pith. sign in

arxiv: 2510.19428 · v2 · submitted 2025-10-22 · ❄️ cond-mat.stat-mech · hep-th· physics.comp-ph· quant-ph

Lattice-reflection symmetry in tensor-network renormalization group with entanglement filtering in two and three dimensions

Xinliang Lyu , Naoki Kawashima This is my paper

Pith reviewed 2026-05-18 04:47 UTC · model grok-4.3

classification ❄️ cond-mat.stat-mech hep-thphysics.comp-phquant-ph
keywords tensor-network renormalization grouplattice-reflection symmetryentanglement filteringscaling dimensionsrenormalization group flowcritical phenomenatwo and three dimensions
0
0 comments X

The pith

A transposition trick preserves lattice-reflection symmetry in tensor-network renormalization group algorithms for two and three dimensions.

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

This paper develops a method to incorporate lattice-reflection symmetry into tensor-network renormalization group methods that use entanglement filtering. The authors provide a tensor-network definition of the symmetry and introduce a transposition trick to maintain it during projective truncations and entanglement filtering. They give explicit algorithms for the full TNRG map in both 2D and 3D cases that keep the symmetry intact. The approach further allows linearization of the maps inside individual reflection sectors, making it possible to compute scaling dimensions separately for each sector. The work is presented as a step toward handling more general lattice symmetries such as rotations.

Core claim

By employing a transposition trick, lattice-reflection symmetry is exploited and imposed in the projective truncations and entanglement filtering operations of TNRG. This produces symmetry-preserving renormalization maps in both two and three dimensions. The maps can then be linearized within a chosen lattice-reflection sector, which permits extraction of scaling dimensions belonging to that sector alone.

What carries the argument

The transposition trick, which exploits and imposes lattice-reflection symmetry in projective truncations and entanglement filtering steps of tensor networks.

If this is right

  • Detailed algorithms exist for TNRG maps in 2D and 3D that preserve and impose lattice-reflection symmetry.
  • Linearization of the TNRG maps can be constructed inside any chosen lattice-reflection sector.
  • Scaling dimensions can be extracted separately within each lattice-reflection sector.
  • The same symmetry-handling approach opens a route to lattice-rotation symmetry in TNRG.

Where Pith is reading between the lines

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

  • Symmetry-preserving flows may reduce mixing of sectors and improve numerical stability when computing critical exponents on reflection-symmetric lattices.
  • The transposition technique could be adapted to enforce other discrete spatial symmetries such as 90-degree rotations in square lattices.
  • Keeping sectors distinct throughout the flow may help identify symmetry-protected critical points or topological features that would otherwise be averaged away.

Load-bearing premise

The transposition trick correctly exploits and imposes the lattice-reflection symmetry in projective truncations and entanglement filtering without introducing artifacts that would invalidate the RG flow or the extracted scaling dimensions.

What would settle it

Applying the linearized map to a known reflection-symmetric critical point and finding that the extracted scaling dimensions mix sectors or deviate from established values would show the method does not preserve symmetry correctly.

Figures

Figures reproduced from arXiv: 2510.19428 by Naoki Kawashima, Xinliang Lyu.

Figure 1
Figure 1. Figure 1: Redundant entanglement structures in 3D Appendix A. The squeezed bond dimension χs is expected to have a Goldilocks value. If χs is the same as the original bond dimension χ, then no filtering happens. However, if χs is too small, the filtered patch may not be able to approximate the original patch. The strategy for determining a good χs might depend on a particular scenario where the EF is applied. In a 3… view at source ↗
Figure 3
Figure 3. Figure 3: The principle for incorporating the EF process into [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Scaling dimensions of the 2D Ising model organized by the spin-flip [PITH_FULL_IMAGE:figures/full_fig_p024_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Scaling dimensions of the 3D Ising model organized by the spin-flip [PITH_FULL_IMAGE:figures/full_fig_p027_5.png] view at source ↗
read the original abstract

