{"id":"bc1424a9-f26c-4923-a534-184219a2f49a","arxiv_id":"2411.09960","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"A new pseudometric on persistence barcodes, based on eigenvalues of normalized overlap matrices, is proposed and shown experimentally to be invariant under similarity transformations.","lead":"Researchers introduce a new distance for comparing persistent homology barcodes that ignores overall scale and position, so it can group objects by shape rather than size. They test it on synthetic point clouds and sound waves, claiming it stays stable under rotations, translations, and rescalings where standard distances fail.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.3 is false: the Hoffman–Wielandt step applies to non-normal similarity-orbit representatives, and an explicit finite matrix pair gives d_F(S(G_B),S(G_C)) < d_S^(2)(B,C).","rationale":"The paper aims to define a similarity-invariant pseudometric on barcodes and to connect it with operator-theoretic similarity orbits. For that connection to hold, Theorem 3.3 must be true. It is not: the proof applies Hoffman–Wielandt to arbitrary representatives of similarity orbits, even though those representatives need not be normal. The explicit 2x2 counterexample is decisive and does not rely on infinite barcodes or on experimental details. It uses normalized Gram spectra (1,0.9) and (1,0.90001), both realizable by disjoint intervals, and gives d_F < d_S by a factor of about 0.71. The reader's separate concern about unbounded H0 bars in the wave experiment is also valid, but the algebraic failure is already fatal for the stated central claim. The experimental comparisons may retain heuristic interest, but they do not support the paper's main theoretical assertion. I therefore agree with the reader's REJECT verdict and see no reason to adjust it.","tokens_in":17438,"tokens_out":25058,"duration_ms":259720,"concrete_test":"Compute both sides of Theorem 3.3 for the pair above, embedded as infinite direct sums: T = [[1,1],[0,0.9]] and S = [[1.000005,1],[-4.99975e-7,0.900005]]. Verify that T has eigenvalues 1,0.9 and S has eigenvalues 1,0.90001, so d_S^(2) = 1e-5, while ||T-S||_F ≈ 7.09e-6, giving d_F(S(G_B),S(G_C)) <= 7.09e-6 < 1e-5. Re-derive the proof's first inequality using only the stated Hoffman–Wielandt theorem to confirm that the normality hypothesis is the invalid step.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central operator-theoretic claim is Theorem 3.3, the equality d_F(S(G_B),S(G_C)) = d_S^(2)(B,C). The proof needs the inequality d_S^(2)(B,C) <= d_F(...), and obtains it by applying Hoffman–Wielandt (Theorem 2.1) to arbitrary T in S(G_B) and S in S(G_C). That theorem requires both matrices to be normal, but similarity to a normal PSD Gram matrix does not make the representative normal. The required spectral-variation bound is false for diagonalizable non-normal matrices. Explicitly, take barcodes whose normalized Gram matrices are G_B = diag(1,0.9) and G_C = diag(1,0.90001); both are realized by two disjoint intervals with those normalized lengths. Let T = [[1,1],[0,0.9]] and S = [[1.000005,1],[-4.99975e-7,0.900005]]. T is similar to G_B and S is similar to G_C, since each has distinct real eigenvalues equal to the diagonal entries. Neither matrix is normal. Their Frobenius distance is approximately 7.09e-6, so d_F(S(G_B),S(G_C)) <= 7.09e-6. But Proposition 3.2 gives d_S^(2)(B,C) = |0.9 - 0.90001| = 1e-5. Thus d_F < d_S, contradicting Theorem 3.3. The second inequality in the proof is fine; the failure is precisely the Hoffman–Wielandt step, and no missing constant repairs the equality.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines a pseudometric d_S^(p) on barcodes of finite bounded type by taking the eigenvalues of the normalized Gram matrix of interval characteristic functions and comparing them in l^p after optimal matching. For p=2 (and p=1) it proves that the optimal matching is the sorting of eigenvalues (Propositions 3.2 and 3.4). The central theoretical claim is Theorem 3.3, which asserts that d_S^(2) equals the Frobenius distance between the similarity orbits of the corresponding Gram matrices. The paper also reports experiments on Vietoris-Rips barcodes of synthetic ellipse data and on H0 sublevel-set barcodes of sound waves, claiming that d_S is stable under conformal linear transformations and independent of frequency and amplitude, unlike bottleneck and Wasserstein distances. The main theorem is false as stated; the proof relies on an invalid application of the Hoffman-Wielandt theorem. Furthermore, the wave experiments use infinite