REVIEW 4 major objections 4 minor 4 cited by
Sequential Circuit as Generalized Symmetry on Lattice
T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read This paper argues that a generalized symmetry on a lattice is fully encoded by the sequential circuit that sweeps its twist; in 1D this forces fusion with the conjugate to contain exactly one identity, and in 2D the Cheshire string needs…
desk verdict A real advance in the sMPO formalism, but Claim 3's identity-block step needs proof and the 2D tensors need to be explicit before this is fully convincing. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the sequential matrix product operator (sMPO), a translation-invariant matrix product operator formed by taking the tensor network of a sequential circuit and closing the bulk tensors periodically. Its canonical block-upper-triangular form is what lets the paper count fusion outcomes: an identity channel corresponds to a diagonal block whose physical indices form the identity matrix and whose virtual dimension is one; the transfer matrix built from the tensor has largest eigenvalue $d$, the local physical dimension, and injectivity makes that eigenvalue nondegenerate, so a second identity block cannot exist. In the 2D case, the same tensor-network decomposition of the Cheshire-string sweep produces dressed 1-form projectors $I+\widetilde W_y$ and $I+\widetilde W_x$, whose product is the full symmetry operator.
What would settle it
Contract the canonical form of any proposed 1D sMPO on finite chains and inspect the transfer matrix obtained by contracting its physical indices: the paper predicts a unique eigenvector with eigenvalue $d$, the local Hilbert space dimension, belonging to the single identity channel. Finding a second eigenvector at eigenvalue $d$, or an eigenvalue larger than $d$, would refute the classification; so would an explicit open-boundary contraction that fails to equal the identity on a chain of some length. For the Cheshire case, one can compare circuits that generate the twist along a deformed line: the paper predicts the difference is a local $B_v$ projector, so a computation showing extra nonlocal content would falsify it.
Extended reading notes
Core claim
The central claim is that the unitary sequential circuit which moves a symmetry twist across the lattice determines the non-invertible part of the symmetry, provided the twist can be pair-created by a finite-depth circuit (in 1D) or by an additional 1D sequential circuit (in 2D). In 1D, closing the matrix product operator of the circuit into a translation-invariant sMPO reproduces the full Kramers-Wannier operator $D=(1+\eta)U_{\rm KW}R(X_1)/\sqrt{2}$ and gives the fusion rule $D^\dagger \times D = I + \eta$. More generally, the canonical block decomposition of the sMPO shows that fusing a simple symmetry with its conjugate contains one and only one identity block, so every 1D generalized symmetry on a tensor-product Hilbert space satisfies $D^\dagger \times D = I + \cdots$ with uniqueness of the dagger, and fusion coefficients are non-negative integers. For the Cheshire string in the toric code, the 2D sweeping circuit yields a dressed 1-form projector $I+\widetilde W_y$, and the 1D generation/annihilation circuit yields $I+\widetilde W_x$; together they give the full symmetry $C=(I+W_{x_0})\prod_i(I+W_{y_i})$, equivalent on the ground space to projection onto the symmetric subspace of the 1-form symmetry, with the isotropic form $C'=\sum_L W_L$ being a short-range-correlated, long-range-entangled $\mathbb{Z}_2$-injective tensor network operator.
Load-bearing premise
The proof of Claim 3 rests on an asserted, unproven structural property: with a suitably chosen open boundary condition, the symmetry's matrix product operator reduces exactly to the identity operator on a chain of any length, and the uniqueness of the identity fusion channel depends on that fact.
Editorial extensions
If this is right
- Every simple generalized symmetry in one dimension on a tensor-product Hilbert space is annihilable: fusing it with its Hermitian conjugate produces the identity exactly once, so non-invertibility in 1D always has a unique inverse channel.
- Starting from only the sequential circuit that sweeps a twist, one can write down the full non-invertible operator by closing the circuit into a translation-invariant sMPO; no extra input about how twists are created or destroyed is needed in 1D.
- The fusion of two simple 1D symmetries closes into an algebra with non-negative integer coefficients, and each simple symmetry has a unique conjugate, mirroring the fusion rules of anyons in 2D topological states.
- For the Cheshire string, the full symmetry is $C=(I+W_{x_0})\prod_i(I+W_{y_i})$ on the toric code, equivalently a projection onto the symmetric subspace of the 1-form symmetry; the isotropic version $C'=\sum_L W_L$ is short-range correlated but long-range entangled.
