}\Bigr)^2\right]\le 0,
\label{eq: Lyapunov}
\end{equation}
so $H_S$ is a Lyapunov function in the broader sense of LaSalle's invariance principle (Theorem~3.15 of Ref.~\cite{hokayem:gallestey:2020})---nonincreasing along trajectories---and, as $t\to +\infty$, the flow relaxes to the Hartree-Fock critical set
defined by $\Pi_P(\frac{\partial H_S}{\partial P})=0$ ($\iff[P,\frac{\partial H_S}{\partial P}]=0$),
that is, the critical points of $H_S(P)$ on $\Gr$,
with Hamiltonian motion quenched at late times.
%
\paragraph*{Generalization to lattice models and emergence of the Dirichlet energy. ---}
We can generalize the microscopic model of Eq.~\eqref{eq: microscopic model}
to a translation-invariant two-dimensional lattice with $N$ internal degrees of freedom per site
(e.g., orbitals, spin, or sublattice).
Now the annihilation and creation operators in Eq.~\eqref{eq: microscopic model} should carry positional indices $\mathbf{r}$ as well as indices for internal degrees of freedom. After performing Fourier transformation to move to momentum space, we denote the Bloch Hamiltonian describing the hopping to be
$T(\mathbf{k})=[t_{ij}(\mathbf{k})]$
and short-range density-density interactions to be
controlled by a potential $v(\mathbf{q})$. The system Hamiltonian in momentum space then reads
\begin{equation}
\widehat{H}_S=\lambda_T\sum_{\mathbf{k}}\sum_{i,j=1}^{N}
c_{\mathbf{k},i}^\dagger t_{ij}(\mathbf{k})c_{\mathbf{k},j}
+\frac{1}{V}\sum_{\mathbf{q}} v(\mathbf{q})\,{:n_{-\mathbf{q}}n_{\mathbf{q}}:}.
\end{equation}
Here $n_{\mathbf{q}}$ denotes the Fourier transform of the density operator,
and $\lambda_T$ sets the overall hopping strength. Assuming the system occupies a Slater determinant obtained by filling a Chern band,
described by an $N\times N$ orthogonal rank-$r$ projector
$P(\mathbf{k})=[P_{ij}(\mathbf{k})]$, and following the same Born-Markov approximation used for the model in Eq.~\eqref{eq: microscopic model} together with an Ohmic-bath assumption homogeneous in momentum (see the appendix), the evolution of the band is governed by a generalization of Eq.~\eqref{eq: metriplectic},
\begin{equation}
\frac{\partial P(\mathbf{k})}{\partial t}
=\ii\!\left[P(\mathbf{k}),\frac{\delta H_S}{\delta P(\mathbf{k})}\right]
-\gamma\!\left[P(\mathbf{k}),
\left[P(\mathbf{k}),\frac{\delta H_S}{\delta P(\mathbf{k})}\right]\right],
\label{eq: metriplectic-flow-Chern}
\end{equation}
where $\frac{\delta H_S}{\delta P(\mathbf{k})}
=[\frac{\delta H_S}{\delta P_{ij}(\mathbf{k})}]$ denotes the functional derivative of the Hartree-Fock energy density functional
\begin{align}
H_S
&=\lambda_T\!\int_{\BZ^2}\!\!\frac{\dd^2\mathbf{k}}{(2\pi)^2}
\,\mathrm{tr}\!\big[T(\mathbf{k})P(\mathbf{k})\big]
\nonumber\\[-2pt]
&\quad
-\!\!\int_{\BZ^2}\!\!\int_{\BZ^2}\!\!
\frac{\dd^2\mathbf{k}\,\dd^2\mathbf{q}}{(2\pi)^4}\,
v(\mathbf{q})\,\mathrm{tr}\!\big[P(\mathbf{k})P(\mathbf{k}-\mathbf{q})\big].
\end{align}
(The appearance of functional derivatives is because we have a continuum of components $P_{ij}(\mathbf{k})$ parametrized by $\mathbf{k}$.)
For a finite lattice with (twisted or not) periodic boundary conditions,
this functional derivative naturally reduces to an ordinary derivative,
since only a finite set of momenta appears in the Fourier decomposition of $P$.

In the limit of small transferred momentum (small~$\mathbf{q}$) for the interaction term,
together with isotropy under a flat—i.e., constant in $\mathbf{k}$—metric $h=h_{ij}\, \mathrm{d}k^i \mathrm{d}k^j$
(interpretable as an inverse mass tensor, and not to be confused with the quantum metric $g$),
we obtain the geometric contribution
\begin{align}
E_{\mathrm D}(P)
=\frac{1}{2}\!\int_{\BZ^2}\!\!\dd^2\mathbf{k}\,\sqrt{\det h}\,h^{ij}\,
\mathrm{tr}\!\!\left(
\frac{\partial P}{\partial k^i}\frac{\partial P}{\partial k^j}
\right),
\end{align}
that is, the Dirichlet energy of the map
$P:\BZ^2\!\to\!\Gr$
between the Riemannian manifolds $(\BZ^2,h)$ and $(\Gr,g_{\mathrm{FS}})$.
The functional above depends only on the conformal class of $h$; since $h$ is flat, this means it is defined only up to an overall scale.
Up to scale, $h$ can be naturally identified with the Brillouin-zone–integrated quantum metric,
\begin{align}
h_{ij}\propto \int_{\BZ^2}\!\frac{\dd^2\mathbf{k}}{(2\pi)^2}\, g_{ij}(\mathbf{k}),
\end{align}
because $g_{ij}(\mathbf{k})$ inherits the lattice symmetries—present in both the hopping matrix $T(\mathbf{k})$ and the interaction potential $v(\mathbf{q})$—
(for example, $C_4$ symmetry in a square lattice implies $h_{ij}\propto \delta_{ij}$).
See the appendix for a more detailed discussion of this point.


In this limit,
\begin{equation}
H_S(P)\simeq
\lambda_T\!\int_{\BZ^2}\!\!\frac{\dd^2\mathbf{k}}{(2\pi)^2}
\mathrm{tr}[T(\mathbf{k})P(\mathbf{k})]
+\lambda_{\mathrm D}\,E_{\mathrm D}[P],
\end{equation}
and
\(
\frac{\delta H_S}{\delta P(\mathbf{k})}
=\lambda_T T(\mathbf{k})
-\lambda_{\mathrm D}\Delta_h P(\mathbf{k}),
\)
where $\Delta_h P=h^{ij}\frac{\partial^2 P}{\partial k^i\partial k^j}$
is the anisotropic Laplacian associated with $h$. The constant $\lambda_{\mathrm{D}}$ (derived in the appendix) is given by $\lambda_{\mathrm{D}} = \int_{0}^{Q}\! \dd q\, v(q)\, q^{3}$, where $Q$ is the momentum cutoff.

%
\paragraph*{Relaxation toward ideal bands. ---}
The Dirichlet energy functional has extrema given by \emph{generalized Landau levels} (nonzero Chern number harmonic maps),
which satisfy $Q(\Delta_h P)P=0$, equivalently $[P,\Delta_h P]=0$.
\emph{Ideal bands} are precisely the minima of $E_{\mathrm D}$;
they saturate the topological bound $E_{\mathrm D}\ge \pi|\mathcal{C}|$,
the global version of the quantum-geometric trace inequality
relating the Berry curvature and quantum metric.
In the limit $\lambda_{\mathrm D}\gg\lambda_T$,
the Hartree-Fock energy $H_S$ is dominated by the Dirichlet term.
In this regime, $E_{\mathrm D}$ decreases monotonically
under the flow of Eq.~\eqref{eq: metriplectic-flow-Chern},
while the Chern number $\mathcal{C}$ is conserved.
The Laplacian term $\gamma [P,[P,\Delta_h P]]$ ensures the smoothness of $P(\mathbf{k})$,
as it effectively yields a convolution with a smoothing kernel—
it generates the harmonic-map heat flow.
By LaSalle’s invariance principle, with $H_S\simeq\lambda_{\mathrm D}E_{\mathrm D}$,
the long-time solutions are therefore \emph{ideal} bands
within the fixed-topology sector labeled by $\mathcal{C}$.

%
\paragraph*{Numerical demonstration: massive Dirac model. ---}
We integrate Eq.~\eqref{eq: metriplectic-flow-Chern} for a two-band Dirac model (see, for instance, Ref.~\cite{mera:goldman:zhang:2022} for more details on the quantum geometry of this model),
$T(\mathbf{k})=\vec{d}(\bf{k})\cdot\vec{\sigma}$ with 
$\vec{d}(\bf{k})=(\sin k_x,\sin k_y,M-\cos k_x-\cos k_y)$, $M=-0.5$ so the lower band has $\mathcal{C}=1$. Writing $P(t,\bf{k})=(I+\vec{n}(t,\bf{k})\cdot\vec{\sigma})/2$, the flow becomes a Landau-Lifshitz-Gilbert equation
\begin{equation}
\frac{\partial\vec{n}}{\partial t}
=-\,\vec{n}\times\!\vec{h}
+\gamma\,\vec{n}\times\!\left(\vec{n}\times\!\vec{h}\right), \quad \vec{h}=\lambda_T\vec{d}-\frac{\lambda_{\mathrm{D}}}{4}\nabla^2\vec{n},
\label{eq: LLG equation}
\end{equation}
with initial condition $\vec{n}(t=0,\bf{k})=-\vec{d}(\bf{k})/|\vec{d}(\bf{k})|$.

Using $(\gamma,\lambda_{\mathrm{D}},\lambda_T)=(1.5,5.0,0.05)$, we show in Fig.~\ref{fig:energy} the energy as a function of time. In Fig.~\ref{fig:energy}(a) we confirm that the Dirichlet energy $E_{\mathrm{D}}(t)$ decreases monotonically to $\pi \mathcal{C}=\pi$, which is the value one should achieve for the ideal Chern bands.
In Fig.~\ref{fig:energy}(b), we compare the Dirichlet energy $E_{\mathrm{D}}(t)$ and the hopping energy $E_{\mathrm{T}}(t)$; we are in a regime where the Dirichlet energy is dominant over the hopping energy, which is the condition we required in order to have the flow evolve toward the ideal Chern bands.

\begin{figure}[h!]
  \includegraphics[width=\columnwidth]{Dirichlet_energy.pdf}
  \includegraphics[width=\columnwidth]{energies.pdf}
  \caption{
    Time evolution of the Dirichlet and hopping energy components during the time evolution according to Eq.~\eqref{eq: LLG equation}.
    The Dirichlet energy monotonically decreases toward the topological lower bound
    $E_{\mathrm D}\ge\pi|\mathcal{C}|$, while the hopping energy $\lambda_T E_T$, with $E_T=\int_{\BZ^2}\frac{\dd ^2\bf{k}}{(2\pi)^2}\vec{d}(\bf{k})\cdot \vec{n}(t,\bf{k})$, remains nearly constant equal to that of the lowest band. The lower and upper band energies are also plotted for reference.
  }
  \label{fig:energy}
\end{figure}

In Fig.~\ref{fig:saturation1}, we show how the trace inequality $\mathrm{tr}(g) \ge |F_{12}|$ saturates as the time evolves. We also show in Fig.~\ref{fig:saturation2} the profile of the trace of the quantum metric $\mathrm{tr}(g)$ and the Berry curvature $F_{12}$ at the initial and the final times. Note the drastic change of the profiles, showing that the ideal Chern band can be obtained with our method starting from a Chern band which is quite far from the ideal condition.

\begin{figure}[t]
  \centering
  \includegraphics[width=0.49\columnwidth]{saturation_initial.pdf}%
  \hfill
  \includegraphics[width=0.49\columnwidth]{saturation_final.pdf}
  
  \caption{\centering Deviation from saturation of the trace inequality at initial ($t=0$) and final ($t=15$) times, showing that the trace inequality becomes nearly saturated as time evolves. Note the difference in the scales between the left and the right figures.}
  \label{fig:saturation1}
\end{figure}

\begin{figure}[t]
  \centering
    \includegraphics[width=0.49\columnwidth]{metric_vs_curvature_initial.pdf}
  \includegraphics[width=0.49\columnwidth]{metric_vs_curvature_final.pdf}
  
  \caption{\centering Trace of the quantum metric and Berry curvature profiles at the initial and final times. 
  }
  \label{fig:saturation2}
\end{figure}

%
\paragraph*{Discussion and outlook. ---}
The microscopic theory developed in this work, where electrons with dominant short-range repulsive density-density interactions weakly couple to a bosonic Ohmic bath, leads to the Kähler metriplectic flow of Eq.~\eqref{eq: metriplectic-flow-Chern}.
Dominated by the harmonic-map heat flow, this provides a microscopic mechanism under which how generic Chern bands relax toward ideal bands in the long-time limit.

