Computing theta-dependent mass spectrum of the 2-flavor Schwinger model in the Hamiltonian formalism

Akira Matsumoto; Etsuko Itou; Yuya Tanizaki

REVIEW 2 major objections 5 minor 16 references

Computing theta-dependent mass spectrum of the 2-flavor Schwinger model in the Hamiltonian formalism

T0 review · 2 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The theta-dependent pion and sigma masses in the 2-flavor Schwinger model follow the bosonization predictions $M_\pi(\theta)=M_\pi(0)|\cos(\theta/2)|^{2/3}$ and $M_\sigma(\theta)=\sqrt{3}M_\pi(\theta)$.

desk verdict A clean proceedings summary of the authors' own JHEP result; no new physics here, but the two-scheme cross-check and the bosonization agreement are real and worth citing via the original paper. read the letter →

arxiv 2501.18960 v2 pith:VXK3C6YE submitted 2025-01-31 hep-lat

Akira Matsumoto , Etsuko Itou , Yuya Tanizaki This is my paper · ORCID

classification hep-lat

keywords 2-flavorSchwingermodelthetadependenceDMRGtensornetworkHamiltonianformalismmesonspectrumbosonization

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Using a tensor-network DMRG computation in the Hamiltonian formalism, this paper determines the $\theta$-dependent meson masses of the 2-flavor Schwinger model, a (1+1)-dimensional gauge theory used as a testbed for non-perturbative methods. It claims that the pion and $\sigma$ meson stay stable for nonzero $\theta$, with masses matching the bosonization formulas $M_\pi(\theta)=M_\pi(0)|\cos(\theta/2)|^{2/3}$ and $M_\sigma(\theta)=\sqrt{3}M_\pi(\theta)$, while the eta meson becomes unstable. Two independent extractions, an improved one-point-function method with operator mixing handled by correlation-matrix diagonalization, and a dispersion-relation method based on momentum-identified excited states, agree with each other across $0\le\theta<\pi$. At $\theta=\pi$ the one-point functions match the $SU(2)_1$ WZW conformal-field-theory forms, indicating a nearly conformal theory. The Hamiltonian approach is free from the sign problem and yields more precise results in the large-$\theta$ region than the reweighted Monte Carlo comparison.

What carries the argument

The machinery that carries the argument is DMRG applied to the gauge-fixed lattice Hamiltonian written as a spin chain via staggered fermions and the Jordan-Wigner transformation, combined with two independent spectral-extraction schemes. The improved one-point-function scheme uses 'wing' boundary sites with different fermion masses to excite a meson from the boundary, defines meson operators by diagonalizing the correlation matrix in Eq. (3) to resolve $\sigma$-$\eta$ mixing, and fits the exponential decay of the one-point function including a second-excited-state term. The dispersion-relation scheme generates excited energy eigenstates with the orthogonality penalty term in Eq. (7), labels them by isospin, and fits $E=\sqrt{K^2+M^2}$ to read off the mass at $K^2\to0$. The analytic formulas $M_\pi(\theta)=M_\pi(0)|\cos(\theta/2)|^{2/3}$ and $M_\sigma=\sqrt{3}M_\pi$ are the bosonization identities that the numerics are checked against.

What would settle it

Repeat the DMRG run at $\theta/2\pi=0.5$ with $N=320$, $a=0.25$, and bond dimension $D=4000$ or larger, and compare the fitted pion mass with the published value; a significant shift would show the agreement with $|\cos(\theta/2)|^{2/3}$ is a truncation artifact. Alternatively, compute the one-point functions at $\theta=\pi$ on a larger lattice, say $N=640$, and check whether the WZW fitting forms in Eq. (6) continue to describe the data in the bulk.

