Bulk-Edge Correspondence in Fractional Quantum Hall States

Bulk-Edge Correspondence in Fractional Quantum Hall States


We substantiate a complete picture of the “bulk-edge correspondence” conjecture for topological phases. By studying the eigenstates in the entanglement spectrum for both the ideal and realistic Coulomb ground state of the fractional quantum Hall system, it is verified that the eigenstates in the universal part of the entanglement spectrum purely lie in the Hilbert space of the edge excitations projected onto the physical Hilbert space of the subsystem itself. Hence, not only the eigenlevels in the entanglement spectrum are in one-to-one correspondence with the eigenenergies of an effective dynamical edge Hamiltonian, but all the eigenstates are confirmed to be the actual (projected) edge excitations of the subsystem. This result also reveals the possibility of extracting the full information of the edge excitations from the state of the subsystem reduced from a geometric cut of the pure ground state of the total system in topological phases.


Since the discovery of the fractional quantum Hall effect Tsui et al. (1982); Gir (1987), a new paradigm of phases of matter which is of a topological nature Wen (1989); Wen and Niu (1990); Nayak et al. (2008) and yet does not fit into the conventional symmetry-breaking picture comes into play. These novel phases of matter exhibit spectacular behaviors, such as fractional charges Laughlin (1983), (non-)Abelian anyons Leinaas and Myrheim (2007); ?; ?, or topologically protected gapless edge excitations. In absence of local order parameters, these phases are usually characterized by certain global properties such as the ground state degeneracy Wen and Niu (1990). Despite intense effort, a complete understanding of such topological ordered phases still remains elusive.

Among various approaches to characterize these intriguing zero-temperature phases of matters, entanglement plays an increasingly crucial role Amico et al. (2008). It is believed that ground states of topological ordered systems are able to build up long range entanglement Chen et al. (2010). One way to study the entanglement in a many-body system is to look at the correlation between a subsystem and its complement in the total system. After the total system has been prepared in a pure ground state , the reduced density matrix of a subsystem is obtained by tracing over the degrees of freedom of the outside subsystem , . The Von Neumann entropy of subsystem A, , usually called the entanglement entropy, captures the entanglement between subsystem A and the rest of the total system. In typical ground states of many-body systems, the entanglement entropy follows an area scaling law Eisert et al. (2010). For a state in a topological phase, besides the usual term that is proportional to boundary area, there emerges a constant sub-leading term that is characteristic of the topological nature of the system. Coined as topological entanglement entropy Hamma et al. (2005); Levin and Wen (2006); Kitaev and Preskill (2006), this quantity is one of the most recognized signatures for states with topological phases.

In subsequent work Li and Haldane (2008), Li and Haldane proposed and numerically substantiated that more information about the underlying topological state can be obtained by studying instead the full spectrum of the reduced density matrix of the subsystem. Writing , the entanglement spectrum is defined as the spectrum of the effective Hamiltonian . The eigenvalues and eigenstates of can be alternatively extracted from the Schmit decomposition of the total state,


Though , in general, is not a real dynamical Hamiltonian, it is ultimately related to the physical edge of the subsystem: In the quantum Hall system, the number of the low-lying universal eigenenergies, which are separated by a finite gap Li and Haldane (2008); Thomale et al. (2010) from other generic levels, are in one-to-one correspondence with the edge modes describable by a macroscopic edge theory, at least for small angular momentum excitations.

Both the entanglement entropy and the counting structure of the entanglement spectrum can be naturally interpreted if a relation between the bulk and the physical edge of the subsystem is established. This relation, termed the bulk-edge correspondence Qi et al. (2012); Swingle and Senthil (2012); Dubail et al. (2012a), states that the entanglement Hamiltonian corresponds to an effective dynamical Hamiltonian in an edge theory. The bulk-edge correspondence, originally sketched in the early work of entanglement entropy Kitaev and Preskill (2006), has been extensively studied in various fields Liu et al. (2013); Santos (2013); Ho et al. (2015); Chen et al. (2016); Ho et al. (2017); Koch-Janusz et al. (2017). This correspondence emerged in part from general arguments based purely on the topological properties of the system and on the standard renormalization-group method Qi et al. (2012), in part from observations of geometric aspects and the Lorentz invariance of the emergent effective theory Swingle and Senthil (2012), and in part through study of model wavefunctions specific to the quantum Hall system constructed from conformal blocks Dubail et al. (2012a). These arguments are only valid provided certain assumptions hold in the thermodynamic limit.

