{"id":"fdb2c8ee-a1a6-4580-a38b-a3ebf03b709f","arxiv_id":"1908.05235","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"A claimed necessary and sufficient condition for simultaneous disturbance decoupling and instantaneous fault detection in Boolean control networks rests on a lemma that fails for multiple fault variables.","lead":"This paper proposes broader definitions of disturbance decoupling for Boolean control networks and algorithms for instantaneous fault detection, including a single feedback controller that decouples disturbances and detects faults at the same time. The main necessary-and-sufficient theorem relies on a false lemma, so the headline result is not established.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 14 is false: for two or more remaining variables, individual reflectivity does not imply joint injectivity, so the full-column-rank condition is not equivalent; Theorems 15 and 17 overclaim necessary and sufficient IFD/DD.","rationale":"The reader's weakest assumption is correct and load-bearing. The paper's headline claim is Theorem 17, and both Theorem 15 and Theorem 17 are proved by invoking Lemma 14 and Lemma 16; once the lemma's equivalence is false, the necessary direction collapses. A minor caveat is that the reader's two-variable XOR example, as written, would set r=0 if both variables are the remaining variables, while Lemma 14 states 0<r≤n. However, padding with one dummy state variable (so r=1) gives the same failure within the lemma's stated range, so the objection stands. This is not a stylistic or presentation issue: Algorithm A.6 requires full column rank 2^t and an invertible or pseudo-invertible map, which is a different and stronger condition than 'every fault variable is reflected separately'; the theorem statements and abstract therefore overclaim. The worked examples do not rescue the general claim, since those examples satisfy the stronger rank condition. The appropriate verdict remains REJECT, in agreement with the reader; no change to the reader's verdict is needed.","tokens_in":38291,"tokens_out":10857,"duration_ms":109788,"concrete_test":"Construct in the paper's STP framework the structure matrix M_G for G(x1, ξ1, ξ2) = (ξ1⊕ξ2, ξ1⊕ξ2), and split it into the two blocks corresponding to x1=0 and x1=1. Verify that each fault variable satisfies Eq. (28) for all settings of the other variables, and compute the rank of each 4×4 block: it will be 2, not 4. Also evaluate the fault-to-output map at (ξ1,ξ2) = (0,0), (0,1), (1,0), (1,1): the outputs are (0,0), (1,1), (1,1), (0,0). This directly contradicts Lemma 14 and falsifies the necessity direction of Theorem 17, because a double fault is undetectable despite every fault variable being individually reflective.","verdict_should_be":"REJECT","load_bearing_attack":"The central necessary-and-sufficient claims rest on Lemma 14 (Section VII-C, Eq. (28)) and Lemma 16, which equate “every fault (disturbance) variable is reflective (redundant)” with a full-column-rank block structure of M_O. That equivalence is false for two or more remaining variables. Take one state variable x1 and two fault variables ξ1, ξ2, with outputs y1 = y2 = ξ1⊕ξ2. Each fault variable flips every output component for every value of the other fault and of x1, so each satisfies Eq. (28). But the map (ξ1,ξ2) ↦ y is not injective: (0,0) and (1,1) both give output (0,0), while (0,1) and (1,0) both give (1,1). The corresponding M_O has two identical column pairs; with x1 fixed, each 4×4 block has rank 2 rather than the required 2^f = 4. Padding with the dummy state variable keeps Lemma 14's premise 0<r≤n satisfied, so the counterexample is within the lemma's stated range. Consequently Theorem 15's claim that separate reflectivity of each fault is necessary and sufficient for instantaneous fault detection is wrong: the double fault (1,1) is completely invisible and the two single faults are indistinguishable. Algorithm A.6 itself demands full rank 2^t and an invertible or pseudo-invertible fault-to-output map, which is strictly stronger than the theorem's stated condition. Lemma 16 and Theorem 17 inherit the same failure, so the abstract's headline necessary-and-sufficient claim with a single controller is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses disturbance decoupling (DD) and instantaneous fault detection (IFD) in Boolean control networks. It proposes relaxed notions of DD (in mapping, in iterations, invariant in output), algorithms for controller synthesis, and claims necessary and sufficient conditions for IFD and for combined DD+IFD. The central theoretical results are Lemma 14, which equates \"reflective\" fault variables with a full-column-rank block condition on the output structure matrix, and Lemma 16, which extends this to simultaneous redundant disturbance variables and reflective fault variables. Theorems 15 and 17 then state that separate reflectivity of each fault variable, together with redundancy of each disturbance variable, is necessary and sufficient for IFD and for combined DD+IFD with a single state-feedback controller.","tokens_in":38652,"tokens_out":6319,"duration_ms":71644,"significance":"If the central conditions were correct, the paper would offer a practically useful, checkable characterization