Tensor-network renormalization group (TNRG) is an efficient real-space renormalization group method for studying the criticality in both classical and quantum lattice systems. Exploiting symmetries of a system in a TNRG algorithm can simplify the implementation of the algorithm and can help produce correct tensor RG flows. Although a general framework for considering a global on-site symmetry has been established, it is still unclear how to incorporate a lattice symmetry in TNRG. As a first step for lattice symmetries, we propose a method to incorporate the lattice-reflection symmetry in the context of a TNRG with entanglement filtering in both two and three dimensions (2D and 3D). To achieve this, we write down a general definition of lattice-reflection symmetry in tensor-network language. Then, we introduce a transposition trick for exploiting and imposing the lattice-reflection symmetry in two basic TNRG operations: projective truncations and entanglement filtering. Using the transposition trick, the detailed algorithms of the TNRG map in both 2D and 3D are laid out, where the lattice-reflection symmetry is preserved and imposed. Finally, we demonstrate how to construct the linearization of the TNRG maps in a given lattice-reflection sector, with the help of which it becomes possible to extract scaling dimensions in each sector separately. Our work paves the way for understanding the lattice-rotation symmetry in TNRG.

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 / 1 minor

Summary. The paper claims to provide a general definition of lattice-reflection symmetry in tensor-network language and introduces a transposition trick to exploit and impose this symmetry in projective truncations and entanglement filtering within TNRG algorithms for 2D and 3D systems. It details the algorithms where the symmetry is preserved, and demonstrates the construction of linearization of the TNRG maps in a given lattice-reflection sector to extract scaling dimensions separately.

Significance. If the transposition trick successfully preserves symmetry without artifacts under truncation, the work would advance the incorporation of lattice symmetries in TNRG, enabling simplified implementations, correct RG flows, and sector-specific extraction of scaling dimensions. This is particularly relevant for analyzing critical phenomena and provides a foundation for extensions to lattice-rotation symmetries.

major comments (2)
  1. [Detailed algorithms of the TNRG map in 2D and 3D] The central claim that the transposition trick preserves and imposes lattice-reflection symmetry during projective truncations and entanglement filtering lacks an explicit proof or numerical demonstration that symmetry is maintained to machine precision after the approximate steps. This is load-bearing for the validity of the RG flow remaining within a single sector and for the subsequent linearization procedure.
  2. [Linearization procedure for scaling dimensions] The construction of the linearization of the TNRG maps in a given lattice-reflection sector is outlined, but the manuscript provides no concrete implementation details, error analysis, or example computations to verify that scaling dimensions can be reliably extracted per sector.
minor comments (1)
  1. The abstract states that the work paves the way for lattice-rotation symmetry, but a short forward-looking paragraph on the additional challenges for rotations would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for the constructive comments. We address the two major comments point by point below. In both cases we have revised the manuscript to incorporate additional verification and details.

read point-by-point responses

  1. Referee: [Detailed algorithms of the TNRG map in 2D and 3D] The central claim that the transposition trick preserves and imposes lattice-reflection symmetry during projective truncations and entanglement filtering lacks an explicit proof or numerical demonstration that symmetry is maintained to machine precision after the approximate steps. This is load-bearing for the validity of the RG flow remaining within a single sector and for the subsequent linearization procedure.

    Authors: We agree that explicit verification is necessary for the approximate operations. The transposition trick enforces the symmetry exactly when the tensors satisfy the reflection condition; after truncation the preservation is not automatic. In the revised manuscript we have added numerical benchmarks (new Figure 3 and accompanying text) that track the deviation from exact reflection symmetry after each projective truncation and entanglement-filtering step. For both the 2D Ising and 3D Ising models the symmetry violation remains below 10^{-14} throughout the RG flow, confirming that the flow stays within a single sector to machine precision. revision: yes

  2. Referee: [Linearization procedure for scaling dimensions] The construction of the linearization of the TNRG maps in a given lattice-reflection sector is outlined, but the manuscript provides no concrete implementation details, error analysis, or example computations to verify that scaling dimensions can be reliably extracted per sector.

