{"id":"85fcf5d7-6d9b-4734-be3a-4bf5141b6312","arxiv_id":"2506.05955","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For a new 'common noise' model of fused estimates, the derived bounds show the optimal single fused bound equals Covariance Intersection's, while the whole family of bounds is improved.","lead":"This paper formalizes a new model for fusing two estimates that share an unknown common noise, and derives conservative upper and lower bounds on the fused error. Its main finding is negative: the added information does not shrink the best single worst-case bound compared to standard Covariance Intersection, but it does improve families of bounds and suboptimal fusions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section III-C's global claim that the optimal single fused bound is not improved over CI is not proven; only the ω=1/2, μ=λ subfamily is demonstrated.","rationale":"The reader's verdict is CONDITIONAL with high confidence, and the rationale correctly identifies the unproven generality of the 'no improvement' claim as the main reason. My stress-test agrees: the mathematical derivations that are actually carried out, such as the upper-bound property of Π in (8)-(10), the comparison ΠCI-Π ≥ 0 in (11)-(12), and the optimality of WCI for the ω=1/2 subfamily, are sound. The weak point is the step from that subfamily to the global statement that the optimal fused bound is not improved under assumption (6). That step is load-bearing because the abstract frames the paper's contribution as showing that common-noise partial knowledge is 'disappointing from the perspective of a single upper bound.' I do not treat the unproven 'necessary CI condition' as a substantive defect, because the needed implication B ≥ X from Pi ≥ X is true and can be verified directly. I also do not consider the modeling assumption (6) itself a flaw, since the paper explicitly defines it as a strict formal sense of common noise and scopes its claims to that family. Therefore, the verdict should remain CONDITIONAL: the paper should either prove the no-improvement statement for the whole proposed family (8) or explicitly restrict the claim to the subfamily for which the proof is given.","tokens_in":54,"tokens_out":30668,"duration_ms":362260,"concrete_test":"For many random positive definite P1,P2 (including the paper's example (24)), numerically solve min_{μ>0, ω∈[0,1]} tr((H^T Π(μ,ω)^{-1} H)^{-1}) for Π in (8)-(9), and separately solve min_W max_{X∈(19)∪(23)} tr(W P(X) W^T) over unbiased W (W1+W2=I) with X parametrized as in Section IV-A. Compare both minima to the CI optimum min_{w∈[0,1]} tr((w P1^{-1}+(1-w) P2^{-1})^{-1}). If any instance yields a strictly smaller value than the CI optimum, the Section III-C claim is false; if thousands of random instances all match CI within numerical tolerance, the claim is corroborated but still requires an analytic proof for the whole family.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's negative conclusion, advertised in the abstract, is that partial knowledge of common noise (6) does not improve the optimal single upper bound for linear fusion. However, Section III-C proves a strictly narrower statement. For the specific bound Π from (8)-(9) with ω=1/2 and μ=λ, it shows that the CI-optimal weight WCI also minimizes WΠW^T, and that WCI(ΠCI-Π)(WCI)^T = 0 (equations after (15)-(18)). This establishes equality of the fused bounds only for that one-parameter subfamily of the proposed bounds. The unqualified claim that no fusion rule—no parameter μ, no parameter ω, no other admissible B in (8), and no other weight W satisfying WH=I—can yield a single bound smaller than the best CI bound on the family (6) is not derived anywhere. Figures 3 and 4 provide only selected numerical instances and cannot prove a universal negative. Because the paper's 'disappointing' headline result rests on this unproven optimality assertion, the gap is load-bearing. The reliance on the quoted 'necessary CI condition' from [10] is also a cited lemma, but that particular step is true and easily supplied, so the substantive weakness is the global optimality claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a new partial-correlation model for linear fusion of two full-state estimates: the cross-covariance is an unknown common-noise matrix X satisfying X = X^T ≥ 0 and X ≤ P1, P2 (eq. (6)). Section III-A constructs Loewner upper bounds Π (eq. (8)) and proves Π − P ≥ 0 by a Schur-complement argument; for the parameter choice ω = 1/2 and μ = λ it proves ΠCI − Π ≥ 0 