{"id":"79ae6dc7-a35b-4720-8a4a-74fd2aab5ec7","arxiv_id":"2508.08255","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A single-photon quantum-walk experiment realizes the unitary almost-Mathieu operator and observes metal-insulator, parity-time symmetry-breaking, and all-imaginary-quasienergy spectral transitions.","lead":"This paper reports the first photonic quantum-walk implementation of the unitary almost-Mathieu operator, a discrete-time simulator of the Aubry-André-Harper model. It observes the predicted metal-insulator transition, a parity-time symmetry-breaking transition, and a second spectral transition where all quasienergies become imaginary.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Non-Hermitian transition evidence rests on reconstructed P(t) using the nominal loss η (Methods Eqs. 8-9); without independent PPBS calibration, the exponential-versus-constant P(t) contrast could be a normalization artifact.","rationale":"The paper is a well-executed photonic quantum-walk experiment with high similarity between measured and simulated normalized distributions; the Hermitian metal-insulator transition at \\lambda_1=\\lambda_2 is supported by clear ballistic-versus-localized dynamics and does not depend on the loss calibration. However, the most novel and load-bearing claims, namely the PT-symmetry-breaking transition and the second spectral transition, are inferred exclusively from the reconstructed total probability P(t). Equation 9 defines P(t) by multiplying the measured surviving-photon fraction by e^{8\\pi\\eta t}, a factor that assumes the PPBS loss is exactly the nominal value. Any deviation between the actual per-step loss and the nominal \\eta produces an exponential trend in the reconstructed P(t) that would be indistinguishable from the predicted PT-broken growth without an independent calibration. Since the Methods do not report a direct measurement of the PPBS reflectivity versus \\eta, and the data are not posted, this assumption is currently unverified. The reader's weakest assumption identifies precisely this point, and I concur. A secondary internal inconsistency (the text's \\eta_{\\rm PT}=-\\log\\lambda_0 versus the correct (\\log\\lambda_0)/(2\\pi) used in the numerics) reinforces the need for a precise calibration statement but does not by itself invalidate the experiment. The proposed calibration check would settle the concern: if the corrected P(t) reproduces the reported constant-versus-growing contrast, the conditional acceptance is warranted; if not, the non-Hmitian claims would need to be withdrawn. Therefore the verdict remains CONDITIONAL, unchanged from the reader's assessment.","tokens_in":14057,"tokens_out":16414,"duration_ms":169006,"concrete_test":"Independently measure the vertical-polarization loss per step of the PPBS assembly for each nominal \\eta used (e.g., \\eta=0.135 and 0.335, plus a PT-unbroken control such as \\eta=0.05) by sending a vertically polarized coherent beam through the full step setup and recording the transmitted intensity. Then recompute P(t) from the raw coincidence counts using in Eq. 9 the measured per-step loss \\eta_{\\rm actual}=-(1/8\\pi)\\ln(1-p) instead of the nominal \\eta. If the corrected P(t) for the PT-unbroken case (\\lambda_1<\\lambda_2, small \\eta) is not constant, or if the exponential growth at \\eta=0.335 disappears, the claimed PT and second spectral transitions are artifacts of the e^{8\\pi\\eta t} normalization; if the contrast survives, the claims stand.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central evidence for both non-Hmitian transitions is the reconstructed total probability P(t) in Figs. 3b,e and 4c,d. Methods Eq. 8 states that the physical walk implements \\tilde{W}=e^{-4\\pi\\eta}W, and Eq. 9 reconstructs P(t) by multiplying the measured surviving-photon fraction by e^{8\\pi\\eta t}. This multiplication assumes the per-step loss of the vertical polarization is exactly 1-e^{-8\\pi\\eta}. If the actual PPBS reflectivity deviates from this nominal value, the reconstructed P(t) acquires a residual exponential factor e^{8\\pi(\\eta-\\eta_{\\rm actual})t}. In the PT-unbroken regime, where theory predicts P(t)=constant, a positive residual would produce a spurious exponential rise that mimics PT breaking; in the broken regime it would corrupt the growth rate. The paper reports no independent calibration of the PPBS