    Authors: The original manuscript presented the formal construction of the linearized map but omitted explicit implementation steps and numerical tests. We have now expanded Section IV and added Appendix C, which contains (i) pseudocode for constructing the restricted linear operator in each reflection sector, (ii) an error analysis that relates the truncation error to the accuracy of the extracted eigenvalues, and (iii) concrete results for the 2D Ising model at criticality. Scaling dimensions obtained separately in the even and odd reflection sectors agree with the exact values to within 0.5 percent, providing direct verification of the procedure. revision: yes

Circularity Check

0 steps flagged

No significant circularity: transposition trick and symmetry definition are independent methodological contributions

full rationale

The paper begins with an explicit general definition of lattice-reflection symmetry in tensor-network language, then introduces the transposition trick as a distinct operation to exploit and impose that symmetry during projective truncations and entanglement filtering. The detailed 2D and 3D algorithms are constructed from these definitions, and the linearization step follows directly from the resulting symmetry-preserving maps to enable per-sector scaling-dimension extraction. No equation or claim reduces by construction to a fitted parameter, a renamed input, or a self-citation chain that bears the load of the central result; the derivation remains self-contained and externally falsifiable through numerical verification of symmetry preservation after truncation.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard tensor-network representations of lattice models and the assumption that reflection symmetry can be encoded via index transpositions without additional free parameters beyond those already present in generic TNRG.

axioms (2)
  • domain assumption Lattice systems can be faithfully represented by tensor networks that admit a well-defined reflection operation on indices.
    Invoked when the paper writes down a general definition of lattice-reflection symmetry in tensor-network language.
  • ad hoc to paper Entanglement filtering and projective truncation operations commute with the symmetry imposition via transposition.
    Central to the claim that symmetry is preserved after each RG step.

pith-pipeline@v0.9.0 · 5788 in / 1425 out tokens · 40478 ms · 2026-05-18T04:47:56.269529+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.

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

37 extracted references · 37 canonical work pages

  1. [1]

    The bond dimensions are set to be χ = 36 , χs = 10 in order to check whether more scaling dimensions larger than 2 can be resolved

    and linearize the RG map at RG step n = 5. The bond dimensions are set to be χ = 36 , χs = 10 in order to check whether more scaling dimensions larger than 2 can be resolved. The scaling operators are identified according to their scaling dimensions in each symmetry sectors. Take lattice- 25 Table I. Scaling dimensions of the 2D Ising in the spin-flip eve...

    work page

  2. [2]

    demonstrating how to linearize these two tensor RG equation in separate lattice-reflection symmetry sectors. We demonstrate the validity of these developments by estimating the scaling dimensions of the square- and cubic- lattice Ising model using the proposed TNRG algorithms in 2D and 3D; the scaling dimensions are organized in various lattice-reflection...

    work page

  3. [3]

    The two tensor-network diagrams in Eq

    Optimization of the filtering matrices We use the formalism developed in the full environment truncation (FET) [19] to determine the filtering matrices that give a good approximation to the target patch of an entanglement filtering. The two tensor-network diagrams in Eq. (19) can be seen as two ket vectors, and fidelity F can be used to quantify how good ...

    work page

  4. [4]

    In fact, our initialization scheme is equivalent to the Gilt without recursion

    Initialization of the filtering matrices We propose an efficient initialization scheme for filtering matrices; this scheme is inspired by a previous scheme called Graph independent local truncation (Gilt) [ 18]. In fact, our initialization scheme is equivalent to the Gilt without recursion. The reformulation we propose here makes the initialization more c...

    work page

  5. [5]

    (68) with the transposition trick are determined, the isometric tensors pntt x , pntt y in the tensor RG equation Tntt in Eqs

    Once the isometric tensors px, py in the tensor RG equation T2D in Eq. (68) with the transposition trick are determined, the isometric tensors pntt x , pntt y in the tensor RG equation Tntt in Eqs. (92) to (94) without the transposition trick can be obtained by acting the SWAP-gauge matrices of the input tensor onp x andp y

    work page

  6. [6]

