pith. sign in

arxiv: 2605.21632 · v1 · pith:75WMSU3Enew · submitted 2026-05-20 · ❄️ cond-mat.str-el · physics.comp-ph

Diagrammatic Monte Carlo for Fermionic R\'enyi Entanglement Entropy

Boyuan Shi This is my paper

Pith reviewed 2026-05-22 08:52 UTC · model grok-4.3

classification ❄️ cond-mat.str-el physics.comp-ph
keywords diagrammatic Monte CarloRenyi entanglement entropylattice fermionsHubbard modelpath integraldeterminant expansionquantum Monte Carloreplicated systems
0
0 comments X

The pith

A diagrammatic Monte Carlo method computes the Rényi entanglement entropy of interacting lattice fermions by mapping the problem to a replicated path integral with mixed boundary conditions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a direct diagrammatic Monte Carlo framework for calculating the Rényi entanglement entropy of interacting lattice fermions. It begins with the fermionic graded-swap representation of the quantity Z_n[A] equal to the trace of the nth power of the reduced density matrix, which converts the entropy computation into a replicated path integral featuring mixed temporal boundary conditions on the entangling region. In this setup the replica momenta become half-shifted at values (2m+1)π/n and the interaction expansion takes a determinant form that permits connected-determinant summation. The authors combine this expansion with a Markov-chain Monte Carlo sampler to generate order-by-order corrections for large systems at high orders, with memory as the primary constraint rather than a sign problem. This approach opens the way to entanglement calculations in regimes where conventional quantum Monte Carlo sampling becomes costly or sign-problem limited.

Core claim

We develop a direct diagrammatic Monte Carlo framework for the Renyi entanglement entropy of interacting lattice fermions. The method starts from the fermionic graded-swap representation of Z_n[A]=Tr_A ρ_A^n, which converts the entropy problem into a replicated path integral with mixed temporal boundary conditions on the entangling region. In this representation the replica momenta are half-shifted, q_m=(2m+1)π/n, and the interaction expansion has a determinant form suitable for connected-determinant summation. We combine this expansion with a many-configuration Markov-chain Monte Carlo sampler to obtain order-by-order corrections for very large systems to very high orders.

What carries the argument

The fermionic graded-swap representation of Z_n[A]=Tr_A ρ_A^n, which converts the entanglement entropy into a replicated path integral with mixed temporal boundary conditions that admits a determinant-form interaction expansion and connected-determinant summation.

If this is right

  • Order-by-order corrections in the interaction strength become available for systems far larger than those treatable by exact diagonalization.
  • Calculations on large periodic lattices with square subregions become practical, with the dominant limit shifting from sign problems to memory.
  • The framework extends diagrammatic techniques to fermionic entanglement observables in parameter regimes where direct quantum Monte Carlo sampling is sign-problem limited.
  • Benchmark comparisons on small clusters establish the method before scaling to production runs on bigger lattices.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same replicated-path-integral construction could be adapted to compute other entanglement measures, such as mutual information, for interacting fermions.
  • High-order results obtained this way might serve as input for resummation techniques that reach non-perturbative coupling strengths.
  • The half-shifted replica momenta suggest that the method could be combined with twisted-boundary-condition tricks used in other many-body diagrammatic calculations.

Load-bearing premise

The graded-swap representation exactly converts the Rényi entropy into a replicated path integral whose interaction expansion can be performed and summed via connected determinants without introducing uncontrolled errors.

What would settle it

Compute the order-by-order interaction corrections for the Rényi entropy on a 3x3 Hubbard cluster using the Monte Carlo sampler and verify whether they match the values obtained from exact diagonalization on the same cluster.

Figures

Figures reproduced from arXiv: 2605.21632 by Boyuan Shi.

