{"id":"0b913bae-0889-42c0-ae1f-9e0c71ddeeb5","arxiv_id":"2608.23539","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For N with a distinct prime factors, box(Γ_E(Z_N)) equals a-1 in three precisely described exponent patterns and equals a otherwise; the square-free case also fixes the boxicity of the disjointness graph of the power set.","lead":"The paper gives the exact boxicity of the compressed zero divisor graph of the ring Z_N for every N, settling two open questions from the authors' earlier work. The answer is a-1 or a depending on the exponents in the prime factorization of N, with a fresh lower-bound method for square-free N.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.9's final N''=q^2 subcase misidentifies the induced graph as K2∨Γ_E(Z_N'); the two added vertices are nonadjacent, so the written lower bound is one short until corrected.","rationale":"After working through the main case analysis, I found no counterexample to Theorem 1.7 and no flaw in the most delicate interval construction, Lemma 4.10. The reader identified the inspection-heavy interval claims there as the weakest assumption, but direct verification of the interval mappings and the adjacency claims in Lemma 4.10 shows that this part of the proof is sound. The genuine soft spot is the final subcase of Lemma 4.9, where the graph induced on the two extra vertices plus B\\{N''} is asserted to be K2∨Γ_E(Z_N'). The two extra vertices N/q and N/q^2 are nonadjacent, so the printed graph identification is false and the displayed lower bound does not follow as written. However, the correct identification, with two nonadjacent universal vertices, gives exactly the required boxicity via Observation 2.8. Thus the central mathematical claim appears to be true, but the proof as written needs a correction in Lemma 4.9 before the lower bound for that subcase is valid. For this reason I recommend conditional acceptance rather than unconditional acceptance.","tokens_in":18363,"tokens_out":53848,"duration_ms":492306,"concrete_test":"Recompute the subgraph in Lemma 4.9 for N=N'q^2: compare the q-adic valuations of N/q=N'q and N/q^2=N'; since they are 1 and 0, the product has q-adic valuation 1, so the two vertices are nonadjacent. Then verify that the corrected graph, with two nonadjacent vertices each universal to B\\{N''}, has boxicity 1 + box(Γ_E(Z_N')) by Observation 2.8, giving exactly a=|P|+3. If the original K2∨Γ_E(Z_N') claim were retained, the lower bound would instead be |P|+2, contradicting Lemma 4.9.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In the last paragraph of Lemma 4.9, when N''=q^2, the proof claims that the graph induced on {N/q, N/q^2} ∪ (B\\{N''}) is K2∨Γ_E(Z_N'). A direct exponent check contradicts this: N/q = N'·q and N/q^2 = N', so the q-adic valuations are 1 and 0, summing to 1 < 2. Hence their product is not divisible by N=N'q^2 and they are nonadjacent. The induced graph is instead the join of Γ_E(Z_N') with two nonadjacent universal vertices, i.e. the complement of K2 joined with Γ_E(Z_N'). Taking the printed K2 claim literally, Observation 2.8 would give box = 0 + box(Γ_E(Z_N')) = |P|+2 = a−1, so the required lower bound box ≥ a does not follow. Replacing the complete pair by an edgeless pair gives box = 1 + |P|+2 = a, so the central theorem survives a one-line correction. This is the one place I found where a stated claim, if taken literally, invalidates a needed lower bound. I also independently checked the interval claims in Lemma 4.10 flagged by the reader: the explicit interval mappings satisfy Claims 1 and 2, and the three nonedge cases in the upper-bound proof appear correct.