reflectivity as a function of \\eta, and the data are only available 'upon request'. The high similarity of normalized P(x,t) distributions does not test this global normalization, because Eq. 9 supplies it separately. Thus the non-Hmitian phase-transition observations are not yet secured against a calibration artifact. A secondary issue: the text states \\eta_{\\rm PT}=-\\log\\lambda_0 in several places, while the correct value used in the numerics is (\\log\\lambda_0)/(2\\pi); the missing 2\\pi is a typo, but it underscores the need for a precise calibration statement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental implementation of the unitary almost-Mathieu operator (UAMO) and its non-unitary (\"pseudo-unitary\") extension using single-photon discrete-time quantum walks with polarization-encoded coin states. In the Hermitian case (η=0) the authors measure six-step probability distributions on both sides of the self-dual line λ1=λ2 and map the standard deviation σ over parameter space to exhibit a metal-insulator transition. In the non-Hermitian case they introduce mode-selective loss through a partially polarizing beam splitter and reconstruct the total probability P(t); they interpret an exponentially growing P(t) as a signature of PT-symmetry breaking, and the persistence of no-loss eigenstates below a second critical η0 as evidence of a novel spectral transition in which all quasienergies become complex. Both non-Hermitian transitions are associated with changes in the spectral winding number of the Floquet operator.","tokens_in":14316,"tokens_out":13980,"duration_ms":144795,"significance":"If the experimental claims hold, this is the first implementation of the UAMO and its non-Hermitian counterpart, providing a concrete photonic platform for quasiperiodic, non-unitary Floquet physics. The paper benefits from exact theoretical predictions: the phase boundaries are taken from published formulas rather than fitted, and the measured normalized spatial distributions agree with simulations at high similarity (S>0.96 in the Hermitian panels and S>0.94 in the non-Hermitian panels). The central new physics, a spectral transition beyond which all quasienergies are imaginary, is well motivated theoretically and, if confirmed, would demonstrate a sharp difference between the discrete-time PUAMO and the continuum non-Hermitian AAH model. However, the non-Hermitian evidence currently rests on a reconstructed global norm whose calibration uncertainty is not quantified, and several published phase-boundary formulas contain sign/factor inconsistencies. These issues are fixable but are load-bearing for the main claims.","major_comments":[{"comment":"The central observable for both non-Hermitian transitions is the reconstructed total probability P(t)=e^{8πηt} Σ_x N(t,x)/[Σ_x N(t,x)+Σ_{t'}Σ_x N_L(t',x)]. This reconstruction explicitly assumes that the only significant loss is the PPBS loss with reflectivity p=1-e^{-8πη} and that the global factor in Eq. (8) is e^{4πη}. The paper reports no independent calibration of p(η) and no systematic-error propagation; if the actual per-step loss rate differs from the nominal η by δ, the reconstructed P(t) acquires a residual factor e^{8πδ t}. In the PT-unbroken case, where P(t) is predicted to be constant, a positive δ would produce a spurious exponential rise that mimics PT breaking; in the broken case it would corrupt the growth rate. The reported spatial similarities S>0.96 do not constrain this global normalization, because Eq. (9) supplies it separately. Please provide calibration data for the PPBS reflectivity as a function of η and the corresponding uncertainty band for P(t), or use a normalization-free observable.","section":"Methods, Eq. (9); Figs. 3b,e and 4c,d"},{"comment":"The critical values stated in the text are inconsistent with the values used in the numerics and with the Appendix. In the localized phase λ1<λ2 one has λ0>1 and hence logλ0>0, so the statement in the main text and Methods that PT symmetry is broken at ηPT=-logλ0, and the unbroken interval logλ0<η<-logλ0, are internally inconsistent. For λ1=0.25, λ2=0.5, logλ0≈0.746, while Fig. 4a reports ηPT=0.119; the latter equals (logλ0)/(2π). The Appendix formula L^♯=max{0,-logλ0+2π|η|-...