{"id":"ca5eb4d4-e266-43bf-bb6d-897606f497a7","arxiv_id":"2608.05432","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In coded time-entanglement QKD reconciliation, infinite diversity (exponential error decay) is achieved if and only if the code's photons per codeword fit within the algebraic decoding radius, or, under soft decoding, if the code is \"MFD deficient\".","lead":"This paper proves conditions under which error-correcting codes used in time-entanglement QKD reconciliation achieve \"infinite diversity\", meaning the reconciliation error probability decays exponentially with signal-to-noise ratio rather than polynomially. The result matters because it suggests that short codes can make QKD reconciliation far more robust to detector timing jitter.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's sufficiency (MFD deficiency => infinite diversity) is false: the proof's 'WLOG zero codeword' ignores that Gray-code adjacency is not translation-invariant; an explicit MFD-deficient [6,2,2] code has a finite-diversity error event.","rationale":"The reader's weakest assumption correctly noted that the MFD property depends on the Gray labeling, but the deeper problem is that the proof of Theorem 2 translates an arbitrary transmitted codeword to the all-zero codeword in the binary label space. This translation changes the physical bins and hence the set of single-bin jumps available to the adversary. A single-bin jump from a nonzero label a lands on a label adjacent to a in the Gray sequence, whose XOR difference with a need not be a neighbor of zero. The provided counterexample is a valid linear code with minimum Hamming distance 2 that satisfies all stated hypotheses of Theorem 2, yet has a pairwise error event formed by two single-bin jumps with probability Θ(γ^{-1}), forcing finite diversity. This is not a minor technical gap in the monotonicity of the DMZ cube; it invalidates the central 'if' direction of the iff characterization. The paper's simulations do not cover this class of codes, so the flaw would not be visible in the displayed examples. A single reproducible simulation or analytical integration over the region [6,7)^2 for the stated code would settle the issue. The verdict should move from CONDITIONAL to REJECT because the headline claim about soft-decision decoding is not correct as stated.","tokens_in":39166,"tokens_out":30202,"duration_ms":285396,"concrete_test":"Compute the exact soft-decision MAP word error rate for C=span(111111,010010) with N=8, m=3, and standard reflected Gray labels 000,001,011,010,110,111,101,100, at SNR γ = 20 to 40 dB, either by Monte Carlo simulation or by numerical integration of the APP decision regions. If the slope of log Pe versus log γ is finite (about -1, as the pairwise lower bound predicts), Theorem 2's 'MFD deficient => infinite diversity' direction is disproved. If the slope diverges, the proposed counterexample is wrong and the concern should be withdrawn.","verdict_should_be":"REJECT","load_bearing_attack":"The load-bearing step is the sufficiency proof of Theorem 2. The proof assumes that a transmitted word c_A can be reduced to the all-zero word by linearity, and then checks only whether a nonzero codeword can be formed from labels adjacent to zero (Definition 3). But physical single-bin jumps from a nonzero label a land on labels adjacent to a in the Gray sequence, and the XOR difference a XOR (adjacent label) is not generally one of the zero-neighboring labels. The channel's APP comparisons depend on physical bin adjacency, not on XOR distance from the zero label, so the reduction to c_A=0 is invalid. Concretely, take m=3 with standard reflected Gray labels 000,001,011,010,110,111,101,100 for bins 0..7, so the zero-neighboring labels are {001,100}. Let C=span(111111,010010)={000000,111111,010010,101101}. Nonzero codeword labels are 111,010,101, none in {001,100}, so C is MFD deficient by Definition 4 and Theorem 2 predicts infinite diversity. But if Alice sends 111111, both photons are in bin 5. For Bob's photon positions y1,y2 in [6,7)^2 (bin 6), the competing word 101101 has higher APP in each coordinate than 111111, and by Proposition 1 each such single-bin jump has probability Θ(γ^{-1/2}). The joint event has probability Θ(γ^{-1}), giving word-error diversity at most 1 under soft-decision decoding. This directly contradicts Theorem 2 statement 1. The theorem could be repaired only by redefining MFD deficiency relative to every possible transmitted label pattern, which is not what Definition 3 states.