We have numerically shown how the ideal Chern band, which is the minimum of the Dirichlet energy functional, can be obtained with this method. Generalized Landau levels, on the other hand, correspond to extrema of the energy functional. We expect that, once the flow reaches an extremum, the system remains to be in the generalized Landau level for a certain period of time, before further relaxing toward the ideal Chern band. Further exploration of precise engineering of generalized Landau levels using our method will be discussed in future works. We have assumed that, in the Hartree-Fock energy, the lowest order in the small-$\mathbf{q}$ expansion is the dominant term; it would be of practical interest to consider the full density-density interaction and/or alternative forms of system-bath coupling and to explore how the results can be affected.

Once the ideal Chern bands and generalized Landau levels are obtained, with further addition of appropriate short-range quantum inter-particle interactions, the ground state of the system should exhibit fractional quantum Hall effects. Our proposal thus opens an avenue toward obtaining fractional Chern insulator phases starting from generic Chern bands.

%

\emph{Acknowledgments---}B.~M. acknowledges support from the Security and Quantum Information Group (SQIG) in Instituto de Telecomunica\c{c}\~{o}es, Lisbon. This work is funded by FCT/MECI through national funds and when applicable co-funded EU funds under UID/50008: Instituto de Telecomunicações (IT). B.~M. further acknowledges the Scientific Employment Stimulus --- Individual Call (CEEC Individual) --- 2022.05522.CEECIND/CP1716/CT0001, with DOI: \href{https://doi.org/10.54499/2022.05522.CEECIND/CP1716/CT0001}{10.54499/2022.05522.CEECIND/CP1716/CT0001}. T.~O. acknowledges suport from JSPS KAKENHI Grant Number JP24K00548, JST PRESTO Grant No. JPMJPR2353.

%
\bibliography{bib.bib}
%
\begin{widetext}

\appendix

%
\section{Mean-field theory and harmonic bands} 
\label{sec: Mean-field theory and harmonic bands}
%
\subsection{Hartree-Fock mean-field theory} 
\label{subsec: Hartree-Fock mean-field theory}
%
We consider the following fermionic Hamiltonian
\begin{align}
\widehat{H}=\sum_{i,j}t_{ij}c_i^\dagger c_j +\frac{1}{2}\sum_{i,j,k,l}v_{ij,kl}c_i^\dagger c_j^\dagger c_l c_k,    
\end{align}
and we wish to find a Slater determinant state $\ket{\psi}$ minimizing the energy
\begin{align}
H=\bra{\psi}\widehat{H}\ket{\psi}.
\end{align}
Above $c_i^\dagger, c_i$ are, respectively, the creation and annihilation operator for an orbital $i=1,\dots,N$. The constants $t_{ij}$ are hoppings describing the kinetic term and $v_{ij,kl}$ the two-body interaction couplings which satisfy the obvious symmetries $v_{ij,kl}=-v_{ji,kl}=-v_{ij,lk}=v_{ji,lk}$.

The Fock space for fermions we work with is nothing but the vector space of the exterior algebra $\Lambda^*(\mathbb{C}^N)$ equipped with the induced inner product coming from that on $\mathbb{C}^N$, exterior product by a vector $v=\sum_{i=1}^{N}v_i e_i\in \mathbb{C}^N$, or in Dirac notation $\ket{v}=\sum_{i=1}^{N}\bra{i}v\rangle  \ket{i}\in \mathbb{C}^N$, is identified with $c_{v}^\dagger=\sum_{i=1}^N\bra{i}v\rangle c_i^\dagger$, and $1\in\mathbb{C}=\Lambda^{0}(\mathbb{C}^N)$ is identified with the vacuum state $\ket{0}$. The grading $\Lambda^*(\mathbb{C}^N)=\bigoplus_{i=0}^{N} \Lambda^i(\mathbb{C}^N)$ is nothing but a decomposition into particle number sectors, i.e., eigenvalues of  $\sum_{i=1}^{N}c_i^\dagger c_i$. A slater determinant state is determined by choosing a collection of linearly independent single-particle orbitals, $\ket{v_1},\dots, \ket{v_r}$---which without loss of generality we may assume are orthogonal---and creating them on top of the vacuum state
\begin{align}
\ket{\psi}=c_{v_1}^\dagger\cdots c^\dagger_{v_{r}}\ket{0}.
\end{align}
In exterior algebra language this a homogeneous vector $v_1\wedge \dots \wedge v_r\in \Lambda^r(\mathbb{C}^N)$. The associated quantum state---the one dimensional line through $v_1\wedge \dots \wedge v_r$---is exactly the image of the dimension $r$ subspace $E=\mathrm{span}\left\{v_1,\dots,v_r\right\}\subset \mathbb{C}^N$ under the Pl\"ucker embedding $\iota :\Gr_{r}(\mathbb{C}^N)\hookrightarrow\mathbb{P}\Lambda^r(\mathbb{C}^N)$, where $\Gr_{r}(\mathbb{C}^N)$ denotes the Grassmanian of $r$-dimensional linear subspaces of $\mathbb{C}^N$ and $\mathbb{P}\Lambda^r(\mathbb{C}^N)$ denotes the projectivization of the vector space $\Lambda^r(\mathbb{C}^N)$. The energy $H=\bra{\psi}\widehat{H}\ket{\psi}$ restricted to those states that are Slater determinants of $r$ orbital then becomes a function $h\in C^{\infty }(\Gr_{r}(\mathbb{C}^N))$. Points in the Grassmannian $\mathrm{Gr}_{r}(\mathbb{C}^N)$ are $r$-dimensional subspaces of $\mathbb{C}^N$ and can be described in terms of slater determinants of $r$-orbitals up to a scale. Alternatively, can also be equivalently be described by rank $r$ orthogonal projection operators in $\mathbb{C}^N$, namely for each subspace $V\subset  \mathbb{C}^N$ we can uniquely associated an orthogonal projection $P_E:\mathbb{C}^N\to E\subset \mathbb{C}^N$. This projection matrix $P_E=[(P_{E})_{ij}]_{1\leq i,j\leq N}$ can be read off from $\ket{\psi}$ as
\begin{align}
(P_E)_{ij}=\langle \psi_E| c^\dagger_{j}c_i |\psi_E\rangle,
\end{align}
where we added the label $E$ on $\ket{\psi_E}$ to denote that the state comes from the image of $E$ under the Pl\"ucker embedding,
i.e., $P_E$ is nothing but the single-particle density matrix of the associated Slater determinant state.
We can use Wick's theorem to evaluate $h$, in particular,
\begin{align}
\bra{\psi_E}c_{i}^\dagger c^\dagger_{j}c_l c_k\ket{\psi_E}=\bra{\psi_E}c_{i}^\dagger c_k \ket{\psi_E}\bra{\psi_E}c^\dagger_{j}c_l \ket{\psi_E}-\bra{\psi_E}c_{i}^\dagger c_l \ket{\psi_E}\bra{\psi_E}c^\dagger_{j}c_k \ket{\psi_E},
\end{align}
to obtain
\begin{align}
H(E)=\bra{\psi_E}\widehat{H}\ket{\psi_E}=\sum_{i,j}t_{ij}(P_{E})_{ji} + \frac{1}{2}\sum_{i,j,k,l}v_{ij,kl}\left[(P_{E})_{ki}(P_{E})_{lj}-(P_{E})_{li}(P_{E})_{kj}\right]
\end{align}
With the identification $E\leftrightarrow P_E$, we will right $H(E)=:H(P_E)$, and we will drop the $E$ in $P_E$, so
\begin{align}
H(P) &=\sum_{i,j}t_{ij}P_{ji} + \frac{1}{2}\sum_{i,j,k,l}v_{ij,kl}\left[P_{ki}P_{lj}-P_{li}P_{kj}\right]\nonumber\\
&=\sum_{i,j}t_{ij}P_{ji} + \frac{1}{2}\sum_{jl}\left(\sum_{k,i}(v_{ij,kl} -v_{ij,lk})P_{ki}\right)P_{lj}\nonumber \\
&=\tr\left[\left(T+\frac{1}{2}V(P)\right)P\right],
\end{align}
where $T=[t_{ij}]_{1\leq i,j\leq N}$ and $V(P)=[\sum_{k,i}(v_{ij,kl} -v_{ij,lk})P_{ki}]_{1\leq j,l\leq N}$. Observe that $V$ contains two terms which differ by a permutation of the last indices---these correspond to the \emph{direct, or Hartree,} term and the \emph{exchange, or Fock,} term. So now we can take the differential of $h$. To do so, we must take into account that $P^2=P$, which implies 
\begin{align}
P\dd PP=0 \quad \text{and} \quad Q\dd PQ=0,
\end{align}
with $Q=I-P$ the orthogonal complement projector, so that
\begin{align}
\dd P=Q\dd PP+P\dd PQ.
\label{eq: differential of P}
\end{align}
One important point is the differential of the second term which simplifies as
\begin{align}
&\frac{1}{2}\sum_{jl}\left(\sum_{k,i}(v_{ij,kl} -v_{ij,lk})\dd P_{ki}\right)P_{lj} +  \frac{1}{2}\sum_{jl}\left(\sum_{k,i}(v_{ij,kl} -v_{ij,lk})P_{ki}\right)\dd P_{lj}\nonumber \\
&=\sum_{jl}\left(\sum_{k,i}(v_{ij,kl} -v_{ij,lk})P_{ki}\right)\dd P_{lj},
\end{align}
due to the symmetry $v_{ij,kl}=v_{ji,lk}$. Thus, we get, using $P^2=P$, $Q^2=Q$, Eq.~\eqref{eq: differential of P} and the cyclic property of the trace 
\begin{align}
\dd H=\tr\left[(T+V(P))\dd P\right]=\tr\left[\left(Q(t+V(P))P\right)P\dd PQ\right]+\operatorname{tr}\left[\left(P(t+V(P))Q\right)Q\dd PP\right].
\end{align}
Now for $\dd h=0$ we have the two equations
\begin{align}
Q(T+V(P))P&=(T+V(P))P -P(T+V(P))P=0, \nonumber \\ 
P(T+V(P))Q&=P(T+V(P)) -P(T+V(P))P=0 
\end{align}
These equations are not independent (they are Hermitian conjugate of each other), and we can simplify to get the simpler form
\begin{align}
 [T+V(P),P]=0,   
\end{align}
which is the self-consistent mean-field equation. It is self-consistent because $V$ itself is dependent on $P$.
%
\subsection{Time-dependent Hartree-Fock mean-field theory} 
\label{subsec: Time-dependent Hartree-Fock mean-field theory}
%
Now we consider the dynamical problem, to do that, we consider the action functional
\begin{align}
S=\int_{0}^1 \dd t\left( \langle \psi(t)|\ii \frac{\dd}{\dd t}\ket{\psi(t)}-\bra{\psi(t)}\widehat{H}\ket{\psi(t)} \right).
\label{eq: action functional}
\end{align}
If we did not restrict the allowed states to be Slater determinants, variation of this action with respect to the curve of states would simply yield the Schr\"odinger equation, thus recovering the standard quantum dynamics. However, if we only allow Slater-determinant states, we get Hamilton's equations on the Grassmannian manifold seen as a symplectic manifold equipped with the Fubini-Study symplectic form. To see this, first observe that 
\begin{align}
\langle \psi(t)|\frac{\dd}{\dd t}\ket{\psi(t)}\dd t,
\end{align}
is the local connection coefficient of the standard connection on the tautological bundle on $\mathbb{P}\Lambda^r (\mathbb{C}^N)$ along the curve of states $\ket{\psi(t)}$. As such, when we vary we should obtain the curvature two-form, which is essentially the Fubini-Study symplectic form. Let us see that this is indeed the case. We want to vary the curve along the image of the Pl\"ucker embedding, i.e., we have a two-parameter Slater determinant state $\ket{\psi(s,t)}$ corresponding two a family of orthogonal projectors $P(s,t)$ with $P(s=0,t)=P(t)$. The variation of the action is
\begin{align}
\frac{\dd  S}{\dd s}=\int_{0}^1 \dd t \left( \langle \frac{\partial}{\partial s}\psi(s,t)|\ii \frac{\partial}{\partial t}\ket{\psi(s,t)}-\langle \frac{\partial}{\partial t}\psi(s,t)|\ii \frac{\partial}{\partial s}\ket{\psi(s,t)}-\dd H(\frac{\partial P}{\partial s}) \right),
\end{align}
where we assumed a parallel transport ``gauge'', i.e., $\bra{\psi(s,t)}\frac{\partial}{\partial t}\ket{\psi(s,t)}=0$ which allowed us to eliminate the boundary term (which can be done, and the curve in the Grassmannian does not care about that). Observe that the first two terms are just
\begin{align}
\omega_{\mathrm{FS}}(\frac{\partial P}{\partial t},\frac{\partial P}{\partial s}),
\end{align}
where we take $\omega_{\mathrm{FS}}$ to have area $2\pi$ on $\mathbb{C}P^1$ (otherwise there would be a factor of $2$ in the formula). So the action is extremal if and only if
\begin{align}
\frac{\dd S}{\dd s}\bigg|_{s=0}=0
\end{align}
for all variations, which yields
\begin{align}
\iota_{\frac{\dd P}{\dd t}}\omega_{\mathrm{FS}}=\dd h,
\end{align}
which is to say that we flow along the Hamiltonian vector field associated with $h$. Additionally, because the Grassmannian manifold is K\"ahlerian, we have
\begin{align}
\omega_{\mathrm{FS}}(X,JY)=g_{\mathrm{FS}}(X,Y),
\end{align}
or equivalently,
\begin{align}
\omega_{\mathrm{FS}}(X,Y)=g_{\mathrm{FS}}(JX,Y),
\end{align}
so we get
\begin{align}
g_{\mathrm{FS}}(J\frac{\dd P}{\dd t},\cdot)=\dd H,    
\end{align}
which is to say that $J\frac{\dd P}{\dd t}$ is the gradient vector field of $h$ with respect to the Fubini-Study metric. But,
\begin{align}
J\frac{\dd P}{\dd t}=\ii Q \frac{\dd P}{\dd t} P -\ii P\frac{\dd P}{\dd t}Q=-\ii[P,\frac{\dd P}{\dd t}]
\end{align}
Using
\begin{align}
\dd H= \operatorname{tr}\left(\left(T+V(P)\right)\dd P\right)=\mathrm{Re}\operatorname{tr}\left(\left(T+V(P)\right)\dd P\right),
\end{align}
and
\begin{align}
g_{\mathrm{FS}}(J\frac{\dd P}{\dd t},\cdot)=\mathrm{Re}\operatorname{tr}\left(J\frac{\dd P}{\dd t}\dd P\right),
\end{align}
we immediately get, from nondegeneracy of $g_{\mathrm{FS}}$
\begin{align}
J\frac{\dd P}{\dd t}=Q(T+V(P))P +P(T+V(P))Q.
\end{align}
Acting with $J$ on both sides yields
\begin{align}
\ii \frac{\dd P}{\dd t}=[Q(T+V(P))P +P(T+V(P))Q,P]=[T+V(P),P],
\end{align}
since $[P,PXP]=0$ and $[P,QXQ]=0$, for any $X$. The equation above is simply the von Neumann equation for the density matrix (in this case the single-particle density matrix), except that the effective Hamiltonian is itself dependent on the density matrix (nonlinear equation).
%
\subsection{Mean-field theory and harmonic bands}
\label{subsec: Mean-field theory and harmonic bands}
%
We now consider the problem of density-density interactions on a translation-invariant system on a lattice $\mathbb{Z}^d$, or, in the finite size case, a discrete torus. We are going to consider Slater determinant states which are associated to fully filled Bloch bands. So in this more general setting a state is parametrized by a map $P:\BZ^d\to \Gr_{r}(\mathbb{C}^N)$ ($N$ here is the total number of bands)---a smooth family of rank $r$ orthogonal projection operators---or, equivalently, by a vector bundle $E\to \BZ^d$ over the Brillouin zone, subbundle of the trivial rank $N$ bundle $E\subset \BZ^d\times \mathbb{C}^N$, and whose fiber at $\bf{k}\in\BZ^d$ is $\mathrm{Im}\; P(\bf{k})$. In other words, when we do mean-field theory, we will be looking at a functional in the space of maps $P:\BZ^d\to\Gr_r(\mathbb{C}^N)$ and when we look at time-dependent mean-field theory we will be looking at a flow in this space.