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper determines the exact boxicity of the compressed zero divisor graph Γ_E(Z_N) for N = ∏_{i=1}^a p_i^{n_i}. Theorem 1.7 gives a complete dichotomy: box(Γ_E(Z_N)) = a−1 exactly in the three cases (one cubed prime with all other exponents 1, all exponents 1 with a ≥ 3, and cube-free numbers with at least one exponent 1 and all other exponents 2); box = 0 for a = 2 with both exponents 1; and box = a in all other cases with a ≥ 2. The paper also determines the threshold dimension (Theorem 1.8), answers an open question on reduced rings (Theorem 1.12), and derives corollaries on disjointness graphs of power sets. The proofs split according to the exponent pattern; lower bounds use Roberts graphs and a new C4-conflict-graph coloring argument, while upper bounds use explicit interval representations and isomorphisms to known cases.","tokens_in":18647,"tokens_out":18209,"duration_ms":157509,"significance":"If the result stands, it fully answers two open questions from Chandran and Sahoo's earlier paper and gives a clean structural dichotomoy for a natural algebraic graph class. The C4-conflict-graph method in Lemma 4.9 is a genuinely new lower-bound technique that goes beyond finding induced subgraphs of large boxicity, and the corollary that the disjointness graph of P([a]) has boxicity a−1 is elegant. The explicit interval assignments in Lemma 4.10 are a strength; I have verified Claims 1 and 2 and the three nonedge cases. The overall argument is coherent and the central claims are sound, but two printed subgraph identifications in the lower-bound proofs are incorrect as written and need local corrections.","major_comments":[{"comment":"The claim that {N/q, N/q^2} ∪ (B\\{N''}) induces K2 ∨ Γ_E(Z_N') is false. The two vertices N/q and N/q^2 are nonadjacent: their product is N'·q, which is not divisible by N = N'·q^2. The induced graph is instead the join of Γ_E(Z_N') with two universal nonadjacent vertices, i.e. (complement of K2) ∨ Γ_E(Z_N'). With the printed K2 reading, Observation 2.8 would give box = 0 + box(Γ_E(Z_N')) = |P|+2 = a−1, not the required lower bound a. With the corrected reading, the complement of K2 on two vertices has boxicity 1, so Observation 2.8 gives 1 + (|P|+2) = a, which restores the intended lower bound. This is a one-line correction, but as written the proof is internally inconsistent.","section":"Section 4, Lemma 4.9, final paragraph"},{"comment":"The verification states that S'_k ∪ S_i and S'_k ∪ S'_k' induce the graph 2K2. A direct exponent check shows that all four cross edges are present, so the induced graph is K_{2,2} (a C4), not 2K2. With the printed claim the union of the pairs would not form the Roberts graph aK2, and the lower bound box(Γ_E(Z_N)) ≥ a would not follow. Replacing '2K2' by 'C4' (or K_{2,2}) makes the intended Roberts-graph construction correct.","section":"Section 4, Lemma 4.5"}],"minor_comments":[{"comment":"Claims 1 and 2 are asserted to be checkable by inspection. I did check them, and they are correct, but a brief indication of how the interval mapping enforces the adjacency conditions would improve readability and make the verification reproducible.","section":"Section 4, Lemma 4.10"},{"comment":"The formula N′ = 2·p_ℓ^{n_1}·N/(2^{n_1}·p_ℓ) is hard to parse. It would be clearer to describe the map as swapping the exponents of 2 and p_ℓ, and to state explicitly that this preserves Γ_E(Z_N) up to isomorphism.","section":"Section 4, Lemma 4.11"},{"comment":"In Claim 4 the sentence 'for r ∈ [a], S_r is an independent set' is followed by a classification of the induced graphs S_r ∪ S_{r'}. The logic that at most one S_r can be an independent set in a threshold supergraph is compressed; expanding it by one sentence would help.","section":"Section 5, Proof of Theorem 1.8"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a direct continuation of the authors' earlier work [21] and uses several results from it as black boxes. That is not circular, but the incremental novelty should be clear in the introduction: the main genuinely new tool is the C4-conflict-graph lower bound in Lemma 4.9 and the interval constructions in Lemma 4.10. The two errors in Lemmas 4.9 and 4.5 are local and easily corrected; I do not regard them as affecting the validity of the main theorem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper computes box(Γ_E(Z_N)) exactly, answering two open questions from the authors' earlier paper, and it derives a crisp corollary for the disjointness graph of the power set of an a-element set. The main theorem is genuinely new, and the proof introduces a lower-bound technique worth noting: the C4-conflict graph, where absent edges are colored by the interval representation, and the chromatic number of this conflict graph lower-bounds boxicity. That goes beyond the usual Roberts graph argument and is what makes the square-free case work.