{"id":"22988735-ea3a-4620-b446-8afa1eebfdd1","arxiv_id":"2506.08478","paper_version":3,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Phase retrievable weaving fusion frames are characterized by a kernel condition on a phase lift operator, while the claimed probabilistic erasure bound is vacuous due to an impossible normalization.","lead":"This paper defines phase retrievable weaving fusion frames, pairs of subspace collections whose interleavings recover a vector up to a global phase from magnitude-only measurements. It gives a kernel characterization of this property and attempts an application to random erasure, but the erasure result rests on an impossible assumption.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.4's proof relies on Theorem 2.6, which is false as stated: in R^2, T=diag(1,2) lies in S_{1,1} but cannot equal f⊗f−g⊗g, so the central characterization is unproven.","rationale":"Reading in good faith, the intended goal is a finite-dimensional phase-lift criterion for weaving fusion frames, and the shape of Theorem 3.4 is standard: a quadratic phase-retrieval condition becomes a linear kernel condition on a lifted operator space. The examples and the unitary-invariance proposition are consistent with that goal. The difficulty is that the specific operator class S_{1,1} used in the statement is the wrong one. The cited Theorem 2.6 is stated with S_{1,1} = {dim R(T) ≤ 2}, but the representation T = f⊗f − g⊗g forces det T ≤ 0 in dimension two (and, more generally, at most one positive eigenvalue). Positive-definite rank-two operators are therefore counterexamples to the cited theorem. Since Theorem 3.4's first implication uses that representation as its only bridge from the kernel condition to a pair f,g, the central equivalence is unsupported. I agree with the reader that the paper should be rejected: Theorem 3.6 also has a homogeneity failure, and Theorem 4.1 assumes an impossible normalization ||P f|| = √n for all f with nontrivial projections. My partial disagreement is that the most load-bearing defect is not the application section; it is the foundation of the main theorem. A corrected version that restricts S_{1,1} to differences of rank-one positive operators, and rechecks the theorems, could still be viable.","tokens_in":12893,"tokens_out":18558,"duration_ms":213801,"concrete_test":"Perform the two-dimensional determinant check. Take T = diag(1,2) in R^2; by the paper's definition of S_{1,1}, T is in S_{1,1}. Solve T = uu^T − vv^T: the determinant identity gives det(T) = −det([u v])^2 ≤ 0, while det(T)=2>0, so no solution exists. This disproves Theorem 2.6. To settle whether Theorem 3.4 survives, one must then either (a) redefine S_{1,1} as {u u^T − v v^T} and prove the equivalence for that class, or (b) prove that a positive-semidefinite rank-two operator can never lie in Ker Fσ for phase-retrievable weaving fusion frames; if (b) fails, exhibit such an operator in an example and the theorem is false.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is Theorem 3.4, and its proof depends on Theorem 2.6(2), which asserts that every T ∈ S_{1,1}(H_n) with dim R(T) ≤ 2 is a difference T1 − T2 of two rank-one positive operators. That assertion is false over R. For H = R^2, T = diag(1,2) satisfies dim R(T) = 2, so T ∈ S_{1,1} by the paper's definition. Yet det(T) = 2 > 0, whereas det(f⊗f − g⊗g) = det([f g]) det(diag(1,−1)) det([f g]^T) = −det([f g])^2 ≤ 0. Hence no f,g ∈ R^2 can represent T. The proof of Theorem 3.4 (1⇒2) invokes Theorem 2.6 to write an arbitrary kernel element T as f⊗f − g⊗g; this step is unjustified for positive-definite rank-two T. The paper gives no separate argument excluding such T from Ker Fσ when the frames are phase retrievable. Thus the kernel characterization, the central mathematical assertion, is not proven as written. The erasure-section problem identified by the reader is real but peripheral; this gap is internal to the main theorem.