We start with the following Hamiltonian with hopping and general density-density interactions for fermions on a lattice
\begin{align}
\widehat{H}=\lambda_T\sum_{\bf{k}}\sum_{i,j=1}^{N}c_{\bf{k},i}^\dagger t_{ij}(\bf{k})c_{\bf{k},j} +\frac{1}{V}\sum_{\bf{q}} v(\bf{q}):n_{-\bf{q}}n_{\bf{q}}:,
\end{align}
where
\begin{align}
n_{\bf{q}}=\sum_{\bf{k}}\sum_{i=1}^{N}c^\dagger_{\bf{k}+\bf{q},i}c_{\bf{k},i}
\end{align}
is the Fourier transform of the particle density and $V$ is the volume, and $:-:$ is normal ordering. Our Slater determinant states parameterized by maps $P:\BZ^d\to\Gr_{r}(\mathbb{C}^N)$ satisfy
\begin{align}
\langle c_{i,\bf{k}}^\dagger c_{j,\bf{k}'}\rangle =\delta_{\bf{k},\bf{k}'}P_{ji}(\bf{k}),
\end{align}
where we denoted $\langle-\rangle$ the average on the state. We then compute $h(P):=\langle H\rangle $. The first term is simply
\begin{align}
\langle \sum_{\bf{k}}\sum_{i,j=1}^{N}c_{\bf{k},i}^\dagger t_{ij}(\bf{k})c_{\bf{k},j}\rangle =\sum_{\bf{k}}\sum_{i,j=1}^{N} t_{ij}(\bf{k})P_{ji}(\bf{k})=\sum_{\bf{k}}\tr \left[T(\bf{k})P(\bf{k})\right],
\end{align}
with $T(\bf{k})=[t_{ij}(\bf{k})]_{1\leq i,j\leq N}$. 
In the thermodynamic limit this goes to
\begin{align}
\frac{V}{(2\pi)^d}\int_{\BZ^d}\dd^d\bf{k}\;\tr\left[T(\bf{k})P(\bf{k})\right].
\end{align}
We evaluate the second term by writing
\begin{align}
\frac{1}{V}\sum_{\bf{q}}\sum_{\bf{k}_1,\bf{k}_2}\sum_{i,j=1}^{N }v(\bf{q}):c^{\dagger}_{\bf{k}_1-\bf{q},i}c_{\bf{k}_1,i}c^{\dagger}_{\bf{k}_2+\bf{q},j}c_{\bf{k}_2,j}:=\frac{1}{V}\sum_{\bf{q}}\sum_{\bf{k}_1,\bf{k}_2}\sum_{i,j=1}^{N }v(\bf{q})c^{\dagger}_{\bf{k}_1-\bf{q},i}c^{\dagger}_{\bf{k}_2+\bf{q},j}c_{\bf{k}_2,j}c_{\bf{k}_1,i}.
\end{align}
Then, Wick's theorem yields
\begin{align}
\langle c^{\dagger}_{\bf{k}_1-\bf{q},i}c^{\dagger}_{\bf{k}_2+\bf{q},j}c_{\bf{k}_2,j}c_{\bf{k}_1,i}\rangle &=\langle c^{\dagger}_{\bf{k}_1-\bf{q},i} c_{\bf{k}_1,i} \rangle \langle c^{\dagger}_{\bf{k}_2+\bf{q},j}c_{\bf{k}_2,j} \rangle -\langle c^{\dagger}_{\bf{k}_1-\bf{q},i}c_{\bf{k}_2,j}\rangle\langle c^{\dagger}_{\bf{k}_2+\bf{q},j}c_{\bf{k}_1,i} \rangle \nonumber \\
&= P_{ii}(\bf{k}_1)P_{jj}(\bf{k}_2) \delta_{\bf{q},0}\delta_{\bf{q},0} - P_{ji}(\bf{k}_2)P_{ij}(\bf{k}_1)\delta_{\bf{k}_1,\bf{k}_2+\bf{q}}\delta_{\bf{k}_1,\bf{k}_2+\bf{q}},
\end{align}
so
\begin{align}
\langle \frac{1}{V}\sum_{\bf{q}} v(\bf{q}):n_{-\bf{q}}n_{\bf{q}}:\rangle &=\frac{1}{V}\sum_{\bf{q}}v(\bf{q})r^2\sum_{\bf{k}_1,\bf{k}_2}(\delta_{\bf{q},0})^2 \nonumber \\
&-\frac{1}{V}\sum_{\bf{q}}v(\bf{q})\sum_{\bf{k}_1,\bf{k}_2}\tr\left[P(\bf{k}_2)P(\bf{k}_1)\right]\delta_{\bf{k}_1,\bf{k}_2+\bf{q}}\delta_{\bf{k}_1,\bf{k}_2+\bf{q}},
\end{align}
in the discrete sum $\delta_{\bf{k},\bf{k}'}^2=\delta_{\bf{k},\bf{k}'}$ (since it is just a Kroenecker delta), so we can simplify this to
\begin{align}
\langle \frac{1}{V}\sum_{\bf{q}} v(\bf{q}):n_{-\bf{q}}n_{\bf{q}}:\rangle=Vr^2\sum_{\bf{q}}v(\bf{q}) -\frac{1}{V}\sum_{\bf{k},\bf{q}}v(\bf{q})\tr\left[P(\bf{k})P(\bf{k}+\bf{q})\right],    
\end{align}
We then conclude that
\begin{align}
\langle \widehat{H}\rangle= \lambda_T\sum_{\bf{k}}\tr \left[T(\bf{k})P(\bf{k})\right] +Vr^2\sum_{\bf{q}}v(\bf{q}) -\frac{1}{V}\sum_{\bf{k},\bf{q}}v(\bf{q})\tr\left[P(\bf{k})P(\bf{k}+\bf{q})\right].  
\end{align}
The second term is independent of $P$ so we may ignore it, and we will simply redefine
\begin{align}
\langle\widehat{H}\rangle=\lambda_T\sum_{\bf{k}}\tr \left[T(\bf{k})P(\bf{k})\right]  -\frac{1}{V}\sum_{\bf{k},\bf{q}}v(\bf{q})\tr\left[P(\bf{k})P(\bf{k}+\bf{q})\right].
\end{align}
In the thermodynamic limit, we actually need to consider $h(P):=\langle \widehat{H}\rangle/V$ so the factors of $V$ drop out,
\begin{align}
H(P)=\lambda_T\int_{\BZ^d}\frac{\dd^d\bf{k}}{(2\pi)^d} \; \tr\left[T(\bf{k})P(\bf{k})\right]-\int_{\BZ^d}\frac{\dd^d\bf{k}}{(2\pi)^d}\int_{\BZ^d}\frac{\dd^d\bf{q}}{(2\pi)^d}\;  v(\bf{q})\tr\left[P(\bf{k})P(\bf{k}+\bf{q})\right].
\end{align}
Now the second term is manifestly quantum geometric in nature. In fact, we have a small-$\bf{q}$ expansion
\begin{align}
-\tr\left[P(\bf{k})P(\bf{k}+\bf{q})\right] &=-r -\frac{1}{2}\sum_{i,j}\tr\left[P(\bf{k})\frac{\partial ^2 P}{\partial k_i \partial k_j}(\bf{k})\right]q_i q_j +\dots \nonumber \\
&=-r +\frac{1}{2}\sum_{i,j}\tr\left[\frac{\partial P}{\partial k_i }(\bf{k})\frac{\partial P}{\partial k_j }(\bf{k})\right]q_i q_j+\dots\nonumber \\
&=-r +\sum_{i,j}g_{ij}(\bf{k})q_iq_j+\dots,
\end{align}
where $g=P^{*}g_{\mathrm{FS}}$ denotes the pullback under $P$ of the Fubini-Study metric, also know in condensed matter as the quantum metric (here we use the normalization convention for the Fubini-Study metric $\int_{\mathbb{C}P^1}\omegaFS=\pi$---compare with different sections). Here $\dots$ denote terms of higher order in the transferred momentum $\bf{q}$. For the lowest Landau level band, truncating this expansion is justified because the above quantities falls off as a Gaussian in the $\bf{q}$-variable. Either way, we may consider the first non-trivial contribution
\begin{align}
v(\bf{q})\sum_{i,j}g_{ij}(\bf{k})q_iq_j.
\end{align}
To take care of $\bf{q}$-integration we now use similar reasoning to that used in Ref.~\cite{onishi:avdoshkin:fu:2025}. Under an isotropy assumption of $v(\bf{q})=v(|\bf{q}|)$, it is reasonable to average over a sphere of radius $q=|\bf{q}|$ and then integrate $q$ until a cutoff high energy momentum $Q$. The result is
\begin{align}
\int_{0}^{Q}\dd q\;q^{d-1}  v(q)\left( q^2 \frac{1}{d}\sum_{i,j}\delta^{ij}g_{ij}(\bf{k})\right),
\end{align}
In other words, we get an effective energy functional which, discarding constant terms and higher order contributions, looks like
\begin{align}
h(P)=\lambda_T\int_{\BZ^d}\frac{\dd^d\bf{k}}{(2\pi)^d} \; \tr\left[T(\bf{k})P(\bf{k})\right] +\frac{\lambda_{\mathrm{D}}}{2}\int_{\BZ^d}\frac{\dd^d\bf{k}}{(2\pi)^d}\sqrt{\det (h)}\sum_{i,j}h^{ij}g_{ij},
\end{align}
where in the second term we introduced a flat metric $h=\sum_{ij}h_{ij}\; \dd k_i \dd k_j$ which accounts for different anisotropy conditions for $v$, so the average on the sphere of before becomes on ellipsoids defined by this metric and $\lambda_{\mathrm{D}}$ is the constant resulting from the integral of $v(q)$. The functional that $\lambda_{\mathrm{D}}$ multiplies is the so-called Dirichlet energy functional---$E_{\mathrm{D}}(P)=\frac{1}2{}\int_{\BZ^d}\dd^d\bf{k}\sqrt{\det (h)}\sum_{i,j}h^{ij}\left(P^{*}g_{\mathrm{FS}}\right)_{ij}$ --- for maps $P:(\BZ^d,h)\to (\Gr_r(\mathbb{C}^N),g_{\mathrm{FS}})$ whose critical points are harmonic maps---referred to in this context as Harmonic bands. If we assume that the band is such that the associated tight-binding energy dispersion, the integrand of the first term, is approximately flat, then we only care about the second terms and the mean-field solution is exactly a harmonic band. In general, extremizing this functional gives
\begin{align}
\lambda_T QTP -\lambda_{\mathrm{D}} Q(\Delta_{h}P) P=0,
\end{align}
with $Q(\Delta_hP) P=\frac{1}{\sqrt{\det (h)}}Q\sum_{i,j}h^{ij}\frac{\partial }{\partial k_i }(\sqrt{\det(h)}\frac{\partial P}{\partial k_j})P=\sum_{i,j}h^{ij}Q\frac{\partial ^2 P}{\partial k_i \partial k_j}P$ (since $h$ is flat).