In this work, we argue that a complete picture of the bulk-edge correspondence should consist of two pieces. One is that the eigenvalues are in one to one correspondence with the eigenvalues of an effective edge Hamiltonian. The other is that the universal eigenstates that appear in the bulk density matrix are all real edge states of the subsystem 2. For the first part of the bulk-edge correspondence in the quantum Hall system, the only direct numerical evidence was provided in Ref. Dubail et al. (2012a), where the spectrum of a perturbative local Hamiltonian of an edge system described by a conformal field theory was computed and shown to match quite well the entanglement spectrum obtained for a real space partition of the total system. To the best of our knowledge, a direct verification of the latter part of the correspondence is still missing. Since an effective edge theory usually describes the degrees of freedom that are distinct from those in the original bulk system, direct comparison of their eigenstates is much harder than comparing their eigenenergies. Thus, typical studies of the entanglement spectrum usually focuses on the counting structure of the eigenlevels. In the present work, by analyzing the detailed information of the entanglement spectrum eigenstates, we complete the verification of the missing piece of the bulk-edge correspondence for both the model wavefunction and the realistic ground state of Coulomb interactions.

The system studied in this work consists of a few spinless fermions in the lowest Landau level on a two-dimensional disk. The plane geometry is a convenient platform for the discussion of edge physics. System partitions are restricted to the subsets of the single particle magnetic orbitals in the symmetric gauge (with unnormalized analytic wavefunction and angular momentum quantum number ). In the lowest Landau level, these orbitals have localized ring shapes. Thus the orbital partition mimics approximately the real space partition. It is worthwhile to mention that another commonly used scheme, based on a cut of particle number Haque et al. (2007); Zozulya et al. (2007); Sterdyniak et al. (2011), has usually been adopted for the study of quasi-hole excitations. The level counting in the orbital entanglement spectrum and the one in the particle entanglement spectrum, which correspond to quasi-hole excitations in the bulk, were shown Chandran et al. (2011) to be ultimately connected. This relation, also termed as “bulk-edge correspondence”, is not to be confused with the one we are considering in the present work.

In the subsystem, the particle number and total angular momentum are still good quantum numbers; Thus, the reduced density matrix of a subsystem contains distinct sectors which do not couple with each other. Fig. 1 compares the standard entanglement spectrum for the ideal Laughlin state with that of the realistic Coulomb ground state of 10 particles in the lowest Landau level. For entanglement spectra of the quantum Hall states in other geometry and partition schemes, see also Refs. Zozulya et al. (2007); Läuchli et al. (2010); Sterdyniak et al. (2011); Dubail et al. (2012b); Sterdyniak et al. (2012); Rodríguez et al. (2012). The subsystem studied in this work involves the inner most 14-orbitals, which is the largest system that is sufficiently far away from the edge of the total system, in the sense that it is still in the linear region of the entanglement entropy, as is indicated in the inset of Fig. 1. In this region, edge effects are supposed to be removed as much as possible. By analyzing the fraction of each eigenstate that resides in the Hilbert space of edge modes (by the procedure explained in the following), we are able to identify all the universal levels apart from the generic levels in the case of Coulomb interactions. It is observed that the universal and generic parts barely mix with each other, albeit at larger angular momentums they start to mix, in contrast to the low angular momenta, where there is a finite gap between them.

