{"id":"c50d2582-7d72-4d8d-9b76-8836dc418434","arxiv_id":"2411.09886","paper_version":3,"verdict":"UNVERDICTED","confidence":"MODERATE","novelty_score":1.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A review of lattice fermion formulations, centered on a non-Hermitian one-sided-difference approach that avoids doublers while preserving chiral symmetry, plus surveys of Wilson, overlap, topological charge, and Monte Carlo methods.","lead":"This preprint is a review of lattice fermion formulations, including a non-Hermitian construction that avoids fermion doubling by dropping Hermiticity while keeping chiral symmetry. It surveys standard material on Wilson and overlap fermions, topological charge, and Monte Carlo methods, and highlights a known (1+1)D equivalence between Wilson and overlap fermions.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The d>1 no-doubling claim rests on an unproved pole-decoupling assertion from Ref. [17]; Sec. 4.5's 'no quadratic term' remark does not establish decoupling, and Eq. (194) has a sum-range typo that obscures the derivation.","rationale":"The reader's weakest_assumption is exactly the concern identified here: the d>1 no-doubling claim depends on an unproved assertion from Ref. [17] about the decoupling of non-physical poles. My stress-test confirms that this is the most load-bearing step in the paper's central argument. The one-dimensional case (Sec. 4.5, Eq. (193)) is explicit and credible, and the two-flavor positive-definite partition function (Sec. 5.5, Eq. (304)) follows from the stated γ5 anticommutation, so those parts are not the main vulnerability. The d>1 extension, however, is quoted rather than derived, and the in-text justification is too terse to be checked. I also note the apparent typo in Eq. (194), which reinforces that the derivation is not self-contained. The reader already marked the paper UNVERDICTED, partly for this reason; my read does not move that verdict. If the manuscript were to be treated as a research claim rather than a review, the d>1 step would need to be independently verified before acceptance, making the claim conditional at best. For now, the appropriate status remains unchanged.","tokens_in":34088,"tokens_out":10953,"duration_ms":104791,"concrete_test":"Independently compute the pole structure of the massive 2D one-sided propagator: solve det(D(p) + m) = 0 for p near (π/2a, -π/2a) with m > 0, and compute the residue of the propagator at the resulting pole. Check whether, as a→0, the would-be doubler either acquires an infinite physical mass or has vanishing residue. Repeat for 4D at the analogous non-physical poles. If any pole yields a finite-mass physical fermion, the no-doubling claim in Sec. 4.5 is false; if all decouple, the concern is resolved. Also re-derive Eq. (195) from Eq. (194) to confirm the displayed inverse operator.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that one-sided non-Hermitian lattice fermions avoid doubling in d>1 hinges on Sec. 4.5, Eqs. (194)-(196). The text asserts that non-physical poles at p1a = -p2a = ±π/2 do not contribute in the continuum limit because the expansion of the denominator around them 'has no quadratic term' [17], and that this generalizes to all d. No derivation is given; the argument is a citation to a 1992 CERN preprint. As written, the claim is not self-evident: absence of a quadratic term in an expansion does not by itself imply that the pole residue vanishes or that the pole moves to infinite physical mass when the mass term is included; one must compute the massive propagator's pole structure and residue. The surrounding formulas also show signs of imprecision: Eq. (194) writes a sum over μ=1,...,4 in a 2D calculation and then equates it to a sum over μ=1,2; if the first sum is taken literally, Eq. (195) is not its inverse. Because the no-doubling conclusion in d>1 is load-bearing for the entire non-Hermitian program (including the MC application via even flavors), and the manuscript supplies no independent check, this is the weakest point. If the decoupling fails, the continuum limit contains extra fermion degrees of freedom and the central claim collapses.