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript defines phase retrievable weaving fusion frames as pairs of fusion frames for which every interleaved measurement map γ_σ is injective on the projective quotient. Its main theoretical result, Theorem 3.4, claims that this property is equivalent to a kernel condition on the associated weaving phase lift operator F_σ intersected with the space S_{1,1}(H_n) of symmetric operators of range dimension at most two. Theorem 3.6 claims an equivalent lower-bound inequality involving the symmetrized rank-one operators [f,g]; Proposition 3.7 asserts unitary invariance; and Section 4 presents a probabilistic erasure estimate. The paper also includes two examples, a Python complement-property check, and several remarks connecting the results to quantum information and distributed processing.","tokens_in":13179,"tokens_out":5439,"duration_ms":70828,"significance":"If Theorem 3.4 were correct, it would provide a compact algebraic characterization of phase retrievability for weaving fusion frames and would be a natural extension of Balan's phase-lift framework. The erasure result, if valid, would give a quantitative robustness statement for weaving fusion frames. However, the soundness defects described below affect the central characterization, the stated equivalent inequality, and the application result. The manuscript has not established its main claims, so its significance as a contribution is currently not realized. The paper does contain some useful definitions and a plausible example in R^2, and the idea of studying phase retrieval in the weaving fusion frame setting is reasonable.","major_comments":[{"comment":"Theorem 2.6(2) is false over the real field, and the proof of Theorem 3.4 depends on it. For H=R^2, the operator T=diag(1,2) belongs to S_{1,1}(H) by the paper's definition, since dim R(T)=2. But det(f⊗f−g⊗g) = −det([f g])^2 ≤ 0 for every f,g∈R^2, so this positive-definite T cannot be written as f⊗f−g⊗g. Thus the step in the proof of Theorem 3.4, implication (1⇒2), in which an arbitrary kernel element T is represented as f⊗f−g⊗g via Theorem 2.6, is unjustified. The authors need either to prove that positive-semidefinite rank-two operators cannot lie in Ker F_σ under phase retrievability, or to replace Theorem 2.6 by a decomposition valid on the relevant subset of S_{1,1}(H_n).","section":"§2, Theorem 2.6(2)"},{"comment":"Theorem 3.6 is false as stated because of scaling. Replacing f,g by t f,t g with t>0 scales the left side of the claimed inequality by t^4, while ||[f,g]||_1 scales by t^2; hence no fixed constant α>0 can satisfy the inequality for all f,g. The proof itself has the same problem: it defines α as a minimum over the unit trace sphere, but then applies the resulting bound to [f,g]/||[f,g]||_1 and multiplies by ||[f,g]||_1^2, while the numerator already contains squared terms. Moreover, implication (2⇒1) is not proved; the text only says it follows by a similar approach. The statement and proof need a consistent normalization, for example a bound of the form ||F_σ([f,g])||^2 ≥ α ||[f,g]||_1^2 with α depending on the frames, not on f,g.","section":"§3, Theorem 3.6"},{"comment":"The hypothesis of Theorem 4.1 is unsatisfiable for n>1. It assumes that for every f∈R^n and every n, ||P_{V_n}f||=√n=||P_{W_n}f||. For an orthogonal projection P, the function f↦||P f|| is constant on all of R^n only when P is the zero operator or the identity, neither of which gives the value √n for every f. Therefore the theorem's assumption can never hold, and the probabilistic erasure application is vacuous. The definitions of δ_n and tilde δ_n/p_n in the proof are also not spelled out, and the displayed expression for \\hat f switches between operators and vectors, so the derivation cannot be checked as written.","section":"§4, Theorem 4.1"},{"comment":"The verification of Example 3.5 is not complete. The text asserts that the frames are phase retrievable and that Ker F_σ(T) ∩ S_{1,1}(R^3)={0}, but it only computes the kernel explicitly for two subsets σ, and even for those cases the conclusion is summarized as matrices that are 'either invertible or zero' rather than as a full rank/range-dimensionality argument. Since the example is offered as validation of Theorem 3.4, all subsets σ should be checked or an algorithmic verification should be provided.","section":"§3, Example 3.5"}],"minor_comments":[{"comment":"The title contains spacing errors ('RETRIEV ABLE', 'WEA VING') and the abstract is generic; it would help to state the main theorems explicitly.","section":"Title and abstract"},{"comment":"The theorem is labeled 'Comlementary Property'; this should be 'Complementary Property'.","section":"§2, Theorem 2.5"},{"comment":"The injectivity verification for σ=∅ says that six of eight sign combinations are 'not well defined' but does not justify this claim. All mixed-sign cases should either be listed or handled by a uniform argument.","section":"§3, Example 3.2"},{"comment":"The example works over C^2 while most of the paper's linear-algebra