{"id":"a6db2b4e-7bab-4b85-9005-37e39691c837","arxiv_id":"2509.16309","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Non-Hermitian couplings are relevant in the disordered interacting Hatano-Nelson chain, producing a random strongly-coupled-pair ground state with negative diverging susceptibility and saturating entanglement entropy.","lead":"The authors derive an asymptotically exact low-energy solution for a disordered, interacting non-Hermitian spin chain, the Hatano-Nelson model, using strong-disorder renormalization. They find that non-Hermiticity drives a quantum-to-classical crossover, producing a negative diverging magnetic susceptibility and a saturation of entanglement entropy.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The γ couplings are exactly removable by a non-unitary gauge transformation, so their SDRG growth is a redundant parameter rather than a relevant perturbation.","rationale":"The reader's weakest assumption, the unproven joint distribution Q_n in SM Eqs. (38)-(39), is a genuine omitted proof and should be flagged. It is secondary, however, because the stated form of Q_n is plausible once one recognizes that γ and ℓ obey identical additive rules; the deeper problem is that the Hamiltonian itself is exactly gauge-equivalent to the Hermitian random XXZ chain. The paper's main result, that non-Hermiticity is relevant in the RG sense, is then a statement about a basis choice rather than about low-energy physics: all energy gaps, the SDRG decimation order, and the J and Δ flows are exactly those of the Hermitian model. The observable predictions are consequently convention-dependent, as the paper itself shows through the RL/RR dichotomy for the entanglement entropy, and the claimed finite-temperature divergence of the transverse susceptibility traces back to unbounded gauge phases in the thermodynamic limit rather than to a collective instability. The omitted Q_n derivation and the fitted non-universal constants are legitimate concerns, but the unreported gauge redundancy undercuts the central interpretation and the claimed novelty of an asymptotically exact solution of a genuinely non-Hermitian interacting problem. The paper should be rejected unless the authors reframe it as a study of basis-dependent observables in a gauge-transformed Hermitian model and retract the relevance claim.","tokens_in":18130,"tokens_out":28959,"duration_ms":280555,"concrete_test":"Verify the similarity identity analytically: define θ_1 = 0 and θ_{i+1} = θ_i + γ_i, set U = ∏_i exp(θ_i S^z_i), and compute U H U^{-1} for the Hamiltonian (1); the result should be the Hermitian XXZ chain with the same J_i and Δ_i. If the identity holds, the additive γ SDRG rule (4) is exactly the composition of these gauge phases. Then recompute the single-pair transverse susceptibility and the RL entanglement after the gauge transformation: the spectrum and S^z correlations contain no γ, and the γ-dependence of S^x S^x and S_pair is precisely the local non-unitary filtering by U. This single check determines whether γ is a relevant physical coupling or a redundant gauge direction.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim is that non-Hermiticity is an RG-relevant perturbation, producing a quantum-to-classical crossover. However, for open/infinite chains the Hamiltonian in Eq. (1) is exactly similar to the Hermitian disordered XXZ chain: taking U = ∏_i exp(θ_i S^z_i) with θ_{i+1} − θ_i = γ_i gives U H U^{-1} = ∑_i J_i [ (S^+_i S^-_{i+1} + S^-_i S^+_{i+1})/2 + Δ_i S^z_i S^z_{i+1} ]. This holds for arbitrarily strong γ and is not an SDRG approximation; the SM's own Eq. (16) shows the two-site eigenvalues are γ-independent. The decimation order and the flows of J and Δ are therefore exactly those of the Hermitian model, and the additive rule for γ in Eq. (4) just accumulates the gauge phase, exactly as the bond length ℓ does. The broadening of P(γ) is kinematical cluster growth, not the flow of a coupling that affects any spectral scale or gauge-invariant long-distance property. The susceptibility divergence and the EE saturation are then basis/convention-dependent properties of the right (RR) eigenvector: the RL scheme gives S_pair = 1 and no saturation, and the single-pair transverse susceptibility is a biorthogonal expectation of a positive operator that becomes negative. The manuscript never acknowledges this exact gauge redundancy, so its main interpretation is not established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a strong-disorder renormalization-group (SDRG) study of the disordered non-Hermitian spin-1/2 XXZ chain, Eq. (1), which is equivalent to the interacting fermionic Hatano-Nelson model. The authors derive decimation rules for the couplings J, Δ, and the non-Hermitian parameter γ, show that J and Δ flow as in the Hermitian case, and find that the distribution of γ broadens without limit. They interpret this broadening as