{"id":"ee3a93b2-1f26-43bd-9bf9-49fd366129a0","arxiv_id":"2412.20306","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Rapid dephasing localizes the eigenmodes of the incoherent propagator in weakly disordered non-Hermitian lattices, causing abrupt jumps between distant sites.","lead":"This paper shows that rapid random phase kicks, normally a delocalizing effect, make waves in a weakly disordered chain with gain and loss suddenly jump between distant sites. The jump times can be predicted from the eigenvalues of the averaged evolution matrix, and the effect should be testable in existing photonic quantum-walk experiments.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Single-realization comparison to ensemble-averaged theory lacks phase-ensemble statistics; the a priori jump prediction may hold only on average.","rationale":"The reader's weakest assumption matches the most load-bearing concern I can identify: the paper derives its predictive formulas for the ensemble-averaged probability but validates them on a single random-phase realization. This is not a manufactured issue; the authors themselves flag it in the text, yet they do not quantify the fluctuations. If different phase sequences produce different jump sequences, then the headline claim that jump positions and durations 'can be analytically predicted a priori' from S is not true for a given experiment—only for the statistical average. That would weaken the central claim from a deterministic prediction to an ensemble statement, which is a significant but not fatal adjustment. The mathematical derivation of S and the localization mechanism in Appendix B are sound for the ensemble average, and the fast-dephasing expansion independently supports the localization of S's eigenmodes. I considered whether the possible presence of negative eigenvalues of S could invalidate Eq. (9) (since ln β_j would be undefined), but in the fast-dephasing regime used in the main figures, S is diagonally dominant with positive diagonal and small off-diagonal entries, so all eigenvalues are expected to be positive; this is a lesser concern. The single-realization-versus-ensemble issue is more direct and is explicitly conceded by the paper. The proposed concrete test—repeating the same disorder realization and initial condition over many phase sequences and measuring the spread of jump positions and rates—would settle whether the a priori prediction survives for individual runs. Given that this concern was already the basis for the reader's CONDITIONAL verdict, and no new objection emerges, the verdict should remain unchanged.","tokens_in":15701,"tokens_out":5065,"duration_ms":56666,"concrete_test":"Fix the disorder realization and initial condition used in Fig. 2. Generate M=100 independent random-phase sequences (different seeds) for the same dephasing period l=0.01. For each realization, extract the jump positions (the z values at which the most probable site changes, or at which the dominant projection onto S eigenvectors crosses) and the piecewise-constant d ln P/dz. Compute the mean and standard deviation over the phase realizations for each jump position and rate. If the spread is much smaller than the typical separation between jumps, the single-realization comparison is justified; if the spread covers multiple jump intervals, the a priori prediction fails for individual realizations and holds only on average.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that jump positions and durations can be predicted a priori from the eigenvalues and eigenvectors of the incoherent propagator S—rests on comparing Eqs. (9) and (12), derived after statistical averaging over dephasing phases, to a single random-phase realization (Figs. 2 and 3). The ensemble-averaged probability obeys P(z)=S^a P(0), giving Eq. (9). But a single realization evolves under the random product ψ(z_a)=D_a U ... D_1 U ψ(0), where D_a are diagonal phase matrices; its instantaneous intensity contains interference cross terms that do not cancel. The authors explicitly acknowledge this in Sec. IV: the wave function and d ln P/dz are 'numerically calculated for a single realization of random phases' although Eqs. (9) and (12) 'were derived under the assumption of statistical averaging.' No phase-ensemble statistics, no random seed, and no code are provided. Without a demonstration that inter-realization fluctuations are small compared with the gaps between ln β_j, the 'a priori' prediction is established only for the ensemble average, not for any particular