Abstract
We develop a nonequilibrium increment method in quantum Monte Carlo simulations to obtain the Rényi entanglement entropy of various quantum manybody systems with high efficiency and precision. To demonstrate its power, we show the results on a few important yet difficult (2 + 1)d quantum lattice models, ranging from the Heisenberg quantum antiferromagnet with spontaneous symmetry breaking, the quantum critical point with O(3) conformal field theory (CFT) to the toric code \({{\mathbb{Z}}}_{2}\) topological ordered state and the Kagome \({{\mathbb{Z}}}_{2}\) quantum spin liquid model with frustration and multispin interactions. In all these cases, our method either reveals the precise CFT data from the logarithmic correction or extracts the quantum dimension in topological order, from the dominant area law in finitesize scaling, with very large system sizes, controlled errorbars, and minimal computational costs. Our method, therefore, establishes a controlled and practical computation paradigm to obtain the difficult yet important universal properties in highly entangled quantum matter.
Similar content being viewed by others
Introduction
Entanglement entropy (EE) is a physical quantity that measures the quantum entanglement in an interacting manybody system. The scaling form of the entanglement entropy contains universal information of the system and can be used to characterize quantum phases and phase transitions in interacting lattice models. For these reasons, the understanding and computation of EE attract much attention from the condensed matter, quantum information, and highenergy communities. Significant theoretical progress have been made to determine the universal terms in the EE in different kinds of quantum phases, including quantum critical systems, gapped phases in (2 + 1)d, and systems with spontaneous continuous symmetry breaking^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23}, and on the computational front the Quantum Monte Carlo methods have proven to be a reliable way to numerically measure the Rényi entanglement entropy^{10,20,22,24,25,26,27,28}.
We now briefly review what has been known about the scaling form of EE in (2 + 1)d, which is the main focus of this work. Suppose the system is partitioned into regions A and \(\bar{A}\), and denote the length of the boundary of A by l. For critical or gapped systems, the entanglement entropy (both von Neumann and Rényi) takes the following form when A is sufficiently large:
Here a is a nonuniversal area law coefficient. The coefficient of the logarithmic correction s is a universal number determined by the geometric properties of the partition (i.e., if A is a rectangle surrounded by \(\bar{A}\)) and intrinsic properties of the underlying physical system (i.e., central charge of the conformal field theory at certain critical points)^{3,4}. When s = 0, the constant γ becomes universal and when the system is gapped, it becomes independent of the shape of A and is known as the topological entanglement entropy (TEE). In that case, the value of γ is given by the logarithm of the total quantum dimension of the underlying topological order^{5,6}.
It is of great interest to obtain the values of the universal coefficients s and γ since they provide access to universal properties of the system that are hard to measure otherwise. Extracting s or γ requires finitesize scaling with l, which is challenging given that the calculation of EE for an interacting manybody system is already a difficult task in general. Using quantum Monte Carlo (QMC) simulations the numerical study of the scaling of 2nd Rényi EE has been carried out in lattice models that exhibit e.g., spontaneous O(N) symmetry breaking and related critical points^{10,17,18,19,20,21,22,23,26,27,29}, and also in models with \({{\mathbb{Z}}}_{2}\) topological order^{12,13,30}. However, despite all the progress numerical computation of EE in QMC remains difficult due to the lack of a stable estimator, especially when lattice models have multispin interactions or frustrations, not to mention the even more complex fermionic models where the computational complexity scales with the system size to a higher power^{25,31}. The high numerical cost of measuring EE mainly stems from the fact that to calculate the Rényi EE one has to enlarge the configuration space to create replicas, and perform partial connection of the partition functions between the replicas during the sampling processes.
These are the difficulties we set to overcome in this work. Here, by combining the nonequilibrium measurements of entanglement entropy based on the Jarzynski equality^{24,28,32} and the increment trick swapping replicas^{10,26}, we developed a new nonequilibrium increment method, that makes use of the divideandconquer procedure of the nonequilibrium process to improve the speed of the simulation and the data quality of the entanglement measurement. We demonstrate the strength and versatility of the method with a few representative examples of (2 + 1)d quantum manybody lattice model systems, in which the EEs are notoriously hard to compute. We explicitly show that previous methods are not able to extract the universal corrections beyond the leading arealaw contribution in the scaling of EE, due to the lack of efficiency and precision in the EE computation.
In the QMC computation of the 2nd Rényi EE, the shape of spacetime configuration, as shown in Fig. 5, is a twosheeted Riemann surface for 1D systems. Such a topological unit provides a stable and efficient way of carrying out the computation both conceptually and technically. In general, the nth order Rényi entropy can be obtained by an nsheeted Riemann surface, which can also help to reconstruct the lowlying entanglement spectrum for quantum manybody systems^{33}.
The structure of the paper is organized in the following manner. Firstly, we show the results of finitesize scaling of EE on three representative nontrivial systems, all in (2 + 1)d, starting from the Néel state of the antiferromagnetic Heisenberg model with spontaneously broken continuous symmetry, to the quantum critical point of O(3) CFT and eventually arriving at the scaling form on toric code toy model with \({{\mathbb{Z}}}_{2}\) topological order and Kagome \({{\mathbb{Z}}}_{2}\) quantum spin liquid state on a torus geometry, where in the former we also compared the results with densitymatrix renormalization group (DMRG) on cylinder geometry. In both cases, the TEE of \(2\gamma =2\ln (2)\) is obtained unambiguously in QMC lattice model simulations. And in all these cases, our algorithm provides reliable data with high efficiency and very large system sizes. Then we conclude the results with some immediate directions for further research. Lastly, we explain in detail the methodology of our algorithm, which is the nonequilibrium increment estimator for the Rényi EE^{24,27} with high efficiency and precision.
Results
In this section, we demonstrate the strength of our algorithm with three representative examples. (i) Entanglement entropy at the quantum phases with spontaneous continuous symmetry breaking; (ii) Entanglement entropy at the quantum critical point with conformal field theory description; (iii) Entanglement entropy at the topological order phase with fractional excitations. In each case we will show that the universal correction (s or γ) can be reliably determined using our method.
Entanglement of spontaneous continuous symmetry breaking state
For systems with spontaneously broken continuous symmetry, it is known that the EE follows the scaling form of Eq. (1) with the coefficient of the logarithmic correction s = − N_{G}(d − 1)/2, where N_{G} is the number of Goldstone modes and d is the spatial dimension^{11}. The logarithmic correction originates from the interplay between gapless Goldstone modes and restoration of the symmetry in a finite system.
We consider a square lattice antiferromagnetic Heisenberg model, whose ground state breaks the SU(2) spin rotational symmetry in the thermodynamic limit. The number of Goldstone modes is N_{G} = 2 and therefore we expect s = −1. We measure \({S}_{A}^{(2)}(L)\), the 2nd Rényi entropy for an entangling region A. The simulation setup is shown in the inset of Fig. 1. The model is defined on a L × L/2 lattice with periodic boundary conditions or a L × L/2 torus, and the entangling region A is chosen to be a L/2 × L/2 cylinder on the torus, in such a way that there are no corners between A and \(\overline{A}\), and the boundary of A is of length (volume) L. \({S}_{A}^{(2)}(L)\) is computed using the finitetemperature stochastic series expansion (SSE) QMC^{34,35} for different L, which is then fitted to Eq. (1). From the fitting with L ∈ [40, 160], we find s = − 1.00(9), in good agreement with the theoretical prediction and a previous numerical measurement with the sequential nonequilibrium method^{24}. However, we emphasize that unlike ref. ^{24} here we do not need to use the valence bond basis in the QMC simulation to suppress the thermal fluctuations. The more standard finitetemperature simulation 1/T = L in the S^{z} basis already suffices to successfully achieve the desired data quality with high efficiency.
