{"id":"4da31e56-72ba-4500-ac39-7c03fbd1cc1f","arxiv_id":"1908.11405","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Structured hidden-Markov processes can have exactly flat power spectra, so flat spectra cannot certify randomness; the paper characterizes, constructs, and demonstrates such processes.","lead":"A flat power spectrum is often read as a signature of white noise, but this paper shows that arbitrarily complex and even fully predictable processes can produce exactly flat spectra, hiding all structure from pairwise statistics. It gives closed-form formulas for the spectra of hidden Markov models and constructive recipes for building such 'fraudulent white noise', with examples in quantum entanglement and crystal diffraction.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the zero-mean flat-spectrum theorem is sound; remaining gaps are presentation-level.","rationale":"The reader's weakest_assumption focuses on nonstationarity, measurement drift, and finite-realization transfer. Those are legitimate caveats for applications, but they are not load-bearing for the central mathematical result, which is an infinite-ensemble statement. The central argument is robust: the autocorrelation identity and the zero-mean corollary are elementary and correct, and the paper's own physical examples are consistent with pairwise whiteness. The genuine soft spots are editorial and presentational: an unfinished sentence in App. O, a missing treatment of singular transition matrices in the polyspectra derivation, and a complex-conjugation oversight in the displayed formulas for complex alphabets. None of these changes the conclusion that flat power spectra can be generated by arbitrarily structured hidden-state processes. Therefore the reader's CONDITIONAL verdict, with its emphasis on polish and omitted cases, remains appropriate; no stronger action is needed.","tokens_in":63394,"tokens_out":21847,"duration_ms":235945,"concrete_test":"Re-derive Eq. (6) with γ(τ)=⟨π|\\bar{Ω}T^{|τ|}Ω|1⟩ for a two-state complex-mean HMM, e.g., deterministic alternation with μ_1=1 and μ_2=i. Verify that the corrected expression reproduces the direct periodogram of the sequence while the printed expression does not, and verify that setting all μ_s=0 still gives a constant spectrum. This settles whether the conjugation omission is merely notational or affects the theorem's force.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing claim is Cor. 1: any finite-state HMM whose every state-conditioned emission distribution has zero mean has zero autocorrelation for all nonzero lags, hence a flat power spectrum. This is correct: for τ > 0, γ(τ)=E[E[X_t|S_t] E[X_{t+τ}|S_{t+τ}]]=0 under conditional independence, regardless of how complex the hidden-state dynamics are. The constructive examples (RRXOR, content-preserving whitening, chaotic crystals) indeed have pairwise whiteness, and the non-finite-realization issue raised by the reader does not attack the theorem, which is explicitly an ensemble statement. The paper itself flags real presentation gaps: the unfinished sentence 'which we out' in App. O and the omitted singular-T case in the polyspectra derivation; there are also leftover editorial notes in Sec. II. These justify the reader's CONDITIONAL verdict but do not undermine the central claim. One genuine technical wart is that Eqs. (4) and (6) suppress complex conjugation in Ω for complex alphabets (the left Ω should be the conjugate-average matrix). This is a notational/proof-presentation gap: Cor. 1 is unaffected, and the nonzero-mean crystal calculations in App. I already use the correct conjugates.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the notion of 'fraudulent white noise': stochastic processes whose power spectral density is exactly flat although the generating mechanism carries arbitrarily complex temporal structure. The central analytical results are closed-form expressions for the autocorrelation and power spectrum of finite-state hidden Markov models (Eqs. (4)-(11)), Theorem 1 (the spectrum depends on state-conditioned emission distributions only through their means), Corollary 1 (any HMM whose state emissions all have zero mean has a flat spectrum), and Theorem 2 (a general sufficient condition for flat spectra in input-dependent and time-varying hidden-state models). The paper further derives a closed form for general polyspectra (Eq. (26)), gives spectral-degeneracy results (Theorem 3), and illustrates the claims with examples from entangled quantum measurements, close-packed chaotic crystals with encoded messages, and potassium ion-channel fluctuations. The central conclusion is that flat spectra are not evidence of structureless randomness and that higher-order or information-theoretic measures are needed to detect the hidden structure.","tokens_in":63603,"tokens_out":6175,"duration_ms":66856,"significance":"If the results are taken as stated, the message is important and broad: pairwise spectral measures are blind to arbitrarily high-order generative structure, even when that structure is entirely predictable. The core derivation, in particular Eq. (6) and Corollary 1, is sound and self-contained: it follows from the HMM definitions, conditional independence, and standard linear algebra, and the numerical examples match the analytic curves rather than being fitted. The constructive examples (zero-mean state