} also gives the transition at |η|=logλ0/(2π). Please correct all occurrences, including the Appendix sentence \"L^♯=0 as long as |η|<logλ0\", which should read |η|<logλ0/(2π), and fix the sign so that the localized-phase transition is at +logλ0/(2π).","section":"Main text, Results; Methods; Appendix"},{"comment":"The no-loss states used to demonstrate the second transition are obtained from spectra computed under periodic boundary conditions, while the experiment is a finite open-boundary quantum walk. Because the model has non-reciprocal hopping, open and periodic boundary-condition spectra can differ substantially in non-Hermitian systems. Please justify that the experimentally prepared initial state has large overlap with a genuine no-loss state of the actual open system and that the flat P(t) at η=0.135 is not a boundary artifact; specify the lattice size and boundary treatment used in the experiment and in the numerical spectra.","section":"Fig. 4b and the Results section on the second spectral transition"}],"minor_comments":[{"comment":"The caption states that panels b and e correspond to the \"PT unbroken and broken phases, respectively,\" which is reversed relative to the text: panel b is the broken phase and panel e is the unbroken phase.","section":"Fig. 3 caption"},{"comment":"The word \"pervious\" in \"pervious non-Hermitian AAH model\" should be \"previous.\"","section":"Results, first paragraph"},{"comment":"References [31] and [54] are the same work (Liu, Zhou, and Chen, PRB 104, 024201) and should be merged.","section":"References"},{"comment":"The caption says the left panel has \"two turning points,\" while the text says there are four turning points at ±logλ0 and ±η0; make the caption consistent with the text.","section":"Fig. 5 caption"},{"comment":"Data and code are available only upon request; for a claims-heavy experimental paper, depositing the datasets and calibration files in a public repository would strengthen reproducibility.","section":"Data and code availability"}],"recommendation":"major_revision","confidential_remarks":"The main gate for acceptance is the calibration of the PPBS loss and the associated systematic uncertainty in P(t). The sign and 2π inconsistencies in the published critical formulas suggest a hasty writing pass; they are fixable. I do not regard the authors' reliance on their own theoretical papers [38,41] as disqualifying, but the experimental novelty should be framed carefully against previous lossy quantum-walk experiments [42,43]. The second-transition evidence is currently based on only two η values; a short η scan, even over three or four points, would considerably strengthen the claim. Overall, major revision is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead the Lin et al. quantum-walk paper. Bottom line: it is a legitimate experimental first for the UAMO, with a clean Hermitian phase diagram and data that genuinely track the predicted dynamics. The non-Hermitian part is plausible but rests on a normalization assumption that needs to be checked before I would call the PT and second transitions fully established.\n\nWhat is new: the UAMO is implemented for the first time, and the experiment sees the Hermitian metal-insulator transition via photon distributions and the standard deviation of position. The similarity numbers (0.96 and higher) are good. The second transition—where all quasienergies leave the unit circle—is a discrete-time effect that cannot happen in the non-Hermitian AAH model, so seeing a dynamical signature of it is the most interesting thing in the paper. No fitted parameters go into the phase boundaries; the predictions come from published formulas. The theory being confirmed is partly co-authored, but that does not bother me; the data are independent.\n\nWhere I would push back:\n\nThe central evidence for both non-Hermitian transitions is the reconstructed total probability P(t), which requires multiplying the measured surviving fraction by e^{8πηt} (Methods Eq. 9). That factor is only correct if the PPBS loss is exactly the nominal e^{-8πη} per step. The paper does not show an independent calibration of that reflectivity as a function of η. If the actual loss is even a few percent off, the PT-unbroken P(t)=const plateau in Fig. 3b/e turns into a spurious exponential, and the growth rates in the broken phase shift accordingly. This is the load-bearing weakness. It can be fixed with a calibration curve and/or a direct measurement of the effective non-unitary factor.