Figure 1: Entanglement spectrum of the 14-orbital subsystem at the 5-particle sector. The total system of 10 particles in totally 28 orbitals is prepared in (a) the ideal Laughlin state and (b) the realistic Coulomb interaction ground state. In (b) the red levels are the universal part identified with high overlaps () with the edge modes of the subsystem (see also Fig. (2) and (3)), while the blue levels are the generic part of Coulomb interaction with nearly zero overlaps. Inset: Entanglement entropy v.s. the square root of the number of orbitals () in subsystem , which is proportional to the length of the boundary. The blue line is our fit to the first 14 points, which are in the linear region away from the edge of the total system. The extracted topological entanglement entropy is , close to the theoretical value ln.

To prove the bulk-edge correspondence, on the “edge” part, we use the well-known series of model wavefunctions Wen (1992) for the (neutral) edges excitations of the Laughlin ground state. These analytic model wavefunctions are generated by the power sum symmetric polynomials , . To build an edge mode with angular momentum (this is the additional angular momentum carried by the edge mode beyond the ground state angular momentum), one generates an integer partition , where are positive integers. The corresponding (unnormalized) edge state is constructed as the product of the Laughlin wavefunction and a symmetric polynomial:


where is the filling factor. For a given , the number of different modes is thus the number of integer partitions of . These states are the only zero energy states for the case of p-wave contact interactions (equivalent to the Haldane pseudo-potential Haldane (1983) in the lowest Landau level), and they describe the gapless edge excitations of the Laughlin ground state. In the thermodynamic limit, the Hilbert space of the edge modes generated by the order- power sum polynomials is identical to that generated by the U(1) Kac-Moody algebra Wen (1990a, b, 1992) in a macroscopic theory of the edges physics. However, for finite size systems, when the orders of the symmetric polynomials become larger than the number of particles, these states are no longer linearly independent. In fact, the number of linearly independent symmetric polynomials of order is the number of partitions of into at most parts, where is the number of particles and the number of variables in the polynomial. Denote this restricted partition as . For each such integer partition , one alternative way to construct the edge state is to use the corresponding order- monomial symmetric polynomial , where the summation is taken over all permutations of such that the polynomial is symmetrized. The leads to a correction of the number of edges modes for finite size systems compared with the macroscopic theory for systems in the thermodynamical limit.

Another finite size effect arises when comparing the edge modes with the eigenstates in the entanglement spectrum. That is, the ideal edge modes might involve single particle orbitals that are outside the region of the subsystem under consideration. (Fig.4 shows the actual spectra of the subsystem and the edge state spectra with open boundary condition.) These additional degrees of freedom need to be traced out in formatting the appropriate Hilbert space of the subsystem Dubail et al. (2012a). Precisely speaking, the Fock space spanned by the free edge states are projected onto a subspace involving only orbitals that are resident in the subsystem. This inevitable projection procedure was also discussed in Ref. Dubail et al. (2012a) for the case the real space partitions. In this manner, the projections of the original linearly independent edge modes might become linearly dependent, such that the projected Fock space could have a smaller dimension than the original Fock space of the free edge states. Table 1 lists the number of free edge modes, the number of independent projected edge modes, and the number of observed energy levels in the entanglement spectrum for the 5- and 4-particle sectors. The observed level counting in the primary 5-particle sector of the entanglement spectrum matches precisely the reduced dimension of edge modes Hilbert space at every angular momentum sector. In the large particle number limit, the restricted integer partition reduces to the regular integer partition. This is in agreement with the original conjecture for the counting structure of the entanglement spectrum in Ref. Li (2009).

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 1 2 3 5 7 10 13 18 23 30 37 47 57 70 84
1 1 2 3 5 7 9 11 14 16 18 19 20 20 19 18
1 1 2 3 5 7 9 11 14 16 18 19 20 20 19 18
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 1 2 3 5 6 9 11 15 18 23 27 34 39 47 54
1 1 2 3 5 6 9 11 14 16 19 20 23 23 24 23
1 1 2 3 4 5 7 7 8 8 8 7 7 5 4 3
Table 1: The number of edge modes at various angular momentum in the 14-orbital subsystem. Upper and lower tables correspond to five and four particles, respectively. is the number of free edge modes, i.e., the number of restricted integer partitions. represents the dimension of the Hilbert space of projected edges modes; labels the observed multiplicity of levels in the entanglement spectrum.
Figure 2: The projection probabilities for the whole entanglement spectrum of the 5-particle sector (left) and the 4-particle sector (right) for the case of Coulomb ground state. Both the cases of ideal edge states (upper) and realistic edges states of Coulomb interaction are examined.
Figure 3: Typical patterns for the projection probabilities. Red squares represent the probabilities for 5 particles at (). Blue dots correspond to the 4-particle sector at (). Projection probabilities are computed using (a) the model edge states for the Laughlin case and (b) the real edge states for Coulomb interactions.