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the diversity order of coded information reconciliation in time-entanglement QKD (TE-QKD). It derives high-SNR transition laws separating single-bin jumps (polynomial probability) from multi-bin jumps (exponential probability), and it formulates two main results: Theorem 1 gives a necessary and sufficient condition for infinite diversity under bounded-distance algebraic decoding in terms of the number of photons per codeword, while Theorem 2 claims that soft-decision decoding achieves infinite diversity if and only if the code is MFD deficient. The paper also presents rate bounds and simulation examples for Golay, Reed-Solomon, BCH, and Reed-Muller codes. The algebraic-decoding result is plausible, but the soft-decision characterization is not correct as stated.","tokens_in":39461,"tokens_out":8833,"duration_ms":84975,"significance":"If the main theorems were correct, the paper would report a striking phenomenon: a finite-diversity channel whose coded soft-decision decoder produces exponentially decaying error probability with finite-length codes. The split of the transition probabilities in Proposition 1 and the algebraic-decoding condition in Theorem 1 are useful and appear sound. However, the central soft-decision claim is false: the proof of Theorem 2 relies on an invalid reduction to the all-zero codeword, and an explicit counterexample shows an MFD-deficient code with finite diversity. The examples and tables based on Theorem 2 therefore do not validate the claimed theory.","major_comments":[{"comment":"The sufficiency proof of Theorem 2 is invalid because it assumes, 'without loss of generality and thanks to the code linearity,' that the transmitted word c_A is the all-zero word. Code linearity translates the set of competing codewords, but it does not translate the Gray-label adjacency relation: a single-bin jump from a nonzero label lands on a label that is Hamming-adjacent to that nonzero label, not necessarily to the zero label. Concretely, take m=3 with the standard reflected Gray code and the binary [6,2,2] code C=span(111111,010010). The nonzero codewords are 111111, 010010, and 101101; none uses only the zero-neighboring labels {001,100}, so C is MFD deficient by Definition 4. Yet if Alice transmits c_A=111111 and Bob's measured positions both fall in bin 6, the competing codeword 101101 has higher APP in each coordinate. This event consists of two single-bin jumps and has probability Θ(γ^{-1}), giving finite diversity at most 1. This directly contradicts Theorem 2 statement 1.","section":"Section V-C, Theorem 2 sufficiency proof"},{"comment":"Even in the zero-word case, the proof does not establish that the cube of side Δ lies entirely in the correct decision region. The derivation of Eq. (51) equates APP(c_A) and APP(c'_A) only at the single point y_ℓ = x_ℓ + 1 + Δ. The proof then asserts that this cube is included in the decision region of c_A without proving monotonicity of the APP ratio over the whole cube. This missing monotonicity argument is load-bearing for the claimed O(e^{-Δ²γ/4}) pairwise-error bound and hence for the sufficiency of MFD deficiency.","section":"Theorem 2 proof, DMZ cube inclusion after Eq. (51)"},{"comment":"The necessity direction inherits the same translation problem. Lemma 6 constructs a finite-diversity lower-bound event using a full-MFD codeword whose labels are neighbors of zero, implicitly taking the transmitted word to be the all-zero word. For a nonzero transmitted word, the physical single-bin jumps are neighbors of the transmitted labels, not neighbors of zero. Thus the statement that full MFD implies finite diversity is not established for general transmissions. Both directions of the claimed equivalence in Theorem 2 are therefore unsupported.","section":"Lemma 6 necessary-condition proof"},{"comment":"The paper states that the results do not depend on the Gray-labeling version chosen. This is contradicted by Definition 3, which counts neighbors of the all-zero label; changing the Gray labeling changes the zero-neighboring set and can change the MFD status of a code. The RM(2,4) and RM(2,5) tables show different MFD distributions for different generator-matrix versions, illustrating the dependence. More fundamentally, a correct diversity condition must depend on the geometric adjacency of the transmitted labels, so the labeling-independence claim is not established.","section":"Section III-B, Gray-labeling independence claim"}],"minor_comments":[{"comment":"The phrases 'shocking result' and 'never encountered in the literature' are overstated and are not appropriate for a technical claim that is not established by the proofs.","section":"Abstract and Introduction"},{"comment":"The notation for exponential terms is inconsistent in a few places: for example, the text after Proposition 2 in [24] is quoted as O(e^{γ/4}), while the correct decaying form is O(e^{-γ/4}). Please harmonize the notation.","section":"Section IV, around Lemma 3 and Proposition 1"},{"comment":"The layout of the Gray-code tables is