\n\nSecond, the transitions are characterized at two parameter points each, by comparing P(t) behavior. That is thin for a phase transition claim. No winding number is measured; the topological attribution is inferred from the spectrum calculations. Fine as supporting evidence, but weaker than the text implies.\n\nThird, the claim that this is the first experimental demonstration of the non-Hermitian AAH model is too strong given the existing Floquet quasicrystal experiment (Weidemann et al., Nature 2022). The first UAMO claim is defensible; the broader one is not.\n\nAlso, the text states ηPT = -log λ0 in several places; the correct critical value used in the numerics is (log λ0)/(2π), which the Appendix confirms. That is a typo, but it is exactly the kind of slip that makes the calibration statement matter. Data and code are only \"available upon request\"—with a normalization-dependent analysis, they should be posted.\n\nIf I were the editor, I would send it to a careful experimental referee and ask specifically for the PPBS calibration. The UAMO implementation and the Hermitian transition are solid enough on their own. The non-Hermitian part is likely right, but it is not yet secured.\n\nRead it if you work on quasicrystal simulators; otherwise skim the Hermitian part.","headline":"A legitimate first UAMO implementation with a clean Hermitian phase diagram; the non-Hermitian transition claims are plausible but depend on a calibration-sensitive normalization and need tightening before they fully land.","tokens_in":14872,"tokens_out":5374,"would_cite":true,"duration_ms":57535,"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 reports the first experimental realization of the unitary almost-Mathieu operator with single-photon quantum walks, observing a metal-insulator transition at $\\lambda_1=\\lambda_2$ and two non-Hermitian spectral transitions tied…","keywords":["Aubry-André-Harper model","unitary almost Mathieu operator","non-Hermitian quantum walk","metal-insulator transition","PT symmetry breaking","spectral winding number","quasiperiodic potential","single-photon quantum walk"],"falsifier":"Measure the number of photons removed at each step directly, the $N_L(t,x)$ counts in Eq. (9), and compare the reconstructed $P(t)$ with the raw survival probability obtained by dividing detected counts by input counts; the claimed PT and all-complex transitions should appear only after the stated $e^{8\\pi\\eta t}$ normalization is applied. Repeating the same walk with a tunable attenuator in place of the loss element and checking that the fitted $\\eta_{\\rm PT}$ and $\\eta_0$ move exactly with the independently measured loss rate would settle whether the transitions are spectral or calibration artifacts.","tokens_in":13842,"feed_emoji":"⚛️","tokens_out":12479,"duration_ms":114233,"temperature":0.7,"pith_summary":"This paper reports the first experimental realization of the unitary almost-Mathieu operator (UAMO), an exactly solvable quantum-walk model that simulates the Aubry-André-Harper (AAH) quasicrystal, using single photons. In the Hermitian limit, the measured photon distributions show the predicted metal-insulator transition at $\\lambda_1=\\lambda_2$, with ballistic spreading on the metallic side and localization on the insulating side. By adding non-reciprocal hopping controlled by a parameter $\\eta$, the authors observe the parity-time (PT) symmetry-breaking transition at $\\eta_{\\rm PT}=-\\log\\lambda_0$, where the total photon probability begins to grow exponentially and the walker drifts directionally. They also identify a second, discrete-time-specific spectral transition at $\\eta_0=\\operatorname{arcsinh}(\\lambda'_1/\\lambda_1)/(2\\pi)$, above which all quasienergies become purely imaginary; both non-Hermitian transitions are accompanied by a change in the spectral winding number. The work matters because it turns a theoretically well-studied but experimentally elusive family of quasicrystal models into a controllable photonic platform and clarifies how localization, symmetry breaking, and topology interact.","feed_headline":"Single photons map a quasicrystal's three phase transitions","feed_subtitle":"First experiment on the unitary almost-Mathieu operator confirms localization, PT breaking, and a new all-complex spectral phase.","key_machinery":"The central object is the Floquet operator $W_{\\lambda_1,\\lambda_2,\\eta}=S_{\\lambda_1,\\eta}Q_{\\lambda_2,\\theta}$, a