{"id":"bba370ee-a74b-43d5-b154-b06732b86004","arxiv_id":"2501.04999","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For almost all finite fields, there is a 3-primitive 1-normal polynomial of degree m with any prescribed last two coefficients, with a finite list of possible exceptions.","lead":"The authors prove a condition that guarantees a finite field contains a special polynomial whose last two coefficients can be chosen in advance, and they list the rare exceptions for a specific case. The result gives complete control over these coefficients for almost all field sizes, which is useful in constructions that need polynomials with prescribed trace and norm.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.7 is false as stated: the norm of every r-primitive element lies in the subgroup of F_q^* of order (q-1)/gcd(r,q-1), so \"for any prescribed b\" cannot hold; for q=7, r=3 the only attainable norm is -1.","rationale":"The reader's weakest assumption—that Lemma 4.7 and the computations are not independently verifiable—is a legitimate concern about the completeness of the finite exception lists. But it presupposes that the analytic theorem is valid. The norm-order identity above shows that Theorem 3.7 is false as stated: the norm of an r-primitive element is constrained to a proper subgroup whenever gcd(r, q-1) > 1. For the paper's main case r=3, q=7 and q=4, this subgroup is nontrivial (order 2 and order 1 respectively), so the claim 'for every b in F_q^*' is impossible. The proof's construction ξ = ε^r with N(ε)=c only yields N(ξ)=c^r, and no argument is given that every prescribed b admits such a c; indeed it generally does not. This is not a matter of external consensus or a missing code artifact; it is an internal contradiction with elementary cyclic-group structure. The character-sum estimates may be sound on their own, but they count objects with a different norm quantifier than the one stated in the theorem and used in the classification. A revision would need to restrict b to the subgroup of order (q-1)/gcd(r,q-1) and then re-check the Section 4 lists, especially the q=4 and q=7 cases; as written, the central existence claim and the resulting classification are false.","tokens_in":19484,"tokens_out":29943,"duration_ms":297186,"concrete_test":"Verify the norm-order identity: if ord(ξ) = (q^m-1)/r, then ord(N_{F_{q^m}/F_q}(ξ)) = (q-1)/gcd(r, q-1). Apply it with q=7, r=3, m=75: ord(ξ) = 2h with h = (7^75-1)/6, so N(ξ) = ξ^h has order 2 and therefore equals -1 in F_7; in particular b = 1 is impossible. This single identity settles that the quantification over b in Theorem 3.7 is false.","verdict_should_be":"REJECT","load_bearing_attack":"Let n = q^m - 1, h = n/(q-1), and w = gcd(r, q-1). If ord(ξ) = n/r, then N(ξ) = ξ^h. Writing q-1 = wL and r = wR with gcd(R,L)=1, the condition r | n forces R | h; hence h = Rh' and n/r = Lh'. Therefore gcd(n/r, h) = h', so ord(N(ξ)) = L = (q-1)/w. The attainable norms of r-primitive elements are exactly the elements of order (q-1)/w, not all of F_q^*. For q=7, r=3, w=3, so every 3-primitive element in every F_{7^m} has norm -1; b = 1 is never attained. The paper's Lemma 4.3 and the surrounding classification put (7,75) in Γ_{a,b}(3,1) for every b, and the argument uses Theorem 3.7, so the theorem cannot be right as quantified. The same obstruction applies to q=4, where w=3 and the only possible norm is 1. This is an internal algebraic contradiction, not merely a missing numerical estimate: the proof counts ε with N(ε)=c and then sets ξ = ε^r, but the required link c^r = b is absent from Theorem 3.7 and is often unsatisfiable.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies r-primitive k-normal elements over finite fields and seeks a sufficient condition for the existence of an r-primitive k-normal polynomial of degree m whose last two coefficients are prescribed. The main tool is a character-sum count, Theorem 3.7, which yields Inequality (8), and a sieve, Proposition 3.9, used to reduce the range in which direct verification is needed. The paper then specializes to r=3, k=1 and claims to determine all possible exceptional pairs (q,m) for m>=7, with separate statements for q<8 and q>=8, plus some results for m<=4. The character-sum derivation is internally detailed and contains no fitted parameters, but the paper's central quantification over all prescribed norms is algebraically incorrect, and the computational classification is not reproducible from the text.","tokens_in":19756,"tokens_out":16946,"duration_ms":167730,"significance":"If corrected, the method would be a useful extension of prior work on primitive normal polynomials with prescribed last two coefficients to r-primitive k-normal polynomials. The derivation of the main character-sum inequality is a genuine strength, and the paper explicitly separates a sufficient condition from the sieve-based reduction. However, the advertised conclusion, namely existence for every prescribed b in F_q^*, is false: the norm of an r-primitive element is constrained to the subgroup of F_q^* of order (q-1)/gcd(r,q-1). This error propagates