garbled and should be redrawn so that the bin number, Gray label, and photon position are clearly aligned.","section":"Table I and Table II"},{"comment":"The proof says 'the calculus details are not shown' for the exact integration leading to Eq. (48). Since Lemma 5 is used in the main theorem, the intermediate steps should be included or a reference provided.","section":"Lemma 5 proof, Eq. (48)"}],"recommendation":"reject","confidential_remarks":"The manuscript's core soft-decision theorem is demonstrably false as stated, and the flaw is not a local presentation issue: it invalidates the advertised contribution and the interpretations of the simulation tables. The authors may be able to repair the theory by defining MFD deficiency relative to every possible transmitted label pattern and by proving cube-inclusion monotonicity, but that is a substantial reworking beyond a minor revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the quick read. The paper's first half is solid: the TE-QKD transition asymptotics (single-bin jumps are Θ(γ^{-1/2}), multi-bin jumps are O(e^{-γ/4})) are derived carefully, and the algebraic decoding condition L ≤ t for infinite diversity (Theorem 1 and Corollary 1) follows cleanly and is likely correct. The MFD vocabulary is a useful way to organize the soft-decision problem, and the code examples are extensive.\n\nThe problem is Theorem 2, the soft-decision if-and-only-if. The sufficiency proof reduces to the all-zero codeword by linearity. That's legitimate in Hamming space, but the TE-QKD channel is not translation-invariant in the XOR sense. Gray-code adjacency depends on the actual label: a label that is one bit flip from a nonzero label is not necessarily one of the zero-neighboring labels. So MFD deficiency (no nonzero codeword built from zero-neighbors) does not rule out a competing codeword built from single-bin jumps away from the transmitted word.\n\nConcrete counterexample: take m=3 with the standard reflected Gray code, and C = span(111111, 010010) over F2. Its nonzero codewords are 111111, 010010, 101101; none of their 3-bit label components are neighbors of zero (001 or 100), so C is MFD deficient. Transmit 111111 (both photons in bin 5). If Bob's photons both land in bin 6, the word 101101 has larger APP in each coordinate, so the MAP decoder chooses 101101. Conditioned on Alice's photons being in bin 5, the joint event has probability Θ(γ^{-1}), giving finite diversity at most 1. That directly contradicts Theorem 2.\n\nThe same issue kills the paper's claim that the results are independent of the Gray-labeling version: the set of zero-neighboring labels changes with the labeling, and the real condition would have to involve the neighbors of every transmitted label pattern, not just the all-zero word. The reader's report flags the DMZ cube inclusion and the unproved uniformity of the APP approximations; those are real but secondary. The translation-invariance flaw is structural.\n\nSo: the algebraic part deserves a serious referee, the soft-decision part needs major repair or a substantial restriction. I'd send it out because the channel analysis and Theorem 1 are worth engaging, but the current version should not be accepted.","headline":"The algebraic-diversity condition is a genuine contribution, but the soft-decision theorem is not: the WLOG zero-codeword step ignores that Gray adjacency is not XOR-translation-invariant.","tokens_in":40042,"tokens_out":5881,"would_cite":false,"duration_ms":48205,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94B05","94B35","81P94"],"pacs":["03.67.Dd"],"model":"deepseek-v4-flash","headline":"This paper proves that coded time-entanglement QKD reconciliation can reach infinite diversity order — error probability decaying exponentially in signal-to-noise ratio — from a channel whose uncoded diversity is only 1/2.","keywords":["time-entanglement quantum key distribution","information reconciliation","diversity order","infinite diversity","maximal finite diversity","soft-decision decoding","bounded-distance decoding","Gray labeling"],"falsifier":"Take the [6,3,3] shortened Hamming code on N = 8 bins (m = 3) and measure the soft-decision word error rate at SNR γ = 20, 25, and 30 dB: if the log-log slope keeps steepening rather than approaching a constant, the infinite-diversity claim is supported. A sharper test uses the same code with m = 2, where the paper's Table III shows an MFD(3,3) term, so the theory predicts finite diversity 3/2 and the slope should flatten near 1.5.","tokens_in":38919,"feed_emoji":"🔑","tokens_out":5741,"duration_ms":53841,"temperature":0.7,"pith_summary":"This paper establishes exactly when an error-correcting code used for information reconciliation in time-entanglement quantum key distribution (TE-QKD) makes the reconciliation error