In order to prove that the universal part of the entanglement spectrum is indeed spanned by the edge modes, we compute the probability for each eigenstate in the entanglement spectrum to reside in this space. The projection probability is defined as


where is the projection operator to the Hilbert space of projected edge modes.

For the ideal Laughlin wavefunction, it is confirmed that all the eigenstates in the entanglement spectrum have identity projection probabilities in the Hilbert space of the projected edge states. For the case of Coulomb ground state, both the ideal edge states and the real edge modes of the subsystem are used to analyze the projection probability Eq.(3). The real edge modes are computed from the Coulomb interaction at the corresponding angular momenta with open boundary condition (without performing single particle orbital cut). In the thermodynamic limit, they are low-lying modes describing the gapless edge excitations. However, in few-body calculations, there are usually no clear gaps separating the edge modes and the bulk excitations. In this case, we identify each edge state with the aid of high (0.97) overlaps with the ideal edge wavefunctions (2). Fig. 2 and Fig. 3 show the projection probability patterns. Both the primary 5-particle sector and the sector with particle number offset (4-particle sector in this case) show clear high probability plateaus for the universal levels in the entanglement spectrum. In the 4-particle sector, since the number of universal levels is smaller than the dimension of the projected edges modes (as was shown in Table.1), the generic levels also have fluctuating non-zero probabilities, in contrast to 5-particle sector, where all the generic levels have nearly zero projection probabilities. They are nevertheless clearly separated from the universal part.

In the primary sector, since all universal eigenstates in the entanglement spectrum have nearly identity probabilities in the corresponding Hilbert space of projected edge modes, it is concluded that the information of all edge excitations is hidden in the pure ground state of the total system, which can be extracted from the bulk state of the subsystem. This can be viewed as a complementary effect to a similar result in Ref. Chandran et al. (2011), where it is observed that the primary sector of the entanglement spectrum under particle partition spans the entire space of quasi-hole excitations of the subsystem.

Figure 4: The spectrum of 5 particles under the Haldane pseudo-potential with an open boundary condition (red levels) and with the same boundary condition as the 14-orbital subsystem (blue levels). The blue levels are shifted slightly to the right from their corresponding angular momenta. For the case of an open boundary condition, the degeneracies of the ground states are , which are identical to the number of the ideal edge states.

To further firm up the evidence of the bulk-edge correspondence, we present in Fig. 4 the actual energy spectrum of 5 particles under Haldane interaction with and without boundary conditions. The degenerate zero-energy states for the latter are model edge states. If we did not use the mechanism of projecting the edge states of the subsystem with open boundaries, the nondegenerate low-lying states of the subsystem alone would have been inadequate to explain all the observed universal levels in the entanglement spectrum (Fig. 1).

In conclusion, this work shows that the universal part of the subsystem reduced density matrix, by orbital partition, of the fractional quantum Hall state lie in the Hilbert space of edge modes of the subsystem itself, hence completes the picture of the bulk-edge correspondence conjecture. This result also reveals the possibility of extracting edge excitations from the state of the subsystem reduced by geometric cut of the pure ground state of the total system in topological phases. In the future, it will be informative to apply this method to other topological phases.

The work of BY and CHG have been supported in part by National Science Foundation grant PHY-1607180. R.R.B. was supported by Purdue University startup funds.


  1. preprint: APS/123-QED
  2. To be precise, the universal levels stands for the ones appear in the entanglement spectrum of the ideal model wavefunctions, as well as the part of the spectrum having similar structures for the case of generic interactions. By comparing the entanglement spectrum eigenstates with the actual edge excitations, we are able to identify every universal state (See Fig. 1 for an example).