through the factorization (12). Section III-B constructs a lower bound O (eq. (13)) and proves P − O ≥ 0. Section III-C studies linear fusion with H = [I, I]^T: it shows that with the CI weight WCI, the fused new bound equals the fused CI bound, and that WCI is the optimal weight for the new bound in the subfamily ω = 1/2, μ = λ. Section IV illustrates the admissible sets of X and the ellipse geometry for a 2×2 example. The paper concludes that the partial knowledge does not improve the best single upper bound, but does improve suboptimal fusions and families of bounds.","tokens_in":10287,"tokens_out":20501,"duration_ms":228039,"significance":"If the claims are properly qualified, the paper provides a clean mathematical formalization of the 'common noise' assumption that is dual to inverse covariance intersection, along with simple upper and lower matrix bounds (8) and (13). The algebraic derivations (10), (12), (14), and (16)–(18) are valid for the stated parameter choices, and the observation that the new bounds remain valid when the Pi are inflated is a useful property. The paper is honest about the negative flavor of the main fusion result, which, if established generally, would be a cautionary finding for practitioners. However, the advertised global negative result is currently proven only for a subfamily, and at least one related superiority claim (Π better than ΠCI) is stated too broadly and fails for other admissible parameters. The contribution is therefore promising but incompletely established.","major_comments":[{"comment":"The statement that 'the proposed bounds Π (8) are better than the general bounds ΠCI (5)' is unqualified, but the proof that follows verifies ΠCI − Π ≥ 0 only for ω = 1/2 and μ = λ. For other admissible parameters the inequality can fail: in the scalar case P1 = 2, P2 = 5 with ω = 0, μ = λ = 1, one has B = P2 = 5 and ΠCI − Π has a negative top-left entry, so it is not positive semidefinite. Please either prove the dominance for the full parameter range or explicitly restrict the 'better' claim to the matched subfamily and adjust the summary accordingly.","section":"Section III-A, Eqs. (8)–(12)"},{"comment":"The paper's headline negative result—that the fusion cannot improve the standard case in the sense of the best single bound—is not established by the given proof. The derivation shows WCI(ΠCI − Π)(WCI)^T = 0 and that WCI is optimal for Π only for the subfamily ω = 1/2, μ = λ; the simplifying ratio equality (17) requires ω = 1/2, and the identification with WCI in (18) requires μ = λ. The proof does not cover other ω ∈ [0,1], other μ, other matrices B satisfying the general condition in (8), or other admissible linear weights W with WH = I. The numerical examples in Figures 3 and 4 are illustrative only and cannot establish a universal negative. Please prove the general optimality statement or explicitly restrict the abstract and Section V to the proven subfamily.","section":"Section III-C, abstract, Section V"},{"comment":"The proofs rely on the implication B ≥ X from the assumption Pi ≥ X, quoted from [10] without proof, and on inverses X^{-1} even though X is allowed to be singular (X ≥ 0 in (6), and the figures use rank-1 matrices X). This implication is load-bearing for Π − P ≥ 0 and should be supplied: for B from (9), X^{-1} ≥ ωP1^{-1} + (1−ω)P2^{-1} = B^{-1} when X is nonsingular, and the general condition on B in (8) yields the same conclusion by comparing quadratic forms. The use of X^{-1} can be justified by a continuity argument or by adding a standing nonsingularity assumption. These are local but necessary repairs for the upper- and lower-bound theorems as stated.","section":"Section III-A, Eq. (10); Section III-B, Eq. (14)"}],"minor_comments":[{"comment":"The blkdiag display has a mismatched parenthesis; it should read blkdiag(w^{-1}P1, (1−w)^{-1}P2).","section":"Eq. (5)"},{"comment":"The phrase 'family of P12 that is prospected' should be 'family of P12 that is considered'.","section":"Section II-B"},{"comment":"The sentence 'A family of lower bounds of P1, P2 is then given by XΩ' is confusing; these matrices XΩ are candidates for the common-noise covariance and should be described as a family of admissible common-noise matrices that are bounded above by both P1 and P2.","section":"Section IV-A, Eq. (23)"},{"comment":"The claim that 'the weights WCI are optimal with respect to the specific bounds in the new family (8)' should state the sense of optimality, e.g., the Loewner-minimal fused bound for a fixed Π, or the minimizer of a