Entanglement at (2 + 1)d quantum critical point
The second example to test the performance of our method is a lattice model that realizes the O(3) quantum critical point (QCP). For this purpose, following previous literature^{27,36,37}, we consider the square lattice J_{1}J_{2} Heisenberg model (the columnar dimer lattice model), illustrated in the inset of Fig. 2. The Hamiltonian reads
where 〈ij〉 denotes the thin J_{1} bonds and \({\langle ij\rangle }^{\prime}\) denotes the thick J_{2} bonds. The QCP located at \({({J}_{2}/{J}_{1})}_{c}=1.90951(1)\)^{36} is known to fall within the (2 + 1)d O(3) universality class.
In the model, the presence of strong J_{2} and weak J_{1} bonds breaks the lattice translation symmetry. Because of the translation symmetry breaking, we choose the entangling region A so that its boundary avoids strong J_{2} bonds to reduce the finitesize error in the scaling behavior of the entanglement entropy. A similar strategy has also been adopted in recent studies of the disorder operator in the same model^{37}.
We employ the torus geometry of L × L and choose a rectangle entangling region of size L/2 × L/2 with four corners(shown in the inset of Fig. 2). We measure \({S}_{A}^{(2)}(L)\) at \({({J}_{2}/{J}_{1})}_{c}\) at 1/T = L and monitor its scaling behavior. The results are shown in Fig. 2. Here again, by fitting with the form in Eq. (1), we obtain the coefficient of the logarithmic correction s = 0.081(4). This value is again consistent with the theoretical prediction for the O(3) CFT and previous numerical results on (2 + 1)d Ising, XY and Heisenberg QCPs^{17,21,22,29,37}. And our method can reach much larger systems sizes and has better data quality than the previous numerical results.
Entanglement of topological ordered state
In a fully gapped ground state, the entanglement entropy S for a region A generally satisfies S = al − γ + ⋯ where γ is the TEE. For simplyconnected regions, \(\gamma =\ln {{{\mathcal{D}}}}\) where \({{{\mathcal{D}}}}\) is the total quantum dimension of the underlying topological order. For \({{\mathbb{Z}}}_{2}\) toric code we have \({{{\mathcal{D}}}}=2\). Extracting TEE from numerical simulations of interacting lattice models has proven to be a challenging task. In the following, we present our results for TEE in two different lattice models exhibiting \({{\mathbb{Z}}}_{2}\) topological order. As will be shown below, our algorithm can go beyond limitations in previous MonteCarlo studies and the expected value of TEE is obtained in the BalentsFisherGirvin kagome spin1/2 model, which unambiguously proves that the ground state is a \({{\mathbb{Z}}}_{2}\) spin liquid. For clarify, in this section, we fix \(\gamma =\ln 2\).
Toy model
To test the efficiency of our algorithm on systems with topological order, we start with a toy model of \({{\mathbb{Z}}}_{2}\) topological phase on a 2d checkerboard lattice^{38,39,40}, with the following Hamiltonian,
with \({B}_{\blacksquare}\,\equiv {\prod }_{i\in \blacksquare}{\sigma }_{i}^{z}\) and ■ being the redcolored plaquettes in Fig. 3a. Note that this Hamiltonian has an extensive number of conserved quantities:
where □ are the white plaquettes in Fig. 3a. In fact, − ∑_{◼}B_{◼} − ∑_{□}A_{□} is the celebrated toric code model.
Below, we will combine finitesize DMRG^{41} and QMC methods, to show the lowenergy physics of this model [c.f. Eq. (3)] is equivalent to the wellknown toric code model under transverse field^{13},
Our DMRG simulations reveal 〈A_{□} 〉 = 1 for the range of h we consider in this section, in the lowtemperature regime of H_{toy}. Thus the ground state wavefunctions of H_{toy} and H_{TCM} are identical for J_{s} > 0. In the toric code model, the transverse field induces dynamics of ■plaquette excitations. As h increases, such excitations condense and drive a transition to a trivial confined phase. Thus, the toy model H_{toy} should also experience a continuous phase transition at h = h_{c} ≃ 0.333^{42}, from a \({{\mathbb{Z}}}_{2}\) deconfined phase to a confined phase.
With DMRG, the ground state properties of H_{toy} are computed on a L_{x} × L_{y} cylinder geometry with open/periodic boundary condition along the horizontal(x)/vertical(y) direction. We can directly calculate the von Neumann entropy
where the reduced density matrix \({\rho }_{{{{\mathcal{A}}}}}={{{{\rm{tr}}}}}_{{{{\mathcal{B}}}}}\left\psi \right\rangle \left\langle \psi \right\) and the subsystems \({{{\mathcal{A}}}}\) and \({{{\mathcal{B}}}}\) are both cylinders of size \(\frac{{L}_{x}}{2}\times {L}_{y}\). Figure 3b shows that the plaquette energy 〈B_{◼}〉s monotonically decrease as the transverse field h grows, and a change of curve can be seen around the dashed vertical line, which highlights the point h = h_{c} ≃ 0.333 (the inset shows \(\frac{{{{\rm{d}}}}\langle {B}_{\blacksquare}\rangle }{{{{\rm{d}}}}h}\) and a dip is found around h ≃ 0.338 for both L_{y} = 10 and 12 systems). In Fig. 3c, for 72 × 10 and 72 × 12 cylinders, S^{vN}s are shown versus h and exhibit peaks around h_{c}. The above DMRG data strongly suggest the expected deconfinementconfinement transition at h = h_{c} ≃ 0.333, consistent with the literature^{13,42}.
To further extract the nature of the lowh phase, we perform finitesize scaling and extrapolate the topological entanglement entropy (TEE), i.e., γ in Eq. (1), both in the cylinder geometry for DMRG and torus geometry with our QMC algorithm. As shown in Fig. 3d, we measure both von Neumann entropy S^{vN} and Rényi entropy S^{(2)} in cylinder geometry via DMRG simulations. With circumferences L_{y} = 6, 8, 10, 12 and fixed length L_{x} = 72, the entanglement entropies are measured and extrapolated with boundary length l = 2L_{y}, where the extrapolated value of γ = 0.69(1) is obtained. For the torus geometry of size L × L, the Rényi entropy versus l = 2L is also measured in QMC simulations, and the extrapolation gives 2γ = 1.35(8). Both of the results are numerically indistinguishable from \(\gamma =\ln 2\approx 0.693\).