emissions, RRXOR, content-preserving whitening, close-packed crystals carrying binary text) make the claim concrete and effectively falsifiable. The paper also gives a genuinely useful spectral framework connecting transition-matrix eigen-spectra, resolvents, and line shapes, and it extends the same machinery to polyspectra. These strengths justify serious consideration, provided the manuscript is brought to a complete and internally consistent state.","major_comments":[{"comment":"For complex alphabets (explicitly allowed since A is a subset of C), the autocorrelation is gamma(tau) = E[conj(X_t) X_{t+tau}], so the left average-observation matrix in Eqs. (4) and (6) should be the conjugate matrix, not Omega. As written, the formulas are incorrect for complex-valued state means; for instance, the right-hand side of Eq. (6) is not guaranteed to be real. This is a load-bearing gap in the claimed closed-form solution for the whole stated class of processes. The correction is local (replace the left Omega by its conjugate), and Appendix I already uses the correct conjugates, but the main-text equations and the statements that follow from them must be corrected.","section":"Sec. II C, Eqs. (4)-(6) and (10)-(11)"},{"comment":"The derivation of the general polyspectrum assumes that the transition matrix T is non-singular and then leaves the singular case with the unfinished sentence '... which we out.' Since Section V A claims a closed-form expression for the polyspectra of HMMs without excluding singular transition matrices, the treatment of the zero-eigenspace is part of the claimed result. The authors must either complete the derivation for singular T or explicitly restrict the theorem's statement and state the restriction in the main text.","section":"Appendix O, before Eq. (O9)"},{"comment":"The manuscript contains multiple explicit editorial notes and unfinished instructions: 'The technical level of results here is bouncing around...', 'Give the (flat) power spectrum for each. The reader needs this.', and 'Need cites to HMM literature: [27-33].' These are not merely stylistic; they show that the text as submitted is incomplete. All such notes must be removed or resolved, and the missing HMM literature citations must be supplied, before the paper can be published.","section":"Sec. II C and Fig. 1 caption"}],"minor_comments":[{"comment":"The sentence 'And, the eigenvalues of T_tau0 and G are simply related by ... [74]' is missing a period and runs into the following paragraph; please fix the punctuation and sentence boundary.","section":"Sec. II E"},{"comment":"The phrase 'indistinguishable from white noise' should be qualified as 'indistinguishable by power-spectral (second-order) measures alone', since the paper itself later shows that pairwise mutual information or higher-order measures can reveal the structure.","section":"Sec. III, Corollary 1"},{"comment":"The construction that encodes arbitrary binary text into a crystal with the same diffraction pattern is striking, but the text should state more explicitly how the concatenated six-layer blocks satisfy the pairwise joint-probability condition of Theorem 3, especially at block boundaries; a one-sentence justification matching the figure panels (a)-(d) would help the reader.","section":"Sec. IV B, Fig. 10"},{"comment":"The notation F(kappa)_K for the set of surjective functions is used in the main text but defined only in Appendix O; please add a definition or forward reference at first use.","section":"Eqs. (20)-(26)"},{"comment":"There are duplicated passages and placeholder figure captions (for example, the repeated text in the Fig. 1 block and the overlapping insets in Fig. 5). A careful editorial pass is needed to remove these artifacts.","section":"Overall manuscript"}],"recommendation":"major_revision","confidential_remarks":"The central mathematical claim is sound and the paper has real value, but the submitted text is clearly a draft: it contains unfinished sentences, editorial self-notes, and an incomplete case in the polyspectra derivation. I would not recommend acceptance until the authors complete the manuscript and fix the complex-conjugation issue in the main spectral formulas. The scope fits cond-mat.stat-mech and related interdisciplinary venues, but the current presentation would likely draw reviewer criticism that obscures the genuine contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: the central claim is right, and it is proven properly. Cor. 1—any finite-state HMM with zero-mean emissions in every state has a flat power spectrum—is exactly the kind of result that should make people stop using flat spectra as a randomness certificate. The closed-form spectrum, Eq. (6), and the accompanying eigen-spectrum interpretation are the real contributions, and they are derived cleanly from the HMM definitions and standard linear algebra. The numerical examples match the analytic curves. The physics examples, especially the crystal that encodes the entire manuscript in a flat diffraction pattern, make the point concrete without overclaiming.\n\nWhat is actually new: the general formula for power spectra of state-emitting HMMs, the broad flat-spectrum conditions in Thms. 2 and 5, and the content-preserving whitening construction. The observation that spectra only see pairwise statistics is not new, but the paper goes beyond that platitude and gives tools. The quantum example is a nice bridge, and the ion channel discussion draws a genuinely useful distinction between conformational switching and state-local noise.