experimental run. The effect may be real, but the paper's strongest claim is under-supported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a one-dimensional tight-binding lattice with complex on-site disorder under periodic random-phase dephasing. It argues that, in contrast to coherent weak-disorder dynamics, rapid dephasing makes the eigenmodes of the ensemble-averaged incoherent propagator S(l) highly localized and produces abrupt jumps between spatially distant regions. The principal analytic results are Eq. (9), ln|d_j| = ln|d_{j,0}| + (ln β_j / l) z, and Eq. (12), d ln P/dz = ln β_m / l, obtained from the spectral decomposition of S. The authors claim that jump positions and durations can be predicted a priori from the potential distribution and initial condition, and they compare these predictions with single-realization numerical simulations in Figs. 2 and 3. The dephasing-rate dependence of the power derivative and of the eigenmode IPR is analyzed in Figs. 4 and 5.","tokens_in":15892,"tokens_out":8692,"duration_ms":92097,"significance":"If the central claim holds, this is a useful counterpoint to the well-known dephasing-induced delocalization in Hermitian disordered systems: in non-Hermitian lattices, dephasing can instead sharpen localization and create abrupt dynamical transfers. The analytical framework is elementary and parameter-free: S is real and symmetric, its Perron-Frobenius eigenvalue controls the long-z evolution, and the fast-dephasing expansion in Eq. (B2) makes the eigenmodes nearly site-localized. The predictions in Eqs. (9) and (12) are falsifiable and can be checked by direct matrix computation. The main weakness is that the a priori prediction is compared only with a single random-phase realization, although Eqs. (9) and (12) are derived after statistical averaging; this gap is acknowledged in the text and needs to be addressed quantitatively.","major_comments":[{"comment":"The paper’s strongest claim, that jump positions and durations can be predicted a priori from the eigenvalues and eigenvectors of S, is supported only at the level of the ensemble-averaged probability, not for the single random-phase realization shown. The text states in Sec. IV that Eqs. (9) and (12) were derived under the assumption of statistical averaging, while the numerical wavefunction and d ln P/dz are computed for a single realization. A single realization evolves by a random product of phase matrices and unitary steps, and its instantaneous intensity contains interference cross terms that do not survive the averaging. The crossing of the averaged coefficients ln|d_j| therefore need not coincide with the crossing in any particular run. The authors should either provide phase-ensemble statistics (for example, distributions of jump positions, jump durations, and d ln P/dz over many phase realizations for the same disorder potential) or give a quantitative self-averaging or concentration argument showing that inter-realization fluctuations are small compared with the gaps between ln β_j. Without such evidence, the a priori prediction is established for the average, not for a single experimental run.","section":"Sec. IV, Eqs. (9) and (12); Figs. 2 and 3"},{"comment":"The quantity plotted in Fig. 2(b) is not fully specified. The projection coefficients d_j in Eq. (9) are defined for the unnormalized, ensemble-averaged probability vector |P(z)>, whereas the wavefunction in Fig. 2(a) is the normalized single-realization wavefunction |φ> = |ψ>/√P(z). Because normalization introduces a z-dependent prefactor and because the single-realization intensity is not the averaged probability, it is unclear whether the plotted ln|d_j| values are computed from the averaged P(z) or from the instantaneous single-realization intensity. The authors should state exactly which quantity is plotted and whether the comparison in Fig. 2(b) is intended to be literal or illustrative.","section":"Sec. IV, Fig. 2(b)"}],"minor_comments":[{"comment":"The heading contains a typo: “Emergenge” should be “Emergence”.","section":"Sec. IV, heading"},{"comment":"The inline label “(7)” in Eq. (10) is not a proper reference; the sentence should read “using Eq. (7)”.","section":"Eq. (10)"},{"comment":"The text invokes a critical dephasing period l_cr(W) but never defines it; the paper should specify the threshold criterion used for “complete localization” and should provide error bars or a statistical measure for the mean IPR over the 1000 realizations.","section":"Sec. V, Fig. 5"},{"comment":"No random