Watch

Extended reading notes

Core claim

The central claim is that the $\theta$-dependent spectrum of the 2-flavor Schwinger model is computable with controlled accuracy in the Hamiltonian formalism, and that the resulting pion and $\sigma$ masses agree with the bosonized model at fermion mass $m/g=0.1$: $M_\pi(\theta)\propto|\cos(\theta/2)|^{2/3}$ and $M_\sigma(\theta)=\sqrt{3}M_\pi(\theta)$. At $\theta=\pi$ the system is nearly conformal, and the one-point functions of the pion and $\sigma$ operators are described by the $SU(2)_1$ WZW CFT forms of Eq. (6). The eta meson, stable at $\theta=0$, becomes unstable for $\theta\neq0$ because the $\theta$ term breaks parity and $G$-parity, causing $\sigma$-$\eta$ mixing and $\eta\to\pi\pi$ decay. The mutual agreement of the two numerical schemes and their consistency with the analytic predictions constitute the paper's evidence.

Load-bearing premise

The computation assumes that bond dimension $D\simeq1400$ at lattice size $N=320$ and spacing $a=0.25$ is sufficient to converge the low-lying states for all $\theta$, including near $\theta=\pi$ where the mass gap is small and the entanglement entropy grows as $(c/3)\log N$.

Editorial extensions

If this is right

The pion mass formula $M_\pi(\theta)\propto|\cos(\theta/2)|^{2/3}$ holds numerically across the full range $0\le\theta<\pi$ at $m/g=0.1$.
The sigma mass is $\sqrt{3}$ times the pion mass within numerical precision, confirming the WKB/bosonization mass ratio.
At $\theta=\pi$ the one-point functions of the sigma and pion follow the $SU(2)_1$ WZW CFT predictions, so the model is nearly conformal there.
The eta meson is not a stable particle for $\theta\neq0$; the spectrum there consists of stable pions and sigma mesons plus scattering states.
The Hamiltonian formulation provides a sign-problem-free route to $\theta$-dependent observables that is more accurate than reweighting Monte Carlo at large $\theta$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The agreement with bosonization at $m/g=0.1$ suggests the analytic formulas remain reliable beyond the strict $m/g\ll1$ regime, which could be tested by pushing to larger fermion masses where the bosonized description should eventually break down.
The dispersion-relation scheme does not rely on local operators, so it may transfer directly to other sign-problematic regimes such as finite density, where identifying the relevant excitations is harder.
The $\theta=\pi$ one-point functions could be used to extract finite-size CFT data such as conformal dimensions, linking the lattice results to the $SU(2)_1$ WZW model more quantitatively.
If the same Hamiltonian approach scales to higher dimensions on quantum computers, it offers a route to $\theta$-dependent spectra in four-dimensional QCD, where conventional Monte Carlo methods face the sign problem.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Bottom line first: this is a LATTICE2024 proceedings that explicitly says it is based on ref. [1], the authors' JHEP paper. As a standalone research preprint it contains no new result. If you are looking for original physics, skip it and read the JHEP version. If you want a concise, readable summary of the DMRG approach to theta-dependent masses in the 2-flavor Schwinger model, this does that job.

What it does well: the central comparison is not circular. The bosonization formulas M_pi(theta) = M_pi(0)|cos(theta/2)|^(2/3) and M_sigma = sqrt(3) M_pi are external analytic predictions, and only the overall scale is fitted at theta=0. Two independent numerical schemes—improved one-point functions and dispersion relations—agree with each other and with these formulas across the whole theta range. That is a genuine cross-check. At theta=pi, the one-point functions are shown to match the SU(2)_1 WZW forms, a nice consistency test. The authors are also careful to say the range of applicability of the bosonized description at m/g=0.1 remains a theoretical question, which is the right amount of humility.

Soft spots, in proportion. The numerical evidence is all at fixed lattice spacing and volume: a=0.25, N=320 for the one-point scheme, and a=0.2, N=100 for the dispersion scheme. There is no continuum extrapolation, no thermodynamic limit, and no bond-dimension convergence check shown. Near theta=pi, where the gap is a small fraction of 1/L and entanglement entropy grows as (c/3) log N, the extracted masses could shift by more than the plotted fitting errors if D=1400 is not fully converged. The agreement between the two schemes is reassuring, but it does not by itself validate the continuum comparison. None of this is fatal for a proceedings; it is the standard caveat of a fixed-lattice calculation. What is more worrisome is the long unmarked block of text in Section 1 (the eta-correlator discussion with 'phase problem' and the figure captions) that appears to be copied from Ref. [5], the 2003 Monte Carlo paper. If that passage is actually in the PDF, it needs to be removed or properly quoted. It obscures which words are the authors' own.