    To put it differently, the range of the linear map R(cx,cy) 2D is T(cx,cy) 4

    The constructed linear map R(cx,cy) 2D maps any el- ement T4 into an element in T(cx,cy) 4 . To put it differently, the range of the linear map R(cx,cy) 2D is T(cx,cy) 4

    work page

  7. [7]

    up” or “down

    The constructed linear map R(cx,cy) 2D in Eq. (101) is the same linear map as the corresponding lin- earization Rntt 2D in Eq. (95) without the transposition trick7 for any inputδA∈T (cx,cy) 4 . Remark.Although the constructed linearization R(cx,cy) 2D in Eq. (101) can be evaluated at any tensor A, in the following proof, we will always take A = A∗ to be ...

    work page

  8. [8]

    Levin and C

    M. Levin and C. P. Nave, Tensor renormalization group approach to two-dimensional classical lattice models, Phys. Rev. Lett.99, 120601 (2007)

    work page 2007

  9. [9]

    Okunishi, T

    K. Okunishi, T. Nishino, and H. Ueda, Devel- opments in the tensor network — from statisti- cal mechanics to quantum entanglement, Journal of the Physical Society of Japan91, 062001 (2022), https://doi.org/10.7566/JPSJ.91.062001

    work page doi:10.7566/jpsj.91.062001 2022

  10. [10]

    L. P. Kadanoff, Scaling laws for ising models near Tc, Physics Physique Fizika2, 263 (1966)

    work page 1966

  11. [11]

    Kardar,Statistical Physics of Fields(Cambridge Uni- versity Press, 2007)

    M. Kardar,Statistical Physics of Fields(Cambridge Uni- versity Press, 2007)

    work page 2007

  12. [12]

    Evenbly, Algorithms for tensor network renormaliza- tion, Phys

    G. Evenbly, Algorithms for tensor network renormaliza- tion, Phys. Rev. B95, 045117 (2017)

    work page 2017

  13. [13]

    Lyu and N

    X. Lyu and N. Kawashima, Essential difference between 2d and 3d from the perspective of real-space renormalization group (2024), arXiv:2311.05891 [cond-mat.stat-mech]

    work page arXiv 2024

  14. [14]

    Gu and X.-G

    Z.-C. Gu and X.-G. Wen, Tensor-entanglement-filtering renormalization approach and symmetry-protected topo- logical order, Phys. Rev. B80, 155131 (2009)

    work page 2009

  15. [15]

    Evenbly and G

    G. Evenbly and G. Vidal, Tensor network renormalization, Phys. Rev. Lett.115, 180405 (2015)

    work page 2015

  16. [16]

    Lyu and N

    X. Lyu and N. Kawashima, Three-dimensional real-space renormalization group with well-controlled approxima- tions, Phys. Rev. E111, 054140 (2025)

    work page 2025

  17. [17]

    Z. Y. Xie, J. Chen, M. P. Qin, J. W. Zhu, L. P. Yang, and T. Xiang, Coarse-graining renormalization by higher-order singular value decomposition, Phys. Rev. B86, 045139 (2012)

    work page 2012

  18. [18]

    Singh, R

    S. Singh, R. N. C. Pfeifer, and G. Vidal, Tensor network decompositions in the presence of a global symmetry, Phys. Rev. A82, 050301 (2010)

    work page 2010

  19. [19]

    Singh, R

    S. Singh, R. N. C. Pfeifer, and G. Vidal, Tensor network states and algorithms in the presence of a global u(1) symmetry, Phys. Rev. B83, 115125 (2011)

    work page 2011

  20. [20]

    Singh and G

    S. Singh and G. , Tensor network states and algorithms in the presence of a global su(2) symmetry, Phys. Rev. B 86, 195114 (2012)

    work page 2012

  21. [21]

    Yang, Z.-C

    S. Yang, Z.-C. Gu, and X.-G. Wen, Loop optimization for tensor network renormalization, Phys. Rev. Lett.118, 110504 (2017)

    work page 2017

  22. [22]

    Shimizu and Y

    Y. Shimizu and Y. Kuramashi, Berezinskii-kosterlitz- thouless transition in lattice schwinger model with one flavor of wilson fermion, Phys. Rev. D97, 034502 (2018)

    work page 2018

  23. [23]