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is a review of lattice chiral fermion formulations organized around the proposal that abandoning Hermiticity, specifically through one-sided lattice differences, evades the Nielsen-Ninomiya theorem while preserving exact chiral symmetry. It reviews path-integral and RG background material, compares Wilson and overlap fermions, discusses the claimed equivalence of the Wilson and overlap formulations in the (1+1)D Hamiltonian setting, and explains how an even number of flavors with opposite one-sided differences yields a non-negative partition function suitable for Monte Carlo simulation. The final sections address the chiral anomaly and the lattice index theorem, with emphasis on subtleties in finite-volume topological charge. The central positive claims are the absence of fermion doubling in d>1 for the non-Hermitian one-sided action and the applicability of HMC through the positive-definite two-flavor determinant; both are presented largely by citation to earlier work.","tokens_in":34357,"tokens_out":6138,"duration_ms":64140,"significance":"If correct, the non-Hermitian construction would provide a local, exactly chiral lattice fermion, bypassing the Nielsen-Ninomiya assumptions at the price of hypercubic symmetry and standard renormalizability, and it would open a practical simulation route when Hermiticity is unavailable. The review is useful as a concise compendium of path-integral, RG, Wilson/overlap, pseudofermion, and index-theorem material, and the two-flavor positivity argument in Sec. 5.5 is explicit and checkable. However, the manuscript contains no new derivation of its two most load-bearing assertions: the d>1 no-doubling claim in Sec. 4.5 and the Wilson-overlap equivalence in Sec. 4.4 are both quoted from Refs. [17] and [20]. The no-doubling argument as written is incomplete and contains a dimension typo in Eq. (194), so the central claim of the review is not established within the manuscript itself.","major_comments":[{"comment":"The d>1 no-doubling claim is the central physical assertion of the non-Hermitian program, but it is not demonstrated in this manuscript. The text states that the non-physical poles at p1a = -p2a = ±π/2 do not contribute in the continuum limit because the expansion around them 'has no quadratic term', citing Ref. [17]. This is not a proof: the absence of a quadratic term in the denominator expansion does not by itself imply that the pole residue vanishes or that the pole decouples once a mass term is included. One needs an explicit computation of the massive propagator's pole structure and residues. In addition, Eq. (194) writes a sum over μ=1,...,4 in what is announced as a 2D calculation and then equates it to a sum over μ=1,2; if the first sum is taken literally, Eq. (195) is not its inverse. The manuscript should either provide a self-contained derivation of the pole decoupling or clearly state that this is an unproved conjecture inherited from Ref. [17].","section":"Sec. 4.5, Eqs. (194)-(196)"},{"comment":"The claim A†(s=0)A(s=0)=1, and hence Dov(s=0)=DW, is presented as the key connection between Wilson and overlap fermions in the (1+1)D Hamiltonian formulation. Yet Eq. (189) is simply quoted from Ref. [20]; no derivation or even a sketch of the ingredients (the definition of A, the role of the Hamiltonian lattice, and the boundary conditions) is provided. Since this equivalence is one of the paper's advertised insights, the review should either give a compact proof or explicitly frame the statement as a summary of a published result and explain why it is nontrivial.","section":"Sec. 4.4, Eq. (189)"},{"comment":"The paper asserts that the loss of hypercubic symmetry implies non-renormalizability in the standard sense and that quenched averaging over lattice directions restores hypercubic symmetry at the level of observables. The averaging prescription in Eq. (197) is written with notation ⟨O(ϵμ)⟩ϵμ that is ambiguous, and no argument is given that the average restores the symmetry in an interacting theory or that it removes the lattice-spacing dependence of physical observables. Because this is the proposed remedy for the continuum-limit problem, a citation to a 4D QED calculation is not sufficient; the manuscript should at least formulate the averaging prescription precisely and explain the conditions under which it is expected to work.","section":"Sec. 4.5 and Eq. (197)"}],"minor_comments":[{"comment":"There is a typo in the exponent: 'xn+1 - xx' should read 'xn+1 - xn'.","section":"Sec. 2.1, Eq. (24)"},{"comment":"The text contains a duplicated word: 'because because Tr(γ5) = 0' should be 'because Tr(γ5) = 0'.","section":"Sec. 1, paragraph on topological charge"},{"comment":"The Ginsparg-Wilson relation is written as γ5D + γ5 = aDγ5D; the second term should be Dγ5, giving γ5D + Dγ5 = aDγ5D as in Eq. (167).","section":"Sec. 6.3, Eq. (319)"},{"comment":"The phrase 'continuum limit (ma, a/x→0)' is confusing; the intended limit is ma→0 with x fixed, or a/x→0 with m fixed, but the two should not be combined this way.","section":"Sec. 4.5, after Eq. (193)"},{"comment":"The mass parameter is written as