through Section 4, where several pairs are asserted to lie in Gamma_{a,b}(3,1) for all b. The exception lists also rely on an imported numerical bound and on undocumented finite scans. The contribution as submitted is therefore not acceptable, although a reformulation restricted to attainable norms may be salvageable.","major_comments":[{"comment":"The quantification 'for any prescribed a in F_q and b in F_q^*' in Theorem 3.7 is false. Write n=q^m-1, h=n/(q-1), w=gcd(r,q-1), r=wR, q-1=wL with gcd(R,L)=1. For an r-primitive element xi of order n/r, the condition r|n forces R|h; writing h=Rh' gives ord(xi)=Lh' and N(xi)=xi^h. Since gcd(Lh', h)=h', the norm has exact order L=(q-1)/w, not arbitrary order dividing q-1. Thus, for q=7 and r=3, every 3-primitive element has norm -1 in every extension, so b=1 is never attainable; consequently Lemma 4.3's assertion that (7,10) lies in Gamma_{a,b}(3,1) for every b is impossible. The proof counts epsilon with N(epsilon)=c for a primitive c, but the required link c^r=b is never imposed in the statement of Theorem 3.7; this is an algebraic obstruction rather than a missing estimate. The theorem and all Section 4 applications must be reformulated for b of exact order (q-1)/w, i.e. for b in the image of the r-th power map on primitive elements. Relatedly, under the Section 2 definition of 'l-free', Lemma 3.4's 'w-free' assertion is inconsistent: for q=31, m=1, r=3, a 3-primitive element has norm of order 10, yet gcd(3,30/10)=3.","section":"Sections 3.1 and 4"},{"comment":"The bound 'if omega(M) >= 9632 then W(M) < M^{1/15}' is imported without proof from [1] and is then used as a threshold to eliminate in one stroke every case with omega(q^m-1) >= 9632 and to justify the later reductions for 7 <= m <= 11. The paper does not state the exact lemma from [1] nor show the deduction of the constant 9632. Since a wrong threshold would change the exception lists in Lemmas 4.8 and 4.9, this point is load-bearing for the claimed exhaustive classification and needs either a self-contained proof or a precise statement of the cited result together with the derivation.","section":"Section 4.2"},{"comment":"The repeated phrases 'after verifying Inequality (10) and (11)' conceal finite computations that are essential to the exception lists. For example, Lemma 4.9 requires checking all prime powers q up to 780097 for m=7, and Lemmas 4.3 and 4.5 require similar checks for small q and m. The manuscript gives no code, no output tables, and no description of the specific choices of e and f in Proposition 3.9 for each pair. These scans should be made reproducible, for instance by including a Sage or PARI script and its output in an appendix or supplementary file; otherwise the classification cannot be audited.","section":"Sections 4.1-4.2"}],"minor_comments":[{"comment":"The sentence 'This gives us (q,9) in Gamma_{a,b}(3,1)' should read '(q,7) in Gamma_{a,b}(3,1)', since the lemma is about m=7.","section":"Section 4.2"},{"comment":"The term 'l-free' is defined for elements of F_{q^m}, but the same terminology is later applied to elements of F_q^* in Lemma 3.2 and Lemma 3.4; the analogous definition for F_q^* should be stated explicitly, since the two possible conventions (with respect to order or with respect to index) give different results.","section":"Section 2"},{"comment":"There are several typos and OCR artifacts, including 'abelain' in Section 2, 'integerdivide' in the display of Lemma 2.3, 'POL YNOMIALS' in the title, 'Fore these values' in Lemma 4.5, and 'if and only of it' in Section 2.","section":"Throughout"},{"comment":"The symbol r is used both for the order of primitivity and as a real parameter in Lemma 4.1; this overloading is confusing in the rows of Table 1, especially since the table uses real values such as r=8.5. A different symbol for the real exponent would improve readability.","section":"Section 4.2"}],"recommendation":"major_revision","confidential_remarks":"The norm obstruction in Theorem 3.7 is decisive and should be fixed before any further consideration. The good news is that the obstruction is algebraic and local: the theorem can be reformulated for b of exact order (q-1)/w, and the character-sum machinery may still go through. However, the current version overclaims substantially, and the exception-list computations are not auditable. If the authors are unwilling or unable either to restrict the main theorem to attainable norms or to supply reproducible computational evidence, I would recommend rejection rather than further revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main theorem is false as stated. For an r-primitive element, the norm lies in the subgroup of F_q^* of order (q-1)/gcd(r,q-1), not all of F_q^*. For r=3, q=7, that means only norms 1 and -1 are possible (the stress-test's claim that only -1 is attainable is wrong for m≥2, but the subgroup restriction stands). Theorem 3.7 asserts existence for every b ∈ F_q^*, which cannot hold. The proof itself quietly restricts to w-free b and then requires b = c^r with c primitive, which is narrower still—for q=7, b=1 never gets counted. That is an internal contradiction with the authors' own Lemma 3.4, and it breaks the Section 4 application: e.g., q=4, w=3, L=1, so only b=1 is ever a norm, and the paper claims (4,m) ∈ Γ_{a,b}(3,1) for all m except a short list.