one-dimensional quantum walk with a two-level coin. The coin rotation at site $x$ is quasiperiodic, with angle $2\\pi(x\\Phi+\\theta)$ and coupling $\\lambda_2$; the shift operator has amplitudes $e^{\\pm 2\\pi\\eta}\\lambda_1$ that break reciprocity when $\\eta\\ne0$. In the Hermitian case this is the unitary almost-Mathieu operator, an exactly solvable simulator of the AAH model. The analysis is carried by three analytic quantities: the self-dual condition $\\lambda_1=\\lambda_2$, the Lyapunov exponent $\\log\\lambda_0$ with $\\lambda_0=\\lambda_2(1+\\lambda'_1)/(\\lambda_1(1+\\lambda'_2))$, and the spectral winding number $\\nu_\\eta(z)$ of Eq. (4), which is quantized to $0$ or $\\pm1$ and changes exactly at the PT-breaking and all-complex transitions.","core_discovery":"The paper's central claim is that a single-photon discrete-time quantum walk can implement the unitary almost-Mathieu operator and its non-Hermitian pseudo-unitary extension, and that dynamical measurements in this platform directly reveal three phase transitions. First, with $\\eta=0$, the walker's spatial distribution undergoes a metal-insulator transition on the self-dual line $\\lambda_1=\\lambda_2$, detected through the standard deviation of the position distribution after six steps. Second, in the localized phase $\\lambda_1<\\lambda_2$, increasing the non-Hermitian parameter $\\eta$ past $\\eta_{\\rm PT}=-\\log\\lambda_0$ spontaneously breaks PT symmetry: some quasienergies acquire imaginary parts, the total probability $P(t)$ grows exponentially rather than staying constant, and the winding number $\\nu_\\eta(z)$ becomes nonzero. Third, at the larger value $\\eta_0=\\operatorname{arcsinh}(\\lambda'_1/\\lambda_1)/(2\\pi)$, a transition occurs that has no analogue in the continuum non-Hermitian AAH model: all quasienergies move off the unit circle and become purely imaginary, so generic initial states exhibit amplification. The authors state that both non-Hermitian transitions are topological in origin because they coincide with quantized changes of the spectral winding number.","pith_inferences":["If the central claim is right, the $\\eta_0$ transition should appear in any discrete-time quantum walk with the same Floquet structure, not just photonic walks; testing it on other platforms would separate the physics from photon-loss calibration.","An independent measurement of the per-step photon loss rate, rather than the reconstructed global attenuation factor, would give a sharp check: the apparent crossover to exponential growth should shift exactly with the measured loss, not with the assumed $\\eta$.","The duality $\\theta\\mapsto\\theta-i\\eta$ exploited here suggests that the experiment effectively realizes complex quasiperiodic phases, offering a possible route to engineering imaginary gauge fields in other synthetic lattices.","If the paper's interpretation is correct, finite-lattice versions should show the same transitions in edge-state or mean-chiral-displacement observables, which longer quantum walks could test."],"forward_implications":["A single-photon quantum walk resolves the AAH metal-insulator boundary at $\\lambda_1=\\lambda_2$ through the spreading width of the photon wave packet.","In the localized phase, the PT-symmetry-breaking transition at $\\eta_{\\rm PT}=-\\log\\lambda_0$ shows up as the crossover from constant to exponentially growing total photon probability, together with directional transport.","The second transition at $\\eta_0=\\operatorname{arcsinh}(\\lambda'_1/\\lambda_1)/(2\\pi)$ marks a regime where all quasienergies are purely imaginary, so generic initial states are amplified; this spectral phase cannot occur in the standard non-Hermitian AAH model.","Both non-Hermitian transitions coincide with quantized changes of the spectral winding number $\\nu_\\eta(z)$, establishing their topological origin.","Above the PT-breaking threshold, mode-selective amplification becomes possible, since individual eigenstates can be addressed and amplified without affecting other modes."],"supporting_citations":[{"why":"Establishes the unitary almost-Mathieu operator as an exactly solvable simulator of the AAH model with the $\\lambda_1=\\lambda_2$ metal-insulator transition.","marker":"[38]"},{"why":"Predicts the PT-symmetry-breaking transition and the second spectral transition at $\\eta_0$, together with the winding-number