seed or code is provided for the single-realization simulations, so the numerical results are not reproducible; providing a seed or a small code repository would allow readers to verify the claimed agreement.","section":"Figs. 1–4"},{"comment":"The derivation of Eq. (5) assumes that the dephasing phases at different sites and steps are independent and uniformly distributed; this assumption should be stated explicitly in the main text rather than only in Appendix A.","section":"Sec. II, Eq. (5)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and the analytical core is sound. The main substantive issue is the single-realization support for the a priori prediction; if the authors supply phase-ensemble statistics or a rigorous concentration/self-averaging argument, I would support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline: this paper has a real idea — rapid dephasing localizes the eigenmodes of the incoherent propagator S in weakly disordered non-Hermitian lattices, producing abrupt jumps that are absent under coherent dynamics. That's a nice counterintuitive result, and the paper gives an elementary, internally consistent account: S is real symmetric, the dominant eigenvalue controls long-z dynamics, and the fast-dephasing expansion makes S nearly diagonal. No parameters are fitted; the jump positions come from the spectral decomposition of S. That is real value.\n\nWhat's new: coherent Anderson jumps are known, and dephasing in non-Hermitian systems has been studied (Longhi's recent work especially), but the specific combination — weak complex disorder plus rapid dephasing localizes eigenmodes and makes jumps analytically predictable — is not in the cited literature. The paper credits the right prior work and doesn't oversell the novelty.\n\nThe soft spot is the one you'd expect: the comparison in Figs. 2 and 3 is between ensemble-averaged theory (Eqs. 9 and 12) and a single realization of random phases. The authors are honest about this — they state it explicitly in Sec. IV. But their central claim is that jump positions and durations can be predicted a priori for a given experiment. A single realization with no phase-ensemble statistics, no seed, and no code does not establish that. The interference cross terms that vanish under averaging may or may not shift the crossing points in a typical run. If the gaps between ln β_j are large relative to realization fluctuations, the prediction is robust; if not, it only holds on average. The paper doesn't show which regime it's in.\n\nThere's also a second, minor gap: the transition from S nearly diagonal to actual localized eigenmodes is sketched via a second-order expansion, and the saturation in Fig. 4 is shown for one potential realization. The disorder-averaged IPR in Fig. 5 covers that for localization, but not for the jump-prediction claim.\n\nBottom line: this is worth engaging with. The mechanism is plausible and the math is transparent. The single-realization issue is a load-bearing caveat but not a refutation — it should be fixed with a phase-ensemble statistical analysis, and probably it will hold. I'd send it to review. A serious referee can ask for the missing statistics without throwing out the idea.\n\nBest,\n[You]","headline":"A plausible and clean mechanism for dephasing-induced jumps, but the a priori prediction is tested on one realization only — fix that and it's solid.","tokens_in":16430,"tokens_out":2212,"would_cite":true,"duration_ms":20422,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.25.Dd","71.23.-k"],"model":"deepseek-v4-flash","headline":"Dephasing, normally a delocalizer, sharpens localization and causes abrupt jumps between distant sites in weakly disordered non-Hermitian lattices.","keywords":["dephasing","Anderson localization","non-Hermitian disorder","Anderson jumps","incoherent propagator","photonic lattice","gain and loss disorder","inverse participation ratio"],"falsifier":"Run the full dephased dynamics for many independent random-phase sequences using the same weak-disorder realization (e.g., $N=50$, $W_R=W_I=1$, single-site excitation at $n=25$, $l=0.01$) and compare each jump's propagation distance with the crossings of the $\\ln|d_j|$ lines computed from $S$; if the distances scatter widely instead of clustering at the predicted values, the single-realization predictability claim fails.","tokens_in":15488,"feed_emoji":"🔀","tokens_out":8758,"duration_ms":81259,"temperature":0.7,"pith_summary":"Dephasing—randomizing the wave function's phase at fixed intervals—is