intervals, which are not covered by Definition 2.","tokens_in":17663,"tokens_out":6123,"duration_ms":60093,"significance":"If Theorem 3.3 were correct, it would establish a new bridge between operator similarity orbits and persistent-homology barcodes, and the proposed pseudometric would offer a genuinely new similarity-invariant comparison tool with computational advantages. The paper includes some sound components: Definition 7 is a genuine pseudometric (Theorem 3.1), the sorting results in Propositions 3.2 and 3.4 are correct, the code is publicly available, and the experimental comparison is clearly described. However, the central operator-theoretic equality is false, and the main application (sublevel-set H0 barcodes of sound waves) is outside the paper's own theoretical framework because it uses infinite bars. These issues invalidate the paper's primary claims.","major_comments":[{"comment":"","section":"Section 3, Theorem 3.3"},{"comment":"","section":"Section 4.2 and Definition 2"}],"minor_comments":[{"comment":"","section":"Examples 1 and 2"},{"comment":"","section":"Theorem 3.1"},{"comment":"","section":"Theorem 3.3 statement"},{"comment":"","section":"Throughout"}],"recommendation":"reject","confidential_remarks":"The counterexample to Theorem 3.3 is simple, explicit, and verifiable; it shows the central equality is not merely unproven but false. The experimental section also depends on infinite intervals outside the formal definition. I see no way to repair the paper within its current scope while preserving the advertised connection to operator-theoretic similarity orbits. Rejection is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know this paper before reading it: the proposed d_S pseudometric is genuinely new and the wave classification experiment is a legitimate application, but the main theoretical claim—Theorem 3.3, the equality with the Frobenius distance between similarity orbits—is false. The stress-test counterexample is correct.\n\nThe construction is nice: for a barcode of finite bounded intervals, take the Gram matrix of characteristic functions, normalize by operator norm, extend by zeros, and compare the sorted eigenvalues in l_p. This gives a scale- and translation-invariant pseudometric on barcodes. The paper correctly proves it is a pseudometric and that the sorted matching achieves the distance (Propositions 3.2 and 3.4 look right). The application to distinguishing piano and tuning-fork waveforms independent of frequency and amplitude is a genuinely new use case, and the timing comparison is useful.\n\nThe problem is Theorem 3.3. The proof applies Hoffman–Wielandt to arbitrary representatives T and S of the similarity orbits of the normalized Gram matrices. That theorem requires normality, and similarity to a normal matrix does not imply normality. The counterexample is explicit: take barcodes with normalized Gram spectra {1, 0.9} and {1, 0.90001}. Then d_S = 10^{-5}. But the non-normal matrices T = [[1,1],[0,0.9]] and S = [[1.000005,1],[-4.99975e-7,0.900005]] lie in the respective similarity orbits, and their Frobenius distance is about 7.09e-6. So d_F(S(G_B),S(G_C)) < d_S(B,C), contradicting the claimed equality. This is not a missing constant; the equality is simply wrong. The correct statement would involve unitary orbits, not similarity orbits.\n\nThere are also smaller issues. The worked examples contain arithmetic errors: the length of (1,3] is 2, not 4, so the Gram matrices in Examples 1 and 2 are wrong. The experiments on sublevel-set H0 barcodes necessarily produce an infinite bar from the global minimum to +∞, but Definition 2 only covers finite bounded intervals and the paper never says how that bar is truncated. These are fixable, but they add to the impression of carelessness.\n\nBottom line: the pseudometric itself is a reasonable idea and the experimental direction is worth pursuing, but the central operator-theoretic result is false. A revised version that corrects Theorem 3.3 (or replaces it with the unitary-orbit statement) and fixes the examples could be a useful paper. As it stands, it should not be accepted. I'd still send it to a knowledgeable referee because the construction is novel and the error is specific and instructive.","headline":"Interesting new barcode pseudometric with a real application, but the central operator-theoretic theorem is false; the paper needs major revision, not acceptance.","tokens_in":18292,"tokens_out":8353,"would_cite":false,"duration_ms":77064,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["55N31","68T09","15A18","47B15"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper defines a pseudometric on persistence barcodes that is invariant under similarity transformations, not only congruence, with an explicit eigenvalue-sorting formula.","keywords":["persistent homology","barcode","similarity pseudometric","Gram matrix","conformal linear transformation","operator theory","Frobenius norm","timbre classification"],"falsifier":"Run the paper's pipeline on two sound waves with identical finite bar content but different vertical offsets, so their H0 barcodes differ only in the starting point of the single infinite interval; because the definition of $d_S$ excludes infinite intervals, the distance is undefined without an added truncation rule, and the value produced will depend on the rule chosen, which can be measured directly.","tokens_in":17131,"feed_emoji":"🎵","tokens_out":13225,"duration_ms":129739,"temperature":0.7,"pith_summary":"The paper introduces a family of pseudometrics $d_S^{(p)}$ on persistence barcodes that compares barcodes up to similarity rather than up to congruence. Its main claim is that $d_S^{(2)}$ is a similarity invariant: replacing a barcode by another that is similar in the operator-theoretic sense leaves the distance unchanged, and the distance is exactly the Frobenius distance between the similarity orbits of the barcodes' Gram matrices. If this holds, persistent homology gains a way to compare the shape of topological features while ignoring scale, translation, rotation, and other conformal changes, which bottleneck and Wasserstein distances cannot do. The paper supports this with experiments on synthetic ellipse point clouds and on slices of piano and tuning-fork sound waves, where $d_S^{(1)}$ and $d_S^{(2)}$ separate the instruments regardless of frequency and amplitude, at computation times below bottleneck distance.","feed_headline":"A new barcode distance ignores scale and rotation","feed_subtitle":"It separates piano and tuning-fork waves regardless of pitch or volume, where bottleneck and Wasserstein distances fail.","key_machinery":"The load-bearing object is the Gram matrix $G_B$ with entries $\\int_{\\mathbb{R}} \\mathbf{1}_{(a_i,b_i]}(x)\\mathbf{1}_{(a_j,b_j]}(x)\\,dx$, built from the indicator functions of the bars in a barcode $B$. Dividing by the operator norm removes scale, padding with zeros makes barcodes with different numbers of bars comparable, and the sorted eigenvalue sequence $\\Lambda_B$ of this normalized padded matrix is what $d_S^{(p)}$ compares. The descending-order matching formula, $d_S^{(2)} = \\sqrt{\\sum_i |\\lambda_i-\\mu_i|^2}$ after sorting, is what makes the distance easy to compute, and the similarity-orbit equality in Theorem 3.3 is the mechanism that gives the distance its invariance.","core_discovery":"On the paper's own terms, the discovery is that a barcode of finite bounded type can be converted into a Gram matrix whose entries are the overlaps of the indicator functions of its intervals; after normalizing by the operator norm and padding with zeros, the eigenvalues of this matrix form a sequence that describes the barcode's shape up to similarity. The pseudometric $d_S^{(2)}$ is the $\\ell^2$ distance between two such eigenvalue sequences after the optimal matching, and Proposition 3.2 shows that sorting the eigenvalues in descending order makes the identity matching optimal, giving a closed-form calculation. Theorem 3.3 then states that $d_S^{(2)}$ equals the Frobenius distance between the similarity orbits of the two normalized Gram matrices, which is the sense in which the pseudometric is a similarity invariant. The paper further claims $d_S^{(1)}$ behaves analogously, and that both distances are stable under conformal linear transformations in the reported experiments.","pith_inferences":["As an editorial extension, replacing the indicator-function Gram kernel with any positive-semidefinite kernel on intervals would produce a family of pseudometrics with the same sorted-eigenvalue formula, potentially extending the construction to barcodes with infinite intervals.","A testable consequence the authors do not spell out is that the choice of how to truncate the infinite bar in an H0 sublevel-set barcode will generally change the Gram matrix, so a user of the published pipeline must adopt an extra convention before the theory applies.","If the similarity-orbit equality extends beyond finite bounded barcodes, $d_S$ could serve as an invariance-preserving loss for representation learning where the model must be insensitive to affine transformations of the input."],"forward_implications":["If $d_S^{(2)}$ is a similarity