Who gets value from this: people working on tensor-network methods for gauge theories, or on theta dependence in (1+1)-d models, will find a compact account of a credible numerical result. It is a fine proceedings contribution, but as a research paper it is a duplicate of the authors' own JHEP work. I would not send it to a referee as a new result; I would cite the JHEP paper for the physics and treat this as a talk summary. If the unmarked quote is real, the authors should fix that before posting.

Referee Report

2 major / 5 minor

Summary. This proceedings paper applies DMRG to the two-flavor Schwinger model at m/g=0.1 and theta in [0,pi]. Section 3 develops an improved one-point-function scheme: meson operators are defined by diagonalizing correlation matrices (Eq. 5) to handle sigma-eta and pion mixing, and masses are extracted from exponential decays of boundary-induced one-point functions. Section 4 uses a dispersion-relation scheme in which excited states are generated by penalty DMRG and the energies are plotted against momentum expectation values (Fig. 5). The two schemes are compared in Fig. 1 with the bosonization predictions M_pi(theta)=M_pi(0)|cos(theta/2)|^(2/3) and M_sigma=sqrt(3) M_pi, with only M_pi(0) fitted. At theta=pi the one-point functions are fitted to SU(2)_1 WZW forms (Eq. 6). The authors report that the eta becomes unstable for theta != 0 and emphasize that the Hamiltonian approach is sign-problem-free and more accurate than reweighting Monte Carlo.

Significance. The main strength is the cross-check of two independent extraction schemes and the comparison with external analytic predictions whose functional form and mass ratio are not fitted. If the agreement survives continuum and thermodynamic control, this is a valuable demonstration of Hamiltonian tensor-network spectroscopy for theta-dependent quantities. The manuscript is, however, a proceedings contribution: the parameter counts in the CFT fits and the absence of systematic-error analysis leave the central quantitative claim only partially supported within this paper.

major comments (2)

[Section 2, Fig. 1] The load-bearing comparison in Fig. 1 is made at a single lattice spacing and volume for each scheme (one-point scheme: a=0.25, N=320; dispersion scheme: a=0.2, N=100), and no DMRG bond-dimension convergence check is presented anywhere in the paper. Because the gap becomes small as theta approaches pi and the entanglement entropy grows as (c/3) log N, finite-volume, cutoff, and truncation effects can plausibly change the extracted masses near theta=pi by more than the plotted fitting errors. The claim in the abstract that the Hamiltonian-formalism calculation 'confirmed' the bosonized theta dependence is therefore stronger than the evidence shown; please add at least one D-dependence check and one a- or L-dependence check, or explicitly present the comparison as preliminary and direct the reader to the companion paper [1] for the controlled analysis.
[Section 3.3, Eq. (6)] The assertion that the theta=pi one-point functions reproduce the SU(2)_1 WZW prediction is supported by fits whose pion form contains an overall amplitude and an unconstrained parameter Delta, and the paper does not report the fitted values or compare Delta with the value required by the boundary condition. With two free parameters available, a good fit is a much weaker test than 'reproduce the expected CFT-like behavior' as stated in the abstract. Please report the fitted parameters (and their predicted values) or soften this claim.

minor comments (5)

[Abstract, Section 1] The abstract contains the typo 'Schwingr' for 'Schwinger', and the string 'thetameson' in Section 1 is missing a space.
[Section 4] The fitted velocity parameter b in Delta E = sqrt(b^2 Delta K^2 + M^2) is not reported; because the mass extraction extrapolates to Delta K^2 -> 0, the b values and their uncertainties should be given or referenced.
[Section 3.2, Fig. 3] The fitting ranges and the size of the second-state gap Delta M used in the two-exponential ansatz are not stated in the text or in Fig. 3; please provide them or cite the corresponding table in [1].
[Section 3.1, Fig. 2] The solid curve for delta_+ is obtained from a bosonized ansatz whose functional form is only described in footnote 1; the form and fitted parameters should be displayed in this paper for the fit to be reproducible.
[Section 3.3] The main text should define Delta appearing in Eq. (6) so that the pion one-point function is self-contained, even if the detailed derivation is deferred to appendix A of ref. [1].