usually thought of as a decoherence that destroys Anderson localization and restores diffusion. This paper argues that in one-dimensional non-Hermitian lattices with weak complex on-site disorder the opposite happens: rapid dephasing sharpens the localization of the eigenmodes and turns the evolution of the normalized wave function into a sequence of abrupt jumps between spatially distant regions. The authors show that the position and duration of each jump can be computed in advance from the eigenvalues and eigenvectors of the averaged incoherent propagator $S$, whose largest eigenvalue eventually dominates. This matters because it suggests that a normally destructive environmental noise can, in open systems with gain and loss, be used as a controllable switching mechanism rather than only as a source of decoherence.","feed_headline":"Dephasing turns weak-disorder lattices into a sequence of jumps","feed_subtitle":"Random phase kicks sharpen eigenmode localization and make the wave function hop between predictable distant regions.","key_machinery":"The load-bearing object is the incoherent propagator $S(l)$, the real symmetric non-negative matrix with entries $S_{kj}=|(e^{iHl})_{kj}|^2$; it advances the ensemble-averaged site probabilities by one dephasing period. Its eigen-equation $S |v_j\\rangle = \\beta_j |v_j\\rangle$ supplies the straight-line projection formula $\\ln|d_j| = \\ln|d_{j0}| + (\\ln \\beta_j / l)\\, z$ and the power-derivative formula $d\\ln P/dz = \\ln \\beta_m / l$, where $m$ labels the dominant eigenvalue. Jumps occur at crossings of the dominant projections, and the fast-dephasing expansion $U \\approx I + iHl - \\tfrac12 H^2 l^2$ shows why: to second order in $l$, $S$ is nearly diagonal with diagonal entries $1 - 2 b_n l + (2 b_n^2 - 2)l^2$ set by the imaginary disorder, making the eigenstates approximate single lattice sites.","core_discovery":"The central claim is that in the fully incoherent regime—phases randomized every propagation step $l$—the ensemble-averaged site probabilities evolve under the non-negative symmetric matrix $S_{kj}=|(e^{iHl})_{kj}|^2$, and the projections $d_j$ onto $S$'s eigenstates obey $\\ln|d_j| = \\ln|d_{j0}| + (\\ln \\beta_j / l)\\, z$. Because the eigenstate with the largest eigenvalue $\\beta_{\\max}$ dominates after a finite distance $z_{\\rm cr}$, the probability density switches sharply between regions of the lattice when one dominant projection overtakes another; the same crossing explains plateaus in $d\\ln P/dz = \\ln \\beta_m / l$. Under weak disorder, coherent dynamics is smooth because eigenstates overlap, but dephasing makes $S$'s eigenstates increasingly single-site localized, so the jumps become abrupt; in the fast-dephasing limit $l\\to 0$ the diagonal entries of $S$ are set by the imaginary (gain/loss) disorder alone, and all eigenstates become single-site localized. The paper verifies these predictions for single phase realizations and shows that the mean inverse participation ratio of $S$'s eigenstates tends to 1 below a critical dephasing period.","pith_inferences":["Because Eqs. (9) and (12) are derived after averaging over phase ensembles, the a priori prediction of jump positions may hold only for the ensemble-averaged probability; whether a single laboratory run with one random phase sequence reproduces those exact jump distances is a separate test the paper leaves open by comparing to only one realization.","The mechanism, a non-negative symmetric propagator whose dominant eigenvector changes with parameters, is generic, so similar dephasing-induced jumps should appear in other non-Hermitian transport models where noise is equivalent to periodic phase randomization, not only the tight-binding chain studied here.","One testable extension: measure the jump-distance distribution over many phase realizations for a fixed disorder sample; a narrow distribution would validate single-shot predictability, while a broad distribution would confine the prediction to averaged dynamics.","The saturation of $d\\ln P/dz$ and $\\langle\\mathrm{IPR}\\rangle$ with decreasing $l$ suggests a crossover line $l_{\\rm cr}(W)$ that could be mapped experimentally as a function of gain/loss disorder strength, providing a phase diagram for jumpy versus smooth incoherent transport."],"forward_implications":["A single-channel excitation in a weakly disordered non-Hermitian lattice with $l < 10^{-4}$ will evolve as a sequence of hops between distant single sites, because all eigenstates of $S$ are then localized at one