%
\section{Microscopic derivation of the metriplectic flow}
Our starting point will be the equation of motion for Heisenberg picture operators
\begin{align}
\frac{\dd \widehat{O}}{\dd t }(t)=\ii [\widehat{H},\widehat{O}],
\end{align}
and then take the average in the density matrix of the composite system.

To be able to properly recover our results in the previous notes, in principle we have to add an additional assumption of having a Slater determinant state for the system at all times, which is to say that we are considering the motion of the system's state only along the manifold of Slater determinants. This is translated, in this language, into $\widehat{\rho}_S(t)$---in the Schr\"odinger picture---being a free-fermion Gaussian density matrix---hence satisfying Wick's theorem---with
\begin{align}
\tr_S(\widehat{\rho}_S(t) c_i^\dagger c_j):=\langle c_i^\dagger c_j\rangle =P_{ji}(t),
\end{align}
here $P=[P_{ji}]$ is an orthogonal projector of rank equal to the total number of particles in the system---i.e., determines a curve in a Grassmannian manifold.

Our results should then be obtainable through the equation of motion
\begin{align}
\frac{\dd }{\dd t} P_{ji}(t)=&\frac{\dd }{\dd t} \tr_{S\cup B}(\widehat{\rho}(t_0) c_i^\dagger (t)c_j(t))=\frac{\dd }{\dd t} \tr_S(\widehat{\rho}_S(t) c_i^\dagger c_j),
\end{align}
where in the first equality we use the Heisenberg picture and in last the Schr\"odinger picture. Actually, as we will see later, if we assume that the system's initial state is a Slater determinant, it turns out that provided the system remains Gaussian---satisfying Wick's theorem--- it will remain a Slater determinant. It is then enough to consider it to be Gaussian.

We will first consider the case of no interaction with the bath, i.e., $H_{I}=0$, that means that it is enough to focus on the first term. We will recover the time-dependent Hartree-Fock mean-field equation. Then we will consider the interaction term.

We will deal with $\widehat{H}=\widehat{H}_S+H_B +\widehat{H}_I$, with the system's piece being
\begin{align}
\widehat{H}_{S}=\sum_{i,j}t_{ij}c_i^\dagger c_j +\frac{1}{2}\sum_{i,j,k,l}v_{ijkl}c_i^\dagger c_j^\dagger c_l c_k,
\end{align}
and the bath piece
\begin{align}
\widehat{H}_B=\sum_{\alpha}\Omega_{\alpha} b_{\alpha}^\dagger b_{\alpha},
\end{align}
and the interaction between the system and bath being
\begin{align}
\widehat{H}_{I}&=\lambda\sum_{\alpha,i,j}\left(b_{\alpha}^\dagger c^\dagger_{i}\left(V_{\alpha}\right)_{ij}c_j +b_{\alpha} c^\dagger_{i}\overline{\left(V_{\alpha}\right)}_{ji}c_j\right)=\lambda \sum_{\alpha}\left(b^\dagger_{\alpha}c^\dagger V_{\alpha}c +b_{\alpha}c^\dagger V_{\alpha}^\dagger c\right)\nonumber \\
&=\lambda \sum_{\alpha}c^\dagger \left(b^\dagger_{\alpha}V_{\alpha} +b_{\alpha} V_{\alpha}^\dagger\right)c,
\end{align}
where $(V_{\alpha})_{ij}$'s are the interaction vertices and $\lambda$ is a real parameter. In the last two lines we used a compact notation where the fermionic indices $ij$ are omitted.

The relevant Heisenberg equations of motion are for the operators $S=S_{ji}(t):=[c_i^\dagger(t) c_j(t)]$ and $b_{\alpha}(t)$. They read
\begin{align}
\frac{\dd S_{ji}}{\dd t}=\ii [\widehat{H},S_{ji}]=\ii [\widehat{H},S_{ji}] +\ii \lambda \sum_{\alpha,k}\left(S_{jk}\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{ki}-\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{jk}S_{ki}\right)
\label{eq: system eom}
\end{align}
and
\begin{align}
\frac{\dd b_{\alpha}}{\dd t}=\ii[\widehat{H},b_{\alpha}]=-\ii \Omega_{\alpha} b_{\alpha} -\ii\lambda \sum_{\alpha} c^\dagger V_{\alpha}c.
\label{eq: bath eom}
\end{align}
Observe that $P_{ji}(t)=\tr_{S\cup B}(\widehat{\rho}(t_0)S_{ji}(t))$. 

We will assume, as it is common in the derivation of quantum master equations in open quantum systems (see Chapter.~3.3. in the book by Breuer-Petruccione~\cite{breuer:petruccione:2002}), that at each time $t$, the system and the bath decouple
\begin{align}
\widehat{\rho}(t)\cong \widehat{\rho}_S (t)\otimes \widehat{\rho}_B(t),
\end{align}
where this equation is in the Schr\"odinger picture. This is the so-called \emph{Born approximation}, in which we assume that the influence of the system on the bath is small because their coupling is weak. We will also assume that, in addition to the system being Gaussian at all times, the bath is Gaussian at initial time $t=t_0$, so
\begin{align}
\langle b_{\alpha}\rangle=\langle b_{\alpha}^\dagger\rangle=0 \text{ and } \langle b_{\alpha}^\dagger b_{\alpha}\rangle=n_B(\Omega_{\alpha}),
\end{align}
with Wick's theorem being satisfied. Here $n_B(\Omega_{\alpha})$ is the Bose-Einstein distribution.

%
\subsection{No interaction, i.e., $H_{I}=0$: time-dependent Hartree-Fock mean-field equation}
\label{subsec: No interaction: time-dependent Hartree-Fock mean-field equation}
%
In this case, the dynamics of the system decouples from that of the bath, and we have 
\begin{align}
\frac{\dd }{\dd t} P_{nm}(t)=&\frac{\dd }{\dd t} \tr_S(\widehat{\rho}_S(t) c_m^\dagger c_n)=\ii \tr_S(\widehat{\rho}_S(t) [\widehat{H}_S,c_m^\dagger c_n]).
\end{align}
Because of the Gaussian density matrix assumption, this expectation value can be expressed in terms of single-particle density matrix, i.e., in terms of the projector $P(t)$. First we evaluate $[H_S,c_m^\dagger c_n]$. The first term is the hopping
\begin{align}
[\sum_{i,j}t_{ij}c_i^\dagger c_j,c_m^\dagger c_n]=\sum_{i} t_{im}c_m^\dagger c_n -\sum_{j}t_{nj}c^{\dagger}_ic_j.
\end{align}
After taking the average on $\rho_S$, this yields
\begin{align}
\Big\langle [\sum_{i,j}t_{ij}c_i^\dagger c_j,c_m^\dagger c_n]\Big\rangle=\sum_{i} t_{im}P_{nm} -\sum_{j}t_{nj}P_{ji}=-[T,P]_{nm},
\end{align}
where $T=[T_{ij}]$ is the hopping matrix. Then we evaluate the quartic term:
\begin{align}
\frac{1}{2}\Biggl[\sum_{i,j,k,l,} v_{ijkl}\, c_i^\dagger c_j^\dagger c_l c_k,\; c_m^\dagger c_n\Biggr]
= \sum_{i, j, l} v_{ijml}\, c_i^\dagger c_j^\dagger c_l c_n
\;-\;
\sum_{i, l, k} v_{inkl}\, c_i^\dagger c_m^\dagger c_l c_k.
\end{align}
Finally, we take the average on the density matrix $\rho_{S}(t)$, simplifying quartic averages according to Wick's theorem,
\begin{align}
\langle c_i^\dagger c_j^\dagger c_l c_k\rangle=P_{ki}P_{lj}-P_{li}P_{kj}.  
\end{align}
The result is
\begin{align}
\Big\langle\frac{1}{2}\Bigl[\sum_{ijkl} v_{ijkl}\, c_i^\dagger c_j^\dagger c_l c_k,\; c_m^\dagger c_n\Bigr]\Big\rangle
&= \sum_{i, j, l} v_{ijml}\,\bigl(P_{ni}P_{lj}-P_{li}P_{nj}\bigr)
-\sum_{i, l, k} v_{inkl}\,\bigl(P_{ki}P_{lm}-P_{li}P_{km}\bigr)\nonumber \\
&=\sum_{i}P_{ni}\left(\sum_{j, l} (v_{ijml}-v_{jiml}) P_{lj}\right)   -\sum_{l}\left(\sum_{i,k}(v_{inkl}-v_{inlk})P_{ki}\right)P_{lm}\nonumber \\ 
&=\sum_{i}P_{ni}\left(\sum_{j, l} (v_{ijml}-v_{ijlm}) P_{lj}\right)   -\sum_{l}\left(\sum_{i,k}(v_{nilk}-v_{nikl})P_{ki}\right)P_{lm}\nonumber \\
&=\sum_{i}P_{ni} V(P)_{im}-\sum_{l}V(P)_{nl}P_{lm}\nonumber \\
&= -\bigl[V(P),\,P\bigr]_{mn}, \qquad
V(P)_{ik}
= \sum_{j,l}\bigl(v_{i j k l}-v_{i j l k}\bigr)\,P_{l j}.
\end{align}
Thus, we find that the equation of motion for $P$ is
\begin{align}
\ii \frac{\dd P}{\dd t}=[T+V(P),P],   
\end{align}
which is consistent with the obtained in the other notes, coming from Hamilton's equations of motion in the Grassmannian
\begin{align}
\iota_{X_H}\omega=\dd H_{S} \iff \ii \frac{\dd P}{\dd t}=[\frac{\partial H_{S}}{\partial P},P]
\end{align}
with $H_{S}=\langle \widehat{H}_S\rangle=\tr\left[P\left(T+\frac{1}{2}V(P)\right)\right]$.
%
\subsection{Interaction with the bath}
\label{subsec: Interaction with the bath}
%
First we formally solve Eq.~\eqref{eq: bath eom} and plug it in Eq.~\eqref{eq: system eom}. Eq.~\eqref{eq: bath eom}, which we reproduce here,
\begin{align}
\frac{\dd b_{\alpha}}{\dd t}=\ii[\widehat{H},b_{\alpha}]=-\ii \Omega_{\alpha} b_{\alpha} -\ii \lambda \sum_{\alpha} c^\dagger V_{\alpha}c,
\end{align}
is a standard linear ODE with a nonhomogeneous term. We can solve it by resorting to the $b_{\alpha}$'s retarded Green's function
\begin{align}
G_{\alpha}(t)=\theta(t)e^{-\ii \Omega_{\alpha}(t)},
\end{align}
namely
\begin{align}
b_{\alpha}(t)&=b_{\alpha}e^{-\ii \Omega_{\alpha}(t-t_0)} -\ii \lambda \int_{t_0}^{t}G_{\alpha}(t-\tau) \left(c^\dagger V_{\alpha}c\right)(\tau) \dd\tau\nonumber\\
&=b_{\alpha}e^{-\ii \Omega_{\alpha}(t-t_0)} -\ii \lambda \int_{t_0}^{t}G_{\alpha}(t-\tau) \sum_{ij}S_{ji}(
\tau)(V_{\alpha})_{ij} \; \dd\tau
\end{align}

