Conjecture on generators with nonsquare quadratic translate over finite fields of odd characteristic

From papers

Let K\mathbb{K} be a finite field of characteristic different from 22, and let K\mathbb{K}^{*} denote its multiplicative group. An element xx is a generator if it generates K\mathbb{K}^{*}, and an element is a square if it is a square in K\mathbb{K}. The finite-field generator–nonsquare conjecture. There exists a generator xx of K\mathbb{K}^{*} such that x2+4Kx^{2}+4_{\mathbb{K}} is not a square in K\mathbb{K}. This is presented as a stronger version of the preceding prime-field conjecture; the source gives no proof or resolution.

Progress summary

Open

A 2025 paper proposes that every finite field of odd characteristic has a multiplicative generator whose square plus four is a nonsquare, but gives no proof or resolution.

The conjecture asks whether, for every finite field K\mathbb{K} with odd characteristic, some generator xx of K\mathbb{K}^{*} satisfies that x2+4x^{2}+4 is not a square. Flavien Mabilat presents it as a stronger finite-field version of a preceding prime-field conjecture.

October 2025 formulation

Mabilat’s paper states the conjecture but explicitly provides no proof or resolution. No retrieved source reports a counterexample, proof, verification, or substantive subsequent progress.

Current status (as of August 2026): The conjecture remains open, with its formulation recorded in Mabilat’s 2025 paper and no publicly verified progress found.

Sources
Sources & referencesView supporting material

Primary source

Flavien Mabilat, “Étude de quelques familles de λ-quiddités et minoration de la taille maximale des λ-quiddités irréductibles sur un corps fini”, arXiv:2510.09219 (2025).

Solutions 1

Proof

Primitive elements with a nonsquare quadratic translate

The statement

For every odd prime power qq, there exists gFq×g\in\mathbb F_q^\times of multiplicative order q1q-1 such that

χq(g2+4)=1,\chi_q(g^2+4)=-1,

where χq(0)=0\chi_q(0)=0, and χq(u)=u(q1)/2{1,1}\chi_q(u)=u^{(q-1)/2}\in\{1,-1\} for u0u\ne0. In particular, g2+4g^2+4 is not zero and is not a square.

This proves Conjecture 2 of Mabilat, arXiv:2510.09219v1, and its prime-field specialization proves Conjecture 1. The generator required here is a generator of the multiplicative group, not merely of the field extension. Neither conjecture excludes q=3q=3 or imposes the restrictions on a parameter appearing in the preceding quiddity theorem.

1. Reduction to 24 fields by an established theorem

We use Booker, Cohen, Sutherland and Trudgian, Primitive values of quadratic polynomials in a finite field, Theorem 1 and equation (4), arXiv:1803.01435v2, published in Mathematics of Computation 88 (2019), 1903–1912, DOI 10.1090/mcom/3390.

Their theorem states that, for a finite field Fq\mathbb F_q and a quadratic

Q(X)=aX2+bX+c,a0,b24ac0,\begin{gathered} Q(X)=aX^2+bX+c,\\ a\ne0,\qquad b^2-4ac\ne0, \end{gathered}

there is a primitive element gg for which Q(g)Q(g) is also primitive, whenever qq is outside their displayed finite exception list. Its odd entries are exactly

E={3,5,7,9,11,13,19,23,25,29,31,37,41,43,49,61,67,71,73,79,121,127,151,211}.(1)\begin{aligned} E=\{&3,5,7,9,11,13,19,23,\\ &25,29,31,37,41,43,49,61,\\ &67,71,73,79,121,127,151,211\}. \end{aligned} \tag{1}

The even entries 2,4,162,4,16 are irrelevant to the present statement. These are exceptions to the theorem's assertion uniformly over all coefficient triples; they are not assertions that the specific polynomial used here fails in those fields.

Apply the theorem with

Q(X)=X2+4.Q(X)=X^2+4.

The leading coefficient is 11, and its discriminant is 16-16, which is nonzero in every odd characteristic. Thus, when qEq\notin E, the theorem supplies primitive gg and primitive g2+4g^2+4.

Every primitive element uu of Fq×\mathbb F_q^\times, for odd qq, is a nonsquare. Indeed, its order is the even number q1q-1, so u(q1)/21u^{(q-1)/2}\ne1. The square of that element is 11; since the field has odd characteristic, it follows that u(q1)/2=1u^{(q-1)/2}=-1. This establishes the desired result for every qEq\notin E.

It remains to exhibit witnesses in the 24 fields in (1). The rest of the proof does exactly that; no extrapolation from a finite search is involved.

2. How to read the finite certificates

Mabilat already observes that g=2g=2 works whenever 22 is primitive in the proof of Proposition 6.9, and supplies witnesses for q=9,25,49,41q=9,25,49,41 in the proof of Proposition 6.16. He also reports computational verification of Conjecture 1 for every prime from 33 to 2,000,0002{,}000{,}000 in §6.2, referring to Appendix D. The following tables provide explicit certificates for the entire set EE.

Put n=q1n=q-1. A nonzero field element gg has order nn if, for every prime divisor \ell of nn,

gn/1.(2)g^{n/\ell}\ne1. \tag{2}

To justify the criterion, the order dd of gg divides nn by Lagrange's theorem. If d<nd<n, some prime \ell divides n/dn/d, and then dn/d\mid n/\ell, contrary to (2).

Each row below gives gg, an odd positive integer kk satisfying

g2+4=gk,(3)g^2+4=g^k, \tag{3}

and all the values required by (2). The notation :v\ell:v in the last column means gn/=vg^{n/\ell}=v. All the listed values differ from 11, and the listed \ell's are precisely the prime divisors of nn.