mF in the action and then as m in the determinant identity; the notation should be harmonized, and the step using γ5-Hermiticity should be written in one line to avoid ambiguity.","section":"Sec. 5.5, Eq. (304)"},{"comment":"The text says eigenvectors with real eigenvalues 0 and 2/a have non-vanishing chirality, but in Eq. (321) the factor (2-aλ) removes the λ=2/a contribution; the wording should be adjusted to say that only the zero modes contribute to the index.","section":"Sec. 6.3, Eq. (321)"},{"comment":"Ref. [17] is a CERN preprint from 1992 with no journal reference or DOI; the authors should provide a more accessible reference or, failing that, state clearly that the no-doubling claim in d>1 is based on an unpublished preprint.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is almost entirely a review, and its two main advertised insights — the d>1 no-doubling property and the Wilson-overlap equivalence — are cited from the authors' own prior papers and from Ref. [20], while the key d>1 decoupling assertion is credited to an old CERN preprint. Editors may wish to consider whether a review with no new technical content and with its central claim left unproved is appropriate for the journal in its present form. The paper would be strengthened by adding at least a sketch of the pole-decoupling computation and by being explicit about which statements are conjectural."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is a review, not a research paper. The non-Hermitian construction (Sec. 4.5), the two-flavor Monte Carlo trick (Sec. 5.5), and the Wilson-overlap equivalence (Sec. 4.4) all come from earlier published work (Refs. [17,20,34,35]). So don't expect a new result.\n\nWhat it does well: the standard material is handled competently. The path integral, RG flow, overlap spectrum, and the Ginsparg-Wilson index theorem sections are mostly solid. The two-flavor determinant identity det(D+m)det(-D†+m)=|det(D+m)|^2 via γ5D+Dγ5=0 is correctly presented and is a genuinely useful observation for simulating the non-Hermitian theory. The review also does not hide the main caveats: loss of hypercubic symmetry and non-renormalizability are stated up front.\n\nThe soft spot is in Sec. 4.5. The claim that one-sided non-Hermitian fermions avoid doubling in d>1 rests entirely on an assertion quoted from Stamatescu–Wu (Ref. [17]): that the non-physical poles at p1a=-p2a=±π/2 do not shift the mass pole because the expansion around them has no quadratic term. That is not a proof, and as written it is not even transparent: Eq. (194) writes a sum over μ=1..4 in a 2D calculation, then equates it to a sum over μ=1,2. If the first sum is taken literally, Eq. (195) is not its inverse. The reader is left without a way to check the decoupling, which matters because this is the load-bearing step for the whole program, including the Monte Carlo application. A review can cite, but here the review presents this as an established result without flagging that it is an imported claim with no derivation in sight.\n\nMinor point: the abstract and introduction describe the non-Hermitian formulation as 'novel' without making clear that the construction itself is from Ref. [17]; for a review, the provenance should be sharper.\n\nWho is this for? Someone wanting a broad survey of lattice chiral fermions with a focus on the non-Hermitian route. It is a reasonable entry point, but the central claim is not self-contained. As a research preprint, it is unverdictable; as a review, it deserves revision more than acceptance as is.\n\nRecommendation: send to peer review if the venue publishes reviews, but ask referees to press on the pole-decoupling argument and to fix the Eq. (194) typo. If it's a research submission, desk reject as not a research paper.","headline":"A competent review of lattice chiral fermions whose central non-Hermitian no-doubling claim is imported from a 1992 preprint and not actually established.","tokens_in":34952,"tokens_out":3900,"would_cite":false,"duration_ms":33782,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["11.15.Ha","11.30.Rd"],"model":"deepseek-v4-flash","headline":"By abandoning Hermiticity and using one-sided lattice differences, a non-Hermitian lattice fermion can keep exact chiral symmetry and avoid fermion doubling, and paired flavors give a positive partition function that standard Monte Carlo…","keywords":["lattice chiral fermion","fermion doubling","Nielsen-Ninomiya theorem","non-Hermitian lattice fermion","one-sided lattice difference","Ginsparg-Wilson relation","overlap fermion","Monte Carlo sign problem"],"falsifier":"Compute the interacting one-loop or two-loop self-energy in the 