\n\nSoft spots are real but mostly presentation-level. There are leftover editorial notes in Sec. II, an unfinished sentence in App. O, and a missing treatment of singular transition matrices in the polyspectra derivation. The stress-test note about complex conjugation in Eqs. (4) and (6) is a genuine notational gap, but it does not touch Cor. 1. None of this undermines the central theorem. The reader's worry about nonstationarity and finite realizations does not land on the theorem itself, which is explicitly an ensemble statement; it is a caution for practitioners, not a flaw in the math.\n\nWho this is for: anyone who uses power spectra to certify randomness—neuroscience, cosmology, quantum measurement, crystallography. The paper deserves a serious referee; it is rough, but the load-bearing content is sound and the tools are widely useful. I would bring it to a reading group and I would cite it. Send it out, but tell the authors to clean up the notes before acceptance.","headline":"The core theorem is correct and the toolkit is useful; the manuscript needs cleanup before it is final, but it deserves a serious referee.","tokens_in":64113,"tokens_out":1309,"would_cite":true,"duration_ms":15512,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["02.50.-r","05.45.Tp","02.50.Ey","02.50.Ga"],"model":"deepseek-v4-flash","headline":"The paper proves that every finite-state hidden Markov process whose states share a common mean output has a perfectly flat power spectrum, so flat spectra cannot certify randomness.","keywords":["hidden Markov models","power spectral density","fraudulent white noise","higher-order correlations","polyspectra","statistical complexity","dependence function","diffraction pattern"],"falsifier":"Take any finite-state HMM whose state-conditioned distributions all have mean zero, simulate a single long realization, and compute the ensemble-averaged periodogram with Welch's method: if any systematic frequency-dependent structure appears beyond sampling noise, the theorem's prediction is violated. A decisive laboratory version is to build the content-preserving whitened crystal of Sec. IV B and measure its diffraction; a diffraction pattern that deviates from the predicted flat background plus two Bragg reflections would falsify the degeneracy claim.","tokens_in":63192,"feed_emoji":"📡","tokens_out":5302,"duration_ms":47654,"temperature":0.7,"pith_summary":"The paper argues that a flat power spectrum is not evidence of structurelessness. It proves that for any finite-state hidden Markov model, the power spectrum depends on the state-conditioned output distributions only through their means; therefore any process whose states all have the same mean output produces exactly the flat spectrum of white noise, no matter how elaborate the hidden dynamics or how rich the higher-order correlations. Sympathetically read, the paper's central claim is that pairwise statistics are generically blind to the structure that complex systems exhibit, so white-noise diagnoses based on spectra should be re-examined. This matters because power spectra are routinely the decisive diagnostic in physics, astronomy, neuroscience, materials science, and communications.","feed_headline":"Complex processes can fake white noise perfectly","feed_subtitle":"A new theorem shows flat power spectra reveal only per-state means, so higher-order structure can hide completely.","key_machinery":"The load-bearing object is the average-observation matrix $\\Omega=\\sum_{s\\in S}\\langle X\\rangle_{p(X|s)}|s\\rangle\\langle s|$, a diagonal matrix that records the conditional mean output of each hidden state. The argument runs by writing the autocorrelation as $\\gamma(\\tau)=\\langle\\pi|\\Omega T^{|\\tau|}\\Omega|1\\rangle$ and then Fourier-transforming it; all frequency structure enters through the resolvent $(e^{i\\omega}I-T)^{-1}$, filtered by $\\Omega$ on both sides. Because $\\Omega$ contains only means, replacing each state's distribution by a delta function at its mean leaves the spectrum unchanged up to a constant offset. The companion concept of fraudulent white noise names processes whose spectrum is flat while their generative structure is arbitrarily complex.","core_discovery":"The central discovery is Theorem 1 and its Corollary 1: every finite-state hidden Markov chain with state-conditioned output distributions all sharing the same mean generates a flat power spectrum indistinguishable from genuine white noise. More generally, the power spectrum of an HMM is insensitive to the shape, support, or higher moments of each state's output distribution; only the per-state averages matter. The paper also gives a closed-form expression for the continuous part of the power spectrum, $P_c(\\omega)=\\langle |x|^2\\rangle + 2\\,\\mathrm{Re}\\langle\\pi|\\Omega T(e^{i\\omega}I-T)^{-1}\\Omega|1\\rangle$, and shows the spectrum is a filtered image of the transition matrix's resolvent on the unit circle. It then demonstrates the same blindness afflicts polyspectra in certain cases, and introduces the dependence function as a tool for detecting $L$-way correlations that spectra cannot see.","pith_inferences":["A direct practical extension the authors leave implicit: spectral flatness alone should