\n\nThe overall proof is a long case analysis and it is largely sound. The upper-bound constructions in Lemma 4.10, which the reader flagged, actually check out: the interval assignments are correct, and the 'easy to check by inspection' adjacencies are there, though not fully written out.\n\nThe one genuine problem is the last paragraph of Lemma 4.9. The paper claims that when N'' = q^2, the set {N/q, N/q^2} ∪ (B \\ {N''}) induces K2 ∨ Γ_E(Z_N'). It doesn't. N/q and N/q^2 are nonadjacent (their q-adic valuations are 1 and 0, so the product has valuation 1 < 2). What you get is the join of Γ_E(Z_N') with two nonadjacent universal vertices. Taken literally, the lower bound box ≥ a does not follow; the printed K2 claim would give box ≥ a-1. The fix is a one-liner: an edgeless pair joined to Γ' contributes 1 to boxicity, exactly the missing a. So the theorem survives, but the proof as written has a load-bearing misstatement in that subcase.\n\nA few other 'easy to check' claims are doing real work, but I didn't find another error. The citation pattern is not a problem; building on [21] and answering their own questions is fine.\n\nThe audience is narrow: people working on boxicity of algebraically defined graphs, or on interval representations of graph classes. They will get real value from the C4-conflict argument. This paper deserves a serious referee. The main result is very likely correct, the technique is reusable, and the fix is straightforward. I'd take it with a minor revision.","headline":"Genuinely new exact result with a reusable C4-conflict lower-bound technique, but Lemma 4.9 has a misstated join that needs a one-line fix.","tokens_in":19177,"tokens_out":3044,"would_cite":true,"duration_ms":29482,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":["05C62","13A99"],"pacs":[],"model":"deepseek-v4-flash","headline":"The compressed zero divisor graph of Z_N has boxicity a−1 exactly for three prime-exponent patterns, and a in every other nontrivial case.","keywords":["zero divisor graphs","compressed zero divisor graph","boxicity","threshold dimension","interval graphs","disjointness graphs","ring of integers modulo N"],"falsifier":"Take $N = p^3 q r$ for three distinct primes. Enumerate every pair of proper divisors $u,v$ of $N$ and check the interval assignments in Lemma 4.10: if any pair with $N \\nmid uv$ overlaps in every interval graph $I_i$ and $I_j$, then the claimed upper bound $box = a-1$ fails. The same computation should verify Claims 1 and 2 for all pairs, not merely pictorially.","tokens_in":18171,"feed_emoji":"📦","tokens_out":13704,"duration_ms":120865,"temperature":0.7,"pith_summary":"The paper determines the exact boxicity of the compressed zero divisor graph of the ring of integers modulo $N$, a graph whose vertices are equivalence classes of zero divisors sharing the same annihilator. It proves that for $N$ with $a$ distinct prime factors, the boxicity is $a-1$ precisely when the exponent pattern is one of three types: one cubed prime and the rest square-free; all exponents $1$ with $a \\geq 3$; or exponents only $1$ and $2$, with at least one of each. In all remaining cases with $a \\geq 2$ the boxicity is $a$, except the single case $N = pq$ where the graph is a clique and boxicity is $0$. This settles the two open questions posed in the preceding paper on zero divisor graphs, and yields the boxicity of the disjointness graph of the power set of an $a$-element set.","feed_headline":"Exact boxicity found for compressed zero-divisor graphs","feed_subtitle":"For Z_N, boxicity is 0, a−1, or a depending only on prime exponents, settling two open