2D non-Hermitian formulation and track the spurious poles $p_1 a=-p_2 a=\\pm\\pi/2$ as $a\\to0$; if such a pole moves to a finite physical momentum or keeps an order-one residue, the decoupling claim fails and the continuum limit would contain doublers, while if all spurious poles stay at lattice-scale momenta with vanishing residue, the no-doubling claim survives.","tokens_in":33835,"feed_emoji":"⚛️","tokens_out":13432,"duration_ms":119324,"temperature":0.7,"pith_summary":"Lattice fermions face a no-go: a local, Hermitian, chirally symmetric Dirac operator necessarily brings spurious extra species, or doublers. This review argues that abandoning Hermiticity and discretizing with one-sided lattice differences sidesteps that no-go: the non-Hermitian Dirac operator stays local and exactly chiral, and in one dimension its propagator has half the poles of the naive fermion. In more dimensions, non-physical poles appear but are claimed to decouple in the continuum limit, leaving a single physical fermion. With an even number of flavors whose one-sided derivatives point in opposite directions, the fermion determinant equals $|\\det(D+m_F)|^2$, making the partition function non-negative and the theory accessible to standard Hybrid Monte Carlo simulation. The price is the loss of hypercubic symmetry and standard renormalizability, with a quenched averaging over directions proposed to restore symmetry at the observable level.","feed_headline":"Non-Hermitian lattice fermion dodges doubling, keeps chirality","feed_subtitle":"Opposite one-sided derivatives yield a positive determinant, so standard Monte Carlo can simulate exact chiral fermions.","key_machinery":"The load-bearing object is the one-sided difference Dirac operator $D(n_1,n_2)=\\gamma_1(\\delta_{n_1+1,n_2}-\\delta_{n_1,n_2})/a$, whose momentum-space inverse is $\\bar D^{-1}(p)=-(i/2a)\\sum_\\mu\\gamma_\\mu e^{ip_\\mu a/2}\\sin(p_\\mu a/2)$ divided by $a^{-2}\\sum_\\nu e^{ip_\\nu a}\\sin^2(p_\\nu a/2)$. This exponential dispersion cuts the number of zeros of the denominator and therefore the number of fermion species. The second mechanism is the even-flavor pairing $S_{FD}=\\bar\\psi_1(D+m_F)\\psi_1+\\bar\\psi_2(-D^\\dagger+m_F)\\psi_2$, whose determinant becomes $|\\det(D+m_F)|^2$ using $\\gamma_5D+D\\gamma_5=0$, giving the positive weight required for Monte Carlo. For the overlap and index-theorem parts, the Ginsparg-Wilson relation $D\\gamma_5+\\gamma_5D=aD\\gamma_5D$ is the central identity: it defines a modified lattice chiral symmetry and encodes the index theorem as $\\mathrm{Tr}[\\gamma_5(1-aD/2)]=n_+-n_-$.","core_discovery":"The central claim is that the Nielsen-Ninomiya obstacle can be bypassed by replacing the symmetric lattice derivative $i\\sin(pa)$ with the one-sided exponential $e^{ipa}-1$. In 1D this halves the number of propagator poles, so the continuum limit returns the free Dirac propagator without doublers. In $d>1$ the inverse Dirac matrix in Eqs. (194) and (195) has non-physical poles such as $p_1 a=-p_2 a=\\pm\\pi/2$, but the expansion of the denominator around them has no quadratic term, and the authors argue, following Ref. [17], that these poles do not shift the physical mass pole and decouple as $a\\to0$. The two-flavor action $S_{FD}$ with forward and backward differences yields $\\det(D+m_F)\\det(-D^\\dagger+m_F)=|\\det(D+m_F)|^2$, a non-negative determinant, so the non-Hermitian theory can in principle be simulated with HMC even though Hermiticity is lost. The review also argues that in the (1+1)D Hamiltonian formulation the Wilson fermion with $s=0$ coincides with the overlap fermion, and it surveys how Ginsparg-Wilson fermions realize or fail to realize the index theorem on finite lattices, including cases where the chiral anomaly appears while the topological charge is zero.","pith_inferences":["Editorial inference: the pole-decoupling argument is perturbative in spirit; a two-loop calculation of the fermion self-energy in 4D non-Hermitian QED would directly test whether radiative corrections drag the non-physical poles toward the continuum.","Editorial inference: the positive-definite even-flavor pairing is structurally similar to simulations of QCD with opposite isospin chemical potentials, so a natural extension is to check whether a real baryon chemical potential can be handled sign-free in this formulation.","Editorial inference: the 1+1D Wilson-overlap equivalence suggests searching for Ginsparg-Wilson-type modified symmetries in other local discretizations, potentially giving a four-dimensional route to exact chirality without nonlocality.","Editorial inference: a clean numerical test would be a continuum