criterion used here.","marker":"[41]"},{"why":"Supplies the alternating-loss single-photon quantum-walk implementation and the reconstruction formula for the total probability $P(t)$.","marker":"[42]"},{"why":"Provides the experimental scheme for observing PT-symmetry breaking and exceptional points in quantum walks.","marker":"[43]"},{"why":"Gives the rigorous metal-insulator transition for the almost Mathieu operator, the continuum result the UAMO simulates.","marker":"[20]"},{"why":"Predicts the topological PT-symmetry-breaking transition in non-Hermitian quasicrystals, the baseline the PUAMO extension builds on.","marker":"[33]"},{"why":"Supplies the Lyapunov-exponent and winding-number formulas for non-Hermitian quasicrystals used to compare the PUAMO with the AAH model.","marker":"[54]"},{"why":"Introduces the Aubry-André duality that underlies the self-dual localization transition.","marker":"[17]"}],"fun_headline_variants":["Single-photon walk maps three phase transitions in quasicrystal","First quantum walk experiment maps three phase transitions","Quantum walk reveals PT breaking and a new all-complex phase","New spectral transition found in non-Hermitian quasicrystal via photons","Quantum walk maps three transitions, including all-complex phase"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusions depend on the per-step loss introduced by the partially polarizing beam splitter and on the formula used to reconstruct the total photon probability matching the nominal value of $\\eta$; if that calibration is inaccurate, the observed change from constant to exponentially growing probability could be an artifact of the reconstruction rather than a spectral transition.","fun_headline_variants_meta":{"raw":{"variants":["Single-photon walk maps three phase transitions in quasicrystal","First quantum walk experiment maps three phase transitions","Quantum walk reveals PT breaking and a new all-complex phase","New spectral transition found in non-Hermitian quasicrystal via photons","Quantum walk maps three transitions, including all-complex phase"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00171,"raw_usage":{"total_tokens":6808,"prompt_tokens":1027,"completion_tokens":5781,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":643,"completion_tokens_details":{"reasoning_tokens":5698}},"tokens_in":643,"tokens_out":5781,"duration_ms":44892,"temperature":1.0,"reasoning_tokens":5698,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:36:30.351728+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the number of photons removed at each step directly, the $N_L(t,x)$ counts in Eq. (9), and compare the reconstructed $P(t)$ with the raw survival probability obtained by dividing detected counts by input counts; the claimed PT and all-complex transitions should appear only after the stated $e^{8\\pi\\eta t}$ normalization is applied. Repeating the same walk with a tunable attenuator in place of the loss element and checking that the fitted $\\eta_{\\rm PT}$ and $\\eta_0$ move exactly with the independently measured loss rate would settle whether the transitions are spectral or calibration artifacts.","supporting_citations":[{"cited_title":"Twenty dry Martinis for the Unitary Almost Mathieu Operator","cited_arxiv_id":"2503.06710","evidence_quote":"Supplies the alternating-loss single-photon quantum-walk implementation and the reconstruction formula for the total probability $P(t)$."},{"cited_title":"Absence of Bound States for Quantum Walks and CMV Matrices via Reflections","cited_arxiv_id":"2402.11024","evidence_quote":"Provides the experimental scheme for observing PT-symmetry breaking and exceptional points in quantum walks."},{"cited_title":"& Andr´ e, G","cited_arxiv_id":null,"evidence_quote":"Gives the rigorous metal-insulator transition for the almost Mathieu operator, the continuum result the UAMO simulates."},{"cited_title":"Phase transitions in a non-Hermitian Aubry-Andr\\'e-Harper model","cited_arxiv_id":"2102.09214","evidence_quote":"Predicts the topological PT-symmetry-breaking transition in non-Hermitian quasicrystals, the baseline the PUAMO extension builds on."},{"cited_title":"Direct observation of a localization transition in quasi-periodic photonic lattices","cited_arxiv_id":"0807.2845","evidence_quote":"Introduces the Aubry-André duality that underlies the self-dual localization transition."}],"review_version":1}