probability decay exponentially with signal-to-noise ratio, an infinite diversity order, even though the uncoded channel has diversity only 1/2 and the code is short. For hard-decision algebraic decoding, the condition is that the number of photons carrying one codeword does not exceed the decoder's correction radius. For soft-decision decoding, the condition is a new code property: no nonzero codeword can be formed entirely from single-bin jumps, which the paper calls MFD deficiency. A sympathetic reader should care because this converts a polynomial error decay into an exponential one, and soft decoding achieves this at twice the coding rate allowed by algebraic decoding. The paper argues this behavior has no counterpart in classical fading channels, where decoding can only multiply a finite diversity order by a finite factor.","feed_headline":"Short codes make QKD reconciliation errors vanish exponentially","feed_subtitle":"Proven conditions on codeword length and correction radius turn polynomial error decay into exponential decay in time-entanglement QKD.","key_machinery":"The load-bearing mechanism is the separation of TE-QKD detector errors into two asymptotic classes: single-bin jumps with probability Θ(γ^(−1/2)) and multi-bin jumps with probability O(e^(−γ/4)). A code attains infinite diversity when its decoder absorbs all configurations of single-bin jumps, leaving only exponentially rare multi-bin events. For soft-decision decoding, the central object is the maximal finite diversity (MFD) property, which records whether a non-zero codeword can be built entirely from labels that are single-bin neighbors of zero. The proof of Theorem 2 uses a demilitarized-zone (DMZ) cube of side Δ, with Δ ≥ 1/(1+√L), placed inside the correct codeword's decision region; leaving that cube costs a factor O(e^(−$Δ^{2}$ γ/4)), which is exponentially small. The same DMZ argument also shows that a full-MFD code has finite diversity, falling back to ω/2 for the offending weight ω.","core_discovery":"The central claim is that infinite diversity in TE-QKD reconciliation occurs precisely when the code eliminates every dominant error pattern composed of single-bin jumps. Proposition 1 splits the uncoded channel errors into two regimes: a jump to a neighboring bin has probability Θ(γ^(−1/2)), while any jump of two or more bins has probability O(e^(−γ/4)). Theorem 1 proves that a bounded-distance algebraic decoder with correction radius t achieves infinite diversity if and only if L = n log2(q)/m ≤ t, where L is the number of photons per codeword. Theorem 2 proves that soft-decision decoding achieves infinite diversity if and only if the code is MFD deficient, meaning no non-zero codeword has a binary image whose Hamming weight ω is realized entirely by ω neighboring labels of zero; a sufficient condition is L < d_Hmin(C_b). The paper also derives Singleton-type rate bounds: algebraic decoding requires R_c ≤ 1 − 2 log2(q)/m, while soft decoding only requires R_c ≤ 1 − log2(q)/m. Examples with Golay, Reed–Solomon, BCH, and Reed–Muller codes confirm the predicted finite-to-infinite transitions.","pith_inferences":["Editorial extension: If the infinite-diversity condition holds at modest SNR, TE-QKD implementations could tolerate larger detector jitter or use fewer bins per frame while still keeping reconciliation errors negligible; the paper states the exponential decay but does not quantify this operational trade-off.","Editorial extension: Because the MFD property depends on which labels are neighbors of zero, the claimed independence of the Gray-labeling version could be tested directly by comparing a standard and a centered Gray code on the same code; if the MFD classification changes, the labeling independence claim would need qualification.","Editorial extension: The same two-scale error structure—polynomial near-boundary events and exponential far-boundary events—may appear in other timing or quantization channels, suggesting that infinite diversity could be engineered wherever a decoder can absorb all low-order boundary-crossing error patterns.","Editorial extension: A practical verification would measure the frame error rate slope for an MFD-deficient code at multiple SNR values; if the slope continues to increase without flattening, the infinite-diversity prediction is confirmed, whereas saturation at a finite slope would indicate a missing dominant error event."],"forward_implications":["If the condition L ≤ t holds, bounded-distance algebraic reconciliation has error probability O(e^(−cγ)) instead of the polynomial decay typical of coded fading channels.","If the code is MFD deficient, soft-decision reconciliation has infinite diversity; conversely, a full-MFD code forces finite diversity ω/2 for some weight ω ≥ d_Hmin(C_b).","Soft reconciliation