- The 2D tensor network operator for this symmetry is not injective but only $\mathbb{Z}_2$-injective, a structure shared with string-net wavefunctions, so 2D generalized symmetries can carry a form of topological order that 1D symmetries cannot.
Reading between the lines
- A practical extension would be to automate fusion-rule extraction: given any 1D sequential circuit, build its sMPO, compute the canonical decomposition, and read off the fusion channels and positive integer coefficients as a numerical tool for discovering non-invertible symmetries in lattice models.
- The paper's proof stops at tensor-product Hilbert spaces; testing the same reconstruction on constrained-space symmetries such as the golden-chain (Fibonacci) symmetry would show whether the annihilable property and the unique-dagger channel survive there, as the authors conjecture.
- The annihilable/unannihilable distinction can be read as a circuit-depth diagnostic: if a symmetry twist cannot be pair-created by a finite-depth circuit, the corresponding symmetry is unannihilable and needs an extra generating circuit, giving a lattice criterion for classifying defects without invoking field theory.
- In higher dimensions the two-step pattern likely iterates: a symmetry twist of dimension $d$ would be moved by a $(d+1)$-dimensional sequential sweep and generated by a $d$-dimensional sequential circuit, with the resulting symmetry operator expected to be a gauge-injective tensor network; an explicit 3D construction would test this.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper argues that a lattice generalized symmetry, understood through the field-theoretic picture of topological defects, is implemented in the bulk by a sequential quantum circuit, and that the full non-invertible symmetry action can be reconstructed from that circuit together with ancillary boundary operations. In 1D, the authors introduce sequential matrix product operators (sMPOs), show that they form a closed algebra, that fusion of simple sMPOs yields nonnegative integer combinations, that the fusion of a simple sMPO with its Hermitian conjugate contains exactly one identity channel (the 'annihilable' property), and that the dagger is unique. These claims are illustrated with the Kramers-Wannier transformation. In 2D, the paper studies a Cheshire-string symmetry twist in the toric code and argues that combining the 2D sweeping circuit with a 1D generation/annihilation circuit produces a projection onto the 1-form symmetric subspace of the toric code.
Significance. If the central claims hold, this gives a concrete lattice-level characterization of generalized symmetries in 1D, explains why non-invertible symmetries such as Kramers-Wannier are annihilable, and distinguishes them from genuinely unannihilable symmetries such as the Cheshire-string symmetry. The derivation is constructive rather than circular: the Kramers-Wannier fusion rule is reproduced from the sMPO tensor calculation rather than used as an input, and the 2D construction explicitly builds the symmetry operator from the sweeping and generating circuits. The proposed terminology and the connection to the canonical form of matrix product operators are potentially useful for future work. The main risk is that the proof of Claim 3 relies on an unproven structural assertion about open boundary conditions, and this assertion is load-bearing for the annihilability theorem and for the uniqueness argument in Claim 4.
major comments (4)
- [Sec. IV, Claim 3 proof] The proof of Claim 3 asserts, in the paragraph beginning 'First, we argue that there has to be at least one identity block on the diagonal,' that 'with a properly chosen open boundary condition, Mα should represent the identity operator I ⊗ ... ⊗ I on a chain of arbitrary length.' This is the only stated reason for the existence of an identity diagonal block in the canonical decomposition of Mα, and it is also reused implicitly in Eq. (44), where Dα† × Dα is said to differ from Uα†Uα = I only by boundary operations. The assertion is not derived from the local gate decomposition in Eq. (12), from the canonical form theorem in Appendix A, or from any explicit construction of boundary vectors. As written, this is an unproven structural assumption rather than a proof step. Since Eq. (42) and Claim 4 both depend on it, please replace this assertion with a proof that for an injective sMPO arising from a sequential circuit there exist boundary vectors l,r such that l Mα^N r = I for all chain lengths N, or state precisely which additional conditions on the sequential circuit are needed and prove them.