non-Hermiticity being a relevant perturbation, leading to a 'quantum-to-classical crossover' and a random strongly coupled pair phase. The paper predicts a negative transverse magnetic susceptibility that diverges at a finite temperature and an entanglement entropy that saturates with subsystem size, and it supports these predictions with numerical SDRG simulations of chains up to 5 million sites.","tokens_in":18391,"tokens_out":10877,"duration_ms":93090,"significance":"The model is well chosen and the SDRG machinery is applied with care: the decimation rules are derived in biorthogonal perturbation theory, and the fixed-point distributions of γ are checked numerically on very large chains. If the central physical interpretation were correct, the paper would be a significant analytical advance in disordered non-Hermitian many-body physics. However, the main claim—that non-Hermiticity is an RG-relevant perturbation—is not supported because the γ couplings can be removed by an exact similarity transformation, and the two headline observables (susceptibility and entanglement entropy) depend, as the paper itself shows, on the chosen biorthogonal/RR convention. The significance is therefore conditional on resolving these foundational issues.","major_comments":[{"comment":"The Hamiltonian in Eq. (1) is exactly similar to the Hermitian disordered XXZ chain. With U = ∏_i exp(θ_i S^z_i) and θ_{i+1} – θ_i = γ_i/2, one has U H U^{-1} = ∑_i J_i[(S^+_i S^-_{i+1} + S^-_i S^+_{i+1})/2 + Δ_i S^z_i S^z_{i+1}], so the full spectrum and the SDRG decimation order are those of the Hermitian model. This is consistent with SM Eq. (16), which shows the two-site eigenvalues are γ-independent, and with the sentence after SM Eq. (25) stating that the decimation 'follows the same hierarchy of decimations as the Hermitian one would.' The additive rule for γ in Eq. (4) therefore accumulates the gauge phase in the same way that the bond length ℓ does. The broadening of P(γ) is a kinematic consequence of cluster formation, not the flow of a coupling that controls any spectral scale or similarity-invariant physical property. The central claim in the abstract and Conclusions that non-Hermiticity is 'relevant in the RG sense' is thus not established. The paper should either identify a similarity-invariant observable that distinguishes the non-Hermitian model from the Hermitian one or substantially revise the interpretation.","section":"Main text, around Eq. (4) and SM Section II"},{"comment":"Both headline predictions are convention-dependent. The negative transverse susceptibility of a single pair, Eq. (8) and SM Eq. (46), is obtained as the biorthogonal matrix element ⟨ψ_L|(S^α)^2|ψ_R⟩ of a positive operator; this quantity can be negative because the right and left eigenstates of a non-Hermitian Hamiltonian are not orthogonal. The paper does not justify why this matrix element is the physical magnetic response of a non-Hermitian thermodynamic system. Similarly, the entanglement-entropy saturation in Fig. 3 is based on the RR reduced density matrix, whereas SM Eq. (69) shows that the RL scheme gives S_pair = 1 for all γ and hence no saturation. Since the claimed quantum-to-classical crossover is identified through these two quantities, the main physical conclusions are not robust to the choice of convention. A physical argument for the RR/biorthogonal choice is required before these results can be interpreted as evidence of a crossover.","section":"Eq. (8) and SM Section IV; Eq. (11) and SM Section V"},{"comment":"The joint fixed-point distribution Q_n(η, x, y) is stated without derivation; the text says the derivation is lengthy and details will be provided elsewhere. This distribution is essential for the entanglement-entropy calculation, since the EE in SM Eq. (72) and the analytical curves in Fig. 3 are computed from it. The 'asymptotically exact' claim for the EE therefore rests on an unverified input. The derivation should be included in the paper, or the EE result should be labeled as a numerical observation.","section":"SM Eqs. (38)–(39)"},{"comment":"The quantitative comparison between the analytical expressions and the numerics is not parameter-free. The constants C_γ and C_ℓ are determined by fitting the numerical SDRG flow to the universal fixed-point distributions (SM Fig. 4), and the same constants enter the analytical curves for χ and S that are overlaid on the numerical data in Figs. 2 and 3. Because the fitted parameters come from the same simulations that are being compared, the agreement demonstrates consistency but does not independently validate the theory. Please separate the fitting step from the test, for example by extracting C_γ and C_ℓ from one quantity and using them to predict another.","section":"SM Fig. 4 and main-text Figs. 2–3"}],"minor_comments":[{"comment":"The text refers to red, yellow, green, and blue curves, but the color scheme in the