Figure 1
Figure 1. Figure 1: FIG. 1. Order-by-order interaction corrections. Left: [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
read the original abstract

We develop a direct diagrammatic Monte Carlo framework for the Renyi entanglement entropy of interacting lattice fermions. The method starts from the fermionic graded-swap representation of Z_n[A]=Tr_A\rho_A^n, which converts the entropy problem into a replicated path integral with mixed temporal boundary conditions on the entangling region. In this representation the replica momenta are half-shifted, q_m=(2m+1)\pi/n, and the interaction expansion has a determinant form suitable for connected-determinant summation. We combine this expansion with a many-configuration Markov-chain Monte Carlo sampler to obtain order-by-order corrections for very large systems to very high orders. As a benchmark, we compare the order-by-order coefficients of a 3*3 Hubbard cluster with exact diagonalization. We then report a production calculation for a large periodic lattice with a square subregions. The dominant system-size limitation is therefore memory rather than a conventional auxiliary-field sign problem. The results provide a step toward diagrammatic calculations of fermionic entanglement observables in regimes where direct quantum Monte Carlo sampling is costly or sign-problem limited.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript develops a direct diagrammatic Monte Carlo framework for the Rényi entanglement entropy of interacting lattice fermions. It starts from the fermionic graded-swap representation of Z_n[A] = Tr_A ρ_A^n, which maps the problem to a replicated path integral with mixed temporal boundary conditions on the entangling region and half-shifted replica momenta q_m = (2m+1)π/n. This yields a determinant-form interaction expansion that is sampled order-by-order via many-configuration Markov-chain Monte Carlo. Benchmarks against exact diagonalization are reported for a 3×3 Hubbard cluster, followed by production runs on larger periodic lattices with square subregions; the dominant limitation is stated to be memory rather than the auxiliary-field sign problem.

Significance. If the graded-swap representation is exact and the sampling is free of uncontrolled biases, the method supplies a new route to fermionic entanglement observables on system sizes where direct quantum Monte Carlo is limited by the sign problem. The combination of connected-determinant summation with high-order diagrammatic expansions on large lattices is a technical strength that could enable calculations inaccessible to other approaches. The work therefore represents a concrete step toward diagrammatic treatment of entanglement in sign-problematic regimes.

major comments (2)
  1. [Methods / graded-swap representation] The central claim that the graded-swap representation converts Z_n[A] into a replicated path integral whose interaction expansion is determinant-form and free of uncontrolled errors rests on the assertion that half-shifted momenta together with mixed temporal boundary conditions produce a precise determinant without extra phase factors upon tracing out the complement. A self-contained derivation or explicit check that anticommutation signs cancel correctly under the partial trace is not supplied in the methods section; without it the connected-determinant summation could accumulate systematic bias even if low-order coefficients match on small clusters.
  2. [Results / benchmark] Benchmark paragraph (comparison with exact diagonalization on the 3×3 Hubbard cluster): no error bars, convergence criteria for the order-by-order expansion, or discussion of possible post-hoc choices in the Markov-chain sampling are reported. This leaves the accuracy of the higher-order coefficients only partially supported and weakens the validation of the production runs on larger lattices.
minor comments (2)
  1. [Abstract] The abstract and introduction use the phrase 'very large systems' without a quantitative definition of system size relative to memory limits; a brief statement of the largest lattice treated would improve clarity.
  2. [Introduction] Notation for the replica momenta q_m is introduced without an immediate reference to the corresponding temporal boundary-condition implementation; adding a short equation or diagram would aid readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments. We address each major comment below and indicate planned revisions.

read point-by-point responses

  1. Referee: [Methods / graded-swap representation] The central claim that the graded-swap representation converts Z_n[A] into a replicated path integral whose interaction expansion is determinant-form and free of uncontrolled errors rests on the assertion that half-shifted momenta together with mixed temporal boundary conditions produce a precise determinant without extra phase factors upon tracing out the complement. A self-contained derivation or explicit check that anticommutation signs cancel correctly under the partial trace is not supplied in the methods section; without it the connected-determinant summation could accumulate systematic bias even if low-order coefficients match on small clusters.

    Authors: We agree that a more explicit verification strengthens the methods section. In the revised manuscript we will add a dedicated subsection with a self-contained derivation of the graded-swap representation. This will include an explicit calculation for n=2 and n=3 demonstrating that anticommutation signs cancel under the partial trace, confirming the determinant form contains no extra phase factors. The low-order benchmarks already match exact diagonalization; the added derivation will rule out systematic bias in the connected-determinant summation. revision: yes

  2. Referee: [Results / benchmark] Benchmark paragraph (comparison with exact diagonalization on the 3×3 Hubbard cluster): no error bars, convergence criteria for the order-by-order expansion, or discussion of possible post-hoc choices in the Markov-chain sampling are reported. This leaves the accuracy of the higher-order coefficients only partially supported and weakens the validation of the production runs on larger lattices.