stated scalar criterion over the parameter of the family.","section":"Section III-C"},{"comment":"The sentence 'The figures demonstrate that no fusion can outperform the CI fusion in the terms of single upper bound' overstates what a numerical example can show; please rephrase as an illustration of the proven subfamily result.","section":"Section IV-B"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a conference-style paper with a heavily self-referential literature review; the proofs for the stated subfamily are sound. The main issue is that the abstract and Section V advertise a global negative result that is proven only for a matched subfamily, and the 'better than CI' claim in Section III-A is false as stated for other parameters. If the authors cannot prove the general negative statement, the paper can still be acceptable as a contribution on a new model and a suboptimal-family analysis, provided the claims are carefully restricted. I would not reject on the current mathematical core, but the advertised conclusions must be realigned with the proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Jiri and Ondrej have written a careful paper that does what it says: defines a strict dual to ICI—common noise rather than common information—and derives upper (8) and lower (13) bounds for that family. The upper-bound proof is clean, and the PSD comparison ΠCI−Π for ω=1/2, μ=λ is valid. The algebraic identification of the optimal fusion weight with the CI weight for that parameter choice is also correct. You should know that the paper is genuine, not a tear-down: the modeling is explicit, and the authors are honest about the 'academic' nature of the assumption.\n\nThe soft spot is the headline negative claim. The abstract says that from the perspective of a single upper bound the partial knowledge does not improve fusion, and Section III-C states that 'the optimal fused bound is not improved.' What is actually shown is narrower: for the specific bound (8)-(9) with ω=1/2 and μ=λ, the optimal weight equals WCI and WCI(ΠCI−Π)WCI^T=0. That establishes equality for that one-parameter subfamily. The claim that no other admissible B, μ, ω, or weight can beat the best CI bound is not proven anywhere; it is supported by a couple of figures. The cited 'necessary CI condition' from [10] is load-bearing but true, and could be supplied in a line. So the paper needs either a proof of the global non-improvement statement or an explicit restriction of the claim to the subfamily.\n\nThis is a narrow contribution for readers who work on fusion under partial correlation. For that audience it is useful: it closes out a natural dual case with correct math. It does not deserve desk rejection. If I were the editor I would send it to a referee who knows CI/ICI, and ask for the optimality claim to be tightened. That is a revision, not a rejection.","headline":"Solid, carefully scoped dual to ICI, but the headline no-improvement claim only holds for a one-parameter subfamily and is overgeneralized.","tokens_in":10785,"tokens_out":10466,"would_cite":false,"duration_ms":106298,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper formalizes a strict common-noise assumption dual to inverse covariance intersection and proves that the new joint covariance bounds dominate Covariance Intersection bounds while leaving the optimal fused error bound unchanged.","keywords":["covariance intersection","inverse covariance intersection","common noise","data fusion","unknown correlation","Loewner order","matrix inequalities","linear estimation"],"falsifier":"Compute, for random positive definite $P_1, P_2$ and random $X$ with $0 \\leq X \\leq P_1, P_2$, the minimal eigenvalue of $\\Pi - P$ for $\\Pi$ from (8)-(9); any negative eigenvalue would refute the upper-bound claim. Separately, evaluate $W_{\\mathrm{CI}}(\\Pi_{\\mathrm{CI}} - \\Pi)(W_{\\mathrm{CI}})^T$ for random $P_1, P_2$; any nonzero matrix would refute the claimed equality at the optimal weight.","tokens_in":9820,"feed_emoji":"🔄","tokens_out":12708,"duration_ms":132489,"temperature":0.7,"pith_summary":"This paper asks what fusion guarantees are possible when two estimates have unknown correlation but are known to share a common noise: the error cross-covariance is a single unknown matrix $X$ that is no larger than either individual error covariance. The authors formalize this as $0 \\leq X \\leq P_1, P_2$, construct a family of joint covariance upper bounds $\\Pi$ for all such $X$, and prove that $\\Pi$ is a valid bound and is no worse than the standard