- [Sec. IV, Claim 2 proof] The proof of Claim 2 asserts that after the canonical decomposition of Mβ × Mα, projecting the ancillas onto the subspace Vγ gives an sMPO λγMγ, and that this fixes the normalization so that no prefactor |λγ| ≠ 1 is allowed. The statement that the projected tensor inherits the sMPO property is not demonstrated: one must show that the resulting bond-dimension-dim(Vγ) operator can be implemented by a sequential circuit with ancillas and non-unitary operations only at the ends, and that its normalization satisfies the same spectral-radius condition. This step is used again in Claim 3 when the identity block is identified with a sequential circuit and its normalization is fixed. Please provide the explicit construction or state it as an additional assumption.
- [Sec. V, tensor network derivation after Fig. 8] The text states that after composing the steps of the 2D sequential circuit, 'the tensors on the horizontal edges are ..., the tensors on the vertical edges are ..., and the tensor in each plaquette is ...', but in the manuscript I reviewed these displayed tensor expressions are missing, as are the corresponding expressions for the generation/annihilation circuit. This makes it impossible to verify the derivation of the projection operator in Eq. (50) and the subsequent claim that the symmetry action projects onto the 1-form symmetric subspace. Please ensure that all tensor network diagrams or explicit tensor formulas are included and that the notation for the loops Ly, L−y, L+y is defined in the text.
- [Sec. V, Eqs. (52)-(53)] The passage from the projector C in Eq. (52) to the loop-summation operator C′ in Eq. (53) is sketched rather than proven. In particular, the claim that projectors onto local Bv terms must be included in the symmetry action, and the assertion that C and C′ 'have the same action' on the toric code ground state, need a more precise statement about the Hilbert space being acted on and about the sense in which C′ is a symmetry operator if it cannot be implemented by the sequential-circuit protocol. Without this clarification, it is difficult to assess whether C′ is a genuine lattice generalized symmetry or only a representative of the same operation on the ground-state subspace.
minor comments (4)
- [Sec. II, paragraph after Eq. (5)] The statement that 0d symmetry twists in 1D 'can always be (pair)-generated from vacuum with local unitary transformations' should be reconciled with the later discussion in Sec. V, where 1D symmetry twists in 2D systems require a sequential circuit for pair-creation; a short remark distinguishing the dimensionality of the twist would prevent confusion.
- [Sec. IV, Eq. (32) and Claim 1] The text says that 'we expect simple symmetries to be represented by injective MPOs' but then uses injectivity as a standing assumption in Claims 3 and 4. Please state explicitly whether injectivity of the sMPO is part of the definition of 'simple' or a separate assumption.
- [Sec. IV, Claim 1 proof] The proof of Claim 1 refers to a 'controlled sequential circuit' for aDα + bDβ and says 'this can be done in a sequential way,' but the gate ordering is not shown; a brief explicit ordering or a small diagram would make the sequential-circuit property transparent.
- [Appendix A, Eq. (A8)] The normalization ∑_ii′ M_ii′ (M_ii′)† = d I for injective sMPOs is stated without derivation; since this normalization is used in Eq. (43) to identify the eigenvalue d, it would be helpful to derive it directly from the sequential-circuit decomposition.
Circularity Check
No significant circularity: the main derivations are constructive and the imported Cheshire-string circuits are prior published inputs, not the target conclusions.
full rationale
The 1D construction starts from a given sequential circuit U, builds the sMPO M by translation-invariant contraction of bulk tensors (Eqs. 12-14), and then computes D†D and D†OD; the Kramers-Wannier fusion rule I+η (Eqs. 19-25) emerges from the tensor calculation and is checked against the known result only afterwards. Claims 1-4 are derived from the sMPO/canonical-form setup; Claim 3's 'properly chosen open boundary condition' assertion (Sec. IV) is a consequence of the sMPO being the bulk of a sequential circuit (so U†U = I) rather than an assumption of the target identity-block statement, and should be read as a rigor gap rather than circularity. The 2D Cheshire-string circuits are imported from Ref. 22 (same authors), but they are explicit constructive inputs reviewed in Eqs. (49)-(51); the paper then computes the tensor-network representation and obtains the projection (Eqs. 50-52) rather than renaming the cited result. There are no fitted parameters called predictions, no uniqueness theorem imported from the authors' prior work as a black box, and no definition that bakes in the claimed fusion rule. The Outlook's admission that the constrained-Hilbert-space proof is left to future study is an honest scope limitation, not a circular step.