    Authors: We acknowledge the omission. In the revised manuscript we will report statistical error bars obtained from the Markov-chain Monte Carlo runs for each perturbative order, state the convergence criteria (higher-order coefficients smaller than the statistical uncertainty), and describe the sampling protocol (including thermalization, autocorrelation checks, and absence of post-hoc selection) to ensure transparency. These additions will strengthen the validation for both the 3×3 benchmark and the larger-lattice production runs. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation adopts external representation then expands independently

full rationale

The paper begins from the fermionic graded-swap representation of Z_n[A] = Tr_A ρ_A^n as an established starting point that converts the problem into a replicated path integral with mixed temporal boundary conditions and half-shifted momenta q_m = (2m+1)π/n. It then performs an independent determinant-form interaction expansion combined with connected-determinant summation and Markov-chain Monte Carlo sampling. Benchmarks against exact diagonalization on small clusters and production runs on large lattices provide external checks. No equations reduce the final observables to fitted parameters, self-defined quantities, or load-bearing self-citations; the central framework remains self-contained against the adopted representation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the graded-swap representation exactly maps the entanglement entropy onto a path integral amenable to determinant expansion; no free parameters or new entities are introduced in the abstract.

axioms (1)
  • domain assumption The fermionic graded-swap representation of Z_n[A]=Tr_A ρ_A^n converts the entropy problem into a replicated path integral with mixed temporal boundary conditions suitable for determinant-form interaction expansion.
    This is the explicit starting point stated in the abstract on which the entire diagrammatic Monte Carlo framework is built.

pith-pipeline@v0.9.0 · 5720 in / 1553 out tokens · 67684 ms · 2026-05-22T08:52:31.336384+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    The method starts from the fermionic graded-swap representation of Z_n[A]=Tr_A ρ_A^n, which converts the entropy problem into a replicated path integral with mixed temporal boundary conditions on the entangling region. In this representation the replica momenta are half-shifted, q_m=(2m+1)π/n, and the interaction expansion has a determinant form suitable for connected-determinant summation.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    The connected part of this expansion gives ln eZ_n[A]. The normalized entropy object then requires the subtraction ln Z_n[A] = ln eZ_n[A] − n ln Z_1.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages

  1. [1]

    Eisert, M

    J. Eisert, M. Cramer, and M. B. Plenio, Rev. Mod. Phys. 82, 277 (2010)

    work page 2010

  2. [2]

    M. M. Wolf, Phys. Rev. Lett.96, 010404 (2006)

    work page 2006

  3. [3]

    Gioev and I

    D. Gioev and I. Klich, Phys. Rev. Lett.96, 100503 (2006)

    work page 2006

  4. [4]

    Swingle, Phys

    B. Swingle, Phys. Rev. Lett.105, 050502 (2010)

    work page 2010

  5. [5]

    Troyer and U.-J

    M. Troyer and U.-J. Wiese, Phys. Rev. Lett.94, 170201 (2005)

    work page 2005

  6. [6]

    M. B. Hastings, I. Gonzalez, A. B. Kallin, and R. G. Melko, Phys. Rev. Lett.104, 157201 (2010)

    work page 2010

  7. [7]

    Humeniuk and T

    S. Humeniuk and T. Roscilde, Phys. Rev. B86, 235116 (2012)

    work page 2012

  8. [8]

    Grover, Phys

    T. Grover, Phys. Rev. Lett.111, 130402 (2013)

    work page 2013

  9. [9]

    Wang and M

    L. Wang and M. Troyer, Phys. Rev. Lett.113, 110401 (2014)

    work page 2014

  10. [10]

    Broecker and S

    P. Broecker and S. Trebst, J. Stat. Mech.2014, P08015 (2014)

    work page 2014

  11. [11]

    Zhao, Y.-C

    J. Zhao, Y.-C. Wang, Z. Yan, M. Cheng, and Z. Y. Meng, Phys. Rev. Lett.128, 010601 (2022)

    work page 2022

  12. [12]