Circularity Check

0 steps flagged · score 2.0 of 10

No substantive circularity: the theta-dependence and sqrt(3) ratio are external bosonization predictions; only the overall mass scale is fitted at theta=0.

full rationale

The central derivation is a comparison of DMRG-determined masses against independent bosonization results. The functional forms M_pi(theta)=M_pi(0)|cos(theta/2)|^(2/3) and M_sigma(theta)=sqrt(3)M_pi(theta) come from Coleman [4]; only the overall coefficient M_pi(0) is fixed by averaging the numerical results at theta=0, which is a one-point normalization and not a fit of the theta dependence or of the mass ratio. The two extraction schemes—improved one-point function (Sec. 3) and dispersion relation (Sec. 4)—are mutually independent and are described in the proceedings; they agree with each other. The theta=pi CFT comparison uses the WZW functional forms of Eq. (6) as external fitting templates, and the authors present it as consistency rather than as a parameter-free prediction. Self-citations to [1] and [3] supply the methods and detailed derivations, but the computational steps are summarized in the text and the validation benchmarks (bosonization, Fukaya-Onogi Monte Carlo) are external. The absence of continuum-limit, infinite-volume, and bond-dimension convergence checks is a systematic-error caveat, which the authors themselves partly acknowledge in Sec. 5 when they say the range of applicability of the bosonized description remains a theoretical question; this affects reliability of the comparison, not circularity of the derivation chain.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on numerical DMRG rather than new entities; no particles, forces, or dimensions are invented. The main pull-from-upstream items are the bosonization and WZW CFT predictions used as benchmarks, plus the method-specific assumptions listed above. The free parameters are normalization and fit parameters, not physical constants.

free parameters (3)

Overall normalization M_pi(0) = determined by averaging the two schemes at theta=0; numeric value not quoted
Used in Fig. 1 to normalize the bosonization curves M_pi(theta)=M_pi(0)|cos(theta/2)|^(2/3); the theta-dependence itself is not fitted.
Parameters in bosonized sigma-eta mixing-angle fit = not specified ('a few unknown parameters')
Section 3.1 fits delta_+ to the analytic form theta/2+omega(theta) from bosonization with unspecified free parameters; this supports consistency of the operator-mixing prescription but is not parameter-free.
Amplitude and possibly Delta in WZW CFT one-point-function fits = not stated
At theta=pi, Eq. (6) is fitted to the numerical one-point functions; the proportionality constants and any fitted scaling exponent are not itemized in this proceedings.

assumptions (4)

domain assumption DMRG with bond dimension D <= 1400 gives converged ground and low-lying excited states for all theta, including near theta=pi where S_EE ~ (c/3) log N.
Section 2 sets N up to 320, a=0.25, D <= 1400 and states the calculation is still doable without showing a convergence check in this paper.
domain assumption The one-point function in the bulk decays as e^{-M x} with the lightest meson mass when the boundary wings act as a source, and two-exponential fits isolate the ground-state mass.
Section 3.2 uses the wall-source analogy and fits <O(x)> ~ A e^{-Mx} + B e^{-(M+Delta M)x}; validity for the constructed boundary states is assumed.
ad hoc to paper The open-boundary expectation value <K^2>, after subtracting the ground-state value, can be treated as the squared momentum and the dispersion relation E=sqrt(K^2+M^2) applies.
Section 4 explicitly states translational invariance is broken and <K^2> is not a genuine quantum number but works well empirically; the whole dispersion-relation scheme rests on this.
domain assumption The lattice Hamiltonian with m_lat = m - N_f g^2 a/8 and staggered fermions correctly represents the continuum 2-flavor Schwinger model.
Section 2 adopts the standard staggered discretization and the Dempsey-Klebanov-Pufu-Zan mass shift; this is standard background in the cited literature.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Computing theta-dependent mass spectrum of the 2-flavor Schwinger model in the Hamiltonian formalism." pith.science (2026). https://pith.science/paper/VXK3C6YE