For readability we reproduce the formula for the equation of motion of $S_{ji}(t)$
\begin{align}
\frac{\dd S_{ji}}{\dd t}=\ii [\widehat{H},S_{ji}]=\ii [\widehat{H},S_{ji}] +\ii \lambda \sum_{\alpha,k}\left(S_{jk}\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{ki}-\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{jk}S_{ki}\right).
\end{align}
We can now take the average of this equation on $\widehat{\rho}(t_0)$ to obtain
\begin{align}
\frac{\dd P_{ji}}{\dd t}&=\ii \langle[\widehat{H},S_{ji}]\rangle=\ii [\widehat{H},S_{ji}] +\ii \lambda \langle\sum_{\alpha,k}\left(S_{jk}\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{ki}-\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{jk}S_{ki}\right)\rangle \nonumber \\
&=\left(\ii [P,\frac{\partial H_{S}}{\partial P}]\right)_{ji}+ \ii \lambda \sum_{\alpha,k}\left(P_{jk}\langle\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{ki}\rangle-\langle\left(b^\dagger_{\alpha}V_{\alpha} +b V_{\alpha}^\dagger\right)_{jk}\rangle P_{ki}\right),
\end{align}
where the second step follows from the result with $\widehat{H}_{I}=0$ and from the factorization of the density matrix (Born approximation). Thus, what we need to plug in this equation is the average in $\widehat{\rho}(t_0)$ of $b_{\alpha}(t)$, which is given by
\begin{align}
b_{\alpha}(t)&= -\ii \lambda \int_{t_0}^{t}G_{\alpha}(t-\tau) \sum_{ij}P_{ji}(
\tau)(V_{\alpha})_{ij} \; \dd\tau=-\ii \lambda \int_{t_0}^{t}G_{\alpha}(t-\tau) \tr(P(\tau)V_{\alpha}) \; \dd\tau
\end{align}



Observe that the $b_{\alpha}(t_0)=b_{\alpha}$'s enter linearly and they will hence average to zero. Thus, only the convolution part from the formal solution remains. We then find
\begin{align}
\sum_{\alpha}\langle\left(b^\dagger_{\alpha}V_{\alpha} +b_{\alpha} V_{\alpha}^\dagger\right)_{ki}\rangle &= \ii \lambda \sum_{\alpha}\left(\left(V_{\alpha}\right)_{ki}\int_{t_0}^{t}\overline{G_{\alpha}}(t-\tau) \tr(P(\tau)V^{\dagger}_{\alpha}) \; \dd\tau -\left(V^\dagger_{\alpha}\right)_{ki}\int_{t_0}^{t}G_{\alpha}(t-\tau) \tr(P(\tau)V_{\alpha}) \; \dd\tau\right)\nonumber\\
&=: \lambda F_{ki},
\end{align}
so we can write the resulting equations of motion compactly in matrix notation as
\begin{align}
\frac{\dd P}{\dd t } =\ii [P,\frac{\partial H_S}{\partial P}] + \ii \lambda^2[P,F].
\end{align}
From this, we can already understand that if the initial state of the system was a Slater determinant, it will remain a Slater determinant. The reason is that
\begin{align}
P\frac{\dd P}{\dd t }P=0 \text{ and } (1-P)\frac{\dd P}{\dd t }(1-P)=0,
\end{align}
provided $P^2=P$. That means that the vector field on the right-hand side restricted to the Grassmannian is tangent to the Grassmannian, and hence it defines a flow inside the Grassmannian. General fermionic Gaussian states do not satisfy $P^2=P$. 
%
\subsection{Markov approximation} 
\label{subsec: Markov approximation}
%
We now wish to perform a Markov approximation, which assumes that the system has little memory with regards to what happened in the past. We only need to focus on studying the memory-dependent term
\begin{align}
F=    \ii\sum_{\alpha}\left(V_{\alpha}\int_{t_0}^{t}\overline{G_{\alpha}}(t-\tau) \tr(P(\tau)V^{\dagger}_{\alpha}) \; \dd\tau -V^\dagger_{\alpha}\int_{t_0}^{t}G_{\alpha}(t-\tau) \tr(P(\tau)V_{\alpha}) \; \dd\tau\right)
\end{align}

We will do the Markov approximation in three steps. The first is to consider that $t_0\to -\infty$ and change the integration variable,
\begin{align}
F&=    \ii\sum_{\alpha}\left(V_{\alpha}\int_{0}^{\infty}\overline{G_{\alpha}}(\tau) \tr(P(t-\tau)V^{\dagger}_{\alpha}) \; \dd\tau -V^\dagger_{\alpha}\int_{t_0}^{t}G_{\alpha}(\tau) \tr(P(t-\tau)V_{\alpha}) \; \dd\tau\right)\nonumber\\
&=\ii\sum_{\alpha}\left(V_{\alpha}\int_{0}^{\infty}e^{i\Omega_{\alpha}\tau} \tr(P(t-\tau)V^{\dagger}_{\alpha}) \; \dd\tau -V^\dagger_{\alpha}\int_{t_0}^{t}e^{-i\Omega_{\alpha}\tau} \tr(P(t-\tau)V_{\alpha}) \; \dd\tau\right) \nonumber \\
&=\ii\int_{0}^{\infty}\sum_{\alpha}e^{i\Omega_{\alpha}\tau} \tr(P(t-\tau)V^{\dagger}_{\alpha})V_{\alpha} \; \dd\tau -\ii \int_{t_0}^{t}e^{-i\Omega_{\alpha}\tau} \tr(P(t-\tau)V_{\alpha})V^\dagger_{\alpha} \; \dd\tau.
\end{align}