attains infinite diversity at rates up to R_c ≤ 1 − log2(q)/m, while algebraic reconciliation is limited to R_c ≤ 1 − 2 log2(q)/m, so the soft-decision rate penalty is half that of algebraic decoding.","Short codes such as the [24,12,8] Golay code and [6,3,3] shortened Hamming code reach infinite diversity under soft decoding at modest frame sizes (m = 3 or more coded bits per photon), while algebraic decoding needs larger frames.","The results imply a sudden, qualitative improvement in reconciliation error rate once the code parameters cross the threshold, an effect not observed in classical Rayleigh or Nakagami fading channels."],"supporting_citations":[{"why":"Supplies the TE-QKD channel model, the transition probabilities, and the high-SNR APP expressions in (13)–(14) that the proofs use as their starting point.","marker":"[24]"},{"why":"Provides the coding-theoretic background, including the Singleton bound and weight enumerators, used to convert the infinite-diversity conditions into rate bounds in Corollaries 2 and 3.","marker":"[35]"},{"why":"Defines bounded-distance algebraic decoding and the correction-radius t formalism on which Theorem 1 and Corollary 1 rest.","marker":"[39]"},{"why":"Establishes the classical fading-channel diversity framework that the TE-QKD results are contrasted against, including the finite-factor diversity improvement after decoding.","marker":"[23]"},{"why":"Provides the photon-efficient time-energy entanglement encoding with high-dimensional bins that motivates modeling raw key symbols as pulse-position-modulation-like bin indices.","marker":"[26]"},{"why":"Provides the Slepian-table coset decoding construction used to illustrate the algebraic decoder in the paper's examples.","marker":"[38]"}],"fun_headline_variants":["Infinite diversity from finite QKD codes, proven","QKD codes: finite length, infinite error decay","Short codes unlock infinite diversity in TE-QKD","Single-bin jumps: the key to infinite QKD diversity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on the high-SNR approximations to the per-photon likelihoods being accurate enough that a small cube of Bob's soft measurements, of fixed width around the bin boundary, lies entirely inside the correct codeword's decision region; if that cube is not fully inside, the exponential error bound does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Infinite diversity from finite QKD codes, proven","QKD codes: finite length, infinite error decay","Short codes unlock infinite diversity in TE-QKD","Single-bin jumps: the key to infinite QKD diversity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000243,"raw_usage":{"total_tokens":1569,"prompt_tokens":1029,"completion_tokens":540,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":645,"completion_tokens_details":{"reasoning_tokens":477}},"tokens_in":645,"tokens_out":540,"duration_ms":5447,"temperature":1.0,"reasoning_tokens":477,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T13:12:44.286008+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the [6,3,3] shortened Hamming code on N = 8 bins (m = 3) and measure the soft-decision word error rate at SNR γ = 20, 25, and 30 dB: if the log-log slope keeps steepening rather than approaching a constant, the infinite-diversity claim is supported. A sharper test uses the same code with m = 2, where the paper's Table III shows an MFD(3,3) term, so the theory predicts finite diversity 3/2 and the slope should flatten near 1.5.","supporting_citations":[{"cited_title":"Time-Entanglement QKD: Secret Key Rates and Information Reconciliation Coding,","cited_arxiv_id":null,"evidence_quote":"Supplies the TE-QKD channel model, the transition probabilities, and the high-SNR APP expressions in (13)–(14) that the proofs use as their starting point."},{"cited_title":"MacWilliams and N.J.A","cited_arxiv_id":null,"evidence_quote":"Provides the coding-theoretic background, including the Singleton bound and weight enumerators, used to convert the infinite-diversity conditions into rate bounds in Corollaries 2 and 3."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines bounded-distance algebraic decoding and the correction-radius t formalism on which Theorem 1 and Corollary 1 rest."},{"cited_title":"Photon-efficient quantum key distribution using time–energy entanglement with high-dimensional encoding,","cited_arxiv_id":null,"evidence_quote":"Provides the photon-efficient time-energy entanglement encoding with high-dimensional bins that motivates modeling raw key symbols as pulse-position-modulation-like bin indices."},{"cited_title":"A class of binary signaling alphabets,","cited_arxiv_id":null,"evidence_quote":"Provides the Slepian-table coset decoding construction used to illustrate the algebraic decoder in the paper's examples."}],"review_version":1}