Minimal entropy state
We now discuss an important subtlety in our prescription for extracting TEE. Since TEE is the correction to the entanglement area law, to directly extract TEE complex prescriptions on planar partitions have been designed^{5,6} to cancel the leading arealaw contributions as well as possible corrections from corners. However, when numerically implemented on lattice such prescriptions typically suffer from severe finitesize effects because they require several nonoverlapping regions. As an example, the LevinWen prescription has been employed to study the TEE of the \({{\mathbb{Z}}}_{2}\) quantum spin liquid state, which in principle should yield a universal correction of \(2\gamma =2\ln 2\). The best results by now only find \(\ln 2\) instead of \(2\ln 2\)^{12,30}. This is because within the allowed system sizes for largescale numerical simulations, the size of each of the four partitions is at most l ~ 10, which is too small compared with the characteristic lengthscale of the problem (inverse vison gap^{43}, as discussed in the next section), and the cancellation of the entanglement entropies between different partitions are too noisy to give the TEE in a controlled manner.
To this end, the kind of entanglement cut adopted in our simulations, i.e., bipartition of a torus into two cylinders, has the advantage of maximally enlarging the entangling region while avoiding corners on the boundary. Naively, the TEE should be 2γ since the boundary has two disconnected components. However, because the boundary curve is topologically nontrivial (noncontractible cycles on a torus), the value of the TEE now depends on which ground state is used for the calculation of the entanglement entropy. More specifically, for \({{\mathbb{Z}}}_{2}\) toric code and a given entanglement cut (e.g., along the y direction), only certain choices of ground states on torus yield the expected value \(2\gamma =2\ln 2\). Any other ground state gives a smaller γ and thus a larger S for the same size. For this reason, these special ground states which saturate \(2\gamma =2\ln 2\) are called minimum entropy states (MES)^{44,45}. Physically, the MES are characterized by a definite anyon flux through the noncontractible entanglement cut, so are in onetoone correspondence with anyon excitations of the topological order.
Numerical simulations in DMRG for the toric code toy model in the previous section and quantum spin liquid system^{13}, have shown that due to the energy minimization process in the DMRG for quasionedimensional systems with finite accuracy, the algorithm actually finds the desired MES with \(\gamma =\ln 2\) for the cylinder geometry (also shown in ref. ^{13}) as shown in Fig. 7d.
Our numerical results in the previous section for \({{\mathbb{Z}}}_{2}\) toric code toy model suggest that the Monte Carlo sampling processes in the our algorithm serve the same purpose in finding the expected TEE of \(2\gamma =2\ln 2\), as shown in Fig. 3d. In the next section, we further strengthen this point using an even more nontrivial test of the kagome \({{\mathbb{Z}}}_{2}\) spin liquid lattice model.
\({{\mathbb{Z}}}_{2}\) quantum spin liquid on kagome lattice
The last and yet the most challenging case for the entanglement entropy measurement is that of the BalentsFisherGirvin (BFG) model with \({{\mathbb{Z}}}_{2}\) quantum spin liquid (QSL) ground state^{12,46,47,48,49,50,51,52,53}. As far as we are aware of, the expected value of the TEE for such Kagome \({{\mathbb{Z}}}_{2}\) QSL model has never been observed in previous QMC computations of the 2nd Rényi EE.
As illustrated in Fig. 4a, b, the model is defined on a Kagome lattice with the following Hamiltonian
where J_{±} is the ferromagnetic transverse nearest neighbor interaction and J_{z} is the antiferromagnetic longitudinal interactions between any two spins in the hexagon of the Kagome plane. The model is known from previous intensive QMC simulations to host a \({{\mathbb{Z}}}_{2}\) QSL and the transition from the ferromagnetic phase to the \({{\mathbb{Z}}}_{2}\) QSL occurs at \({({J}_{\pm }/{J}_{z})}_{c}=0.07076(2)\)^{47,52}. The identification of a \({{\mathbb{Z}}}_{2}\) QSL ground state for \(({J}_{\pm }/{J}_{z}) < {({J}_{\pm }/{J}_{z})}_{c}\) has been supported by several different types of measurements, such as the unconventional quantum phase transition between ferromagnetic phase and the \({{\mathbb{Z}}}_{2}\) QSL belonging to the (2 + 1)d XY^{*} (instead of XY) universality class^{47,48,51}, signifying the existence of fractional anyon excitations in the \({{\mathbb{Z}}}_{2}\) QSL phase, and the visonpair spectra with translation symmetry fractionalization^{50}, the vestigial anyon condensation transition towards other topological ordered state^{53} and the fractional conductivity at the (2 + 1)d XY^{*} critical point^{52}.
This system is also relevant for ongoing experimental efforts in synthesizing and identifying QSL materials in the Kagome based quantum magnets such as Znparatacamite Zn_{x}Cu_{4−x}(OH)_{6}Cl_{2}(0 ≤ x ≤ 1) with Herbertsmithite its full Zn end^{54,55,56,57,58} and Zndoped Barlowite Zn_{x}Cu_{4−x}(OH)_{6}FBr and Zndoped Claringbullite Zn_{x}Cu_{4−x}(OH)_{6}FCl (0 ≤ x ≤ 1)^{59,60,61,62,63} with NMR^{59}, neutron scattering^{64,65}, μSR^{65} and thermodynamic measurements^{66,67}. Better theoretical characterization of the \({{\mathbb{Z}}}_{2}\) QSL states with fundamental probe such as the EE, and its future connection with the experimental reality such as the existences of magnetic impurities^{67}, will certainly help to eventually reveal the existence of fractionalized excitations and anyonic statistics in Kagome based quantum magnets^{68,69}.
However, previous attempts of measuring the second Rényi EE in this model^{12} and other similar frustrated Kagome spin model^{30} using the LevinWen prescription, were not successful in revealing the actual value of the TEE \(2\gamma =2\ln 2\) in Eq. (1). As discussed in the previous section, although the LevinWen construction^{5,6} can in principle remove the area law contribution and expose the universal constant correction of 2γ, what has been observed at best is a plateau of approximately γ for finitesize systems (e.g., 3 × 8 × 8 with 3 sites per unit cell of the Kagome lattice), at an intermediate temperature below the spinon energy scale ~ J_{z}, but still comparable or higher than the vison energy scale of \(\sim {J}_{\pm }^{3}/{J}_{z}^{2}\). It is understood that the data quality and the computational complexity in simulating even larger sizes and lower temperatures prohibit more precise determination of the 2nd Rényi EE of the system^{12}.
We found that our our algorithm successfully overcomes these difficulties. We consider two kinds of geometry as shown in Fig. 4a, b, with periodic boundary conditions in both lattice directions and aspect ratios L_{x}/L_{y} = 2 and L_{x}/L_{y} = 1. We choose a cylindrical entangling region A without corners but with two disconnected boundaries. In this way, the second Rényi entropy (for a MES) should scale as
where 2L_{y} is now the total length of the boundary and the TEE is \(2\gamma =2\ln 2\approx 1.386\). We carried out the nonequilibrium increment measurement with N = 240 parallel pieces and gradually increase the system sizes L_{x} = L (so the two torus are L × L/2 and L × L, respectively), with J_{±}/J_{z} = 0.0625 (inside the QSL phase) and T = 1/480 (below the anyon gaps)^{50,52}.