figure as rendered is not consistent; please verify the labels against the published figure.","section":"Fig. 2 caption and text"},{"comment":"The quantity Γ_T is used in Eq. (9) but only defined in the following sentence; please define it before first use.","section":"Main text, after Eq. (9)"},{"comment":"Please clarify the dimensions and definitions of C_γ and C_ℓ explicitly, and state whether they are dimensionless constants depending on the initial distributions.","section":"SM Eqs. (26)–(28)"},{"comment":"The phrase 'infrared stable fixed points of the Δ distribution are Δ_i = 0 ... and Δ_i → ∞' is imprecise; the Δ → ∞ fixed point is the Ising fixed point, and the flow to Δ = 0 occurs for 0 < Δ_i < 1, while Δ_i > 1 flows to the Ising fixed point. Please rephrase for clarity.","section":"Main text, paragraph after Eq. (5)"},{"comment":"Several references are recent preprints; please update citation data where journal versions are available, and define 'biorthogonal basis' on first use in the main text.","section":"Introduction and references"}],"recommendation":"reject","confidential_remarks":"The similarity-transformation issue is the core problem. If the authors can show that some physical observable (e.g., for periodic boundary conditions with a net imaginary flux, or a quantity in the Lindblad evolution) is genuinely sensitive to γ, a resubmission focused on that observable could be viable. As it stands, the paper's main physical conclusion appears to be an artifact of gauge choice, and the two central predictions are explicitly shown to be convention-dependent."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my honest read of the Hatano-Nelson SDRG paper. The headline result—non-Hermiticity is an RG-relevant perturbation producing a quantum-to-classical crossover—does not hold up. For open or infinite chains, the Hamiltonian in Eq. (1) is exactly similar to the Hermitian disordered XXZ chain via D = ∏_i exp(θ_i S^z_i) with θ_{i+1} − θ_i = γ_i. This is the standard imaginary gauge transformation of Hatano-Nelson, not an approximation: the hoppings become symmetric, the Δ terms are untouched, and the spectrum is identical. The flows of J and Δ and the decimation hierarchy are therefore exactly those of Fisher's XXZ solution. The broadening of P(γ) is kinematical—the same accumulation as bond lengths—and does not affect any spectral scale or gauge-invariant long-distance property.\n\nCredit where due: the biorthogonal perturbation theory that produces the decimation rules is careful, and the joint fixed-point distributions are nontrivial and backed by numerical SDRG on large chains. The authors even note that the two-site eigenvalues are γ-independent, which is the telltale sign of the gauge redundancy. But the physical conclusions are convention-bound. The negative x,y susceptibility is the biorthogonal expectation of a positive operator; it is negative because the left and right ground states differ. The entanglement entropy saturates only in the RR scheme; the RL scheme gives S_pair = 1 for all γ. Neither observable is invariant under the similarity transformation, and the manuscript never flags this. The RR density matrix is a legitimate choice in the non-Hermitian literature, but a claim that the model 'becomes classical' should not rest on a basis-dependent measure when the Hamiltonian itself is exactly similar to the Hermitian problem.\n\nTwo lesser issues: the joint distribution Q_n is stated without derivation (deferred to 'elsewhere'), and the constants Cγ and Cℓ are fitted from the same numerics used to test the analytical curves. These are minor next to the gauge problem.\n\nWho is this for? People working on non-Hermitian SDRG will want to read it, because the mistake is instructive. But the claim to have solved the disordered interacting Hatano-Nelson chain is overreach. I would not cite it for the crossover. I would still send it to a serious referee—the gauge question deserves an expert ruling, and if the authors can defend their scheme as physically meaningful, there may be a salvageable paper. As is, I would not accept it.","headline":"The paper's central claim of a non-Hermitian relevant perturbation collapses under the model's exact similarity to the Hermitian XXZ chain; the headline signatures are basis-dependent.","tokens_in":18946,"tokens_out":6540,"would_cite":false,"duration_ms":57218,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes an asymptotically exact low-energy solution of the disordered interacting Hatano-Nelson spin chain, showing that non-Hermitian couplings grow under renormalization and turn the ground state into classicalized…","keywords":["non-Hermitian many-body systems","Hatano-Nelson model","disordered XXZ spin chain","strong-disorder renormalization group","random singlet phase","entanglement entropy","magnetic susceptibility","quantum-to-classical crossover"],"falsifier":"Run the SDRG on