{"id":"fb03ec08-9025-49f6-83f8-d39e9af114fe","arxiv_id":"2411.16988","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Real-valued discrete quaternionic Gabor frames are characterized by pointwise matrix inequalities; within Parseval frames, orthonormal bases occur exactly when N^2 = LM^2.","lead":"This math paper proves conditions for quaternion-valued Gabor systems on the integer lattice to be frames, orthonormal bases, dual frames, and stable systems. The clean criteria apply to real-valued windows and reduce the frame question to pointwise matrix inequalities.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2 and Lemma 6 drop quaternionic conjugates, so their exact identities are false for H-valued test functions; Theorem 1 survives, but Theorem 3 and Theorem 7 rely on these lemmas and need correction or a real-window restriction.","rationale":"The reader's weakest assumption correctly identifies the real-valued window hypothesis as central and notes the boundedness of the infinite matrices. In my reading, the boundedness concern is not fatal: the operator inequality in Theorem 1 itself entails boundedness, and the frame-to-matrix direction produces the missing bound from the frame inequality. Theorem 1's proof is built on Lemma 5 and the direct phase-collapse argument, not on Lemma 2, and it appears mathematically sound for real windows. The more serious issue is the systematic omission of quaternionic conjugation in Lemma 2 and Lemma 6: as stated, those identities are false for non-real H-valued test vectors. This affects the Parseval and duality characterizations, including Theorem 7, whose statement additionally omits the real-window hypothesis required by Lemma 6. These problems are repairable, and the central frame characterization stands, so I do not move the verdict away from CONDITIONAL; I would keep the paper conditional pending correction of these algebraic identities and restriction of Theorem 7 to real windows.","tokens_in":20816,"tokens_out":54066,"duration_ms":522668,"concrete_test":"Evaluate both sides of Lemma 2 for M=2, N=1, with a real window g=delta_0+delta_2 supported in one coordinate and h(0)=1, h(2)=i, all other h(k)=0. The displayed F2 term gives M^2 h(0)h(2)=4i, making the claimed identity conflict with the real-valued left-hand side, while the corrected term M^2 h(0)overline{h(2)}=-4i combines with the p=-1 term to give a real contribution. Repeating this check for Lemma 6 with f(0)=1, phi(2)=i isolates the same missing conjugation. If the corrected identities still imply Theorems 2, 3, and 8, and if Theorem 7 is restricted to real windows, the conditional verdict can stand; otherwise those results should be weakened or removed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 2, Lemma 2 claims that for h in l0(Z^2,H), the Gabor analysis sum equals F1(h)+F2(h) with F2(h)=M^2 sum_k sum_{p neq 0} h(k) h(k+pM) (sum_l M_g(k) M_g^t(k))_{0,p}. Expanding |sum_k overline{Eg(k)} h(k)|^2 correctly gives the off-diagonal factor h(k) overline{h(k+pM)}, not h(k) h(k+pM). For non-real h these differ, so the stated identity is false. The same loss of conjugation appears in Lemma 6 in Section 4, where the cross term is written as f(k) phi(k+pM) instead of overline{f(k)} phi(k+pM). Because g_l and h_l are real, these errors do not affect arguments that only use real-valued test vectors, and Theorem 1's proof goes through Lemma 5 rather than Lemma 2. However, Theorem 3 (Parseval characterization), Theorem 2, and Theorem 8 use Lemma 2, and Theorem 7 is stated for general quaternionic windows even though its proof invokes Lemma 6, which explicitly requires real-valued windows. Thus the central Theorem 1 is not sunk, but the paper's Parseval and duality claims are not established as written. The real-window hypothesis is essential, and the conjugation error must be repaired before the lemma-based results can be accepted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies discrete multi-window Gabor systems in the quaternionic Hilbert space ℓ²(Z×Z,H), with modulation by left-multiplication by e^{2πi m₁/M} and right-multiplication by e^{2πj m₂/M}. The central claim is Theorem 1: for real-valued windows g_l, the system is a frame with bounds A≤B if