Assumptions & free parameters
assumptions (5)
- domain assumption A lattice generalized symmetry twist can be moved by local unitary steps, so the bulk sweep is a sequential circuit.
- standard math The standard canonical form of translation-invariant MPS and MPO applies, including the block upper triangular decomposition and spectral radius properties.
- ad hoc to paper For a sequential MPO Mα, a properly chosen open boundary condition makes it represent the identity operator on a chain of arbitrary length.
- ad hoc to paper Projecting ancillas onto the subspace Vγ in the canonical decomposition yields an sMPO with fixed normalization.
- domain assumption The Cheshire-string generation, sweeping and annihilation circuits of Ref. 22 are correct and have the stated effect on the toric code.
Cite this review
Pith. "Pith review of Sequential Circuit as Generalized Symmetry on Lattice." pith.science (2026). https://pith.science/paper/N6EXBUJZ
@misc{pith2026250722394,
author = {Pith},
title = {Pith review of: Sequential Circuit as Generalized Symmetry on Lattice},
year = {2026},
howpublished = {\url{https://pith.science/paper/N6EXBUJZ}},
note = {Machine review of arXiv:2507.22394}
}
read the original abstract
Generalized symmetry extends the usual notion of symmetry to ones that are of higher-form, acting on subsystems, non-invertible, etc. The concept was originally defined in the field theory context using the idea of topological defects. On the lattice, an immediate consequence is that a symmetry twist is moved across the system by a sequential quantum circuit. In this paper, we ask how to obtain the full, potentially non-invertible symmetry action from the unitary sequential circuit and how the connection to sequential circuit constrains the properties of the generalized symmetries. We find that for symmetries that contain the trivial symmetry operator as a fusion outcome, which we call annihilable symmetries, the sequential circuit fully determines the symmetry action and puts various constraints on their fusion. In contrast, for unannihilable symmetries, like that whose corresponding twist is the Cheshire string, a further 1D sequential circuit is needed for the full description. Matrix product operator and tensor network operator representations play an important role in our discussion.
Figures
Figures from the paper (3 more)
Forward citations
Cited by 4 Pith papers
-
Symmetry Spans and Enforced Gaplessness
Symmetry spans enforce gaplessness when a symmetry E embedded into two larger symmetries C and D has no compatible gapped phase that restricts from both.
-
Invariants of Sequential Circuits and Generalized Non-Abelian Statistics
Sequential circuit invariants detect non-invertible symmetry anomalies and characterize non-Abelian fermionic loops plus a new mixed topological order in (3+1)D.
-
Non-Invertible Symmetries on Tensor-Product Hilbert Spaces and Quantum Cellular Automata
Any weakly integral fusion category admits a QCA-refined realization on tensor-product Hilbert spaces with QCA and symmetry indices fixed by the categorical data under defect assumptions.
-
Lattice Gauging Interfaces and Noninvertible Defects in Higher Dimensions
Explicit lattice constructions of gauging interfaces and condensation defects are given for higher-dimensional systems with higher-form symmetries, using movement operators to manage constrained Hilbert spaces.
Reference graph
Works this paper leans on
-
[1]
Place one near the first physical qubit and one near the last
Introduce two ancilla qubits in the maximally entangled state |v0⟩ = 1√ 2 |00⟩ + 1√ 2 |11⟩. Place one near the first physical qubit and one near the last
-
[2]
Apply a controlled operation controlled by the first ancilla qubit on the first physical qubit |0⟩⟨0| ⊗I + |1⟩⟨1| ⊗Z
-
[3]
Apply the sequential unitaries UKW = uN −1,N ...u2,3u1,2
-
[4]
Apply a controlled operation controlled by the second ancilla qubit on the last physical qubit |0⟩⟨0| ⊗R(X) + |1⟩⟨1| ⊗iZR(X)
-
[5]
Project the two ancilla qubits back into the maximally entangled state |v0⟩ = 1√ 2 |00⟩ + 1√ 2 |11⟩. By following the evolution of the state through this procedure, we see how full symmetry action is achieved. In fact, when using the ancilla to simulate the contraction of the virtual indices, we miss by an overall factor of √ 2. To correct for this, we mo...