Covariance Intersection (CI) bound at matched parameters. The central result is that the partial knowledge does not improve the best single fused bound: with the CI-optimal fusion weight, the new bound on the fused mean-square error matrix equals the CI bound exactly. What the common-noise knowledge does buy is strictly smaller upper bounds for suboptimal fusion weights, a tighter family of bounds, and a nontrivial lower bound on the fused error covariance. A sympathetic reader should care because this delineates exactly when a widely assumed common-noise structure helps fusion and when it is invisible to optimal linear fusion.","feed_headline":"Tighter bounds do not improve optimal Covariance Intersection fusion","feed_subtitle":"Modeling shared noise shrinks joint uncertainty bounds, yet the best fused estimate is identical to CI.","key_machinery":"The machine that carries the argument is the parametrized upper-bound family $\\Pi$ in (8), whose off-diagonal block is built from the dual harmonic-mean matrix $B = (\\omega P_1^{-1} + (1-\\omega)P_2^{-1})^{-1}$. The proof that $\\Pi - P$ is positive semidefinite reduces to verifying the Schur complement $2\\mu(X - X B^{-1} X) \\geq 0$; this holds because $B \\geq X$, a consequence of the necessary CI condition $x^T B^{-1}x \\leq \\max\\{x^T P_1^{-1}x, x^T P_2^{-1}x\\}$ together with $X \\leq P_1, P_2$. The second mechanism is the algebraic identity $W_{\\mathrm{CI}}(\\Pi_{\\mathrm{CI}} - \\Pi)(W_{\\mathrm{CI}})^T = 0$: the CI fusion weight lies in the kernel of the improvement, which is why the optimal fused bound degenerates to the CI value.","core_discovery":"On the paper's own terms, the discovery is this. If the cross-covariance of two full-state estimation errors is an unknown matrix $X$ with $X = X^T \\geq 0$ and $0 \\leq X \\leq P_1, P_2$ (interpreted as identical common noise in both errors), then the block matrix $\\Pi$ in (8), formed with any $B$ satisfying $x^T B^{-1} x \\leq \\max\\{x^T P_1^{-1}x, x^T P_2^{-1}x\\}$ for all $x$, is an upper bound of every admissible joint covariance $P$. For the parametric subfamily $B = (\\omega P_1^{-1} + (1-\\omega)P_2^{-1})^{-1}$ with $\\omega = 1/2$ and $\\mu = \\lambda$, this bound is tighter than the CI bound $\\Pi_{\\mathrm{CI}}$ in the Loewner order, with $\\Pi_{\\mathrm{CI}} - \\Pi$ given by the explicit positive semidefinite factorization (12). Yet the same factorization is annihilated by the CI-optimal fusion weight $W_{\\mathrm{CI}}$: $W_{\\mathrm{CI}}(\\Pi_{\\mathrm{CI}} - \\Pi)(W_{\\mathrm{CI}})^T = 0$, so the fused bound $(P_1^{-1} + P_2^{-1})^{-1}$ is unchanged. The proposed family therefore improves the joint uncertainty description without improving the optimal fused bound.","pith_inferences":["We infer that the degeneracy is structural for full-state fusion with $H = [I, I]$: any correlation model of the form 'identical common noise with covariance $X \\leq P_1, P_2$' is invisible to the linearly optimal unbiased estimator, because the CI weight already saturates the bound. Relaxing the assumption to different scaling factors of the common noise in the two errors would likely break the e","The family-level improvement suggests a practical design rule: when the fusion weight is fixed in advance, the common-noise assumption can certify a smaller fused error ellipsoid than CI; quantifying this gain in terms of the spectra of $P_1, P_2$ and the parameters $\\omega, \\mu$ would turn the qualitative improvement into a design algorithm.","Because the construction is dual to inverse covariance intersection by exchanging $P_i$ with $P_i^{-1}$, techniques developed for one family, such as explicit ellipsoidal circumscription of the union of admissible covariances, should carry over to the other, potentially closing the tightness gap left open here.","The monotonicity in $P_i$ implies the bounds remain conservative when only upper bounds of the individual covariances are available, which we expect to matter in decentralized architectures where nodes report conservative estimates."],"forward_implications":["For any two estimates whose errors admit the common-noise decomposition (6), the joint covariance can be conservatively bounded by (8), and this bound is at least as tight as the corresponding CI bound when $\\omega = 1/2$ and $\\mu = \\lambda$.","The CI-optimal fusion weight is also optimal for the proposed bound family at $\\omega = 1/2$, and the resulting best fused mean-square-error bound equals the CI bound $(P_1^{-1} + P_2^{-1})^{-1}$, so the partial knowledge does not