extrapolation of the chiral condensate in the 2D Gross-Neveu-Yukawa model simulated with this fermion; the review reports initial results [35] but does not claim that extrapolation."],"forward_implications":["A local, exactly chiral lattice fermion exists without the nonlocal operator of the overlap construction, at least as far as the free propagation is concerned.","Even-flavor non-Hermitian fermions admit standard Hybrid Monte Carlo simulation, because opposite one-sided differences produce a non-negative determinant.","Observables become hypercubic-symmetric after quenched averaging over the $2^d$ choices of one-sided direction; without this averaging, lattice-spacing artifacts persist.","Since the non-Hermitian action is exactly chiral, its Dirac operator has no topological zero modes on a finite lattice; the index theorem and nonzero topological charge must be recovered only through a subtle infinite-volume limit.","In $1+1$ dimensions the Wilson and overlap fermions coincide at $s=0$, showing that modified chiral symmetry is not exclusive to the overlap construction."],"supporting_citations":[{"why":"The Nielsen-Ninomiya no-go theorem proved in these papers is the obstacle the non-Hermitian formulation is designed to bypass.","marker":"[4, 5, 7]"},{"why":"Supplies the one-sided-difference non-Hermitian formulation and the pole-decoupling assertion on which the no-doubling claim rests.","marker":"[17]"},{"why":"Sets up the free two-flavor fermion with opposite one-sided differences and demonstrates the positive-definite partition function used for Monte Carlo.","marker":"[34]"},{"why":"Simulates the interacting 2D Gross-Neveu-Yukawa model with non-Hermitian lattice fermions, the main interacting evidence that the method works.","marker":"[35]"},{"why":"Defines the overlap fermion, the exact-chiral alternative construction against which the non-Hermitian approach is compared.","marker":"[6]"},{"why":"Gives the Ginsparg-Wilson relation and the modified lattice chiral symmetry underlying the overlap and index-theorem discussion.","marker":"[8, 9]"},{"why":"Shows the 1+1D Hamiltonian equivalence of Wilson and overlap fermions at s=0, a central claimed connection.","marker":"[20]"},{"why":"Documents Ginsparg-Wilson Dirac operators with anomaly but zero topological charge, anchoring the topological-charge discussion.","marker":"[30, 31]"}],"fun_headline_variants":["Dropping Hermiticity dodges lattice fermion doublers","One-sided derivative kills doublers, keeps chirality","Non-Hermitian fermions: HMC works, chirality intact","Wilson-overlap link via non-Hermitian lattice fermions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The no-doubling conclusion in more than one dimension rests on the assertion taken from Ref. [17] and not proved in this review that the spurious poles at $p_1 a=-p_2 a=\\pm\\pi/2$ never shift the physical mass pole because the expansion around them has no quadratic term; if that decoupling fails, the continuum limit would contain extra fermion species.","fun_headline_variants_meta":{"raw":{"variants":["Dropping Hermiticity dodges lattice fermion doublers","One-sided derivative kills doublers, keeps chirality","Non-Hermitian fermions: HMC works, chirality intact","Wilson-overlap link via non-Hermitian lattice fermions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000162,"raw_usage":{"total_tokens":1281,"prompt_tokens":1026,"completion_tokens":255,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":181}},"tokens_in":642,"tokens_out":255,"duration_ms":3487,"temperature":1.0,"reasoning_tokens":181,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:12:29.864800+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the interacting one-loop or two-loop self-energy in the 2D non-Hermitian formulation and track the spurious poles $p_1 a=-p_2 a=\\pm\\pi/2$ as $a\\to0$; if such a pole moves to a finite physical momentum or keeps an order-one residue, the decoupling claim fails and the continuum limit would contain doublers, while if all spurious poles stay at lattice-scale momenta with vanishing residue, the no-doubling claim survives.","supporting_citations":[{"cited_title":"A New formulation of lattice gauge theory with fermions,","cited_arxiv_id":null,"evidence_quote":"Supplies the one-sided-difference non-Hermitian formulation and the pole-decoupling assertion on which the no-doubling claim rests."},{"cited_title":"Non-Hermitian Lattice Fermions in 2D GNY Model","cited_arxiv_id":"2404.18441","evidence_quote":"Simulates the interacting 2D Gross-Neveu-Yukawa model with non-Hermitian lattice fermions, the main interacting evidence that the method works."}],"review_version":1}