    Zhao, B.-B

    J. Zhao, B.-B. Chen, Y.-C. Wang, Z. Yan, M. Cheng, and Z. Y. Meng, npj Quantum Mater.7, 69 (2022)

    work page 2022

  13. [13]

    G. Pan, Y. D. Liao, W. Jiang, J. D’Emidio, Y. Qi, and Z. Y. Meng, Phys. Rev. B108, L081123 (2023)

    work page 2023

  14. [14]

    D’Emidio, R

    J. D’Emidio, R. Orús, N. Laflorencie, and F. de Juan, Phys. Rev. Lett.132, 076502 (2024)

    work page 2024

  15. [15]

    Jiang, G

    W. Jiang, G. Pan, Z. Wang, B.-B. Mao, H. Shen, and Z. Yan, arXiv:2409.20009 10.48550/arXiv.2409.20009 (2025), arXiv:2409.20009

    work page doi:10.48550/arxiv.2409.20009 2025

  16. [16]

    Z. Wang, Z. Wang, Y.-M. Ding, B.-B. Mao, and Z. Yan, Nat. Commun.16, 5880 (2025)

    work page 2025

  17. [17]

    Kozik, K

    E. Kozik, K. Van Houcke, E. Gull, L. Pollet, N. Prokof’ev, B. Svistunov, and M. Troyer, EPL90, 10004 (2010)

    work page 2010

  18. [18]

    Y. Deng, E. Kozik, N. V. Prokof’ev, and B. V. Svistunov, 5 EPL110, 57001 (2015)

    work page 2015

  19. [19]

    Wu, M.Ferrero, A.Georges, andE

    W. Wu, M.Ferrero, A.Georges, andE. Kozik,Phys.Rev. B96, 041105 (2017)

    work page 2017

  20. [20]

    W. Wu, M. S. Scheurer, S. Chatterjee, S. Sachdev, A. Georges, and M. Ferrero, Phys. Rev. X8, 021048 (2018)

    work page 2018

  21. [21]

    Šimkovic, R

    F. Šimkovic, R. Rossi, A. Georges, and M. Ferrero, Sci- ence385, eade9194 (2024)

    work page 2024

  22. [22]

    Kozik, E

    E. Kozik, E. Burovski, V. W. Scarola, and M. Troyer, Phys. Rev. B87, 205102 (2013)

    work page 2013

  23. [23]

    Garioud, F

    R. Garioud, F. Šimkovic, R. Rossi, G. Spada, T. Schäfer, F. Werner, and M. Ferrero, Phys. Rev. Lett.132, 246505 (2024)

    work page 2024

  24. [24]

    Lenihan, A

    C. Lenihan, A. J. Kim, F. Šimkovic, and E. Kozik, Phys. Rev. Lett.129, 107202 (2022)

    work page 2022

  25. [25]

    Rossi, Phys

    R. Rossi, Phys. Rev. Lett.119, 045701 (2017)

    work page 2017

  26. [26]

    Šimkovic and R

    F. Šimkovic and R. Rossi, arXiv:2102.05613 10.48550/arXiv.2102.05613 (2021), arXiv:2102.05613

    work page doi:10.48550/arxiv.2102.05613 2021

  27. [27]

    Shi, arXiv:2507.12349 10.48550/arXiv.2507.12349 (2025), arXiv:2507.12349

    B. Shi, arXiv:2507.12349 10.48550/arXiv.2507.12349 (2025), arXiv:2507.12349

    work page doi:10.48550/arxiv.2507.12349 2025

  28. [28]

    Moitra and R

    S. Moitra and R. Sensarma, Phys. Rev. B108, 174309 (2023)

    work page 2023

  29. [29]

    Islam, R

    R. Islam, R. Ma, P. M. Preiss, M. E. Tai, A. Lukin, M. Rispoli, and M. Greiner, Nature528, 77 (2015)

    work page 2015

  30. [30]

    Pichler, L

    H. Pichler, L. Bonnes, A. J. Daley, A. M. Laeuchli, and P. Zoller, New J. Phys.15, 063003 (2013)

    work page 2013

  31. [31]

    Elben, B

    A. Elben, B. Vermersch, M. Dalmonte, J. I. Cirac, and P. Zoller, Phys. Rev. Lett.120, 050406 (2018)

    work page 2018