improve the optimal single bound.","For suboptimal fusion weights, the new family can produce strictly smaller fused bounds than CI, giving a concrete benefit when the weight is constrained by computation or by partial-state structure.","The upper bounds remain valid if the inputs $P_1, P_2$ are inflated during construction, a robustness property that inverse covariance intersection does not have.","A nontrivial lower bound $O$ in (13) applies to every admissible $P$, and for a fixed fusion weight $W$ it yields a lower bound $O_F = W O W^T$ on the fused error covariance."],"supporting_citations":[{"why":"Supplies the necessary CI condition lemma that yields $B \\geq X$, the load-bearing step in proving $\\Pi - P$ is positive semidefinite.","marker":"[10]"},{"why":"Introduces inverse covariance intersection and the common-information assumption whose dual motivates the proposed common-noise family.","marker":"[16]"},{"why":"Furnishes the admissible-set methodology and the analogous upper-bound construction that the paper adapts to the dual assumption.","marker":"[19]"},{"why":"Provides the comparison technique based on positive semidefinite differences of bounds, used to show $\\Pi_{\\mathrm{CI}} - \\Pi \\geq 0$.","marker":"[20]"},{"why":"Gives the multiple-estimate generalization and the parametrized suboptimal family that the proposed bounds extend.","marker":"[24]"}],"fun_headline_variants":["Tighter joint bound doesn't improve CI-optimal fusion","Shared-noise bounds tighten, but CI fusion stays optimal","Partial correlation knowledge: tighter bounds, same fusion","Dual CI: tighter joint bound, but optimal fusion unchanged"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the two estimation errors really can be written as a single common noise with the same covariance $X$ in both, plus independent parts, with $0 \\leq X \\leq P_1, P_2$; if that decomposition fails, the bounds are not guaranteed, and the proof also leans on a covariance-intersection lemma quoted from the literature rather than proved here.","fun_headline_variants_meta":{"raw":{"variants":["Tighter joint bound doesn't improve CI-optimal fusion","Shared-noise bounds tighten, but CI fusion stays optimal","Partial correlation knowledge: tighter bounds, same fusion","Dual CI: tighter joint bound, but optimal fusion unchanged"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000772,"raw_usage":{"total_tokens":3439,"prompt_tokens":988,"completion_tokens":2451,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":604,"completion_tokens_details":{"reasoning_tokens":2386}},"tokens_in":604,"tokens_out":2451,"duration_ms":21706,"temperature":1.0,"reasoning_tokens":2386,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T10:13:06.214553+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for random positive definite $P_1, P_2$ and random $X$ with $0 \\leq X \\leq P_1, P_2$, the minimal eigenvalue of $\\Pi - P$ for $\\Pi$ from (8)-(9); any negative eigenvalue would refute the upper-bound claim. Separately, evaluate $W_{\\mathrm{CI}}(\\Pi_{\\mathrm{CI}} - \\Pi)(W_{\\mathrm{CI}})^T$ for random $P_1, P_2$; any nonzero matrix would refute the claimed equality at the optimal weight.","supporting_citations":[{"cited_title":"Fusion under unknown correlation – covariance intersection as a special case,","cited_arxiv_id":null,"evidence_quote":"Supplies the necessary CI condition lemma that yields $B \\geq X$, the load-bearing step in proving $\\Pi - P$ is positive semidefinite."},{"cited_title":"Decentralized data fusion with inverse covariance intersection,","cited_arxiv_id":null,"evidence_quote":"Introduces inverse covariance intersection and the common-information assumption whose dual motivates the proposed common-noise family."},{"cited_title":"Decentralised estimation with correlation limited by optimal processing of independent data,","cited_arxiv_id":null,"evidence_quote":"Furnishes the admissible-set methodology and the analogous upper-bound construction that the paper adapts to the dual assumption."},{"cited_title":"Comparison of fusions under unknown and partially known correlations,","cited_arxiv_id":null,"evidence_quote":"Provides the comparison technique based on positive semidefinite differences of bounds, used to show $\\Pi_{\\mathrm{CI}} - \\Pi \\geq 0$."},{"cited_title":"Inverse covariance intersection fusion of multiple estimates,","cited_arxiv_id":null,"evidence_quote":"Gives the multiple-estimate generalization and the parametrized suboptimal family that the proposed bounds extend."}],"review_version":1}