for fault detection and disturbance rejection in BCNs without requiring full observability, and it would provide explicit controller construction algorithms with finite termination. The paper contains several worked examples and algorithmic procedures (A.2, A.4, A.6, A.7, A.8). However, the load-bearing Lemma 14 is false, and the main necessary-and-sufficient claims therefore are not established. The paper also introduces interesting conceptual distinctions among reachability notions and between DD in mapping versus in iterations, though these are not the main technical contribution.","major_comments":[{"comment":"The above counterexample respects the lemma's premise 0 < r ≤ n, so the error is not an artifact of an excluded boundary case.","section":"Section VII-C, Lemma 14, Eq. (28)"},{"comment":"The necessity direction of Theorem 15 also relies on the false Lemma 14, so the error is load-bearing for both directions of the claimed characterization.","section":"Section VII-D, Theorem 15"},{"comment":"The redundancy part of Lemma 16 (all disturbance sub-blocks identical) is not affected by this counterexample, but the reflective part is essential to the combined theorem, so the combined result is not supported.","section":"Section VIII-B, Lemma 16, and Section VIII-C, Theorem 17"}],"minor_comments":[{"comment":"The same observation applies to Theorems 12 and 13, whose proofs say the statements follow from the reachability definitions; these results are presented as characterizations but are essentially definitions of the constructed sets.","section":"Section V-B, Algorithm A.2 and Theorem 11"},{"comment":"These do not affect the mathematical content but make the manuscript difficult to read.","section":"Throughout"},{"comment":"A reader cannot easily verify the claimed equivalence in this section without reconstructing the argument from scratch.","section":"Section V-D, Eq. (22) and surrounding text"}],"recommendation":"reject","confidential_remarks":"The manuscript is a preprint with many presentation problems, and the central necessary-and-sufficient results rest on a false lemma. The counterexample in the report is simple and directly within the lemma's scope, so the error is not a matter of tightening an edge case. Even though the paper proposes interesting algorithmic ideas and a useful distinction among DD notions, the main theoretical contribution is not recoverable without reformulating Lemma 14, Lemma 16, and Theorems 15 and 17 to require joint injectivity of all fault variables together rather than separate reflectivity of each. That is a substantive change to the paper's central claims, so a major revision would not be sufficient; rejection is appropriate. I would note, however, that the algorithmic sections (A.6, A.7) may contain usable heuristics if the conditions are corrected, so a future revised version could be reconsidered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper identifies a real gap: no one has treated disturbance decoupling and instantaneous fault detection with a single state-feedback controller in Boolean control networks. The problem formulation is new, and Algorithms A.6 and A.7 are concrete enough to test on examples. The authors also engage honestly with the existing DD and fault-detection literature. That is the good part.\n\nThe bad part is that the central theorem is wrong. Lemma 14 claims that each fault variable being \"reflective\" (flipping it changes every output bit for every assignment of the other variables) is equivalent to the corresponding block of the structure matrix having full column rank. Neither direction holds. Full column rank means distinct columns, which only requires a difference in at least one output bit, not every output bit. So reflectivity is not necessary for full rank. Conversely, two variables can each be reflective individually while the joint map collapses: with outputs y1 = y2 = ξ1⊕ξ2, flipping either fault flips both outputs, so both are reflective, but the fault combinations (0,0) and (1,1) give the same output. The corresponding 4×4 block has rank 2, not 4. The stress-test counterexample is correct and lands exactly on the load-bearing assumption. Theorems 15 and 17, and the abstract's necessary-and-sufficient claim, are therefore unsubstantiated.\n\nThere is also a definitional circularity in Propositions 8–10 and Theorems 11–13: the algorithms test exactly the property they claim to characterize, so the proofs reduce to restating the definition. Several sections re-derive standard semi-tensor-product material, and the excursion into counting Boolean networks feels like a different paper bolted on.\n\nThe salvageable idea is the algorithmic one: if you check the joint fault-to-output block for full column rank (or invertibility/pseudo-invertibility), rather than individual reflectivity, the controller construction in A.7 is a reasonable sufficient condition. That is worth a short note, not a theorem claiming necessity and sufficiency.