  1. D C Tsui, H L Stormer,  and A C Gossard, “Two-Dimensional Magnetotransport in the Extreme Quantum Limit,” Physical Review Letters 48, 1559–1562 (1982).
  2. The Quantum Hall Effect, edited by R. E. Prange and S. M. Girvin (Springer-Verlag, Berlin, 1987).
  3. Xiao-Gang Wen, “Vacuum degeneracy of chiral spin states in compactified space,” Physical Review B 40, 7387–7390 (1989).
  4. Xiao-Gang Wen and Q Niu, “Ground-state degeneracy of the fractional quantum Hall states in the presence of a random potential and on high-genus Riemann surfaces,” Physical Review B 41, 9377–9396 (1990).
  5. Chetan Nayak, Steven H Simon, Ady Stern, Michael Freedman,  and Sankar Das Sarma, “Non-Abelian anyons and topological quantum computation,” Reviews of Modern Physics 80, 1083–1159 (2008).
  6. R B Laughlin, “Anomalous Quantum Hall Effect: An Incompressible Quantum Fluid with Fractionally Charged Excitations,” Physical Review Letters 50, 1395–1398 (1983).
  7. J M Leinaas and J Myrheim, “On the theory of identical particles,” Il Nuovo Cimento B (1971-1996) 37, 1–23 (2007).
  8. Frank Wilczek, “Magnetic Flux, Angular Momentum, and Statistics,” Physical Review Letters 48, 1144–1146 (1982a).
  9. Frank Wilczek, “Quantum Mechanics of Fractional-Spin Particles,” Physical Review Letters 49, 957–959 (1982b).
  10. Luigi Amico, Rosario Fazio, Andreas Osterloh,  and Vlatko Vedral, “Entanglement in many-body systems,” Reviews of Modern Physics 80, 517–576 (2008).
  11. Xie Chen, Zheng-Cheng Gu,  and Xiao-Gang Wen, “Local unitary transformation, long-range quantum entanglement, wave function renormalization, and topological order,” Physical Review B 82, 155138 (2010).
  12. J Eisert, M Cramer,  and M B Plenio, “Colloquium: Area laws for the entanglement entropy,” Reviews of Modern Physics 82, 277–306 (2010).
  13. Alioscia Hamma, Radu Ionicioiu,  and Paolo Zanardi, “Ground state entanglement and geometric entropy in the Kitaev model,” Physics Letters A 337, 22–28 (2005).
  14. Michael Levin and Xiao-Gang Wen, “Detecting Topological Order in a Ground State Wave Function,” Physical Review Letters 96, 110405 (2006).
  15. Alexei Kitaev and John Preskill, “Topological Entanglement Entropy,” Physical Review Letters 96, 110404 (2006).
  16. Hui Li and F D M Haldane, “Entanglement Spectrum as a Generalization of Entanglement Entropy: Identification of Topological Order in Non-Abelian Fractional Quantum Hall Effect States,” Physical Review Letters 101, 010504 (2008).
  17. R Thomale, A Sterdyniak, N Regnault,  and B Andrei Bernevig, “Entanglement Gap and a New Principle of Adiabatic Continuity,” Physical Review Letters 104, 180502 (2010).
  18. Xiao-Liang Qi, Hosho Katsura,  and Andreas W W Ludwig, “General Relationship between the Entanglement Spectrum and the Edge State Spectrum of Topological Quantum States,” Physical Review Letters 108, 196402 (2012).
  19. Brian Swingle and T Senthil, “Geometric proof of the equality between entanglement and edge spectra,” Physical Review B 86, 045117 (2012).
  20. J Dubail, N Read,  and E H Rezayi, ‘‘Edge-state inner products and real-space entanglement spectrum of trial quantum Hall states,” Physical Review B 86, 245310 (2012a).
  21. Zhao Liu, D L Kovrizhin,  and Emil J Bergholtz, “Bulk-edge correspondence in fractional Chern insulators,” Physical Review B 88, 081106 (2013).
  22. Raul A Santos, “Bulk-edge correspondence of entanglement spectrum in two-dimensional spin ground states,” Physical Review B 87, 035141 (2013).