The third step will require further assumptions about the bath. It is useful to note that $\tr(A^\dagger B)=\langle A|B\rangle$ determines an inner product in the complex linear space of matrices. We can then write $F$ in a perhaps more appealing form
\begin{align}
F=\ii \int_{0}^{\infty}\sum_{\alpha}e^{i\Omega_{\alpha}\tau }\bra{V_{\alpha}} P(t-\tau)\rangle V_{\alpha}\dd\tau + \text{H.c.}
\end{align}
We introduce the bath's spectral density (super-)operator as
\begin{align}
 J(\omega)=\sum_{\alpha} \delta(\omega-\Omega_{\alpha})\ket{V_{\alpha}}\bra{V_{\alpha}},
\end{align}
so that we can write
\begin{align}
F=\ii \int_{0}^{\infty}\left(\int_{0}^{\infty}e^{i\omega\tau }J(\omega)\left[P(t-\tau)\right]\dd\omega \right)\dd\tau + \text{H.c.}    
\end{align}
It is common in physics to assume that the bath is Ohmic with exponential cutoff. Here this will translate to
\begin{align}
J(\omega)= \eta \;\omega e^{-\tfrac{\omega}{\omega_c}},
\end{align}
where the identity (super-)operator is omitted to simplify the notation and $\eta$ is a friction constant. Using
\begin{align}
J(\tau):=\int_{0}^{\infty}J(\omega)e^{\ii\tau \omega} \dd\omega =\int_{0}^{\infty}   \eta \;\omega e^{-\left(\tfrac{1}{\omega_c}-\ii \tau\right)\omega}\dd\omega= \frac{\eta}{\left(\tfrac{1}{\omega_c}- \ii\tau\right)^2}
\end{align}
this yields
\begin{align}
F= \ii \int_{0}^{\infty}\frac{\eta}{\left(\tfrac{1}{\omega_c}- \ii\tau\right)^2}P(t-\tau)d\tau   -\mathrm{H.c.}
\end{align}
The Markovian limit is to assume that $\omega_c\to \infty$. Recall that
\begin{align}
\frac{1}{\tau \pm \ii 0^{+}}=\mathrm{P.V.} \left(\frac{1}{\tau}\right) \mp\ii \pi \delta(\tau),
\end{align}
so 
\begin{align}
\frac{\dd}{\dd\tau}\left(\frac{1}{\tau +\ii 0^{+}}\right)=-\frac{1}{\left(\tau +\ii 0^{+}\right)^2}=-\mathrm{P.V.} \left(\frac{1}{\tau^2}\right) \mp\ii \pi \delta'(\tau)    
\end{align}
so that---as a distribution---we have
\begin{align}
J(\tau)=\frac{\eta}{\left(\tfrac{1}{\omega_c}-\ii\tau\right)^2}=-\frac{\eta}{\left(\tau+\ii\tfrac{1}{\omega_c}\right)^2}\to -\frac{\eta}{\left(\tau+\ii 0^{+}\right)^2}= -\eta \mathrm{P.V.} \left(\frac{1}{\tau^2}\right) -\ii \pi \eta \delta'(\tau)  ,
\end{align}
observing that for any test function $\varphi(\tau)$
\begin{align}
\int_{0}^{\infty} \varphi(t-\tau)\delta'(\tau)\dd\tau =\varphi'(t),    
\end{align}
we have
\begin{align}
F&=2\pi \eta \frac{\dd P}{\dd\tau} +\left(-\ii \eta\;  \mathrm{P.V.}\int_{0}^{\infty}\frac{\eta}{\tau^2}P(t-\tau)d\tau   -\mathrm{H.c.}\right)\nonumber \\
&=2\pi \ii \eta [P,\frac{\partial H_S}{\partial P}] +\left(-\ii \eta\;  \mathrm{P.V.}\int_{0}^{\infty}\frac{\eta}{\tau^2}P(t-\tau)d\tau   -\mathrm{H.c.}\right)+\mathrm{O}(\lambda^2).
\end{align}
The last term is still non-local in $P(t)$, so we may do one last Markovian approximation $P(t-\tau)\cong P(t)$, which produces a force term which is proportional to $P(t)$ and hence after the commutator, it yields zero so we may neglect it. So effectively, we may consider just
\begin{align}
F\approx 2\pi \ii \eta [P,\frac{\partial H_S}{\partial P}].
\end{align}
Hence, we obtain the equation
\begin{align}
\frac{\dd P}{\dd t}=\ii[ P,\frac{\partial H_S}{\partial P}]- \gamma[P,[P,\frac{\partial H_S}{\partial P}]],
\end{align}
which describes relaxation toward the minima of $H_S$, due to the second term, with relaxation constant $\gamma= 2\pi \eta\lambda^2$ which is second order in the coupling.
%
\subsection{Relaxation toward the minima of $H_S$}
\label{subsec: Relaxation toward the minima of the system Hamiltonian}
%
The equation we obtained
\begin{align}
\frac{\dd P}{\dd t}=\ii[ P,\frac{\partial H_S}{\partial P}]- \gamma[P,[P,\frac{\partial H_S}{\partial P}]],
\end{align}
describes a K\"ahler metriplectic flow on the Grassmannian. What this means is that the Grassmannian manifold $\Gr$ (omitting rank and ambient dimension for brevity) is a K\"ahler manifold equipped with the Fubini-Study symplectic form $\omegaFS=-\ii\tr(P\dd P\wedge \dd P)$ and the Fubini-Study metric $\gFS=2\tr (P\dd P\odot \dd P)$. These two nondegenerate bilinear forms define tensor fields (bivectors) $\omegaFS^{-1}\in \Gamma(\Lambda^{2}T\Gr)$ and $\gFS^{-1}\in \Gamma(S^{2}T\Gr)$ by considering the inverse matrices of their local representatives. In particular, if in local coordinates we have $\omegaFS=\frac{1}{2}\sum_{ij}\omega_{ij}dx^i\wedge  dx^j$ and $\gFS=\sum_{ij}g_{ij}dx^i\odot dx^j$, we have $\omegaFS^{-1}=\frac{1}{2}\sum_{ij}\omega^{ij}\frac{\partial}{\partial x^i}\wedge \frac{\partial}{\partial x^j}$ and $\gFS^{-1}=\sum_{ij}g^{ij}\frac{\partial}{\partial x^i}\odot \frac{\partial}{\partial x^j}$, where $[\omega^{ij}]$ and $[g^{ij}]$ are the inverse matrices of $[\omega_{ij}]$ and $[g_{ij}]$, respectively. The flow equation can therefore be expressed intrinsically in terms of the symplectic and metric bivectors as
\begin{align}
X=\omegaFS^{-1}(\dd H_S,\cdot)-\gamma\gFS^{-1}(\dd H_S,\cdot).
\end{align}
The vector field $\gFS^{-1}(\dd H_S,\cdot)$ is referred to as the Riemannian gradient of $H_S$ and denoted $\nabla H_S$, because it is the unique vector field satisfying
\begin{align}
\gFS(\nabla H_S,\cdot)=\dd H_S.
\end{align}
In local coordinates $\nabla H_S=\sum_{j}g^{ji}\frac{\partial H_S}{\partial x^i}\frac{\partial}{\partial x^j}$.
Also, $X_{H_S}=\omegaFS^{-1}(\dd H_S,\cdot)$ is the Hamiltonian vector field for $H_S$, and it is the unique vector field satisfying
\begin{align}
\omegaFS(X_{H_S},\cdot)=\dd H_S.   
\end{align}
In local coordinates $X_{H_S}=\sum_{j}\omega^{ji}\frac{\partial H_S}{\partial x^i}\frac{\partial}{\partial x^j}$.
To use the K\"ahler structure more explicitly, observe that due to compatibility with respect to the complex structure $J$ (which at the matrix level is just $J(X)=-\ii[P,X]$) 
\begin{align}
\gFS=\omegaFS(\cdot,J\cdot)\implies \gFS(\nabla H_S,\cdot)=-\omegaFS(J\nabla H_S,\cdot)=\dd H_S=\omegaFS(X_{H_S},\cdot)
\end{align}
from which it follows that $X_{H_S}=-J\nabla H_S$, which is consistent with
\begin{align}
-J([P,[P,\frac{\partial H_S}{\partial P}]])=i[P,([P,[P,\frac{\partial H_S}{\partial P}]])]=i[P,\frac{\partial H_S}{\partial P}].
\end{align}
The vector field $X$ can then be compactly written in two ways
\begin{align}
X=J\nabla H_S -\gamma \nabla H_S=X_{H_S}-\gamma JX_{H_{S}},
\end{align}
the first is in terms of the Riemannian gradient, hence a Riemannian picture; and the second in terms of the Hamiltonian vector field, hence a symplectic picture. Both pictures are naturally related by the complex structure $J$, with $J^2=-1$. This also shows once and for all that
\begin{align}
\gFS(\nabla H_S,X_{H_{S}})=-\gFS(\nabla H_S,J\nabla H_S)=\omegaFS(\nabla H_S,JX_{H_S})=\omegaFS(\nabla H_S,\nabla H_S)=0,
\end{align}
so the two vector fields are orthogonal to each other. Alternatively, $\gFS(\nabla H_S,X_{H_{S}})=\dd H_S(X_{H_{S}})=\omegaFS(X_{H_{S}},X_{H_{S}})=0$. This last method shows that this identity holds for arbitrary nondegenerate metriplectic systems.

Notably,
\begin{align}
\dd H_S (X)&=X\cdot H_S=\mathcal{L}_XH_S=\omegaFS^{-1}(\dd H_S,\dd H_S)-\gamma\gFS^{-1}(\dd H_S,\dd H_S)\nonumber \\
&=-\gamma \gFS^{-1}(\dd H_S,\dd H_S)=-\gamma ||\dd H_S||^2\leq 0,
\end{align}
which follows from skew-symmetry of $\omegaFS$ and, consequently, of $\omegaFS^{-1}$ and positive-definiteness of $\gFS$ and, hence, of $\gFS^{-1}$.
It follows that if $t\mapsto P(t)$ is an integral curve of $X$, we have
\begin{align}
\frac{\dd}{\dd t}(H_S(P(t)))=- \gamma ||\dd H_S||^2(P(t))\leq 0 \implies & H_S(P(t))-H_S(P(t_0))\nonumber \\
&=-\gamma \int_{t_0}^{t}\dd s \; ||\dd H_S||^2(P(s))\leq 0,
\end{align}
so that the energy decreases with time along the integral curves of $X$. 

\remark[Local coordinates version]{ In local coordinates the integral curve of $P(t)$ reads, using the Einstein summation convention,
\begin{align}
\frac{\dd x^i}{\dd t}=\omega^{ji}\frac{\partial H}{\partial x^j} -\gamma g^{ji}\frac{\partial H}{\partial x^j}.
\end{align}
Multiplying by  $\frac{\partial H}{\partial x^i}$ and summing over $i$,
\begin{align}
\frac{\dd H_S}{\dd t}=\frac{\partial H_S}{\partial x^i}\frac{\dd x^i}{\dd t}=\omega^{ji}\frac{\partial H}{\partial x^j}\frac{\partial H_S}{\partial x^i} -\gamma g^{ji}\frac{\partial H}{\partial x^j}\frac{\partial H_S}{\partial x^i}=-\gamma g^{ji}\frac{\partial H}{\partial x^j}\frac{\partial H_S}{\partial x^i}\leq 0,
\end{align}
where the first term vanishes due to $\omega^{ij}=-\omega^{ji}$. The previous result then follows.
}

\remark[Orthogonal projector version]{ Take the trace of the equation of motion with $\frac{\partial H_S}{\partial P}$,
\begin{align}
\frac{\dd H_S}{\dd t}&=\tr\left(\frac{\partial H_S}{\partial P}\frac{\dd  P}{\dd t}\right)=\ii \tr\left(\frac{\partial H_S}{\partial P}[P,\frac{\partial H}{\partial P}]\right)-\gamma \tr\left(\frac{\partial H_S}{\partial P}[P,[P,\frac{\partial H}{\partial P}]]\right)\nonumber \\
&=-\gamma\tr\left(\Pi(\frac{\partial H_S}{\partial P})^2\right)=-\gamma||\dd H_S||^2(P(t)),
\end{align}
where we used the cyclic property of the trace and where
\begin{align}
 \Pi(\frac{\partial H_S}{\partial P})=[P,[P,\frac{\partial H_S}{\partial P}]]=(1-P)\frac{\partial H_S}{\partial P}P +P\frac{\partial H_S}{\partial P}(1-P),
\end{align}
denotes the projection onto the tangent space of the Grassmannian at $P$ (with respect to the Hilbert–Schmidt inner product), from which the previous result follows. This projection appears naturally because $\frac{\partial H_S}{\partial P}$ is just a derivative in the conventional sense---it is a matrix whose components are the derivatives of the function $H_S$ with respect to the matrix components of $P$---and it does not capture the constraint $P^2=P$ and $P^\dagger=P$ which determines that $P$ lives in $\Gr$.

Thus, the weak coupling to an Ohmic environment induces a metriplectic vector field on the Grassmannian, realizing dissipative relaxation as a gradient descent in the Fubini–Study metric while preserving the underlying Kähler geometry.
}
%
\subsection{Harmonic band heat flow}
\label{subsec: Harmonic band heat flow}
%
If we do the same reasoning for a Bloch band with short-range density-density interactions and a translation-invariant interaction with a bath of bosonic oscillators---modes  $b_{\alpha,\bf{k}}$ with frequencies $\Omega_{\alpha}(\bf{k})$---of the form
\begin{align}
\lambda \sum_{\bf{k}}\sum_{\alpha}c_{\bf{k}}^\dagger \left(b^\dagger_{\alpha,\bf{k}}V_{\alpha}(\bf{k}) +b_{\alpha,\bf{k}} V_{\alpha}^\dagger(\bf{k})\right)c_{\bf{k}},    
\end{align}
the same reasoning will apply, provided we have now
\begin{align}
J(\omega,\bf{k})=\sum_{\alpha} \delta(\omega-\Omega_{\alpha}(\bf{k}))\ket{V_{\alpha}(\bf{k})}\bra{V_{\alpha}(\bf{k})}\sim \eta \omega e^{-\tfrac{\omega}{\omega_c}}.
\end{align}
The short-range density-density interactions will yield $H_S$, to lowest order in the transferred momentum, proportional to the Dirichlet energy, as discussed earlier. If this is the main source of system energy $H_S$, then the resulting equation of motion for the band projector $P(\bf{k})$ is of the form
\begin{align}
\frac{\partial P(\bf{k})}{\partial t}=-\ii[ P,\lambda_{\mathrm{D}}\Delta_h P(\bf{k})]+ \gamma[P,[P,\lambda_{\mathrm{D}}\Delta_h P(\bf{k})]],
\end{align}   
with $\Delta_h$ a Laplacian with respect to a flat metric $h$ which describes the anisotropy of the system (essentially the inverse mass tensor). In the absence of the first term this is the Harmonic map heat flow. It's presence forces the band to relax to a harmonic one.

%
\section{Symmetries of the hopping matrix, the quantum metric and $h$}
%
In the main text, we claimed that $h$, interpreted as the inverse mass tensor, can be identified up to scale with the integrated quantum metric.
Here we show that the integrated quantum metric indeed inherits the same lattice symmetries—present in both the hopping matrix $T(\mathbf{k})$ and the interaction potential $v(\mathbf{q})$—and hence can be identified with $h$ up to an overall scale.


Suppose a discrete group of symmetries $G$ preserves the free part of the Hamiltonian. For each $R\in G$, we have
\begin{align}
T(R\!\cdot\!\mathbf{k}) = U(R)^{-1} T(\mathbf{k}) U(R),
\end{align}
where $U(R)$ is a unitary or antiunitary representation of the symmetry.
The $U(R)$'s are required to satisfy the group law projectively, i.e.,
$U(R_1 R_2)=U(R_1)U(R_2)e^{i\alpha(R_1,R_2)}$ for some $U(1)$ phase $\mathrm{e}^{i\alpha(R_1,R_2)}$.
The inverse-mass tensor $h$ must therefore transform compatibly with these symmetries:
\begin{align}
h_{kl}=h_{ij}R^{i}_{\ k}R^{j}_{\ l},
\end{align}
where $R=[R^{i}_{\ j}]$ is the matrix representation of the symmetry acting on quasimomenta~$\mathbf{k}$.