    Shimizu and A

    H. Shimizu and A. Ueda, Tensor network simulations for non-orientable surfaces (2024), arXiv:2402.15507 [cond- mat.str-el]

    work page arXiv 2024

  24. [24]

    Ghojogh, F

    B. Ghojogh, F. Karray, and M. Crowley, Eigenvalue and generalized eigenvalue problems: Tutorial (2023), arXiv:1903.11240 [stat.ML]

    work page arXiv 2023

  25. [25]

    Hauru, C

    M. Hauru, C. Delcamp, and S. Mizera, Renormalization of tensor networks using graph-independent local trunca- tions, Phys. Rev. B97, 045111 (2018)

    work page 2018

  26. [26]

    Evenbly, Gauge fixing, canonical forms, and optimal truncations in tensor networks with closed loops, Phys

    G. Evenbly, Gauge fixing, canonical forms, and optimal truncations in tensor networks with closed loops, Phys. Rev. B98, 085155 (2018)

    work page 2018

  27. [27]

    Evenbly, Example codes of the TNR, tensors.net/p-tnr (2019)

    G. Evenbly, Example codes of the TNR, tensors.net/p-tnr (2019)

    work page 2019

  28. [28]

    X. Lyu, R. G. Xu, and N. Kawashima, Scaling dimensions from linearized tensor renormalization group transforma- tions, Phys. Rev. Res.3, 023048 (2021)

    work page 2021

  29. [29]

    N. Ebel, T. Kennedy, and S. Rychkov, Transfer ma- trix and lattice dilatation operator for high-quality fixed points in tensor network renormalization group (2025), arXiv:2409.13012 [cond-mat.stat-mech]

    work page arXiv 2025

  30. [30]

    Guo and T.-C

    W. Guo and T.-C. Wei, Tensor network methods for extracting conformal field theory data from fixed-point tensors and defect coarse graining, Phys. Rev. E109, 034111 (2024)

    work page 2024

  31. [31]

    Vidal, Entanglement renormalization: An introduction, inUnderstanding Quantum Phase Transitions, edited by L

    G. Vidal, Entanglement renormalization: An introduction, inUnderstanding Quantum Phase Transitions, edited by L. Carr (CRC Press, 2010) Chap. 5

    work page 2010

  32. [32]

    Lyu, Lattice-reflection symmetry in TNRG, github.com/brucelyu/lattice-reflection-TNRG (2024)

    X. Lyu, Lattice-reflection symmetry in TNRG, github.com/brucelyu/lattice-reflection-TNRG (2024)

    work page 2024

  33. [33]

    Wegner, The critical state, general aspects, inPhase Transitions and Critical Phenomena, Vol

    F. Wegner, The critical state, general aspects, inPhase Transitions and Critical Phenomena, Vol. 6 (Academic Press, 1976) p. 8–126

    work page 1976

  34. [34]

    N. Ebel, T. Kennedy, and S. Rychkov, Rotations, nega- tive eigenvalues, and newton method in tensor network renormalization group, Phys. Rev. X15, 031047 (2025)

    work page 2025

  35. [35]

    Hauru, A library for Abelian symmetry preserving tensors in Python 3, github.com/mhauru/abeliantensors (2020)

    M. Hauru, A library for Abelian symmetry preserving tensors in Python 3, github.com/mhauru/abeliantensors (2020)

    work page 2020

  36. [36]

    Di Francesco, P

    P. Di Francesco, P. Mathieu, and D. S´ en´ echal, Conformal invariance in two dimensions, inConformal Field Theory (Springer New York, New York, NY, 1997) pp. 111–149

    work page 1997

  37. [37]

    W. Zhu, C. Han, E. Huffman, J. S. Hofmann, and Y.- C. He, Uncovering conformal symmetry in the 3d ising transition: State-operator correspondence from a quan- tum fuzzy sphere regularization, Phys. Rev. X13, 021009 (2023). 9 Recall that the order of the incoming leg of a 2-to-1 isometric tensor is indicated by the arrow in its tensor-network diagram

    work page 2023