  23. Wen Wei Ho, Lukasz Cincio, Heidar Moradi, Davide Gaiotto,  and Guifre Vidal, “Edge-entanglement spectrum correspondence in a nonchiral topological phase and Kramers-Wannier duality,” Physical Review B 91, 125119 (2015).
  24. Xiao Chen, Apoorv Tiwari,  and Shinsei Ryu, “Bulk-boundary correspondence in (3+1)-dimensional topological phases,” Physical Review B 94, 045113 (2016).
  25. Wen Wei Ho, Lukasz Cincio, Heidar Moradi,  and Guifre Vidal, “Universal edge information from wave-function deformation,” Physical Review B 95, 235110 (2017).
  26. Maciej Koch-Janusz, Kusum Dhochak,  and Erez Berg, “Edge-entanglement correspondence for a gapped topological phase with symmetry,” Physical Review B 95, 205110 (2017).
  27. To be precise, the universal levels stands for the ones appear in the entanglement spectrum of the ideal model wavefunctions, as well as the part of the spectrum having similar structures for the case of generic interactions. By comparing the entanglement spectrum eigenstates with the actual edge excitations, we are able to identify every universal state (See Fig. 1 for an example).
  28. Masudul Haque, Oleksandr Zozulya,  and Kareljan Schoutens, “Entanglement Entropy in Fermionic Laughlin States,” Physical Review Letters 98, 48 (2007).
  29. O S Zozulya, M Haque, K Schoutens,  and E H Rezayi, “Bipartite entanglement entropy in fractional quantum Hall states,” Physical Review B 76, 125310 (2007).
  30. A Sterdyniak, N Regnault,  and B Andrei Bernevig, “Extracting Excitations from Model State Entanglement,” Physical Review Letters 106, 100405 (2011).
  31. Anushya Chandran, M Hermanns, N Regnault,  and B Andrei Bernevig, “Bulk-edge correspondence in entanglement spectra,” Physical Review B 84, P09012 (2011).
  32. Andreas M Läuchli, Emil J Bergholtz, Juha Suorsa,  and Masudul Haque, “Disentangling Entanglement Spectra of Fractional Quantum Hall States on Torus Geometries,” Physical Review Letters 104, 156404 (2010).
  33. J Dubail, N Read,  and E H Rezayi, “Real-space entanglement spectrum of quantum Hall systems,” Physical Review B 85, 115321 (2012b).
  34. A Sterdyniak, A Chandran, N Regnault, B Andrei Bernevig,  and Parsa Bonderson, “Real-space entanglement spectrum of quantum Hall states,” Physical Review B 85, 125308 (2012).
  35. Iván D Rodríguez, Steven H Simon,  and J K Slingerland, “Evaluation of Ranks of Real Space and Particle Entanglement Spectra for Large Systems,” Physical Review Letters 108, 256806 (2012).
  36. Xiao-Gang Wen, “Theory of the edge states in fractional quantum Hall effects,” International Journal of Modern Physics B 6, 1711–1762 (1992).
  37. F Duncan M Haldane, “Fractional quantization of the hall effect: a hierarchy of incompressible quantum fluid states,” Physical Review Letters 51, 605 (1983).
  38. Xiao-Gang Wen, “Electrodynamical properties of gapless edge excitations in the fractional quantum hall states,” Physical review letters 64, 2206 (1990a).
  39. Xiao-Gang Wen, “Chiral luttinger liquid and the edge excitations in the fractional quantum hall states,” Physical Review B 41, 12838 (1990b).
  40. Hui Li, “Numerical Studies of Quantum Entanglement in Fractional Quantum Hall Effect Systems,” Dissertation at Princeton University  (2009).
This is a comment super asjknd jkasnjk adsnkj
The feedback cannot be empty
Comments 0
The feedback cannot be empty
Add comment

You’re adding your first comment!
How to quickly get a good reply:
  • Offer a constructive comment on the author work.
  • Add helpful links to code implementation or project page.