chains of millions of sites over many disorder realizations—the same numerical experiment the paper uses for its histograms—and compare the measured joint distribution of scaled couplings, lengths, and gamma at the cutoff with Eqs. (38)-(39); failure to collapse onto those forms would remove the basis for the predicted entropy saturation. A complementary check is to compute the right-right pair entropy of a two-site problem at large gamma and compare with Eq. (11).","tokens_in":17875,"feed_emoji":"🔗","tokens_out":8352,"duration_ms":67910,"temperature":0.7,"pith_summary":"This paper gives an asymptotically exact low-energy solution of the disordered interacting Hatano-Nelson model, which is the same as the non-Hermitian spin-1/2 XXZ chain. Its central claim is that non-Hermitian couplings are relevant in the renormalization-group sense: under the strong-disorder flow they grow in both strength and statistical width, driving a quantum-to-classical crossover. The ground state is a random collection of strongly coupled spin pairs, but each pair is a mixture of the singlet and the $M=0$ triplet rather than a pure singlet. Two observable signatures follow: the $xy$ magnetic susceptibility becomes negative and diverges at a finite small temperature, and the entanglement entropy of a size-$L$ partition saturates instead of growing as $\\ln L$. If correct, this turns a regime previously treated mainly numerically—disorder, interactions, and non-Hermiticity together—into an analytically controlled problem.","feed_headline":"Non-Hermitian disorder saturates entanglement in spin chains","feed_subtitle":"Asymptotically exact low-energy solution predicts spin pairs become classical and xy susceptibility diverges.","key_machinery":"The engine is the strong-disorder renormalization group, a real-space scheme that repeatedly finds the two neighboring spins with the largest local excitation gap, freezes them in their biorthogonal ground state, and replaces them by one effective bond between the outer neighbors. The load-bearing identity is the decimation rule $\\tilde{\\gamma}_{i-1} = \\gamma_{i-1}+\\gamma_i+\\gamma_{i+1}$, identical in form to the rule for bond length $\\tilde{\\ell}_{i-1} = \\ell_{i-1}+\\ell_i+\\ell_{i+1}$ and independent of the flows of $J$ and $\\Delta$. This decoupling lets the authors transfer known fixed-point distributions for bond lengths to $\\gamma$, and gives the joint fixed-point distribution $Q_n(\\eta=0,x,y)$ of scaled couplings, lengths, and $\\gamma$ (subscript $n=1$ for symmetric, $n=2$ for asymmetric initial distributions) from which the susceptibility and the entropy follow.","core_discovery":"The paper claims that at low energies the disordered non-Hermitian XXZ chain flows to an infinite-disorder fixed point whose ground state is a random strongly coupled pair phase. In each decimated pair, the right ground state is $|GS^{(R)}\\rangle = \\cosh(\\gamma/2)|0,0\\rangle + \\sinh(\\gamma/2)|1,0\\rangle$ (and the left one has $\\gamma\\to -\\gamma$), so pairs are coherent mixtures of singlet and $M=0$ triplet; in the Hermitian limit $\\gamma=0$ they reduce to singlets. Because the decimation rule for $\\gamma_i$ is additive and decoupled from $J_i$ and $\\Delta_i$, the $\\gamma$ distribution broadens without bound, in an even universal form for symmetric initial distributions and a one-sided universal form for asymmetric ones. The broadening makes low-energy pairs nearly classical, which produces two concrete predictions: $T\\chi_{x,y}$ from already decimated pairs is negative, proportional to $-\\sinh^2(\\gamma/2)$, and diverges at a finite temperature $T=\\Omega_0 e^{-\\pi C_\\gamma}$ in the symmetric case; and, using the right-right reduced density matrix, the entanglement entropy of a partition saturates for large $L$, in sharp contrast to the $\\ln L$ growth of the Hermitian random-singlet chain.","pith_inferences":["If the claimed fixed-point distribution is right, the saturation of entanglement entropy should be visible in numerical simulations of finite chains with modest disorder: the crossover scale in $L$ should be controlled by $C_\\gamma$, so systems with broader initial $\\gamma$ distributions should saturate earlier.","The negative, diverging susceptibility is an equilibrium thermodynamic signature of non-Hermiticity; because the paper shows it is not self-averaging, any experiment or numerical estimate would need many disorder realizations before claiming a finite-temperature divergence.","The RR versus RL density-matrix ambiguity suggests a sharper test: measuring entanglement with different state-vector normalizations should give either maximal pair entropy (RL) or $\\gamma$-dependent suppression (RR); the paper's saturation picture only holds for RR."],"forward_implications":["The ground state is a random strongly coupled pair phase, not a random singlet phase: pairs of arbitrary length at random positions are