The Bloch projector satisfies an equivalent covariance property,
\begin{align}
P(R\!\cdot\!\mathbf{k}) = U(R)^{-1} P(\mathbf{k}) U(R),
\end{align}
which implies, using the invariance of the Fubini--Study metric under projective unitary and antiunitary transformations, that
\begin{align}
g_{ij}(R\!\cdot\!\mathbf{k})R^{i}_{\ k}R^{j}_{\ l}=g_{kl}(\mathbf{k}).
\end{align}
It then follows that
\begin{align}
\int_{\BZ^2}\frac{\dd^2\mathbf{k}}{(2\pi)^2}\, g_{kl}(\mathbf{k})
&=\int_{\BZ^2}\frac{\dd^2\mathbf{k}}{(2\pi)^2}\, g_{ij}(R\!\cdot\!\mathbf{k})R^{i}_{\ k}R^{j}_{\ l}\nonumber\\
&=\Bigl(\int_{\BZ^2}\frac{\dd^2\mathbf{k}}{(2\pi)^2}\, g_{ij}(\mathbf{k})\Bigr)R^{i}_{\ k}R^{j}_{\ l},
\end{align}
where we used invariance of the Brillouin-zone integral under diffeomorphisms,
together with the fact that $R$ acts linearly on quasimomenta and is $\mathbf{k}$-independent
(note that, to define a diffeomorphism of the Brillouin zone, $R$ must belong to $\mathrm{GL}(2;\mathbb{Z})$
and hence satisfies $\det R = \pm 1$).
The claim then follows.
In particular, for $C_4$ symmetry we have
$R=\begin{bmatrix}0 & -1\\ 1 & 0\end{bmatrix}$,
and it follows that the integrated quantum metric—and hence $h$—is proportional to $\delta_{ij}$, as stated in the main text.






In the main text, we claimed that $h$, interpreted as the inverse mass tensor, can be, up to scale, interpreted as the integrated quantum metric. Here we show that the integrated quantum matric has the correct symmetries, i.e., the symmetries of the hopping matrix $T(\bf{k})$ and so we can identify it, up to an overall scale with $h$. Suppose a discrete group of symmetries $G$ preserves the free part of the Hamiltonian. That means for $R\in G$, we have
\begin{align}
T(R\cdot \bf{k})=U(R)^{-1} T(\bf{k})U(R),
\end{align}
where $U(R)$ is a unitary or anti-unitary matrix. The $U(R)$'s are only required to satisfy group homomorphism property projectively, i.e., $U(R_1R_2)=U(R_1)U(R_2)e^{i\alpha (R_1,R_2)}$ for some $U(1)$ cocycle $e^{i\alpha(R_1,R_2)}$. The inverse mass tensor, is then required to be a tensor which is compatible with these symmetries, namely, 
\begin{align}
h_{kl}=h_{ij}R^{i}_{\, k}R^{j}_{\, l},
\end{align}
where $R=[R^{i}_{\ j}]$ is the matrix representation of the symmetry acting on the quasimomenta $\bf{k}$.

The Bloch projector will satisfy equivalent symmetry properties 
\begin{align}
P(R\cdot \bf{k})=U(R)^{-1} P(\bf{k})U(R),
\end{align}
which implies that, due to invariance of the Fubini-Study metric under projective unitary and anti-unitary transformations,
\begin{align}
g_{ij}(R\cdot\bf{k})R^{i}_{\, k}R^{j}_{\, l}=g_{kl}(\bf{k}).
\end{align}
 It then follows that
\begin{align}
\int_{\BZ^2}\frac{\dd ^2\bf{k}}{(2\pi)^2}\; g_{kl}(\bf{k})&=\int_{\BZ^2}\frac{\dd ^2\bf{k}}{(2\pi)^2}\; g_{ij}(R\cdot\bf{k})R^{i}_{\, k}R^{j}_{\, l}\nonumber \\
&=\left(\int_{\BZ^2}\frac{\dd ^2\bf{k}}{(2\pi)^2}\; g_{ij}(R\cdot\bf{k})\right)R^{i}_{\, k}R^{j}_{\, l}\nonumber \\
&=\left(\int_{\BZ^2}\frac{\dd ^2\bf{k}}{(2\pi)^2}\; g_{ij}(\bf{k})\right)R^{i}_{\, k}R^{j}_{\, l},
\end{align}
where we used invariance of the integral under diffeomorphisms. The claim then follows. In particular for $C_4$-symmetry we have $R=\begin{bmatrix}
0 & -1\\
1 & 0
\end{bmatrix}$ and it follows that the integrated quantum metric, and hence $h$, is proportional to $\delta_{ij}$, as written in the main text.
\end{widetext}
%
%\bibliography{bib.bib}
%