The results are shown in Fig. 4c. One can see that as L increases, \({S}_{A}^{(2)}(L)\) clearly develops a linear behavior, with a converged slope (α in Eq. (6)) and more importantly, converged intercept 1.4(2), for both systems with different aspect ratios. It is worth noting that the converged behavior with system sizes starts to emerge at around l ~ 12, which may explain why previous QMC works on the same system^{12} fail to extract the value of \(2\ln 2\). This observation shows the importance of having access to sufficiently large system sizes to be able to observe the full TEE. Such largescale computation of entanglement entropy only becomes possible because of our improved algorithm, which can be easily and robustly implemented for such complicated models, resulting in significantly improved performance. Our results thus unambiguously demonstrate the existence of \({{\mathbb{Z}}}_{2}\) QSL in the system.
As mentioned in the previous section, despite potential complications known theoretically^{44,45}, owing to the sampling selection mechanism built into the Monte Carlo processes, our algorithm, based on the numerical evidence, indeed finds the MES TEE, similar to the DMRG application on the other \({{\mathbb{Z}}}_{2}\) spin liquid model^{13}.
Discussion
The measurement of entanglement entropy differs fundamentally from more traditional probes such as order parameters, structure factors and various correlation functions, being able to reveal more subtle global information hidden in the wave function of quantum manybody systems. However, the precise detection of EE has been very hard for the past decades. Here we show that previously difficult task can be greatly optimized and improved via the nonequilibrium increment method, and in this way, one can investigate the scaling behavior of EE in many (2 + 1)d quantum manybody systems with very large system sizes, controlled errorbars, and minimal computational costs. Starting from the three representative cases shown here, one can foresee the implementation and measurement of EE via our algorithm for other topological ordered phases and phase transitions, interacting fermionic systems such as the Gross–Neveu QCPs with critical Dirac fermions^{70}, the deconfined QED_{3} problems of gauge fields coupled to fermion matter fields^{71,72,73} and the more complicated situations of nonFermiliquid and quantum critical metals^{74,75,76,77,78,79,80,81,82} and hopefully make further suggestions to the ongoing experimental search for these strongly entangled quantum matter.
At last we want to mention that, except for the measurement of EE, in recent years new measurements such as the symmetry domain walls or field lines of emergent gauge field and disorder operators to probe and characterize phases and phase transitions and the associated condensation of extended objects, spontaneously breaking of the socalled higherform symmetry^{29,37,83,84,85,86,87,88,89,90,91,92,93} have been successfully designed and implemented in many exotic quantum manybody systems. For example, the recent measurement of disorder operator at the deconfined quantum critical point (DQCP) has unambiguously exhibited the difference between the DQC and other QCPs with unitary CFT description^{27,37}, as is also seen in the measurement of entanglement entropy^{27}. These new measurements are very hopeful to extract more fundamental information and lead to a deeper understanding of the interacting many body systems.
Methods
Replica trick
In a quantum manybody system, the entanglement of a subsystem A with the rest of the system \(\overline{A}\) is most commonly identified by the vonNeumann entropy \({S}_{A}^{(vN)}={{{\rm{Tr}}}}{\rho }_{A}\ln {\rho }_{A}\). Here \({\rho }_{A}={{{{\rm{Tr}}}}}_{\overline{A}}\rho\) is a reduced density matrix defined as the partial trace of the total density matrix ρ over a complete basis of subsystem \(\overline{A}\). The calculation of the vonNeumann entropy directly from the reduced density matrix in QMC simulations is difficult as usually one does not have the wave function of a generic quantum manybody system, say, in (2 + 1)d.
However, the Rényi entropy^{2,3,94}
which can be regarded as a generalization of the vonNeumann entropy, approaching the latter when n → 1, can be calculated by the QMC method. As \({{{{\mathcal{Z}}}}}^{(n)}={[{{{\rm{Tr}}}}({e}^{\beta {{{\mathcal{H}}}}})]}^{n}\) is the ordinary partition function of n replicas of the system while \({{{{\mathcal{Z}}}}}_{A}^{(n)}\) is a modified partition function with the boundary conditions of area A of the n replicas changed according to the value of n. In the case of n = 2 and in the 1D model, \({{{{\mathcal{Z}}}}}_{A}^{(2)}\) is two replicas with area A glued together as depicted in Fig. 5. However, for sites in \(\overline{A}\) for each replica, the usual periodical boundary condition is maintained.
Based on Eq. (7), various estimators have been introduced to calculate the Rényi entanglement entropy^{10,20,26}. However, the data quality of these estimators has severely limited their usage in higher dimensions and larger sizes. And for all the previous methods mentioned, the obtainable system sizes of the 2nd Rényi entropy of spin lattice models are very limited. There is one empirical explanation for this limitation. All the previous methods somehow measure the observable \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}}{{{{{\mathcal{Z}}}}}^{(n)}}\) directly to obtain the Rényi entropy in the equilibrated ensemble. For interacting spin lattice models, \({S}_{A}^{(n)}\) typically obeys an area law with respect to the boundary length l, so \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}}{{{{{\mathcal{Z}}}}}^{(n)}}\) should decay exponentially with the system size L (typically we choose l ∝ L in 2D systems). As a result, the observable \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}}{{{{{\mathcal{Z}}}}}^{(n)}}\) will be too small to be obtained precisely and efficiently when we increase the system size.
The recent proposal of the nonequilibrium method^{24,28}, which borrows Jarzynski’s equality^{32} of a nonequilibrium process, has made substantial progress in this regard by conducting the measurement out of the equilibrium protocol. It can obtain the 2nd Rényi entanglement with improved data quality for larger system sizes. The success of this method, as we will discuss in the next section, might owe to the fact that in the nonequlibrium protocol we are not taking \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}}{{{{{\mathcal{Z}}}}}^{(n)}}\) as a direct observable in an equilibrated ensemble, but measuring the total work \({W}_{A}^{(n)}\) done in a nonequilibrium tunning process which is of the same order of \({S}_{A}^{(n)}\) according to Eq. (12). But the disadvantage of this method, as we will show in the following sections, is the tunning process should not be too fast. However, this shortcoming can be overcome by the increment trick which gives rise to a nonequilibrium increment method, making it possible to obtain the Rényi entropy more precisely and efficiently for the more complicated quantum manybody systems with frustration, multispin interactions, etc. We will give more details of the nonequilibrium protocol and the comparison of our method with the original proposal in the subsequent sections.