and only if the matrix-valued function ∑_l M_{g_l}(k)M_{g_l}^t(k) satisfies the pointwise operator inequality A/M² I ≼ … ≼ B/M² I on an N-periodic set of k. Subsequent sections characterize Parseval frames, orthonormal bases, dual frames, and stability under perturbations, again in terms of the matrix-valued functions M_g.","tokens_in":94,"tokens_out":14499,"duration_ms":199006,"significance":"If the main theorem is correct, it gives a transparent, checkable characterization of quaternionic Gabor frames for real windows, extending a body of complex Gabor theory to a noncommutative setting where the usual commutativity arguments fail. The paper is self-contained in its reductions, uses no fitted parameters, and provides constructive examples for Parseval frames and orthonormal bases. However, the significance is currently limited by several local but load-bearing gaps: an incorrect identity in Lemma 2, a missing real-window hypothesis in Theorem 7, and an unstated boundedness condition for the infinite matrices in Theorem 1. These are repairable, but the results are not established as written.","major_comments":[{"comment":"Lemma 2 is false as stated for H-valued h. In the expansion of |∑_k ⟨E^m_M T_{nN} g_l, h⟩|², the off-diagonal term must contain a quaternionic conjugate, e.g. ̅h(k)h(k+pM) (with the paper's inner-product convention), not h(k)h(k+pM). The real-valuedness of g_l makes the window commute with the exponential factors, but it does not remove the conjugate from h. This invalidates the stated identity in Lemma 2 and consequently the proofs that rely on it: Theorem 2, Theorem 3, Proposition 4, and, via Lemma 6, Theorem 7. Most of the arguments can be repaired by inserting the missing conjugate — for the Bessel bound in Theorem 2 only the modulus is used, and in Theorem 3 the two-point test still gives 1 — but the manuscript must present correct identities before these results can be accepted. I note that Theorem 1 itself proceeds through Lemma 5 rather than Lemma 2, so its proof is not invalidated by this particular error.","section":"Section 2, Lemma 2 and F2(h)"},{"comment":"Theorem 7 is stated for arbitrary g_l,h_l ∈ ℓ²(Z×Z,H), but its proof invokes Lemma 6, which is proved only under the hypothesis g_l,h_l ∈ ℓ²(Z×Z,R). The real-valued hypothesis is essential for the commutativity step that collapses the double modulation sum into M_{g_l}(k)M_{h_l}^t(k). The theorem must either be restricted to real-valued windows or be given a genuinely new proof for H-valued windows; as written, the duality characterization for quaternionic windows is unsupported.","section":"Section 4, Theorem 7 and Lemma 6"},{"comment":"Condition 2 of Theorem 1 asserts an operator inequality for the bi-infinite matrices ∑_l M_{g_l}(k)M_{g_l}^t(k) acting on ℓ²(Z²). For an arbitrary l² window, M_g(k) is not automatically a bounded operator on ℓ²(Z²); for M=N=1 it is a convolution operator with an l² kernel, which can be unbounded. The theorem needs an explicit boundedness hypothesis on these matrices, or the inequality should be formulated as a uniform quadratic-form estimate on finitely supported sequences and then extended by density. This issue affects Theorem 1 and all later matrix-characterization theorems (Theorems 2, 3, 5, 7, and 8), since the matrix inequality is the object being verified.","section":"Section 2, Theorem 1 and the definition of M_g in Section 1"},{"comment":"The proof of Theorem 4 contains a logical gap: after deriving ∑_l ‖g_l‖² = L and observing that ‖g_l‖ ≤ 1 for every l, the paper concludes that the system is an orthonormal basis. Unit-norm vectors in a Parseval frame do form an orthonormal basis, but this requires proof; one needs, for example, to show that the Gram matrix is idempotent and has unit diagonal, forcing all off-diagonal entries to vanish. Without this step, Theorem 4, and hence Lemma 4 and Theorem 6 which depend on it, are not fully established.","section":"Section 3, Theorem 4 proof"}],"minor_comments":[{"comment":"In the proof of Lemma 2, the line 'the fact that p2 = k2 + 2M' appears to be a typo; it should read p2 = k2 + q2M (or p2 = k2 + qM in the second coordinate).","section":"Section 2, Lemma 2 proof"},{"comment":"The display contains the