\n\nWhat survives? The exponential sum estimates (Lemmas 3.5 and 3.6) and the character-sum decomposition are genuinely new, and the constant accounting in Inequality (8) checks out. The idea to extend Fan-Wang to r-primitive k-normal with last two coefficients is a fair problem. But the main quantitative claim and all the exception lists are overclaims; they need to be re-stated with the norm restricted to the correct subgroup, and the enumeration would then look different.\n\nThe unproved Lemma 4.7 (ω(M) ≥ 9632 implies W(M) < M^{1/15}) is a separate concern—it is imported without proof from [1] and it props up the Section 4 reduction. The finite scans are also not reproducible: no code or data, just 'we verify Inequality (10)'. That would be a minor issue in a healthy paper; here it compounds the broken main theorem.\n\nWho is this for? Finite-field people working on primitive normal bases. They would get value from a corrected version, not this one. I would send it to a referee rather than desk reject, because the character-sum core is nontrivial and the mistake is instructive, but I would expect the referee to require a major rewrite: fix the norm subgroup quantifier, prove or properly import Lemma 4.7, and provide reproducible computation.","headline":"Norm obstruction kills Theorem 3.7 as stated; the character-sum machinery is real but the paper overclaims existence for all b.","tokens_in":20341,"tokens_out":12805,"would_cite":false,"duration_ms":112488,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["12E20","11T23"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves a sufficient condition that forces finite fields to contain r-primitive k-normal polynomials with any prescribed last two coefficients, and it lists the few possible exceptions for the 3-primitive 1-normal case.","keywords":["r-primitive elements","k-normal elements","finite fields","character sums","prescribed coefficients","trace and norm","prime sieve","primitive polynomials"],"falsifier":"Compute $M$ as the product of the first 9632 primes, count its square-free divisors ($2^{9632}$), and compare with $M^{1/15}$; if $2^{9632}$ is not smaller, then Lemma 4.7 is false and the exception lists for $7\\le m\\le11$ cannot be considered exhaustive.","tokens_in":19259,"feed_emoji":"🧮","tokens_out":15442,"duration_ms":118725,"temperature":0.7,"pith_summary":"The paper asks when a finite field $\\mathbb{F}_{q^m}$ contains an element whose minimal polynomial has two prescribed last coefficients, where the element is simultaneously $r$-primitive (multiplicative order $(q^m-1)/r$) and $k$-normal (its conjugates span a space of codimension $k$). The main result is a sufficient condition: if $q^{m/2-k-2}$ exceeds $r$ times a product of square-free divisor counts attached to $q^m-1$ and to $x^m-1$, then such an element exists for every prescribed pair of coefficients. Specializing to $r=3$, $k=1$, and degree $m\\ge 7$, the condition, together with a prime sieve, leaves only finitely many exceptional pairs $(q,m)$, all explicitly listed; for $m=1,2,3$ the paper shows no such polynomial exists, and for $m=4$ it gives a necessary condition. The interest is that prescribing the last two coefficients is equivalent to prescribing the norm and the trace of $\\xi^{-1}$, so the theorem produces irreducible polynomials with controlled coefficients, a tool relevant to coding and cryptography.","feed_headline":"One inequality forces prescribed-coefficient finite-field polynomials","feed_subtitle":"Large finite fields realize every prescribed pair of last coefficients by an r-primitive k-normal polynomial.","key_machinery":"The engine is a character-sum count $N_{r,k,a,b,c}(Q_e,f)$ for elements $\\epsilon$ that are $Q_e$-free, have $\\epsilon^r = g\\circ\\beta$ for an $f$-free $\\beta$, satisfy $\\operatorname{Tr}(\\epsilon^{-r})=ab^{-1}$ and $\\operatorname{Norm}(\\epsilon)=c$. Writing $r$-primitiveness as $\\xi=\\epsilon^r$ for primitive $\\epsilon$ (Lemma 3.3) and $k$-normality as $\\xi=g\\circ\\beta$ for normal $\\beta$ (Lemma 2.1), the paper expresses this count as a sum of four character sums $T_1,T_2,T_3,T_4$; an orthogonality identity (Lemma 2.6) evaluates the $\\beta$-sum, and Weil-type mixed exponential sum bounds (Lemma 2.4) estimate the terms, yielding Inequality (8). The prime sieve (Lemma 3.8 and Proposition 3.9) then removes some large prime divisors of $Q$ and some irreducible factors of $x^m-1$, producing the cheaper test Inequality (9) used to cover borderline pairs.","core_discovery":"The central claim is Theorem 3.7: with $q$ a prime power, $r\\mid q^m-1$, and $k r\\,W(Q)\\,W\\big((x^m-1)/g\\big)$, then for every $a\\in\\mathbb{F}_q$ and $b\\in\\mathbb{F}_q^*$ there is an $r$-primitive $k$-normal $\\xi\\in\\mathbb{F}_{q^m}$ with $\\operatorname{Tr}_{\\mathbb{F}_{q^m}/\\mathbb{F}_q}(\\xi^{-1})=ab^{-1}$ and $\\operatorname{Norm}(\\xi)=b$; because $k