Nonequilibrium measurement
In this section the nonequilibrium method^{24} will be reviewed as the foundation of our further development. The first step of the nonequilibrium method is to introduce a partition function \({{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )\) parameterized by λ. \({{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )\) is defined as the sum of a collection of partition functions \({Z}_{B}^{(n)}\) weighted by a binomial distribution \({g}_{A}\left(\lambda ,{N}_{B}\right)={\lambda }^{{N}_{B}}{(1\lambda )}^{{N}_{A}{N}_{B}}\). According to this definition, \({{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )\) can be written as
where \({{{{\mathcal{Z}}}}}_{A}^{(n)}(1)={{{{\mathcal{Z}}}}}_{A}^{(n)}\) and \({{{{\mathcal{Z}}}}}_{A}^{(n)}(0)={{{{\mathcal{Z}}}}}_{\varnothing }^{(n)}\). With this definition, the Rényi entanglement entropy can be rewritten as \({S}_{A}^{(n)}=\frac{1}{n1}\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}(1)}{{{{{\mathcal{Z}}}}}_{A}^{(n)}(0)}\), which leads to an integral expression
where \(\frac{\partial \ln {{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )}{\partial \lambda }\) can be measured by \({\langle \frac{\partial \ln {g}_{A}}{\partial \lambda }\rangle }_{\lambda }={\langle ({N}_{B}/\lambda )({N}_{A}{N}_{B})/(1\lambda )\rangle }_{\lambda }\). Ref. ^{24} puts forward a nonequilibrium process where the system evolves from a configuration of \({{{{\mathcal{Z}}}}}_{\varnothing }^{(n)}({{{{\mathcal{Z}}}}}_{A}^{(n)}(0))\) to a configuration of \({{{{\mathcal{Z}}}}}_{A}^{(n)}({{{{\mathcal{Z}}}}}_{A}^{(n)}(1))\) through tunnning from λ = 0 to λ = 1. The total work done in this process is
where λ(t_{i}) = 0 and λ(t_{f}) = 1.
According to the Jarzynski’s equality^{32}, the free energy difference ΔF accumulated in such a path can be extracted by the ensemble average of nonequilibrium measurement by the following relation:
the overbar refers to the average of an ensemble of nonequilibrium paths. As the free energy for a canonical ensemble can be expressed by \(F={\beta }^{1}\ln {{{\mathcal{Z}}}}\), the Rényi entanglement entropy can then be estimated by
where \({S}_{A}^{(n)}\) is rewritten in the free energy difference of ensemble \({{{{\mathcal{Z}}}}}_{A}^{(n)}(0)\) with ensemble \({{{{\mathcal{Z}}}}}_{A}^{(n)}(1)\).
Note that although λ(t) can take different forms, in this paper, we only consider the case where λ varies uniformly with t. In this case, \({W}_{A}^{(n)}\) can be calculated by accumulating \({{\Delta }}\ln {g}_{A}(\lambda ({t}_{m}),{N}_{B}({t}_{m}))=\ln {g}_{A}(\lambda ({t}_{m+1}),{N}_{B}({t}_{m}))\ln {g}_{A}(\lambda ({t}_{m}),{N}_{B}({t}_{m}))\) or alternatively recording the value of \(\frac{{g}_{A}(\lambda ({t}_{m+1}),{N}_{B}({t}_{m}))}{{g}_{A}(\lambda ({t}_{m}),{N}_{B}({t}_{m}))}\) at each time and multiplying them at the end of the measurement.
Nonequilibrium increment method
Above is the basic concept of the nonequilibrium measurement, our optimization of this method takes advantage of the fact that the partition function \({{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )\) is well defined at every λ ∈ [0, 1], so that we can insert a set of identity elements \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}(kh)}{{{{{\mathcal{Z}}}}}_{A}^{(n)}(kh)}(k=1,2,...,N1)\) and rewrite \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}(1)}{{{{{\mathcal{Z}}}}}_{A}^{(n)}(0)}\) as
where N is an integer and \(h=\frac{1}{N}\). The increment trick is discussed in ref. ^{26} as a way to overcome the limitation of equilibrium measurement of the Rényi entanglement entropy with large entangling regions. Our algorithm is the nonequilibrium version of the increment trick. According to Eqs. (13) and (9) the Rényi entanglement entropy can be reexpressed as
Jarzynski’s equality^{32} can be carried out on each piece \(\frac{{{{{\mathcal{Z}}}}}_{A}^{(n)}((k+1)h)}{{{{{\mathcal{Z}}}}}_{A}^{(n)}(kh)}(k=1,2,...,N1)\) and the corresponding small piece of integral \(\int\nolimits_{kh}^{(k+1)h}d\lambda \frac{\partial \ln {{{{\mathcal{Z}}}}}_{A}^{(n)}(\lambda )}{\partial \lambda }\). Then we obtain
where \({W}_{k,A}^{(n)}\)s are defined same as Eq. (10) but λ(t_{i}) = kh and λ(t_{f}) = (k + 1)h for each small piece in the nonequilibrium process, as shown in Fig. 6. In this way, our algorithm follows the protocol:

1.
We perform quantum MonteCarlo simulation on the partition function \({{{\mathcal{Z}}}}\) and store the thermalized QMC configuration (one replica) for later use.

2.
Prepare two replica configurations as the thermalized configuration of \({{{{\mathcal{Z}}}}}_{\varnothing }^{(2)}\) and then send them to N parallel processes.

3.
As shown in Fig. 6, for process k we set the initial value of λ to be λ(t_{i}) = kh. The value of λ controls the probability of sites in A (the entanglement region) joining or leaving the glued geometry of the replicas, with the following probabilities:
$${P}_{{{{\rm{join}}}}}=\min \left\{\frac{\lambda }{1\lambda },1\right\}\quad {P}_{{{{\rm{leave}}}}}=\min \left\{\frac{1\lambda }{\lambda },1\right\}.$$(16)Each MC sweep then consists of the following steps:

Each site in A can choose whether to stay or leave the region according to Eq. (16). After the decision is made, change the topology, i.e., update the connectivity of the entangling region A.

After the trace structure is determined, carry out the MC updates on the replicas.


4.
At this step, we fix λ = kh for process k and conduct several MC sweeps to thermalize the configuration at the beginning of the nonequilibrium measurement.

5.
Start the nonequilibrium measurement. Increase the value of λ by Δλ and record the value of \(\frac{{g}_{A}(\lambda ({t}_{m+1}),{N}_{B}({t}_{m}))}{{g}_{A}(\lambda ({t}_{m}),{N}_{B}({t}_{m}))}\). Here λ(t_{m}) = kh + mΔλ and λ(t_{0}) = λ(t_{i} = kh). Then carry out a MC sweep. Repeat this process until λ(t_{m}) reaches the value of (k + 1)h. For each process, \(\left\langle {e}^{\beta {W}_{k,A}^{(n)}}\right\rangle =\left\langle \mathop{\prod }\nolimits_{m = 0}^{h/{{\Delta }}\lambda 1}\frac{{g}_{A}(\lambda ({t}_{m+1}),{N}_{B}({t}_{m}))}{{g}_{A}(\lambda ({t}_{m}),{N}_{B}({t}_{m}))}\right\rangle\).

6.
In the end, we collect the observable from all the N parallel processes and sum them to get the Rényi entanglement entropy according to Eq. (15).