questions.","key_machinery":"The central object is $\\Gamma_E(\\mathbb{Z}_N)$, the graph on annihilator-equivalence classes of zero divisors of $\\mathbb{Z}_N$; in this ring the classes correspond to proper divisors of $N$, with adjacency meaning the product is divisible by $N$. The proof is carried by three mechanisms: the classical characterization (Theorem 2.1) that boxicity equals the minimum number of interval supergraphs whose edge intersection is the original graph; two lower-bound gadgets, the induced graph $aK_2$ obtained by removing a perfect matching from a complete graph on $2a$ vertices, and the $C_4$-conflict graph whose chromatic number forces boxicity from below; and, for the hardest upper bound, an explicit family of interval supergraphs indexed by the prime factors and built from the $p$-adic valuation function $f(u,q)$. The square-free case instead pushes boxicity down by forbidding certain induced subgraphs of interval graphs inside $\\Gamma_E(\\mathbb{Z}_N)$.","core_discovery":"For $N = \\prod_{i=1}^a p_i^{n_i}$ and $P = \\{i : n_i = 1\\}$, Theorem 1.7 states that $box(\\Gamma_E(\\mathbb{Z}_N)) = a-1$ if and only if (i) $|P| = a-1$ and the unique index outside $P$ has $n = 3$; (ii) $a \\geq 3$ and $P = [a]$; or (iii) $\\emptyset \\subsetneq P \\subsetneq [a]$ and every index outside $P$ has $n = 2$. The graph is a clique, with boxicity $0$, exactly when $a = 2$ and $n_1 = n_2 = 1$; all other cases with $a \\geq 2$ have boxicity $a$. The same proof gives Theorem 1.8: the threshold dimension is $a-1$ only in the square-free case with $a \\geq 3$, is $0$ for the complete graphs, and is $a$ otherwise. Consequentially, for any finite commutative reduced ring whose zero divisor graph has chromatic number $k \\geq 3$, the compressed graph has boxicity $k-1$, so the zero divisor graph itself has boxicity between $k-1$ and $k$, and the lower bound is tight.","pith_inferences":["The trichotomy suggests that boxicity is governed by the number of independent layers in the annihilator poset; a testable extension is whether an analogous exponent formula holds for other finite principal ideal rings, such as quotient rings of polynomial rings over finite fields.","The $C_4$-conflict graph coloring bound may be a general lower-bound technique for boxicity of algebraically defined graphs; one could check whether boxicity equals the chromatic number of the conflict graph for larger families than the modular rings treated here.","A concrete micro-test of the upper bound: for $N = p^3 q r$, locate which pair of non-adjacent vertices violates the interval representation if one perturbs the valuation-based interval lengths in Lemma 4.10, since the paper's proof of the two-cubed-primes case shows the $a-1$ representation must fail there.","Since the exponent pattern is a certificate of embedding dimension, algorithms that pre-process compressed zero divisor graphs could use the formula as an $O(a)$ structural test before attempting any geometric representation."],"forward_implications":["For any $N$, the boxicity of $\\Gamma_E(\\mathbb{Z}_N)$ can be read directly from the prime exponents of $N$, with no graph construction needed.","In the square-free case with $a \\geq 3$, the disjointness graph of the non-empty proper subsets of $[a]$ has boxicity $a-1$, and even the subgraph on the 1-element and 2-element sets already forces boxicity $a-1$.","The open question on the tightness of the lower bound for boxicity of zero divisor graphs of reduced rings is settled: for chromatic number $k \\geq 3$ the bound is $k-1$, not merely $\\lfloor k/2 \\rfloor$, and it is attained.","The threshold dimension of $\\Gamma_E(\\mathbb{Z}_N)$ is now exactly known in every case, matching the boxicity in the square-free and complete-graph cases and exceeding it by one elsewhere.","Adding the singleton sets to the disjointness graph of the 2-element subsets of $[a]$ raises its boxicity from $a-2$ to $a-1$."],"supporting_citations":[{"why":"Establishes the boxicity of the un-compressed zero divisor graph $\\Gamma(\\mathbb{Z}_N)$ and poses the two