site.","The propagation distance at which the wave function jumps can be fixed by the disorder realization and dephasing rate, so dephasing becomes a tunable trigger for abrupt switching.","The long-term growth or decay rate of optical power is set by $\\ln \\beta_{\\max} / l$ and saturates as $l$ shrinks, giving a bounded, predictable power response.","Imaginary on-site disorder (gain/loss) alone suffices to produce the localization and jumps, since the real potential does not enter $S$'s diagonal to second order in $l$.","The predictions apply to existing photonic discrete-time quantum walk experiments that can implement periodic random phase kicks and measure the normalized wave function."],"supporting_citations":[{"why":"Defines the non-Hermitian jumps in disordered lattices under coherent conditions that this work shows are absent at weak disorder and revived by dephasing.","marker":"[102]"},{"why":"Reports sudden Anderson jumps in non-Hermitian open systems, the phenomenon the dephasing mechanism is compared against.","marker":"[100]"},{"why":"Supplies the incoherent propagator formalism and the fast-dephasing expansion used to derive $S$ and the single-site localization limit.","marker":"[132]"},{"why":"Demonstrates the photonic discrete-time quantum-walk platform where the predicted jumps could be observed.","marker":"[101]"},{"why":"Provides the standard phase-randomization model of dephasing on which the equation of motion is based.","marker":"[111]"},{"why":"Establishes the Anderson-localization baseline whose Hermitian dephasing response contrasts with the paper's non-Hermitian result.","marker":"[74]"}],"fun_headline_variants":["Dephasing makes weak-disorder lattices jump between distant sites","Weak disorder + dephasing = abrupt jumps in non-Hermitian lattices","Dephasing triggers abrupt hops in non-Hermitian disordered lattices","Random phase kicks cause sudden jumps in non-Hermitian lattices","How dephasing turns weak disorder into abrupt leaps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's predictions assume that one particular sequence of random phases behaves like the statistical average over all phase sequences, so that the eigenvalues and eigenvectors of $S$ accurately locate the jumps in a single experimental run.","fun_headline_variants_meta":{"raw":{"variants":["Dephasing makes weak-disorder lattices jump between distant sites","Weak disorder + dephasing = abrupt jumps in non-Hermitian lattices","Dephasing triggers abrupt hops in non-Hermitian disordered lattices","Random phase kicks cause sudden jumps in non-Hermitian lattices","How dephasing turns weak disorder into abrupt leaps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000732,"raw_usage":{"total_tokens":3249,"prompt_tokens":892,"completion_tokens":2357,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":508,"completion_tokens_details":{"reasoning_tokens":2265}},"tokens_in":508,"tokens_out":2357,"duration_ms":15783,"temperature":1.0,"reasoning_tokens":2265,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:23:51.989005+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the full dephased dynamics for many independent random-phase sequences using the same weak-disorder realization (e.g., $N=50$, $W_R=W_I=1$, single-site excitation at $n=25$, $l=0.01$) and compare each jump's propagation distance with the crossings of the $\\ln|d_j|$ lines computed from $S$; if the distances scatter widely instead of clustering at the predicted values, the single-realization predictability claim fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the non-Hermitian jumps in disordered lattices under coherent conditions that this work shows are absent at weak disorder and revived by dephasing."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports sudden Anderson jumps in non-Hermitian open systems, the phenomenon the dephasing mechanism is compared against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the incoherent propagator formalism and the fast-dephasing expansion used to derive $S$ and the single-site localization limit."},{"cited_title":"Lifshitz tail states in non-Hermitian disordered photonic lattices","cited_arxiv_id":"2412.09106","evidence_quote":"Provides the standard phase-randomization model of dephasing on which the equation of motion is based."},{"cited_title":"Xiao and Q.-B","cited_arxiv_id":null,"evidence_quote":"Establishes the Anderson-localization baseline whose Hermitian dephasing response contrasts with the paper's non-Hermitian result."}],"review_version":1}