broken LaTeX string '/greaterorequalslant' and the word 'Combinig' should be 'Combining'; also, in the lower-bound half of the proof, the variable f is used where h was introduced, which should be corrected for readability.","section":"Section 2, Theorem 2"},{"comment":"The example in Remark 1 is hard to follow because the notation q is used for a modulation index, while the vector (1/M,1/M) is also written as a quaternionic exponential argument; the computation should be rewritten with clearer indexing.","section":"Section 1, Remark 1"},{"comment":"The proof of Theorem 1 invokes Lemma 5 before that lemma is stated in Section 4; the author should either state Lemma 5 earlier or include an explicit forward reference so the reader can verify the identity in equation (2).","section":"Section 2, Theorem 1 proof"}],"recommendation":"major_revision","confidential_remarks":"The paper's central Theorem 1 is plausible and likely correct once the boundedness of the matrices is made explicit, and the other issues appear to be localized and repairable. I see no concerns about novelty or citation behavior; the main risks are the false conjugation identity and the missing hypothesis in Theorem 7. A careful revision should be sufficient."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is a proof-based extension of multi-window Gabor frame theory to quaternionic-valued sequences on Z^2, with the main positive result restricted to real-valued windows. Theorem 1, the centerpiece, says that for real windows the Gabor system is a frame with bounds A and B exactly when the pointwise, N-periodic matrix inequality A/M^2 I <= sum_l M_g_l(k) M_g_l^t(k) <= B/M^2 I holds. That is a genuine result, and the proof via periodizing over the modulation lattice is standard and plausible. The multi-window real-valued quaternionic combination appears to be new; the explicit Parseval-frame and orthonormal-basis constructions in Examples 1 and 2 are concrete and checkable. The density condition N^2 <= L M^2 and the perfect-square window count for ONBs also match the expected discrete Gabor pattern.\n\nNow the soft spots, in proportion. Lemma 2 and Lemma 6 drop quaternionic conjugates in the cross terms: the off-diagonal term should be h(k) overline{h(k+pM)}, not h(k)h(k+pM), and similarly in the duality identity. For H-valued test functions the written identities are false. The good news is that Theorem 1 does not use Lemma 2, so its proof survives. The bad news is that Theorem 2 and Theorem 3 rely on Lemma 2, and Theorem 7 is stated for general quaternionic windows while its proof invokes Lemma 6, which explicitly requires real-valued windows. The Parseval-frame and duality characterizations therefore need either a repaired conjugation argument or an explicit real-window hypothesis. This is a load-bearing gap, but it looks repairable. Separately, the bi-infinite matrices in Theorem 1 are treated as bounded operators without a stated condition; arbitrary l^2 windows need not give bounded matrix multiplication. The body never cites or compares with [17] and [18], which appear to cover single-window quaternionic Gabor frames, so the novelty boundary is not clearly drawn. Typos, malformed symbols, and forward references (Theorem 1 uses Lemma 5 before it is stated) are numerous but secondary.\n\nVerdict: this paper deserves a serious referee because the central characterization is likely correct and useful to people working on quaternionic frame theory and quaternionic signal processing. I would not cite it in its current form, but I would read a corrected version. Send it out, expect major revision, and ask the author to fix the conjugation, add the missing hypotheses, and compare explicitly with [17] and [18].","headline":"A useful real-window multi-window quaternionic Gabor frame characterization, but the paper ships lemmas with dropped conjugates and a duality theorem stated too broadly; Theorem 1 survives, the rest needs repair.","tokens_in":21623,"tokens_out":2355,"would_cite":false,"duration_ms":23599,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42C15","42C40","51F30"],"pacs":[],"model":"deepseek-v4-flash","headline":"For real-valued windows, a discrete quaternionic Gabor system is a frame exactly