statements are phrased for real symmetric operators; the authors should state explicitly which results are intended over C and which over R.","section":"§3, Example 3.3"},{"comment":"The notation δ_n is used for both a 'standard dirac delta function' and a Bernoulli-type random variable; the tilde quantities are undefined. The displayed derivation also confuses the operator E with the vectors PV_n f⊗S^{-1}PV_n f.","section":"§4"},{"comment":"The Python code is difficult to read because of line breaks and missing whitespace; it should be reformatted as a proper appendix or supplementary listing.","section":"Listing 1"}],"recommendation":"reject","confidential_remarks":"The manuscript's core theorem is unproven because Theorem 2.6(2) is false over R, and the application theorem rests on an unsatisfiable hypothesis. These are not merely presentation defects; they undermine the two advertised contributions. The topic is reasonable and the definitions are natural, so a substantially revised manuscript with corrected statements and a valid erasure model might be worth reconsidering, but the current version is not publishable in its present form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The paper introduces a definition of phase retrievable weaving fusion frames and states a kernel characterization (Theorem 3.4) in the style of Balan's S_{1,1} condition. That is a sensible question. But the proof is built on a false lemma, and the other two main results are broken, so the note is not refereeable as written.\n\nThe definition (3.1) is natural, and the kernel condition in Theorem 3.4 is the right kind of criterion to hope for. Proposition 3.7 (invariance under unitaries) is correct and easy. The examples do illustrate the definition, though the verification is hand-wavy and incomplete (Example 3.2 only checks two of eight subsets σ).\n\nThe load-bearing flaw is Theorem 2.6(2), cited from Balan. It claims every T ∈ S_{1,1}(H_n) is T1−T2 with T1,T2 rank-one positive. Over R^2, T=diag(1,2) has range dimension 2, but det(T)=2 while det(f⊗f−g⊗g)=−|det([f g])|^2 ≤0. So the lemma is false, and the proof of (1⇒2) in Theorem 3.4, which writes a kernel element as f⊗f−g⊗g, does not go through. The theorem might be true — one could try to rule out positive semidefinite kernel elements first — but the paper does not do that.\n\nTheorem 3.6 fails by homogeneity: replacing f,g by tf,tg scales the left side of the claimed inequality as t^4 and the right side as t^2, so no fixed α>0 can satisfy it for all f,g. Theorem 4.1 assumes ||P_{V_n} f||=√n for every f∈R^n; for a proper projection the norm of Pf varies with f, so the hypothesis is impossible and the probabilistic erasure application is vacuous. There are also minor issues: the examples are only partially checked, and the Python code is decorative rather than evidence.\n\nIf the kernel characterization can be given a correct proof, the paper could be a useful short contribution for frame theorists working on weaving fusion frames and phase retrieval. As it stands, the main theorem is unproven and the applications are empty.\n\nMy recommendation: desk reject. The flaws are elementary and central; a corrected version limited to Theorem 3.4 with a proper proof would be worth another look.","headline":"A natural definition and a plausible kernel condition, but the main proof relies on a false lemma and the application section is vacuous; not refereeable as written.","tokens_in":13686,"tokens_out":9625,"would_cite":false,"duration_ms":112987,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42C15","46A32","47A05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Two interwoven subspace frames are phase retrievable exactly when every weaving phase-lift operator avoids rank-two symmetric operators.","keywords":["phase retrieval","weaving fusion frames","fusion frames","phase lift operator","rank-two symmetric operators","probabilistic erasure","frame theory","complement property"],"falsifier":"Compute $F_\\sigma(x\\otimes x-y\\otimes y)$ for the pair in Example 3.3 with $\\sigma=\\{1,2\\}$, where $x=(1,i)$ and $y=(i,1)$ are non-parallel yet share all magnitude measurements; the theorem predicts the resulting nonzero rank-two operator lies in $\\operatorname{Ker} F_\\sigma$. A direct search over small $n$ and random subspace pairs that finds any nonzero rank-two operator in $\\operatorname{Ker} F_\\sigma$ for a genuinely phase-retrievable pair would disprove the iff, and exhaustively checking all $\\sigma$ would either confirm or reject the characterization computationally.","tokens_in":12672,"feed_emoji":"🧵","tokens_out":13682,"duration_ms":150697,"temperature":0.7,"pith_summary":"Two collections of closed subspaces of an $n$-dimensional Hilbert space are called weaving fusion frames when every way of interleaving them still produces a frame. The paper asks when such a pair is phase retrievable: when the lengths of the projections of a vector onto the chosen subspaces determine the vector up to a global unimodular factor, and do so for every interleaving. The central result, Theorem 3.4, says this happens exactly when the weaving phase lift operator $F_\\sigma$ has trivial intersection with $S_{1,1}(H_n)$, the space of symmetric operators of rank at most two, for every subset $\\sigma$. That converts a nonlinear injectivity question into a linear-algebraic kernel condition. The paper also derives a quantitative stability lower bound, proves invariance under a common unitary, and applies the setup to probabilistic erasure of measurements.","feed_headline":"Phase retrievability reduces to a rank-two kernel check","feed_subtitle":"For interwoven subspace frames, injectivity of every magnitude map is equivalent to a trivial rank-two kernel intersection.","key_machinery":"The central object is the weaving phase lift operator $F_\\sigma:\\operatorname{Sym}(H_n)\\to\\mathbb R^m$, defined by $(F_\\sigma T)_i=v_i^2\\operatorname{tr}(P_{V_i}T)$ for $i\\in\\sigma$ and $w_i^2\\operatorname{tr}(P_{W_i}T)$ for $i\\notin\\sigma$. It linearizes magnitude measurements because $\\operatorname{tr}(P_V(f\\otimes f))=\\|P_V f\\|^2$. The companion space $S_{1,1}(H_n)=\\{T\\in\\operatorname{Sym}(H_n):\\dim R(T)\\le 2\\}$ is exactly the collection of differences of two rank-one operators $f\\otimes f-g\\otimes g$; by the cited decomposition theorem [14], every such operator is of the form $\\frac12(f\\otimes g+g\\otimes f)$. The equivalence holds because two vectors produce identical magnitude data precisely when their rank-one difference lies in the kernel of $F_\\sigma$, so a trivial kernel intersection with $S_{1,1}(H_n)$ is exactly injectivity of $\\gamma_\\sigma$ on the quotient $H_n/\\!\\sim$.","core_discovery":"The paper's central assertion is Theorem 3.4: for two weaving fusion frames $\\{(V_i,v_i)\\}_{i=1}^m$ and $\\{(W_i,w_i)\\}_{i=1}^m$ in $H_n$, the pair is phase retrievable for every interleaving if and only if $\\operatorname{Ker} F_\\sigma \\cap S_{1,1}(H_n)=\\{0\\}$ for every $\\sigma\\subseteq\\{1,\\dots,m\\}$. Here $F_\\sigma$ is the weaving phase lift operator and $S_{1,1}(H_n)$ is the set of symmetric operators whose range has dimension at most two. In the author's framing, the nonlinear problem of deciding whether magnitude measurements distinguish vectors up to phase collapses to a linear problem: checking whether any rank-two symmetric operator is annihilated by all selected measurement traces. The paper proves this equivalence by representing every rank-two symmetric operator as a difference of two rank-one operators, $T=f\\otimes f-g\\otimes g$, and showing that equality of the magnitude data $\\gamma_\\sigma(f)=\\gamma_\\sigma(g)$ is exactly the condition $F_\\sigma(f\\otimes f-g\\otimes g)=0$.","pith_inferences":["The kernel condition is the weaving analogue of the complement property for ordinary frames: instead of asking that one side of every subset spans, it asks that no rank-two symmetric operator is invisible to all interleaved measurements, a weaker and more linear condition.","Theorem 4.1's hypothesis $\\|P_{V_n}f\\|=\\sqrt n=\\|P_{W_n}f\\|$ for every $f\\in\\mathbb R^n$ cannot be met by proper orthogonal projections when $n>1$, since a projection's norm varies with $f$ unless it is $0$ or $I$; the probabilistic erasure theorem as stated is therefore vacuous, although this does not affect Theorems 3.4 and 3.6.","The converse direction of Theorem 3.6 is not fully written out; the proof says it follows 'using a similar approach,' so the claimed equivalence of phase retrievability with the lower-bound stability condition rests partly on a sketch.","Because $F_\\sigma$ acts on traces of symmetric operators, the criterion transfers to quantum-state discrimination: the same traces are intensities measured in von Neumann measurements, so Theorem 3.4 gives a condition for distinguishing pure states from intensity data across interleaved measurement configurations."],"forward_implications":["Phase retrievability of a weaving fusion