To test the efficiency of our algorithm with the original nonequilibrium proposal, we carry out the measurement of the second Rényi EE of a 2D antiferromagnetic Heisenberg model on a L_{x}: L_{y} = 32 × 16 torus, with the partition of A and \(\overline{A}\) chosen to be the same as the inset of Fig. 1. Note that when the total number N of small processes equals 1, our algorithm reduces to the original proposal of the last section. Figure 7 shows the comparison of the results of \({S}_{A}^{(2)}\) versus the total quench time for N = 1 and N = 24. For the N = 1 case, one sees that as t_{f} − t_{i} becomes larger, the second Rényi EE gradually converges. When t_{f} − t_{i} is not sufficiently large a shift of the Rényi EE from the converged value of \({S}_{A}^{(2)}=4.84(1)\) manifests. We find that this deviation is systematic and is probably caused by two reasons. First, when the quench is not slow enough, not all sites in A will join the topology of glued geometry at the end of the nonequilibrium process, so the final state of the tunning process is not a configuration of \({{{{\mathcal{Z}}}}}_{A}^{2}(1)\) which gives rise to a bias in Eq. (12). Second, the observable of the nonequilibrium protocol is actually an integral defined in Eq. (10), and in the algorithm the sample mean method is used to estimate this integral, which naturally brings a systematic error that decays when the quench time increases. Although the errors mentioned above can always be eliminated by increasing the quench time, in practice it is not an ideal way because of the limited computing time. This defies the controlled computation of Rényi EE of larger and more complicated systems such as the Kagome spin model.
For the N = 24 case, the 2nd Rényi entropy converges faster and has smaller errorbars than the results of N = 1 as the quench time t_{f} − t_{i} increases. This is because the total quench time from λ = 0 to λ = 1 is actually 24 × (t_{f} − t_{i}) in the simulation. The parallelization of the nonequilibrium process increases the total quench time by N times, which actually cures the problem arising from not enough quench time. One thing we want to stress is, for the comparison in Fig. 7, the total number of the bins for N = 1 case is 24 times bigger than that of N = 24 case, this ensures that the total computing power and simulation time for the two cases are nearly the same. The comparison shows that our method cures the limitation of the original algorithm and has smaller errorbars and better convergence.
We would like to end this section by mentioning that the nonquilibrium increment measurement of entanglement, implemented here and in ref. ^{24}, could also be applied on other physics systems beyond quantum manybody lattice model and even condensed matter physics^{95,96}.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Code availability
All numerical codes in this paper are available upon request to the authors.
References
Cardy, J. L. & Peschel, I. Finitesize dependence of the free energy in twodimensional critical systems. Nucl. Phys. B 300, 377–392 (1988).
Calabrese, P. & Cardy, J. Entanglement entropy and quantum field Theory. J. Stat. Mech.: Theory Exp. 2004, P06002 (2004).
Fradkin, E. & Moore, J. E. Entanglement entropy of 2d conformal quantum critical points: hearing the shape of a quantum drum. Phys. Rev. Lett. 97, 050404 (2006).
Casini, H. & Huerta, M. Universal terms for the entanglement entropy in 2+1 dimensions. Nucl. Phys. B 764, 183–201 (2007).
Kitaev, A. & Preskill, J. Topological entanglement entropy. Phys. Rev. Lett. 96, 110404 (2006).
Levin, M. & Wen, X.G. Detecting topological order in a ground state wave function. Phys. Rev. Lett. 96, 110405 (2006).
Wolf, M. M. Violation of the entropic area law for fermions. Phys. Rev. Lett. 96, 010404 (2006).
Lin, Y.C., Iglói, F. & Rieger, H. Entanglement entropy at infiniterandomness fixed points in higher dimensions. Phys. Rev. Lett. 99, 147202 (2007).
Yu, R., Saleur, H. & Haas, S. Entanglement entropy in the twodimensional random transverse field Ising model. Phys. Rev. B 77, 140402 (2008).
Hastings, M. B., González, I., Kallin, A. B. & Melko, R. G. Measuring Renyi entanglement entropy in quantum Monte Carlo simulations. Phys. Rev. Lett. 104, 157201 (2010).
Metlitski, M. A. & Grover, T. Entanglement entropy of systems with spontaneously broken continuous symmetry. Preprint at https://arxiv.org/abs/1112.5166 (2011).
Isakov, S. V., Hastings, M. B. & Melko, R. G. Topological entanglement entropy of a BoseHubbard spin liquid. Nat. Phys. 7, 772–775 (2011).
Jiang, H.C., Wang, Z. & Balents, L. Identifying topological order by entanglement entropy. Nat. Phys. 8, 902–905 (2012).
Casini, H. & Huerta, M. Positivity, entanglement entropy, and minimal surfaces. J. High Energy Phys. 2012, 87 (2012).
Swingle, B. & Senthil, T. Structure of entanglement at deconfined quantum critical points. Phys. Rev. B 86, 155131 (2012).
Kovács, I. A. & Iglói, F. Universal logarithmic terms in the entanglement entropy of 2d, 3d and 4d random transversefield Ising models. EPL 97, 67009 (2012).
Inglis, S. & Melko, R. G. WangLandau method for calculating Rényi entropies in finitetemperature quantum Monte Carlo simulations. Phys. Rev. E 87, 013306 (2013).
Inglis, S. & Melko, R. G. Entanglement at a twodimensional quantum critical point: a T = 0 projector quantum Monte Carlo study. New J. Phys 15, 073048 (2013).
Kallin, A. B., Hyatt, K., Singh, R. R. P. & Melko, R. G. Entanglement at a twodimensional quantum critical point: A numerical linkedcluster expansion study. Phys. Rev. Lett. 110, 135702 (2013).
Luitz, D. J., Plat, X., Laflorencie, N. & Alet, F. Improving entanglement and thermodynamic rényi entropy measurements in quantum monte carlo. Phys. Rev. B 90, 125105 (2014).
Kallin, A. B., Stoudenmire, E. M., Fendley, P., Singh, R. R. P. & Melko, R. G. Corner contribution to the entanglement entropy of an O(3) quantum critical point in 2 + 1 dimensions. J. Stat. Mech. 2014, 06009 (2014).
Helmes, J. & Wessel, S. Entanglement entropy scaling in the bilayer heisenberg spin system. Phys. Rev. B 89, 245120 (2014).
Laflorencie, N. Quantum entanglement in condensed matter systems. Phys. Rep. 646, 1–59 (2016).
D’Emidio, J. Entanglement entropy from nonequilibrium work. Phys. Rev. Lett. 124, 110602 (2020).
Grover, T. Entanglement of interacting fermions in quantum Monte Carlo calculations. Phys. Rev. Lett. 111, 130402 (2013).
Humeniuk, S. & Roscilde, T. Quantum Monte Carlo calculation of entanglement Rényi entropies for generic quantum systems. Phys. Rev. B 86, 235116 (2012).
Zhao, J., Wang, Y.C., Yan, Z., Cheng, M. & Meng, Z. Y. Scaling of entanglement entropy at deconfined quantum criticality. Phys. Rev. Lett. 128, 010601 (2022).
Alba, V. Outofequilibrium protocol for Rényi entropies via the Jarzynski equality. Phys. Rev. E 95, 062132 (2017).
Zhao, J., Yan, Z., Cheng, M. & Meng, Z. Y. Higherform symmetry breaking at Ising transitions. Phys. Rev. Research 3, 033024 (2021).
Block, M. S., D’Emidio, J. & Kaul, R. K. Kagome model for a \({{\mathbb{Z}}}_{2}\) quantum spin liquid. Phys. Rev. B 101, 020402 (2020)..