when a finite list of pointwise matrix inequalities holds.","keywords":["quaternionic Gabor frames","multi-window Gabor systems","quaternionic Hilbert spaces","frame characterization","Parseval frames","Gabor orthonormal bases","duality of Gabor frames","stability of Gabor frames"],"falsifier":"Take the pure quaternionic window g = iδ_0 with M=N=1: G(δ_0,1,1,1) is an orthonormal basis with frame bounds 1, yet M_g(0)M_g^t(0) = −I, so the inequality of Theorem 1 fails, showing the real-valued hypothesis is essential. For the real-valued theorem itself, compute numerically the extremal eigenvalues of Σ_l M_{g_l}(k)M_{g_l}^t(k) over k ∈ N_N² for a randomly chosen real window and compare them with the frame bounds of G(g,L,M,N) obtained by a direct finite-lattice computation; any mismatch would refute the equivalence.","tokens_in":20426,"feed_emoji":"🧮","tokens_out":8600,"duration_ms":77400,"temperature":0.7,"pith_summary":"This paper proves that, when the windows are real-valued, the frame property of a discrete quaternionic multi-window Gabor system on ℓ²(ℤ², ℍ) is completely determined by finite matrix data. For a system G(g,L,M,N) with windows g_l, the frame bounds A ≤ B are equivalent to the pointwise operator inequalities A/M² I ≤ Σ_l M_{g_l}(k) M_{g_l}^t(k) ≤ B/M² I holding for every k in the N-periodic residue classes. The same matrix machinery characterizes Parseval frames, orthonormal bases, duality, and stability. This matters because it converts an infinite-dimensional frame question in a noncommutative setting into a check on finitely many bi-infinite matrices.","feed_headline":"A pointwise matrix inequality decides quaternionic Gabor frames","feed_subtitle":"Frame bounds come from N-periodic matrix products M_g(k)M_g^t(k) when windows are real.","key_machinery":"The matrix-valued function M_h : ℤ² → M(ℍ) associated with a window h, whose (p,n)-entry is h(k + pM − nN). This function converts translation-modulation sums into products of bi-infinite matrices; for real-valued windows the products M_g(k)M_g^t(k) are N-periodic in k, so the entire frame analysis reduces to checking operator inequalities on the N² residue classes modulo N.","core_discovery":"The central discovery is that real-valued windows commute with the quaternionic modulation factors, which collapses the frame operator into a clean two-part decomposition and lets the frame bounds be read off from the spectra of the matrices S(k) := Σ_l M_{g_l}(k) M_{g_l}^t(k). Theorem 1 states that G(g,L,M,N) is a frame with bounds A ≤ B if and only if A/M² I ≤ S(k) ≤ B/M² I for all k ∈ ℤ² (equivalently, k ∈ N_N²). Theorem 3 sharpens this for Parseval frames: S(k) must equal (1/M²)I on the diagonal and vanish off it. Within the Parseval class, Theorem 4 shows the system is an orthonormal basis precisely when N² = L M², which forces L to be a perfect square; Theorem 6 proves such bases exist exactly under that parameter condition and constructs them from indicator windows. Theorem 7 gives the dual-frame analogue S_{g,h}(k) = (1/M²)I, and Theorem 8 shows that sufficiently small perturbations of the window matrices preserve the frame property with quantitative bounds.","pith_inferences":["The real-valued hypothesis is likely removable only if one tracks noncommutativity more carefully; a two-sided modulation construction might allow complex or quaternionic windows to satisfy an analogous but more complicated matrix inequality.","The reduction to N-periodic matrix inequalities suggests a direct algorithmic route: construct quaternionic Gabor frames by solving finite-dimensional matrix inequalities on the residue classes, which could be useful in quaternionic signal processing.","The parameter condition N² = L M² is a density-type statement that mirrors the classical density theorem for Gabor systems, hinting that a full quaternionic density theory may be within reach.","The same matrix-based approach could extend to other noncommutative coefficient algebras, such as Clifford algebras, whenever the window algebra commutes with the modulation factors."],"forward_implications":["Frame verification for real-window quaternionic Gabor systems becomes a finite computation: check the operator inequalities on the N² residue