\end{document}

%apsrev4-2.bst 2019-01-14 (MD) hand-edited version of apsrev4-1.bst
%Control: key (0)
%Control: author (8) initials jnrlst
%Control: editor formatted (1) identically to author
%Control: production of article title (0) allowed
%Control: page (0) single
%Control: year (1) truncated
%Control: production of eprint (0) enabled
\begin{thebibliography}{27}%
\makeatletter
\providecommand \@ifxundefined [1]{%
 \@ifx{#1\undefined}
}%
\providecommand \@ifnum [1]{%
 \ifnum #1\expandafter \@firstoftwo
 \else \expandafter \@secondoftwo
 \fi
}%
\providecommand \@ifx [1]{%
 \ifx #1\expandafter \@firstoftwo
 \else \expandafter \@secondoftwo
 \fi
}%
\providecommand \natexlab [1]{#1}%
\providecommand \enquote  [1]{``#1''}%
\providecommand \bibnamefont  [1]{#1}%
\providecommand \bibfnamefont [1]{#1}%
\providecommand \citenamefont [1]{#1}%
\providecommand \href@noop [0]{\@secondoftwo}%
\providecommand \href [0]{\begingroup \@sanitize@url \@href}%
\providecommand \@href[1]{\@@startlink{#1}\@@href}%
\providecommand \@@href[1]{\endgroup#1\@@endlink}%
\providecommand \@sanitize@url [0]{\catcode `\\12\catcode `\$12\catcode `\&12\catcode `\#12\catcode `\^12\catcode `\_12\catcode `\%12\relax}%
\providecommand \@@startlink[1]{}%
\providecommand \@@endlink[0]{}%
\providecommand \url  [0]{\begingroup\@sanitize@url \@url }%
\providecommand \@url [1]{\endgroup\@href {#1}{\urlprefix }}%
\providecommand \urlprefix  [0]{URL }%
\providecommand \Eprint [0]{\href }%
\providecommand \doibase [0]{https://doi.org/}%
\providecommand \selectlanguage [0]{\@gobble}%
\providecommand \bibinfo  [0]{\@secondoftwo}%
\providecommand \bibfield  [0]{\@secondoftwo}%
\providecommand \translation [1]{[#1]}%
\providecommand \BibitemOpen [0]{}%
\providecommand \bibitemStop [0]{}%
\providecommand \bibitemNoStop [0]{.\EOS\space}%
\providecommand \EOS [0]{\spacefactor3000\relax}%
\providecommand \BibitemShut  [1]{\csname bibitem#1\endcsname}%
\let\auto@bib@innerbib\@empty
%</preamble>
\bibitem [{\citenamefont {Roy}(2014)}]{rahul:2014}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Roy}},\ }\bibfield  {title} {\bibinfo {title} {{Band geometry of fractional topological insulators}},\ }\href {https://doi.org/10.1103/PhysRevB.90.165139} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {90}},\ \bibinfo {pages} {165139} (\bibinfo {year} {2014})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Claassen}\ \emph {et~al.}(2015)\citenamefont {Claassen}, \citenamefont {Lee}, \citenamefont {Thomale}, \citenamefont {Qi},\ and\ \citenamefont {Devereaux}}]{claassen:2015}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Claassen}}, \bibinfo {author} {\bibfnamefont {C.~H.}\ \bibnamefont {Lee}}, \bibinfo {author} {\bibfnamefont {R.}~\bibnamefont {Thomale}}, \bibinfo {author} {\bibfnamefont {X.-L.}\ \bibnamefont {Qi}},\ and\ \bibinfo {author} {\bibfnamefont {T.~P.}\ \bibnamefont {Devereaux}},\ }\bibfield  {title} {\bibinfo {title} {{Position-Momentum Duality and Fractional Quantum Hall Effect in Chern Insulators}},\ }\href {https://doi.org/10.1103/PhysRevLett.114.236802} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {114}},\ \bibinfo {pages} {236802} (\bibinfo {year} {2015})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ozawa}\ and\ \citenamefont {Mera}(2021)}]{ozawa:mera:2021}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}}\ and\ \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}},\ }\bibfield  {title} {\bibinfo {title} {{Relations between topology and the quantum metric for Chern insulators}},\ }\href {https://doi.org/10.1103/PhysRevB.104.045103} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {104}},\ \bibinfo {pages} {045103} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mera}\ and\ \citenamefont {Ozawa}(2021{\natexlab{a}})}]{mera:ozawa:2021:relations}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}}\ and\ \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}},\ }\bibfield  {title} {\bibinfo {title} {{K\"ahler geometry and Chern insulators: Relations between topology and the quantum metric}},\ }\href {https://doi.org/10.1103/PhysRevB.104.045104} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {104}},\ \bibinfo {pages} {045104} (\bibinfo {year} {2021}{\natexlab{a}})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mera}\ and\ \citenamefont {Ozawa}(2021{\natexlab{b}})}]{mera:ozawa:2021:engineering}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}}\ and\ \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}},\ }\bibfield  {title} {\bibinfo {title} {{Engineering geometrically flat Chern bands with Fubini-Study K\"ahler structure}},\ }\href {https://doi.org/10.1103/PhysRevB.104.115160} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {104}},\ \bibinfo {pages} {115160} (\bibinfo {year} {2021}{\natexlab{b}})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Wang}\ \emph {et~al.}(2021)\citenamefont {Wang}, \citenamefont {Cano}, \citenamefont {Millis}, \citenamefont {Liu},\ and\ \citenamefont {Yang}}]{wang:cano:millis:liu:yang:2021}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Wang}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Cano}}, \bibinfo {author} {\bibfnamefont {A.~J.}\ \bibnamefont {Millis}}, \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Liu}},\ and\ \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Yang}},\ }\bibfield  {title} {\bibinfo {title} {{Exact Landau Level Description of Geometry and Interaction in a Flatband}},\ }\href {https://doi.org/10.1103/PhysRevLett.127.246403} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {127}},\ \bibinfo {pages} {246403} (\bibinfo {year} {2021})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ledwith}\ \emph {et~al.}(2023)\citenamefont {Ledwith}, \citenamefont {Vishwanath},\ and\ \citenamefont {Parker}}]{ledwith:vishwanath:parker:2023}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {P.~J.}\ \bibnamefont {Ledwith}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Vishwanath}},\ and\ \bibinfo {author} {\bibfnamefont {D.~E.}\ \bibnamefont {Parker}},\ }\bibfield  {title} {\bibinfo {title} {{Vortexability: A unifying criterion for ideal fractional Chern insulators}},\ }\href {https://doi.org/10.1103/PhysRevB.108.205144} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {108}},\ \bibinfo {pages} {205144} (\bibinfo {year} {2023})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Estienne}\ \emph {et~al.}(2023)\citenamefont {Estienne}, \citenamefont {Regnault},\ and\ \citenamefont {Cr\'epel}}]{estienne:regnault:valentin:2023}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Estienne}}, \bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Regnault}},\ and\ \bibinfo {author} {\bibfnamefont {V.}~\bibnamefont {Cr\'epel}},\ }\bibfield  {title} {\bibinfo {title} {{Ideal Chern bands as Landau levels in curved space}},\ }\href {https://doi.org/10.1103/PhysRevResearch.5.L032048} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Res.}\ }\textbf {\bibinfo {volume} {5}},\ \bibinfo {pages} {L032048} (\bibinfo {year} {2023})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mera}\ and\ \citenamefont {Ozawa}(2024)}]{mera:ozawa:2023}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}}\ and\ \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}},\ }\bibfield  {title} {\bibinfo {title} {{Uniqueness of Landau levels and their analogs with higher Chern numbers}},\ }\href {https://doi.org/10.1103/PhysRevResearch.6.033238} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Res.}\ }\textbf {\bibinfo {volume} {6}},\ \bibinfo {pages} {033238} (\bibinfo {year} {2024})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Regnault}\ and\ \citenamefont {Bernevig}(2011)}]{bernevig:2011}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Regnault}}\ and\ \bibinfo {author} {\bibfnamefont {B.~A.}\ \bibnamefont {Bernevig}},\ }\bibfield  {title} {\bibinfo {title} {{Fractional Chern Insulator}},\ }\href {https://doi.org/10.1103/PhysRevX.1.021014} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. X}\ }\textbf {\bibinfo {volume} {1}},\ \bibinfo {pages} {021014} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Neupert}\ \emph {et~al.}(2011)\citenamefont {Neupert}, \citenamefont {Santos}, \citenamefont {Chamon},\ and\ \citenamefont {Mudry}}]{titus:2011}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Neupert}}, \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Santos}}, \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Chamon}},\ and\ \bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Mudry}},\ }\bibfield  {title} {\bibinfo {title} {Fractional quantum hall states at zero magnetic field},\ }\href {https://doi.org/10.1103/PhysRevLett.106.236804} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {106}},\ \bibinfo {pages} {236804} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sheng}\ \emph {et~al.}(2011)\citenamefont {Sheng}, \citenamefont {Gu}, \citenamefont {Sun},\ and\ \citenamefont {Sheng}}]{sheng:2011}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {D.~N.}\ \bibnamefont {Sheng}}, \bibinfo {author} {\bibfnamefont {Z.-C.}\ \bibnamefont {Gu}}, \bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Sun}},\ and\ \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Sheng}},\ }\bibfield  {title} {\bibinfo {title} {Fractional quantum hall effect in the absence of landau levels},\ }\href {https://doi.org/10.1038/ncomms1380} {\bibfield  {journal} {\bibinfo  {journal} {Nature Communications}\ }\textbf {\bibinfo {volume} {2}},\ \bibinfo {pages} {389} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Sun}\ \emph {et~al.}(2011)\citenamefont {Sun}, \citenamefont {Gu}, \citenamefont {Katsura},\ and\ \citenamefont {Das~Sarma}}]{sun:2011}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {K.}~\bibnamefont {Sun}}, \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Gu}}, \bibinfo {author} {\bibfnamefont {H.}~\bibnamefont {Katsura}},\ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Das~Sarma}},\ }\bibfield  {title} {\bibinfo {title} {{Nearly Flatbands with Nontrivial Topology}},\ }\href {https://doi.org/10.1103/PhysRevLett.106.236803} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {106}},\ \bibinfo {pages} {236803} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Tang}\ \emph {et~al.}(2011)\citenamefont {Tang}, \citenamefont {Mei},\ and\ \citenamefont {Wen}}]{wen:2011}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Tang}}, \bibinfo {author} {\bibfnamefont {J.-W.}\ \bibnamefont {Mei}},\ and\ \bibinfo {author} {\bibfnamefont {X.-G.}\ \bibnamefont {Wen}},\ }\bibfield  {title} {\bibinfo {title} {{High-Temperature Fractional Quantum Hall States}},\ }\href {https://doi.org/10.1103/PhysRevLett.106.236802} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {106}},\ \bibinfo {pages} {236802} (\bibinfo {year} {2011})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ledwith}\ \emph {et~al.}(2020)\citenamefont {Ledwith}, \citenamefont {Tarnopolsky}, \citenamefont {Khalaf},\ and\ \citenamefont {Vishwanath}}]{ledwith:tarnopolsky:khalaf:vishwanath:2020}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {P.~J.}\ \bibnamefont {Ledwith}}, \bibinfo {author} {\bibfnamefont {G.}~\bibnamefont {Tarnopolsky}}, \bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Khalaf}},\ and\ \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Vishwanath}},\ }\bibfield  {title} {\bibinfo {title} {{Fractional Chern insulator states in twisted bilayer graphene: An analytical approach}},\ }\href {https://doi.org/10.1103/PhysRevResearch.2.023237} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Research}\ }\textbf {\bibinfo {volume} {2}},\ \bibinfo {pages} {023237} (\bibinfo {year} {2020})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Wang}\ and\ \citenamefont {Liu}(2022)}]{wang:liu:2022}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Wang}}\ and\ \bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Liu}},\ }\bibfield  {title} {\bibinfo {title} {{Hierarchy of Ideal Flatbands in Chiral Twisted Multilayer Graphene Models}},\ }\href {https://doi.org/10.1103/PhysRevLett.128.176403} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {128}},\ \bibinfo {pages} {176403} (\bibinfo {year} {2022})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Ledwith}\ \emph {et~al.}(2022)\citenamefont {Ledwith}, \citenamefont {Vishwanath},\ and\ \citenamefont {Khalaf}}]{ledwith:vishwanath:khalaf:2022}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {P.~J.}\ \bibnamefont {Ledwith}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Vishwanath}},\ and\ \bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Khalaf}},\ }\bibfield  {title} {\bibinfo {title} {{Family of Ideal Chern Flatbands with Arbitrary Chern Number in Chiral Twisted Graphene Multilayers}},\ }\href {https://doi.org/10.1103/PhysRevLett.128.176404} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {128}},\ \bibinfo {pages} {176404} (\bibinfo {year} {2022})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Liu}\ \emph {et~al.}(2025)\citenamefont {Liu}, \citenamefont {Mera}, \citenamefont {Fujimoto}, \citenamefont {Ozawa},\ and\ \citenamefont {Wang}}]{liu:mera:fujimoto:ozawa:wang:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {Z.}~\bibnamefont {Liu}}, \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}}, \bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Fujimoto}}, \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}},\ and\ \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Wang}},\ }\bibfield  {title} {\bibinfo {title} {{Theory of Generalized Landau Levels and Its Implications for Non-Abelian States}},\ }\href {https://doi.org/10.1103/1zg9-qbd6} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. X}\ }\textbf {\bibinfo {volume} {15}},\ \bibinfo {pages} {031019} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Fujimoto}\ \emph {et~al.}(2025)\citenamefont {Fujimoto}, \citenamefont {Parker}, \citenamefont {Dong}, \citenamefont {Khalaf}, \citenamefont {Vishwanath},\ and\ \citenamefont {Ledwith}}]{fujimoto:parker:dong:khalaf:vishwanath:ledwith:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {M.}~\bibnamefont {Fujimoto}}, \bibinfo {author} {\bibfnamefont {D.~E.}\ \bibnamefont {Parker}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Dong}}, \bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Khalaf}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Vishwanath}},\ and\ \bibinfo {author} {\bibfnamefont {P.}~\bibnamefont {Ledwith}},\ }\bibfield  {title} {\bibinfo {title} {{Higher Vortexability: Zero-Field Realization of Higher Landau Levels}},\ }\href {https://doi.org/10.1103/PhysRevLett.134.106502} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. Lett.}\ }\textbf {\bibinfo {volume} {134}},\ \bibinfo {pages} {106502} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Paiva}\ \emph {et~al.}(2025)\citenamefont {Paiva}, \citenamefont {Wang}, \citenamefont {Ozawa},\ and\ \citenamefont {Mera}}]{paiva:wang:ozawa:mera:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {C.}~\bibnamefont {Paiva}}, \bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Wang}}, \bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Ozawa}},\ and\ \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}},\ }\bibfield  {title} {\bibinfo {title} {{Geometrical responses of generalized Landau levels: Structure factor and the quantized Hall viscosity}},\ }\href {https://doi.org/10.1103/fbjs-tnm6} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {112}},\ \bibinfo {pages} {045106} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Onishi}\ \emph {et~al.}(2025)\citenamefont {Onishi}, \citenamefont {Avdoshkin},\ and\ \citenamefont {Fu}}]{onishi:avdoshkin:fu:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {Y.}~\bibnamefont {Onishi}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Avdoshkin}},\ and\ \bibinfo {author} {\bibfnamefont {L.}~\bibnamefont {Fu}},\ }\bibfield  {title} {\bibinfo {title} {{Geometric bound on the structure factor and a harmonic condition on band geometry}},\ }\href {https://doi.org/10.1103/8ng1-bwf6} {\bibfield  {journal} {\bibinfo  {journal} {Phys. Rev. B}\ }\textbf {\bibinfo {volume} {112}},\ \bibinfo {pages} {035158} (\bibinfo {year} {2025})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Yu}\ \emph {et~al.}(2025)\citenamefont {Yu}, \citenamefont {Lian},\ and\ \citenamefont {Ryu}}]{yu:lian:ryu:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {J.}~\bibnamefont {Yu}}, \bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Lian}},\ and\ \bibinfo {author} {\bibfnamefont {S.}~\bibnamefont {Ryu}},\ }\href {https://arxiv.org/abs/2509.05410} {\bibinfo {title} {{Wilson-Loop-Ideal Bands and General Idealization}}} (\bibinfo {year} {2025}),\ \Eprint {https://arxiv.org/abs/2509.05410} {arXiv:2509.05410 [cond-mat.mes-hall]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Fukui}(2025)}]{fukui:2025}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {T.}~\bibnamefont {Fukui}},\ }\href {https://arxiv.org/abs/2511.07855} {\bibinfo {title} {{Gradient flow and Bogomolny bounds for quantum metric actions}}} (\bibinfo {year} {2025}),\ \Eprint {https://arxiv.org/abs/2511.07855} {arXiv:2511.07855 [cond-mat.mes-hall]} \BibitemShut {NoStop}%
\bibitem [{\citenamefont {Morrison}(1986)}]{morrison:1986}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {P.~J.}\ \bibnamefont {Morrison}},\ }\bibfield  {title} {\bibinfo {title} {{A paradigm for joined Hamiltonian and dissipative systems}},\ }\href@noop {} {\bibfield  {journal} {\bibinfo  {journal} {Physica D: Nonlinear Phenomena}\ }\textbf {\bibinfo {volume} {18}},\ \bibinfo {pages} {410} (\bibinfo {year} {1986})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Hokayem}\ and\ \citenamefont {Gallestey}(2020)}]{hokayem:gallestey:2020}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {P.~A.}\ \bibnamefont {Hokayem}}\ and\ \bibinfo {author} {\bibfnamefont {E.}~\bibnamefont {Gallestey}},\ }\href {https://people.ee.ethz.ch/~apnoco/Lectures2020/2020_NLSC_lecture_notes.pdf} {\bibinfo {title} {{Lecture Notes on Nonlinear Systems and Control}}},\ \bibinfo {howpublished} {Lecture notes, ETH Zurich, Spring Semester 2020} (\bibinfo {year} {2020})\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Mera}\ \emph {et~al.}(2022)\citenamefont {Mera}, \citenamefont {Zhang},\ and\ \citenamefont {Goldman}}]{mera:goldman:zhang:2022}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {B.}~\bibnamefont {Mera}}, \bibinfo {author} {\bibfnamefont {A.}~\bibnamefont {Zhang}},\ and\ \bibinfo {author} {\bibfnamefont {N.}~\bibnamefont {Goldman}},\ }\bibfield  {title} {\bibinfo {title} {{Relating the topology of Dirac Hamiltonians to quantum geometry: When the quantum metric dictates Chern numbers and winding numbers}},\ }\href {https://doi.org/10.21468/SciPostPhys.12.1.018} {\bibfield  {journal} {\bibinfo  {journal} {SciPost Phys.}\ }\textbf {\bibinfo {volume} {12}},\ \bibinfo {pages} {18} (\bibinfo {year} {2022})}\BibitemShut {NoStop}%
\bibitem [{\citenamefont {Breuer}\ and\ \citenamefont {Petruccione}(2002)}]{breuer:petruccione:2002}%
  \BibitemOpen
  \bibfield  {author} {\bibinfo {author} {\bibfnamefont {H.-P.}\ \bibnamefont {Breuer}}\ and\ \bibinfo {author} {\bibfnamefont {F.}~\bibnamefont {Petruccione}},\ }\href {https://doi.org/10.1093/acprof:oso/9780199213900.001.0001} {\emph {\bibinfo {title} {{The Theory of Open Quantum Systems}}}}\ (\bibinfo  {publisher} {Oxford University Press},\ \bibinfo {address} {Oxford, UK},\ \bibinfo {year} {2002})\BibitemShut {NoStop}%
\end{thebibliography}%