@misc{pith2026250118960,
  author       = {Pith},
  title        = {Pith review of: Computing theta-dependent mass spectrum of the 2-flavor Schwinger model in the Hamiltonian formalism},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VXK3C6YE}},
  note         = {Machine review of arXiv:2501.18960}
}

abstract

We compute the $\theta$-dependent mass spectrum of the 2-flavor Schwingr model using the tensor network (DMRG) in the Hamiltonian formalism. The pion and the sigma meson are identified as stable particles of the model for nonzero $\theta$ whereas the eta meson becomes unstable. The meson masses are obtained from the one-point functions, using the meson operators defined by diagonalizing the correlation matrix to deal with the operator mixing. We also compute the dispersion relation directly by measuring the energy and momentum of the excited states, where the mesons are distinguished by the isospin quantum number. We confirmed that the meson masses computed by these methods agree with each other and are consistent with the calculation by the bosonized model. Our methods are free from the sign problem and show a significant improvement in accuracy compared to the conventional Monte Carlo methods. Furthermore, at the critical point $\theta = \pi$, the mesons become almost massless, and the one-point functions reproduce the expected CFT-like behavior.

Figures

Figures reproduced from arXiv: 2501.18960 by the authors.

**Figure 17.** FIG. 17. The effective mass plot of the full [PITH_FULL_IMAGE:figures/full_fig_p002_17.png] view at source ↗

**Figure 2.** The mixing angles 𝛿− of the triplet and 𝛿+ of the singlets are plotted against 𝜃/2𝜋. The dashed line depicts the expected linear behavior of 𝜃/2 for the triplet sector. The solid curve denotes the fitting result 𝜃/2 + 𝜔(𝜃) for the singlet sector. a rotation angle 𝛿. Then 𝛿 corresponds to the mixing angle, and the eigenvalues are the correlation functions of the meson operators. For example, the relation for the iso-… view at source ↗

Figures from the paper (3 more)

**Figure 3.** Figure 3: The effective masses of the sigma meson (left) and pion (right) computed from their one-point functions are plotted against the distance 𝑥 from the boundary. The solid curves depict the fitting results, where the fitting range is between the vertical dashed lines. meso…

**Figure 4.** Figure 4: The one-point functions at 𝜃 = 𝜋 of the sigma meson (left) and pion (right) are plotted against 𝑥. The fitting results based on the analytic forms (6) are depicted by the solid curves, where the fitting range is between the vertical dashed lines. We compare the one-poi…

**Figure 5.** Figure 5: The energy gap Δ𝐸ℓ is plotted against the square of total momentum Δ𝐾 2 ℓ for each meson. The dashed lines depict the fitting results of the dispersion relations. The resulting meson masses are indicated by the markers at the left endpoints with error bars. In the calc…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 9 canonical work pages

[1]

E. Itou, A. Matsumoto and Y. Tanizaki,DMRG study of the theta-dependent mass spectrum in the 2-flavor Schwinger model, JHEP 09(2024) 155 [2407.11391]

arXiv 2024
[5]

Fukaya and T

H. Fukaya and T. Onogi,Lattice study of the massive Schwinger model with theta term under Luscher’s ’admissibility’ condition,Phys. Rev. D68(2003) 074503 [hep-lat/0305004]

arXiv 2003
[2]

Flavour Lattice A veraging Group (FLAG)collaboration, FLAG Review 2021, Eur. Phys. J. C82(2022) 869 [2111.09849]

arXiv 2022
[3]

E. Itou, A. Matsumoto and Y. Tanizaki,Calculating composite-particle spectra in Hamiltonian formalism and demonstration in 2-flavor QED1+1𝑑,JHEP 11 (2023) 231 [2307.16655]

arXiv 2023
[4]

Coleman,More About the Massive Schwinger Model, Annals Phys.101 (1976) 239

S.R. Coleman,More About the Massive Schwinger Model, Annals Phys.101 (1976) 239

work page 1976
[6]