classes rather than analyzing the whole system.","The Parseval characterization gives an explicit construction of Parseval frames from indicator windows whenever the parameter condition N² ≤ L M² holds.","Orthonormal Gabor bases exist exactly when N² = L M², and the number of windows L must be a perfect square, so the parameter pairs (L,M,N) that admit such bases are completely classified.","Duality between two real-window Gabor systems is characterized by the same matrix condition with cross products, extending the classical discrete-periodic duality theory to the quaternionic setting.","The stability theorem supplies quantitative frame bounds for perturbed systems, showing that small changes in the window matrices cannot destroy the frame property."],"supporting_citations":[{"why":"Supplies the quaternionic frame operator facts (Proposition 1) and the perturbation lemma (Lemma 7) used in the stability section.","marker":"[15]"},{"why":"Earlier quaternionic Gabor frame characterization and density theorem that this paper extends to the multi-window, real-window setting.","marker":"[17]"},{"why":"Provides the duality framework for quaternionic Gabor frames that Theorem 7 refines.","marker":"[18]"},{"why":"Time-domain characterization of multi-window Gabor systems on discrete periodic sets, the prototype for the matrix characterization used here.","marker":"[13]"},{"why":"Source of the cross-Gramian duality condition adapted in Theorem 7.","marker":"[14]"}],"fun_headline_variants":["Real windows turn quaternionic frames into matrix checks","Matrix rule pins down quaternionic Gabor frames","Quaternionic Gabor frames collapse to matrix spectra","A matrix inequality diagnoses quaternionic Gabor frames"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The equivalence rests on the windows being real-valued so they commute with the quaternionic modulation factors, and on the implicit assumption that the bi-infinite matrices M_g(k) and their sums are bounded operators on ℓ²(ℤ²).","fun_headline_variants_meta":{"raw":{"variants":["Real windows turn quaternionic frames into matrix checks","Matrix rule pins down quaternionic Gabor frames","Quaternionic Gabor frames collapse to matrix spectra","A matrix inequality diagnoses quaternionic Gabor frames"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000197,"raw_usage":{"total_tokens":1424,"prompt_tokens":1062,"completion_tokens":362,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":678,"completion_tokens_details":{"reasoning_tokens":299}},"tokens_in":678,"tokens_out":362,"duration_ms":5137,"temperature":1.0,"reasoning_tokens":299,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:41:26.712312+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the pure quaternionic window g = iδ_0 with M=N=1: G(δ_0,1,1,1) is an orthonormal basis with frame bounds 1, yet M_g(0)M_g^t(0) = −I, so the inequality of Theorem 1 fails, showing the real-valued hypothesis is essential. For the real-valued theorem itself, compute numerically the extremal eigenvalues of Σ_l M_{g_l}(k)M_{g_l}^t(k) over k ∈ N_N² for a randomly chosen real window and compare them with the frame bounds of G(g,L,M,N) obtained by a direct finite-lattice computation; any mismatch would refute the equivalence.","supporting_citations":[{"cited_title":"Sharma and S","cited_arxiv_id":null,"evidence_quote":"Supplies the quaternionic frame operator facts (Proposition 1) and the perturbation lemma (Lemma 7) used in the stability section."},{"cited_title":"Banach J","cited_arxiv_id":null,"evidence_quote":"Earlier quaternionic Gabor frame characterization and density theorem that this paper extends to the multi-window, real-window setting."},{"cited_title":"Quaternionic Subspace Gabor Frames and Th eir Duals","cited_arxiv_id":null,"evidence_quote":"Provides the duality framework for quaternionic Gabor frames that Theorem 7 refines."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Time-domain characterization of multi-window Gabor systems on discrete periodic sets, the prototype for the matrix characterization used here."},{"cited_title":"Lian and Y.-Z","cited_arxiv_id":null,"evidence_quote":"Source of the cross-Gramian duality condition adapted in Theorem 7."}],"review_version":1}