singlet-triplet mixtures, so spin-spin correlations and response functions differ from the Hermitian chain.","The $xy$ magnetic susceptibility turns negative at low temperature and diverges at a finite temperature set by the initial distribution ($T=\\Omega_0 e^{-\\pi C_\\gamma}$ in the symmetric case), while the $z$ susceptibility keeps the Hermitian quasi-Curie form.","The entanglement entropy of a partition saturates at large $L$ instead of growing as $\\ln L$; the saturated value is nonuniversal and controlled by the same constants $C_\\gamma$ and $C_\\ell$ that set the flow.","The one-sided growth of the $\\gamma$ distribution at low energies gives a many-body, interacting explanation of the non-Hermitian skin effect: long bonds have strongly one-sided hopping, while symmetric initial $\\gamma$ distributions show no skin effect.","The same SDRG framework can be applied to other disordered non-Hermitian models, such as a non-Hermitian random transverse-field Ising chain, where a similar quantum-to-classical crossover is expected."],"supporting_citations":[{"why":"Supplies the two-site spectrum, biorthogonal SDRG decimation rules, the joint fixed-point distribution $Q_n$ used for later observables, and the derivations of susceptibility and entropy.","marker":"[45]"},{"why":"Establishes the strong-disorder fixed point of the Hermitian XXZ chain and the joint distribution of couplings and bond lengths whose flow the $\\gamma$ distribution mirrors.","marker":"[47]"},{"why":"Defines the fermionic Hatano-Nelson model whose interacting disordered version is the object of study.","marker":"[39]"},{"why":"Provides the infinite-disorder fixed-point concept that makes the SDRG predictions asymptotically exact.","marker":"[44]"},{"why":"Gives the method for summing entanglement contributions of strongly coupled pairs that cross a partition boundary.","marker":"[57]"},{"why":"Supplies the joint distribution at the cutoff and the summation procedure used for the entanglement entropy calculation.","marker":"[58]"},{"why":"Justifies the biorthogonal perturbation theory used to derive the decimation rules for non-Hermitian Hamiltonians.","marker":"[51]"},{"why":"Documents the erratic non-Hermitian skin localization that the paper's one-sided $\\gamma$ flow explains in the interacting case.","marker":"[20]"}],"fun_headline_variants":["Classical pairs from non-Hermitian disorder in spin chains","Entanglement saturation from non-Hermitian disorder","Negative susceptibility from classical pairs","Non-Hermitian spin pairs become classical","Entanglement entropy saturates in non-Hermitian chain"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis depends on an unproven joint fixed-point distribution of couplings, bond lengths, and non-Hermitian parameters, which the paper says will be derived elsewhere; the entanglement-entropy saturation is computed from it, so an error there would invalidate that central prediction.","fun_headline_variants_meta":{"raw":{"variants":["Classical pairs from non-Hermitian disorder in spin chains","Entanglement saturation from non-Hermitian disorder","Negative susceptibility from classical pairs","Non-Hermitian spin pairs become classical","Entanglement entropy saturates in non-Hermitian chain"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000765,"raw_usage":{"total_tokens":3433,"prompt_tokens":1022,"completion_tokens":2411,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":638,"completion_tokens_details":{"reasoning_tokens":2339}},"tokens_in":638,"tokens_out":2411,"duration_ms":15056,"temperature":1.0,"reasoning_tokens":2339,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:49:05.699872+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the SDRG on chains of millions of sites over many disorder realizations—the same numerical experiment the paper uses for its histograms—and compare the measured joint distribution of scaled couplings, lengths, and gamma at the cutoff with Eqs. (38)-(39); failure to collapse onto those forms would remove the basis for the predicted entropy saturation. A complementary check is to compute the right-right pair entropy of a two-site problem at large gamma and compare with Eq. (11).","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the two-site spectrum, biorthogonal SDRG decimation rules, the joint fixed-point distribution $Q_n$ used for later observables, and the derivations of susceptibility and entropy."},{"cited_title":"Measurement-Induced Phase Transition in a Disordered XX Spin Chain: A Real-Space Renormalization Group Study","cited_arxiv_id":"2507.11957","evidence_quote":"Defines the fermionic Hatano-Nelson model whose interacting disordered version is the object of study."},{"cited_title":"Iglói and C","cited_arxiv_id":null,"evidence_quote":"Justifies the biorthogonal perturbation theory used to derive the decimation rules for non-Hermitian Hamiltonians."}],"review_version":2}