invariant as claimed, objects related by rotation, scaling, and translation will have the same or nearly the same $d_S$ distance, so clustering can be done without alignment or normalization preprocessing.","The equality with the similarity-orbit Frobenius distance connects persistence barcodes to operator theory, making spectral approximation and orbit-based perturbation tools available for topological data analysis.","On sound waves, the claimed independence from frequency and amplitude means timbre classification can be performed directly on persistence barcodes of sublevel sets, without pitch correction or amplitude normalization.","Because $d_S^{(1)}$ and $d_S^{(2)}$ only require sorting eigenvalues and taking an $\\ell^p$ difference, pairwise distance matrices are reported to be computable faster than bottleneck distance and comparably to accelerated Wasserstein distance."],"supporting_citations":[{"why":"introduces topological persistence, the construction that produces the barcodes the pseudometric compares.","marker":"[7]"},{"why":"introduces persistence modules and establishes the algebraic framework for barcode representations.","marker":"[8]"},{"why":"gives the definitions of Wasserstein and bottleneck distances with matching costs that $d_S$ is compared against.","marker":"[34]"},{"why":"introduced the p-Wasserstein distance for arbitrary p, the family that $d_S^{(p)}$ parallels.","marker":"[33]"},{"why":"supplies the persistence computations (sublevel-set and Vietoris-Rips filtrations) used in the experiments.","marker":"[42]"},{"why":"provides the accelerated Wasserstein distance implementation used for the running-time comparisons.","marker":"[41]"},{"why":"supplies the piano sound wave used in the timbre-classification experiment.","marker":"[5]"},{"why":"supplies the tuning-fork sound wave used in the timbre-classification experiment.","marker":"[6]"}],"fun_headline_variants":["Barcode metric ignores scale and rotation","Similarity-invariant barcode distance for shapes","New distance stably separates waves by waveform","Scale-free barcode pseudometric outperforms bottleneck"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire construction requires barcodes whose intervals are all finite, so the Gram matrix has finite trace and discrete eigenvalues; the paper's sound-wave experiments use H0 sublevel-set barcodes that always contain an interval from the global minimum to $+\\infty$, and the paper does not state how that unbounded bar is removed or truncated before $d_S$ is applied.","fun_headline_variants_meta":{"raw":{"variants":["Barcode metric ignores scale and rotation","Similarity-invariant barcode distance for shapes","New distance stably separates waves by waveform","Scale-free barcode pseudometric outperforms bottleneck"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000164,"raw_usage":{"total_tokens":1266,"prompt_tokens":984,"completion_tokens":282,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":238}},"tokens_in":600,"tokens_out":282,"duration_ms":3312,"temperature":1.0,"reasoning_tokens":238,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:10:32.558189+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's pipeline on two sound waves with identical finite bar content but different vertical offsets, so their H0 barcodes differ only in the starting point of the single infinite interval; because the definition of $d_S$ excludes infinite intervals, the distance is undefined without an added truncation rule, and the value produced will depend on the rule chosen, which can be measured directly.","supporting_citations":[{"cited_title":"Dis- crete Comput","cited_arxiv_id":null,"evidence_quote":"introduces persistence modules and establishes the algebraic framework for barcode representations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"introduced the p-Wasserstein distance for arbitrary p, the family that $d_S^{(p)}$ parallels."},{"cited_title":"In: 2016 Proceedings of the Eighteenth Workshop on Algorithm Engineering and Experiments (ALENEX), pp","cited_arxiv_id":null,"evidence_quote":"provides the accelerated Wasserstein distance implementation used for the running-time comparisons."},{"cited_title":"https://freesound.org/people/Teddy Frost/sounds/334536/ (2016)","cited_arxiv_id":null,"evidence_quote":"supplies the piano sound wave used in the timbre-classification experiment."},{"cited_title":"https://freesound.or g/people/jmuehlhans/sounds/220747/ (2014)","cited_arxiv_id":null,"evidence_quote":"supplies the tuning-fork sound wave used in the timbre-classification experiment."}],"review_version":1}