open questions answered here; its $a$ upper bound is reused throughout the proof.","marker":"[21]"},{"why":"Initiated the study of boxicity of zero divisor graphs and supplied the earlier general upper bound that the present paper refines for the compressed graph.","marker":"[29]"},{"why":"Provides the lemma identifying the compressed zero divisor graph of a reduced ring with that of a Boolean ring $\\mathbb{Z}_2^k$, the step that turns the modular result into a statement about all finite reduced rings.","marker":"[4]"},{"why":"States that boxicity equals the minimum number of interval supergraphs whose edge intersection is the graph; this is the framework for all upper-bound constructions.","marker":"[35]"},{"why":"Supplies the lower-bound lemma (with its corollary on 6-vertex obstructions) used to force boxicity beyond 1 in the forbidden-induced-subgraph arguments.","marker":"[26]"}],"fun_headline_variants":["Exact boxicity for compressed zero-divisor graphs of Z_N","Boxicity of Γ_E(Z_N) pinned: 0, a-1, or a","Two open questions solved: boxicity of compressed zero-divisor graphs","Compressed zero-divisor boxicity modulo N fully determined","For Z_N, compressed zero-divisor boxicity takes only three values"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The most delicate upper bound, in the case where $N$ has exactly one cubed prime factor and all others square-free, depends on the claim that the two interval graphs built by explicit formulas in Lemma 4.10 have exactly the adjacencies drawn and asserted in Figure 2; that claim is verified only by inspection.","fun_headline_variants_meta":{"raw":{"variants":["Exact boxicity for compressed zero-divisor graphs of Z_N","Boxicity of Γ_E(Z_N) pinned: 0, a-1, or a","Two open questions solved: boxicity of compressed zero-divisor graphs","Compressed zero-divisor boxicity modulo N fully determined","For Z_N, compressed zero-divisor boxicity takes only three values"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00109,"raw_usage":{"total_tokens":4771,"prompt_tokens":1382,"completion_tokens":3389,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":998,"completion_tokens_details":{"reasoning_tokens":3293}},"tokens_in":998,"tokens_out":3389,"duration_ms":26160,"temperature":1.0,"reasoning_tokens":3293,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-28T00:13:25.204999+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $N = p^3 q r$ for three distinct primes. Enumerate every pair of proper divisors $u,v$ of $N$ and check the interval assignments in Lemma 4.10: if any pair with $N \\nmid uv$ overlaps in every interval graph $I_i$ and $I_j$, then the claimed upper bound $box = a-1$ fails. The same computation should verify Claims 1 and 2 for all pairs, not merely pictorially.","supporting_citations":[{"cited_title":"Sunil Chandran and Suraj Kumar Sahoo","cited_arxiv_id":null,"evidence_quote":"Establishes the boxicity of the un-compressed zero divisor graph $\\Gamma(\\mathbb{Z}_N)$ and poses the two open questions answered here; its $a$ upper bound is reused throughout the proof."},{"cited_title":"Kavaskar","cited_arxiv_id":null,"evidence_quote":"Initiated the study of boxicity of zero divisor graphs and supplied the earlier general upper bound that the present paper refines for the compressed graph."},{"cited_title":"Anderson and John D","cited_arxiv_id":null,"evidence_quote":"Provides the lemma identifying the compressed zero divisor graph of a reduced ring with that of a Boolean ring $\\mathbb{Z}_2^k$, the step that turns the modular result into a statement about all finite reduced rings."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States that boxicity equals the minimum number of interval supergraphs whose edge intersection is the graph; this is the framework for all upper-bound constructions."},{"cited_title":"Cozzens and Fred S","cited_arxiv_id":null,"evidence_quote":"Supplies the lower-bound lemma (with its corollary on 6-vertex obstructions) used to force boxicity beyond 1 in the forbidden-induced-subgraph arguments."}],"review_version":1}