frame pair can be certified by finite linear algebra: for each of the $2^m$ subsets $\\sigma$, build the matrix of $F_\\sigma$ and test whether its kernel contains any nonzero rank-two symmetric operator.","Every obstruction has a concrete witness: a nonzero $T\\in\\operatorname{Ker}F_\\sigma\\cap S_{1,1}(H_n)$ yields, through the decomposition $T=f\\otimes f-g\\otimes g$, two vectors that are not unimodular multiples yet have identical magnitude data.","When the pair is phase retrievable, Theorem 3.6 provides a uniform lower bound $\\sum_{i\\in\\sigma}\\|v_i^2\\operatorname{tr}(P_{V_i}[f,g])\\|^2+\\sum_{i\\notin\\sigma}\\|w_i^2\\operatorname{tr}(P_{W_i}[f,g])\\|^2\\ge\\alpha\\|[f,g]\\|_1$, giving a stability margin $\\alpha>0$ that does not depend on $f,g$.","Applying a common unitary $Q$ to all subspaces preserves phase retrievability and the weaving bounds, so the property is unchanged under a change of orthonormal basis (Proposition 3.7).","If the uniform-tightness hypothesis of Theorem 4.1 were satisfiable, random erasure of weaving measurements would give expected reconstruction error controlled by $\\epsilon$ whenever $\\epsilon^2\\ge \\frac{n}{m}\\log n$."],"supporting_citations":[{"why":"Introduces weaving fusion frames and the universal bound definition that Definition 3.1 extends to the phase-retrieval setting.","marker":"[11]"},{"why":"Supplies the weaving fusion frame formalism and operator framework used in the proofs.","marker":"[12]"},{"why":"Provides the rank-two symmetric operator decomposition theorem (Theorem 2.6) that carries the kernel-characterisation proof of Theorem 3.4.","marker":"[14]"}],"fun_headline_variants":["Phase retrieval for weaving frames: rank-two kernel test","All interleavings phase retrievable iff rank-two kernel zero","Weaving frame phase retrievability reduces to rank-two kernel check","Nonlinear phase retrieval to linear kernel check for weaving frames","Weaving fusion frames: phase retrievable iff rank-two kernel is trivial"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The equivalence assumes that every failure of phase retrieval is visible as a rank-two symmetric operator $f\\otimes f-g\\otimes g$ lying in the kernel of $F_\\sigma$, so checking that the kernel meets $S_{1,1}(H_n)$ only at zero is sufficient to rule out all failures.","fun_headline_variants_meta":{"raw":{"variants":["Phase retrieval for weaving frames: rank-two kernel test","All interleavings phase retrievable iff rank-two kernel zero","Weaving frame phase retrievability reduces to rank-two kernel check","Nonlinear phase retrieval to linear kernel check for weaving frames","Weaving fusion frames: phase retrievable iff rank-two kernel is trivial"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000981,"raw_usage":{"total_tokens":4132,"prompt_tokens":881,"completion_tokens":3251,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":497,"completion_tokens_details":{"reasoning_tokens":3164}},"tokens_in":497,"tokens_out":3251,"duration_ms":24055,"temperature":1.0,"reasoning_tokens":3164,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:10:33.206788+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $F_\\sigma(x\\otimes x-y\\otimes y)$ for the pair in Example 3.3 with $\\sigma=\\{1,2\\}$, where $x=(1,i)$ and $y=(i,1)$ are non-parallel yet share all magnitude measurements; the theorem predicts the resulting nonzero rank-two operator lies in $\\operatorname{Ker} F_\\sigma$. A direct search over small $n$ and random subspace pairs that finds any nonzero rank-two operator in $\\operatorname{Ker} F_\\sigma$ for a genuinely phase-retrievable pair would disprove the iff, and exhaustively checking all $\\sigma$ would either confirm or reject the characterization computationally.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces weaving fusion frames and the universal bound definition that Definition 3.1 extends to the phase-retrieval setting."},{"cited_title":"Bhandari, A note on weaving fusion Frames, New York J","cited_arxiv_id":null,"evidence_quote":"Supplies the weaving fusion frame formalism and operator framework used in the proofs."},{"cited_title":"Balan, Stability of Phase Retrievable Frames, SPIE Optical Engineering and Applications, International Society for Optics and Photonics, (2013), DOI: 10.1117/12.2026135","cited_arxiv_id":null,"evidence_quote":"Provides the rank-two symmetric operator decomposition theorem (Theorem 2.6) that carries the kernel-characterisation proof of Theorem 3.4."}],"review_version":1}