Kogut and L

J.B. Kogut and L. Susskind,Hamiltonian Formulation of Wilson’s Lattice Gauge Theories, Phys. Rev. D11(1975) 395

work page 1975
[7]

Susskind,Lattice Fermions, Phys

L. Susskind,Lattice Fermions, Phys. Rev. D16 (1977) 3031

work page 1977
[8]

R.Dempsey, I.R.Klebanov, S.S.PufuandB.Zan, Discretechiralsymmetryandmassshiftin the lattice Hamiltonian approach to the Schwinger model,Phys. Rev. Res.4(2022) 043133 [2206.05308]

arXiv 2022

Show all 16 references

[9]

White,Density matrix formulation for quantum renormalization groups,Phys

S.R. White,Density matrix formulation for quantum renormalization groups,Phys. Rev. Lett. 69 (1992) 2863

1992
[10]

White,Density-matrix algorithms for quantum renormalization groups,Phys

S.R. White,Density-matrix algorithms for quantum renormalization groups,Phys. Rev. B48 (1993) 10345

1993
[11]

Schollwöck,The density-matrix renormalization group,Rev

U. Schollwöck,The density-matrix renormalization group,Rev. Mod. Phys.77(2005) 259

2005
[12]

U.Schollwöck, Thedensity-matrixrenormalizationgroupintheageofmatrixproductstates , Annals Phys.326(2011) 96

2011
[13]

Dempsey, I.R

R. Dempsey, I.R. Klebanov, S.S. Pufu, B.T. Søgaard and B. Zan,Phase Diagram of the Two-Flavor Schwinger Model at Zero Temperature, Phys. Rev. Lett.132 (2024) 031603 [2305.04437]

2024 arXiv
[14]

Fishman, S.R

M. Fishman, S.R. White and E.M. Stoudenmire,The ITensor Software Library for Tensor Network Calculations,SciPost Phys. Codebases(2022) 4

2022
[15]

Wall and L.D

M.L. Wall and L.D. Carr,Out-of-equilibrium dynamics with matrix product states, New J. Phys. 14(2012) 125015

2012
[16]

Bañuls, K

M.C. Bañuls, K. Cichy, K. Jansen and J.I. Cirac,The mass spectrum of the Schwinger model with Matrix Product States, JHEP 11(2013) 158 [1305.3765]. 9

2013 arXiv

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

E. Itou, A. Matsumoto and Y. Tanizaki,DMRG study of the theta-dependent mass spectrum in the 2-flavor Schwinger model, JHEP 09(2024) 155 [2407.11391]

arXiv 2024

[2] [5]

Fukaya and T

H. Fukaya and T. Onogi,Lattice study of the massive Schwinger model with theta term under Luscher’s ’admissibility’ condition,Phys. Rev. D68(2003) 074503 [hep-lat/0305004]

arXiv 2003

[3] [2]

Flavour Lattice A veraging Group (FLAG)collaboration, FLAG Review 2021, Eur. Phys. J. C82(2022) 869 [2111.09849]

arXiv 2022

[4] [3]

E. Itou, A. Matsumoto and Y. Tanizaki,Calculating composite-particle spectra in Hamiltonian formalism and demonstration in 2-flavor QED1+1𝑑,JHEP 11 (2023) 231 [2307.16655]

arXiv 2023

[5] [4]

Coleman,More About the Massive Schwinger Model, Annals Phys.101 (1976) 239

S.R. Coleman,More About the Massive Schwinger Model, Annals Phys.101 (1976) 239

work page 1976

[6] [6]

Kogut and L

J.B. Kogut and L. Susskind,Hamiltonian Formulation of Wilson’s Lattice Gauge Theories, Phys. Rev. D11(1975) 395

work page 1975

[7] [7]

Susskind,Lattice Fermions, Phys

L. Susskind,Lattice Fermions, Phys. Rev. D16 (1977) 3031

work page 1977

[8] [8]

R.Dempsey, I.R.Klebanov, S.S.PufuandB.Zan, Discretechiralsymmetryandmassshiftin the lattice Hamiltonian approach to the Schwinger model,Phys. Rev. Res.4(2022) 043133 [2206.05308]

arXiv 2022