Assaad, F. F. Stable quantum Monte Carlo simulations for entanglement spectra of interacting fermions. Phys. Rev. B 91, 125146 (2015).
Jarzynski, C. Nonequilibrium equality for free energy differences. Phys. Rev. Lett. 78, 2690–2693 (1997).
Yan, Z. & Meng, Z. Y. Extract lowlying entanglement spectrum from quantum Monte Carlo simulation. Preprint at https://arxiv.org/abs/2112.05886 (2021).
Sandvik, A. W. Stochastic series expansion method with operatorloop update. Phys. Rev. B 59, R14157–R14160 (1999).
Syljuåsen, O. F. & Sandvik, A. W. Quantum Monte Carlo with directed loops. Phys. Rev. E 66, 046701 (2002).
Ma, N. et al. Anomalous quantumcritical scaling corrections in twodimensional antiferromagnets. Phys. Rev. Lett. 121, 117202 (2018).
Wang, Y.C., Ma, N., Cheng, M. & Meng, Z. Y. Scaling of disorder operator at deconfined quantum criticality. Preprint at https://arxiv.org/abs/2106.01380 (2021).
Assaad, F. F. & Grover, T. Simple fermionic model of deconfined phases and phase transitions. Phys. Rev. X 6, 041049 (2016).
Chen, C., Xu, X. Y., Qi, Y. & Meng, Z. Y. Metal to orthogonal metal transition. Chin. Phys. Lett. 37, 047103 (2020).
Chen, C., Yuan, T., Qi, Y. & Meng, Z. Y. Fermi arcs and pseudogap in a lattice model of a doped orthogonal metal. Phys. Rev. B 103, 165131 (2021).
White, S. R. Density matrix formulation for quantum renormalization groups. Phys. Rev. Lett. 69, 2863–2866 (1992).
Wu, F., Deng, Y. & Prokof’ev, N. Phase diagram of the toric code model in a parallel magnetic field. Phys. Rev. B 85, 195104 (2012).
Yan, Z., Wang, Y.C., Ma, N., Qi, Y. & Meng, Z. Y. Topological phase transition and single/multi anyon dynamics of Z_{2} spin liquid. npj Quantum Mater. 6, 39 (2021).
Dong, S., Fradkin, E., Leigh, R. G. & Nowling, S. Topological entanglement entropy in ChernSimons theories and quantum Hall fluids. J. High Energy Phys. 2008, 016–016 (2008).
Zhang, Y., Grover, T., Turner, A., Oshikawa, M. & Vishwanath, A. Quasiparticle statistics and braiding from groundstate entanglement. Phys. Rev. B 85, 235151 (2012).
Balents, L., Fisher, M. P. A. & Girvin, S. M. Fractionalization in an easyaxis kagome antiferromagnet. Phys. Rev. B 65, 224412 (2002).
Isakov, S. V., Kim, Y. B. & Paramekanti, A. Spinliquid phase in a spin1/2 quantum magnet on the kagome lattice. Phys. Rev. Lett. 97, 207204 (2006).
Isakov, S. V., Melko, R. G. & Hastings, M. B. Universal signatures of fractionalized quantum critical points. Science 335, 193–195 (2012).
Wang, Y.C., Fang, C., Cheng, M., Qi, Y. & Meng, Z. Y. Topological spin liquid with symmetryprotected edge states. Preprint at https://arxiv.org/abs/1701.01552 (2017).
Sun, G.Y. et al. Dynamical signature of symmetry fractionalization in frustrated magnets. Phys. Rev. Lett. 121, 077201 (2018).
Wang, Y.C., Zhang, X.F., Pollmann, F., Cheng, M. & Meng, Z. Y. Quantum spin liquid with even Ising gauge field structure on kagome lattice. Phys. Rev. Lett. 121, 057202 (2018).
Wang, Y.C., Cheng, M., WitczakKrempa, W. & Meng, Z. Y. Fractionalized conductivity and emergent selfduality near topological phase transitions. Nat. Commun. 12, 5347 (2021).
Wang, Y.C., Yan, Z., Wang, C., Qi, Y. & Meng, Z. Y. Vestigial anyon condensation in kagome quantum spin liquids. Phys. Rev. B 103, 014408 (2021).
Shores, M. P., Nytko, E. A., Bartlett, B. M. & Nocera, D. G. A structurally perfect S = 1/2 kagomé antiferromagnet. J. Am. Chem. Soc. 127, 13462–13463 (2005).
Helton, J. S. et al. Spin dynamics of the spin1/2 kagome lattice antiferromagnet ZnCu_{3}(OH)_{6}Cl_{2}. Phys. Rev. Lett. 98, 107204 (2007).
Han, T. H. et al. Fractionalized excitations in the spinliquid state of a kagomelattice antiferromagnet. Nature 492, 406–410 (2012).
Fu, M., Imai, T., Han, T.H. & Lee, Y. S. Evidence for a gapped spinliquid ground state in a kagome heisenberg antiferromagnet. Science 350, 655–658 (2015).
Norman, M. R. Colloquium: Herbertsmithite and the search for the quantum spin liquid. Rev. Mod. Phys. 88, 041002 (2016).
Feng, Z. et al. Gapped spin1/2 spinon excitations in a new kagome quantum spin liquid compound Cu_{3}Zn(OH)_{6}FBr. Chin. Phys. Lett. 34, 077502 (2017).
Wen, X.G. Discovery of fractionalized neutral spin1/2 excitation of topological order. Chin. Phys. Lett. 34, 090101 (2017).
Feng, Z. et al. Effect of zn doping on the antiferromagnetism in kagome Cu_{4−x}Zn_{x}(OH)_{6}FBr. Phys. Rev. B 98, 155127 (2018).
Feng, Z. et al. From claringbullite to a new spin liquid candidate Cu_{3}Zn(OH)_{6}FCl. Chin. Phys. Lett. 36, 017502 (2019).
Wen, J.J. & Lee, Y. S. The search for the quantum spin liquid in kagome antiferromagnets. Chin. Phys. Lett. 36, 050101 (2019).
Wei, Y. et al. Evidence for a \({{\mathbb{Z}}}_{2}\) topological ordered quantum spin liquid in a kagomelattice antiferromagnet. Preprint at https://arxiv.org/abs/1710.02991 (2017).
Wei, Y. et al. Magnetic phase diagram of Cu_{4−x}Zn_{x}OH_{6}FBr studied by neutrondiffraction and μsr techniques. Chin. Phys. Lett. 37, 107503 (2020).
Wei, Y. et al. Antiferromagnetism in the kagomelattice compound α − Cu_{3}Mg(OH)_{6}Br_{2}. Phys. Rev. B 100, 155129 (2019).
Wei, Y. et al. Nonlocal effects of lowenergy excitations in quantumspinliquid candidate Cu_{3}Zn(OH)_{6}FBr. Chin. Phys. Lett. 38, 097501 (2021).
Wen, X.G. Choreographed entanglement dances: Topological states of quantum matter. Science 363, eaal3099 (2019).
Broholm, C. et al. Quantum spin liquids. Science 367, eaay0668 (2020).