-
[6]
This normalization is important to match the fusion rule D† × D= I + η, where the coefficient before each symmetry is a positive integer. IV. 1D GENERALIZED SYMMETR Y AS MA TRIX PRODUCT OPERA TORS The Kramers-Wannier example discussed in the last section captures all the key features of generalized symmetries in 1D. The bulk of their action is composed of...
-
[7]
‘Simple’ objects in the space of symmetry operators are short-range correlated operators
1D generalized symmetries form a closed algebra under composition (fusion) and linear superposition. ‘Simple’ objects in the space of symmetry operators are short-range correlated operators
-
[8]
The fusion of two simple symmetries can result in the sum of more than one simple symmetry, each with a non-negative integer coefficient
Show all 40 references
-
[9]
The result of fusion of a simple symmetry with its Hermitian conjugate contains one and only one summand that is identity
-
[10]
The fusion of the Hermitian conjugate of a symmetry with a different symmetry cannot contain the identity channel. Claim 1. 1D generalized symmetries, represented by Sequential Matrix Product Operators, form a closed algebra under composition (fusion) and linear superposition....
-
[11]
(47) where × represents anyon fusion and e is the trivial anyon
Each anyons α in 2D topological states always have an inverse-anyon ¯α such that α × ¯α = e + ... (47) where × represents anyon fusion and e is the trivial anyon. e shows up only once in the decomposition
-
[12]
The inverse of an anyon is unique
-
[13]
The fusion of two anyons can result in the sum of multiple anyons with non-negative integer coefficient. This is of course nothing surprising because when realized in the Symmetry Topological Field Theory formalism, non- invertible 1D symmetries are represented by the string 1...
-
[14]
Gaiotto, A
D. Gaiotto, A. Kapustin, N. Seiberg, and B. Willett, Generalized global symmetries, Journal of High Energy Physics 2015, 172 (2015)
2015
-
[15]
McGreevy, Generalized symmetries in condensed matter, Annual Review of Condensed Matter Physics 14, 57 (2023)
J. McGreevy, Generalized symmetries in condensed matter, Annual Review of Condensed Matter Physics 14, 57 (2023)
2023
-
[16]
Sch¨ afer-Nameki, Ictp lectures on (non-)invertible generalized symmetries, Physics Reports 1063, 1 (2024), iCTP lectures on (non-)invertible generalized symmetries
S. Sch¨ afer-Nameki, Ictp lectures on (non-)invertible generalized symmetries, Physics Reports 1063, 1 (2024), iCTP lectures on (non-)invertible generalized symmetries
2024
-
[17]
Shao, What’s done cannot be undone: Tasi lectures on non-invertible symmetries (2024), arXiv:2308.00747 [hep-th]
S.-H. Shao, What’s done cannot be undone: Tasi lectures on non-invertible symmetries (2024), arXiv:2308.00747 [hep-th]
2024 arXiv
-
[18]
Feiguin, S
A. Feiguin, S. Trebst, A. W. W. Ludwig, M. Troyer, A. Kitaev, Z. Wang, and M. H. Freedman, Interacting anyons in topological quantum liquids: The golden chain, Phys. Rev. Lett. 98, 160409 (2007)
2007
-
[19]
Inamura, On lattice models of gapped phases with fusion category symmetries, Journal of High Energy Physics 2022, 36 (2022)
K. Inamura, On lattice models of gapped phases with fusion category symmetries, Journal of High Energy Physics 2022, 36 (2022)
2022
-
[20]
Garre-Rubio, L
J. Garre-Rubio, L. Lootens, and A. Moln´ ar, Classifying phases protected by matrix product operator symmetries using matrix product states, Quantum 7, 927 (2023)
2023
-
[21]
Fechisin, N
C. Fechisin, N. Tantivasadakarn, and V. V. Albert, Noninvertible symmetry-protected topological order in a group-based cluster state, Phys. Rev. X 15, 011058 (2025)
2025
-
[22]
Seifnashri and S.-H
S. Seifnashri and S.-H. Shao, Cluster state as a non- invertible symmetry protected topological phase (2024), arXiv:2404.01369 [cond-mat.str-el]