\n\nRecommendation: reject. If this lands on your desk, don't send it to referees as is; the counterexample will surface immediately. Tell the authors to fix Lemma 14 or downgrade the claims to sufficient conditions.","headline":"The paper's headline necessary-and-sufficient claim is false: Lemma 14's reflectivity/rank equivalence fails in both directions, and Theorems 15 and 17 inherit the error.","tokens_in":39161,"tokens_out":3933,"would_cite":false,"duration_ms":45624,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["93B52","93C55","94C10"],"pacs":[],"model":"deepseek-v4-flash","headline":"In Boolean control networks, a single state-feedback controller can decouple disturbances and detect faults in the next step exactly when every fault variable is reflected in the output and every disturbance variable is redundant.","keywords":["Boolean control networks","disturbance decoupling","instantaneous fault detection","semi-tensor product","state feedback","reflectivity","redundancy","reachability"],"falsifier":"Take two fault variables $a,b$ and two outputs both equal to $a \\oplus b$. Each single variable is reflective by definition (28), yet the map $(a,b)\\mapsto(a\\oplus b,a\\oplus b)$ is not injective and the corresponding $2\\times 4$ block of $M_O$ has rank 2, not 4. Checking such a system against Theorem 17 would settle whether the stated full-column-rank condition is actually necessary.","tokens_in":38114,"feed_emoji":"⚙️","tokens_out":7296,"duration_ms":74285,"temperature":0.7,"pith_summary":"This paper studies Boolean control networks, models in which states, inputs, and outputs are binary and update in discrete time. It argues that the usual way of defining disturbance decoupling is too strict, because it forces the system onto a trajectory that ignores disturbances rather than letting the system evolve naturally while feedback cancels their effect. The central claim is that a single state-feedback controller can decouple disturbances and detect faults in the very next time step exactly when the output map has a block structure: every fault variable must be reflective (flipping it changes every output component for every setting of the other variables) and every disturbance variable must be redundant (flipping it leaves the output unchanged). This matters because full observability of Boolean control networks is NP-hard, and the paper's rank-based conditions give a structural test that avoids expensive state reconstruction.","feed_headline":"One controller can decouple disturbances and detect faults instantly","feed_subtitle":"New necessary and sufficient conditions use the output's block structure and skip costly state reconstruction.","key_machinery":"The machinery is the semi-tensor product representation, which rewrites Boolean dynamics as ordinary matrix equations, together with two rank conditions on the output map $M_O$. Lemma 14 identifies reflective variables with full column rank of the corresponding sub-block of the structure matrix; Lemma 16 adds redundant disturbance variables as identical sub-blocks. Algorithm A.7 turns these conditions into a state-feedback matrix $M_x$ by selecting, for each state index, an input block whose sub-block has the required identical-full-rank structure, and the number of valid controllers is the product $\\prod_j |C_j|$. The reachability-graph constructions in Section V extend the same logic to decoupling in mapping, in finitely many iterations, and invariant-output decoupling.","core_discovery":"The paper's main result, Theorem 17, states that for a Boolean control network $x^+ = L u x \\xi_d \\xi_f$, $y = H x$, with state feedback $u = M_x x$, instantaneous fault detection together with disturbance decoupling is achievable exactly when the composed output map $M_O = H L M_x \\psi_n$ satisfies Lemma (Fault + Disturbance): split $M_O$ into $2^n$ blocks of size $2^p \\times 2^{d+f}$; each block must split into $2^d$ identical sub-blocks, and each sub-block must have full column rank. In words, disturbances must be invisible to the output and each fault must change the output in a distinguishable way. The paper also widens the notion of disturbance decoupling itself, introducing decoupling in mapping, in finitely many iterations, and invariant-output decoupling, so that systems that fail the classical rank test can still be decoupled in a weaker, useful sense.","pith_inferences":["The paper treats reflectivity one variable at a time; a natural strengthening would require every simultaneous fault combination to have its own output signature, which would make the condition more robust when two faults occur in the same step.","If the full-column-rank characterization is repaired for multi-variable blocks, the same block-decomposition idea could be applied to asynchronous or probabilistic Boolean networks, where 'instantaneous' would mean the output distribution changes immediately after a fault.","The decoupling-in-iteration notion is a testable practical extension: one could count how many published Boolean gene regulatory networks become decouplable under Algorithm A.8 even though they fail the classical conditions.","An alternative to the rank test is a direct injectivity check on the columns of each sub-block, which is a weaker premise and would make the necessary-and-sufficient claim easier to verify by hand on small networks."],"forward_implications":["Systems that fail the classical row-wise rank test for disturbance decoupling may still be decouplable, because the paper's condition is imposed on the composed output map $M_O$ rather than on every row of the transition matrix.","A single state-feedback controller $M_x$ can achieve both disturbance decoupling and instantaneous fault detection, so no observer or state reconstruction is needed during the fault-detection phase.","The number of controllers that achieve the combined goal is finite and computable as $\\prod_j |C_j|$, allowing the designer to enumerate all valid state-feedback laws.","The same block condition makes the system fault-output invertible, so the occurrence of a fault is uniquely identifiable from the next output.","For $s$ disturbances and $t$ faults, the criterion becomes $2^s$ identical divisions of each sub-block with an invertible or pseudo-invertible fault-to-output map."],"supporting_citations":[{"why":"Supplies the semi-tensor product linear representation on which the entire matrix-block argument is built.","marker":"[3]"},{"why":"Defines observability and reconstructability for Boolean control networks, the baseline the paper's structural fault detection avoids.","marker":"[13]"},{"why":"Establishes NP-hardness of observability, motivating the search for cheap structural conditions.","marker":"[15]"},{"why":"Introduces the output-friendly subspace and the classical disturbance decoupling problem that the paper broadens.","marker":"[22]"},{"why":"Provides the rank-condition controller design for disturbance decoupling that the paper extends and shows to be conservative.","marker":"[23]"},{"why":"Defines completeness-based fault detection in Boolean control networks, the fault-detection framework the paper's instantaneous notion refines.","marker":"[26]"},{"why":"Defines reachable sets used in the clean, definite, and indefinite reachability constructions for decoupling in iteration.","marker":"[29]"},{"why":"Supplies fault-output invertibility, used to argue unique identifiability of faults in the combined theorem.","marker":"[32]"}],"fun_headline_variants":["One controller decouples disturbances and detects faults instantly","Block structure unlocks simultaneous disturbance decoupling and fault detection","New conditions make disturbance decoupling and fault detection instantaneous","Wider disturbance decoupling notion enables instant fault detection","Single feedback law achieves disturbance decoupling and instant fault detection"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof leans on the claim that a variable being reflective—flipping it changes every output component for every setting of the other variables—is exactly the same as the corresponding block of the output map having full column rank; for two or more variables together that equivalence can fail, and the main theorem's necessity argument depends on it.","fun_headline_variants_meta":{"raw":{"variants":["One controller decouples disturbances and detects faults instantly","Block structure unlocks simultaneous disturbance decoupling and fault detection","New conditions make disturbance decoupling and fault detection instantaneous","Wider disturbance decoupling notion enables instant fault detection","Single feedback law achieves disturbance decoupling and instant fault detection"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000767,"raw_usage":{"total_tokens":3432,"prompt_tokens":1012,"completion_tokens":2420,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":628,"completion_tokens_details":{"reasoning_tokens":2343}},"tokens_in":628,"tokens_out":2420,"duration_ms":18015,"temperature":1.0,"reasoning_tokens":2343,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:24:52.997219+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take two fault variables $a,b$ and two outputs both equal to $a \\oplus b$. Each single variable is reflective by definition (28), yet the map $(a,b)\\mapsto(a\\oplus b,a\\oplus b)$ is not injective and the corresponding $2\\times 4$ block of $M_O$ has rank 2, not 4. Checking such a system against Theorem 17 would settle whether the stated full-column-rank condition is actually necessary.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines observability and reconstructability for Boolean control networks, the baseline the paper's structural fault detection avoids."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes NP-hardness of observability, motivating the search for cheap structural conditions."},{"cited_title":"S2 /equalx{xs∈ X1−s/S1| ∃i∈{ 1,..., 2m} & L′ui xs∈ S1}","cited_arxiv_id":null,"evidence_quote":"Introduces the output-friendly subspace and the classical disturbance decoupling problem that the paper broadens."},{"cited_title":"The difference is evident from the necessary condition, the method presented here, requires ˜Lir to satisfy the given condition, but the individual rows in the sum need not","cited_arxiv_id":null,"evidence_quote":"Defines completeness-based fault detection in Boolean control networks, the fault-detection framework the paper's instantaneous notion refines."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines reachable sets used in the clean, definite, and indefinite reachability constructions for decoupling in iteration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies fault-output invertibility, used to argue unique identifiability of faults in the combined theorem."}],"review_version":1}