never be used to certify a random-number generator; a block-entropy or dependence-function test would be cheap to add.","The construction is testable in the lab: encode a known binary message via content-preserving whitening into a physical stacking process or photonic sequence, then confirm that the measured spectrum matches the predicted flat background while a nonlinear decoder recovers the message.","The same logic applies to biomedical signal analysis: EEG or neural spike-train 'white noise' bands could conceal functionally relevant high-order correlations that current pairwise diagnostics ignore."],"forward_implications":["A flat power spectrum is necessary but not sufficient for genuine white noise; any empirical claim of 'no structure' based on a flat spectrum is now formally incomplete.","The number of resolvable peaks in a power spectrum lower-bounds the number of hidden states of any generative model, since each peak emanates from an eigenvalue of the transition matrix.","For a broad class of stochastic generators, power spectra of different mechanisms coincide whenever the joint statistics of state-averaged outputs agree, so diffraction patterns cannot uniquely determine stacking structure.","Sequential measurements of entangled quantum states can produce fraudulent white noise, so pairwise tests on measurement records can fail to certify quantum randomness.","Polyspectra inherit some of these blind spots; the cumulant bispectrum is flat for processes with equal per-state means, pointing to information-theoretic probes like myopic entropy rates and the dependence function."],"supporting_citations":[{"why":"Supplies the spectral decomposition of nonnormal and nondiagonalizable operators used to derive the closed-form power spectrum.","marker":"[63]"},{"why":"Introduces the coronal-spectrogram representation, the power-of-pairwise-information (POPI) spectrum, and the RRXOR example that later serves as a canonical fraudulent white noise process.","marker":"[62]"},{"why":"Establishes the coronal spectrogram and diffraction-pattern methods that connect eigen-spectra to observed spectra.","marker":"[65]"},{"why":"Provides the computational-mechanics notion of statistical complexity used to show content-preserving whitening preserves or increases generative complexity.","marker":"[81]"},{"why":"Formulates the phase problem for diffraction, the inverse problem the paper argues is fundamentally degenerate.","marker":"[100]"},{"why":"Defines the general polyspectra framework whose closed-form expressions the paper derives and whose blind spots it exposes.","marker":"[115]"},{"why":"Provides the experimental and numerical ion-channel analysis with which the paper contrasts its result that state-conditioned IID noise cannot produce 1/f features.","marker":"[110]"},{"why":"Gives the earlier theoretical calculation of potassium channel Lorentzian spectra that the paper's rate-matrix derivation recovers.","marker":"[112, 113]"}],"fun_headline_variants":["Flat power spectra can hide any complexity","White noise faked by complex processes","Power spectra blind to higher-order correlations","Complex processes can mimic white noise exactly"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The flatness results assume a finite-state hidden Markov model with a stationary distribution and conditionally independent outputs given the hidden state; a real system with nonstationary statistics or memory in the measurement apparatus need not show a flat spectrum even if its hidden dynamics are complex.","fun_headline_variants_meta":{"raw":{"variants":["Flat power spectra can hide any complexity","White noise faked by complex processes","Power spectra blind to higher-order correlations","Complex processes can mimic white noise exactly"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000208,"raw_usage":{"total_tokens":1439,"prompt_tokens":1017,"completion_tokens":422,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":633,"completion_tokens_details":{"reasoning_tokens":371}},"tokens_in":633,"tokens_out":422,"duration_ms":4580,"temperature":1.0,"reasoning_tokens":371,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:16:40.778703+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take any finite-state HMM whose state-conditioned distributions all have mean zero, simulate a single long realization, and compute the ensemble-averaged periodogram with Welch's method: if any systematic frequency-dependent structure appears beyond sampling noise, the theorem's prediction is violated. A decisive laboratory version is to build the content-preserving whitened crystal of Sec. IV B and measure its diffraction; a diffraction pattern that deviates from the predicted flat background plus two Bragg reflections would falsify the degeneracy claim.","supporting_citations":[{"cited_title":"Van Raamsdonk","cited_arxiv_id":null,"evidence_quote":"Formulates the phase problem for diffraction, the inverse problem the paper argues is fundamentally degenerate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the general polyspectra framework whose closed-form expressions the paper derives and whose blind spots it exposes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the experimental and numerical ion-channel analysis with which the paper contrasts its result that state-conditioned IID noise cannot produce 1/f features."}],"review_version":1}