2024
-
[23]
Bhardwaj, L
L. Bhardwaj, L. E. Bottini, S. Schafer-Nameki, and A. Tiwari, Lattice models for phases and transitions with non-invertible symmetries (2025), arXiv:2405.05964 [cond- mat.str-el]
2025 arXiv
-
[24]
Bhardwaj, L
L. Bhardwaj, L. E. Bottini, D. Pajer, and S. Sch¨ afer- Nameki, Gapped phases with non-invertible symmetries: (1+1)d, SciPost Phys. 18, 032 (2025)
2025
-
[25]
Bhardwaj, L
L. Bhardwaj, L. E. Bottini, D. Pajer, and S. Schafer- Nameki, Categorical landau paradigm for gapped phases (2023), arXiv:2310.03786 [cond-mat.str-el]
2023 arXiv
-
[26]
Chen, Z.-X
X. Chen, Z.-X. Liu, and X.-G. Wen, Two-dimensional symmetry-protected topological orders and their pro- tected gapless edge excitations, Phys. Rev. B 84, 235141 (2011)
2011
-
[27]
Levin and Z.-C
M. Levin and Z.-C. Gu, Braiding statistics approach to symmetry-protected topological phases, Physical Review B 86, 10.1103/physrevb.86.115109 (2012)
2012 doi
-
[28]
X. Chen, A. Dua, M. Hermele, D. T. Stephen, N. Tantivasadakarn, R. Vanhove, and J.-Y. Zhao, Sequential quantum circuits as maps between gapped phases, Phys. Rev. B 109, 075116 (2024)
2024
-
[29]
Bhardwaj, L
L. Bhardwaj, L. E. Bottini, L. Fraser-Taliente, L. Gladden, D. S. W. Gould, A. Platschorre, and H. Tillim, Lectures on generalized symmetries (2023), arXiv:2307.07547 [hep-th]
2023 arXiv
-
[30]
H. A. Kramers and G. H. Wannier, Statistics of the two-dimensional ferromagnet. part i, Phys. Rev. 60, 252 (1941)
1941
-
[31]
Ignacio Cirac, D
J. Ignacio Cirac, D. Perez-Garcia, N. Schuch, and F. Verstraete, Matrix product unitaries: structure, symmetries, and topological invariants, Journal of Statistical Mechanics: Theory and Experiment 2017, 083105 (2017)
2017
-
[32]
M. B. S ¸ahino˘ glu, S. K. Shukla, F. Bi, and X. Chen, Matrix 15 product representation of locality preserving unitaries, Phys. Rev. B 98, 245122 (2018)
2018
-
[33]
Styliaris, R
G. Styliaris, R. Trivedi, D. Perez-Garcia, and J. I. Cirac, Matrix-product unitaries: Beyond quantum cellular automata, Quantum 9, 1645 (2025)
2025
-
[34]
However, this operation also requires a sequential circuit that scales with the size of the loop
The fact that these defects are unannihilable does not contradict the fact that a closed loop of such a defect can be shrunken to the vacuum. However, this operation also requires a sequential circuit that scales with the size of the loop
-
[35]
Tantivasadakarn and X
N. Tantivasadakarn and X. Chen, String operators for cheshire strings in topological phases, Phys. Rev. B 109, 165149 (2024)
2024
-
[36]
Swingle and X.-G
B. Swingle and X.-G. Wen, Topological properties of tensor network states from their local gauge and local symmetry structures (2010), arXiv:1001.4517 [cond- mat.str-el]
2010 arXiv
-
[37]
Schuch, I
N. Schuch, I. Cirac, and D. P´ erez-Garc ´ ıa, Peps as ground states: Degeneracy and topology, Annals of Physics 325, 2153 (2010)
2010
-
[38]
Perez-Garcia, F
D. Perez-Garcia, F. Verstraete, M. M. Wolf, and J. I. Cirac, Matrix product state representations, Quant. Inf. Comput. 7, 401 (2007), arXiv:quant-ph/0608197. Appendix A: Canonical form of matrix product state In this section, we review the canonical form of the matrix product ...
2007 arXiv
-
[39]
P i Am† i ΛmAm i = Λ m, for some diagonal positive and full-rank matrices Λ m
-
[40]
Note that replacing all the upper triangular blocks F m,m′ i by 0 does not change the physical wavefunction of the MPS
I is the only fixed point of the operator Tm(X) =P i Am i XA m† i . Note that replacing all the upper triangular blocks F m,m′ i by 0 does not change the physical wavefunction of the MPS. In our calculations in this paper, we work with Matrix Product Operator (MPO) representat...
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.