Liu, Y., Wang, W., Sun, K. & Meng, Z. Y. Designer Monte Carlo simulation for the grossneveuyukawa transition. Phys. Rev. B 101, 064308 (2020).
Xu, X. Y. et al. Monte Carlo study of lattice compact quantum electrodynamics with fermionic matter: the parent state of quantum phases. Phys. Rev. X 9, 021022 (2019).
Wang, W., Lu, D.C., Xu, X. Y., You, Y.Z. & Meng, Z. Y. Dynamics of compact quantum electrodynamics at large fermion flavor. Phys. Rev. B 100, 085123 (2019).
Janssen, L., Wang, W., Scherer, M. M., Meng, Z. Y. & Xu, X. Y. Confinement transition in the QED_{3}GrossNeveuXY universality class. Phys. Rev. B 101, 235118 (2020).
Xu, X. Y., Sun, K., Schattner, Y., Berg, E. & Meng, Z. Y. NonFermi liquid at (2 + 1)D ferromagnetic quantum critical point. Phys. Rev. X 7, 031058 (2017).
Xu, X. Y. et al. Revealing fermionic quantum criticality from new Monte Carlo techniques. J. Phys. Condens. Matter 31, 463001 (2019).
Liu, Z. H., Pan, G., Xu, X. Y., Sun, K. & Meng, Z. Y. Itinerant quantum critical point with fermion pockets and hotspots. Proc. Natl. Acad. Sci. U.S.A. 116, 16760–16767 (2019).
Xu, X. Y., Klein, A., Sun, K., Chubukov, A. V. & Meng, Z. Y. Identification of nonFermi liquid fermionic selfenergy from quantum Monte Carlo data. npj Quantum Mater. 5, 65 (2020).
Klein, A., Chubukov, A. V., Schattner, Y. & Berg, E. Normal state properties of quantum critical metals at finite temperature. Phys. Rev. X 10, 031053 (2020).
Shen, B. et al. Strangemetal behaviour in a pure ferromagnetic kondo lattice. Nature 579, 51–55 (2020).
Jiang, W. et al. Pseudogap and superconductivity emerging from quantum magnetic fluctuations: a Monte Carlo study. Preprint at https://arxiv.org/abs/2105.03639 (2021).
Wu, Y. et al. Anisotropic c − f hybridization in the ferromagnetic quantum critical metal CeRh_{6}Ge_{4}. Phys. Rev. Lett. 126, 216406 (2021).
Liu, Y. et al. The dynamical exponent of a quantum critical itinerant ferromagnet: a Monte Carlo study. Preprint at https://arxiv.org/abs/2106.12601 (2021).
Nussinov, Z. & Ortiz, G. Sufficient symmetry conditions for Topological Quantum Order. Proc. Natl. Acad. Sci. U.S.A. 106, 16944–16949 (2009).
Nussinov, Z. & Ortiz, G. A symmetry principle for topological quantum order. Annals Phys. 324, 977–1057 (2009).
Gaiotto, D., Kapustin, A., Seiberg, N. & Willett, B. Generalized global symmetries. J. High Energ. Phys. 2015, 1–62 (2015).
Ji, W. & Wen, X.G. Categorical symmetry and noninvertible anomaly in symmetrybreaking and topological phase transitions. Preprint at https://arxiv.org/abs/1912.13492 (2019).
Kong, L., Lan, T., Wen, X.G., Zhang, Z.H. & Zheng, H. Algebraic higher symmetry and categorical symmetry – a holographic and entanglement view of symmetry. Phys. Rev. Research 3, 043086 (2020).
Wu, X.C., Ji, W. & Xu, C. Categorical symmetries at criticality. Preprint at https://arxiv.org/abs/2012.03976 (2020).
Wang, Y.C., Cheng, M. & Meng, Z. Y. Scaling of the disorder operator at (2 + 1)d u(1) quantum criticality. Phys. Rev. B 104, L081109 (2021).
Estienne, B., Stéphan, J.M. & WitczakKrempa, W. Cornering the universal shape of fluctuations. Preprint at https://arxiv.org/abs/2102.06223 (2021).
Wu, X.C., Jian, C.M. & Xu, C. Universal features of higherform symmetries at phase transitions. SciPost Phys. 11, 33 (2021).
Chen, B.B., Tu, H.H., Meng, Z. Y. & Cheng, M. Topological disorder parameter. Preprint at https://arxiv.org/abs/2203.08847 (2022).
Yan, Z. & Meng, Z. Y. Extract lowlying entanglement spectrum from quantum monte carlo simulation. Preprint at https://arxiv.org/abs/2112.05886v3 (2021).
Tagliacozzo, L., Evenbly, G. & Vidal, G. Simulation of twodimensional quantum systems using a tree tensor network that exploits the entropic area law. Phys. Rev. B 80, 235127 (2009).
Shirts, M. R., Bair, E., Hooker, G. & Pande, V. S. Equilibrium free energies from nonequilibrium measurements using maximumlikelihood methods. Phys. Rev. Lett. 91, 140601 (2003).
Palassini, M. & Ritort, F. Improving freeenergy estimates from unidirectional work measurements: Theory and experiment. Phys. Rev. Lett. 107, 060601 (2011).
Acknowledgements
J.R.Z., B.B.C., Z.Y., and Z.Y.M. would like to thank Jonathan D’Emidio for encouraging and fruitful discussion on the details of the quench time of the Jarzynski estimator, and its applications in other subjects. They also thank Chenjie Wang for valuable discussions on the theoretical meaning of MES and related issues. They acknowledge support from the RGC of Hong Kong SAR of China (Grant Nos. 17303019, 17301420, 17301721, and AoE/P701/20), the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB33000000), the K. C. Wong Education Foundation (Grant No. GJTD202001) and the Seed Funding “QuantumInspired explainableAI” at the HKUTCL Joint Research Center for Artificial Intelligence. Y.C.W. acknowledges the supports from the NSFC under Grant Nos. 11804383 and 11975024, and the Fundamental Research Funds for the Central Universities under Grant No. 2018QNA39. M.C. acknowledges support from NSF under award number DMR1846109. We thank the Computational Initiative at the Faculty of Science and the Information Technology Services at the University of Hong Kong and the Tianhe platforms at the National Supercomputer Centers in Tianjin and Guangzhou for their technical support and generous allocation of CPU time. The authors also acknowledge Beijng PARATERA Tech CO., Ltd. (https://www.paratera.com/) for providing HPC resources that have contributed to the research results reported within this paper.
Author information
Authors and Affiliations
Contributions
J.Z., Z.Y.M, and M.C. initiated the work. J.Z. developed the algorithm and carried out the Quantum Monte Carlo simulations of all the models mentioned. B.B.C. conducted the DMRG calculations of the toy model. All authors contributed to the analysis of the results and the preparation and revision of the draft. M.C. and Z.Y.M. supervised the project.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Zhao, J., Chen, BB., Wang, YC. et al. Measuring Rényi entanglement entropy with high efficiency and precision in quantum Monte Carlo simulations. npj Quantum Mater. 7, 69 (2022). https://doi.org/10.1038/s41535022004760
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41535022004760
This article is cited by

Unlocking the general relationship between energy and entanglement spectra via the wormhole effect
Nature Communications (2023)