Consequently gg is primitive in every row. In particular, the entry for =2\ell=2 is 1-1, and (3), with odd kk, gives the exact quadratic-character certificate

(g2+4)(q1)/2=(g(q1)/2)k=(1)k=1.(4)\begin{aligned} (g^2+4)^{(q-1)/2} &=\bigl(g^{(q-1)/2}\bigr)^k\\ &=(-1)^k=-1. \end{aligned} \tag{4}

Thus the tables certify both required properties, including the exclusion of zero as a purported nonsquare.

3. The twenty prime fields

In this table every equality is computed modulo the prime qq.

qqggkkOrder tests
3322112:12:-1
5522332:12:-1
7733332:12:-1, 3:23:2
111122332:12:-1, 5:45:4
131322332:12:-1, 3:33:3
191922332:12:-1, 3:73:7
232377112:12:-1, 11:311:3
292922332:12:-1, 7:167:16
31313311112:12:-1, 3:253:25, 5:165:16
373722332:12:-1, 3:263:26
41417713132:12:-1, 5:375:37
434355552:12:-1, 3:363:36, 7:167:16
616122332:12:-1, 3:473:47, 5:95:9
676722332:12:-1, 3:373:37, 11:6411:64
71717723232:12:-1, 5:545:54, 7:457:45
73735535352:12:-1, 3:83:8
79797725252:12:-1, 3:553:55, 13:1813:18
1271276687872:12:-1, 3:1073:107, 7:647:64
1511517787872:12:-1, 3:323:32, 5:85:8
21121122332:12:-1, 3:1963:196, 5:1075:107, 7:1717:171

These are direct modular-power certificates, readily checked by repeated squaring. For example, modulo 3131,

315=1,310=25,36=16,311=13=32+4.\begin{aligned} 3^{15}&=-1,&3^{10}&=25,\\ 3^6&=16,&3^{11}&=13=3^2+4. \end{aligned}

Since 30=23530=2\cdot3\cdot5, these tests prove order 3030; the odd exponent 1111 proves the nonsquare property. The smallest field is covered as well: in F3\mathbb F_3, g=2=1g=2=-1 has order 22 and g2+4=2g^2+4=2.

4. The four quadratic extensions

Use the field models

Fp2=Fp[T]/(T2d),α=Tmod(T2d),\begin{gathered} \mathbb F_{p^2}=\mathbb F_p[T]/(T^2-d),\\ \alpha=T\bmod(T^2-d), \end{gathered}

with the following values of pp and dd:

qpd93225524973121112(5)\begin{array}{c|c|c} q&p&d\\ \hline 9&3&2\\ 25&5&2\\ 49&7&3\\ 121&11&2 \end{array} \tag{5}

These are fields: the square sets in the four prime fields are

p{a2:aFp}3{0,1}5{0,1,4}7{0,1,2,4}11{0,1,3,4,5,9}\begin{array}{c|l} p&\{a^2:a\in\mathbb F_p\}\\ \hline 3&\{0,1\}\\ 5&\{0,1,4\}\\ 7&\{0,1,2,4\}\\ 11&\{0,1,3,4,5,9\} \end{array}

In each case dd is absent, so T2dT^2-d has no root and is irreducible. Every element has a unique form a+bαa+b\alpha, and the arithmetic is explicitly

(a+bα)(c+eα)=(ac+dbe)+(ae+bc)α,(6)\begin{aligned} &(a+b\alpha)(c+e\alpha)\\ &\qquad=(ac+dbe)+(ae+bc)\alpha, \end{aligned} \tag{6}

with coefficients reduced modulo pp. Any field of the same cardinality is isomorphic to this model over its prime field; multiplicative order and the equation g2+4=gkg^2+4=g^k are preserved by that isomorphism.

qqggkkOrder tests
991+α1+\alpha332:12:-1
25252+α2+\alpha332:12:-1, 3:2+3α3:2+3\alpha
49491+α1+\alpha11112:12:-1, 3:43:4
1211212+α2+\alpha21212:12:-1, 3:5+10α3:5+10\alpha, 5:45:4

The translated values in these four rows are:

qqg2+4g^2+4
991+2α1+2\alpha
25254α4\alpha
49491+2α1+2\alpha
12112110+4α10+4\alpha

For instance, in F25\mathbb F_{25}, where α2=2\alpha^2=2,

g2=1+4α,g3=4α=g2+4,g8=2+3α,g12=1.\begin{aligned} g^2&=1+4\alpha,\\ g^3&=4\alpha=g^2+4,\\ g^8&=2+3\alpha,\\ g^{12}&=-1. \end{aligned}

As 24=23324=2^3\cdot3, the last two equalities establish that gg has order 2424, while the first two give the odd-power certificate. The other rows follow from the same explicit multiplication rule (6).

There is also an independent short character check for these four rows. For u=a+bαu=a+b\alpha,

NFp2/Fp(u)=a2db2,u(p21)/2=N(u)(p1)/2.\begin{aligned} N_{\mathbb F_{p^2}/\mathbb F_p}(u)&=a^2-db^2,\\ u^{(p^2-1)/2}&=N(u)^{(p-1)/2}. \end{aligned}

For the four displayed values of u=g2+4u=g^2+4, their norms are

qN(u)Fp922534931212\begin{array}{c|c} q&N(u)\in\mathbb F_p\\ \hline 9&2\\ 25&3\\ 49&3\\ 121&2 \end{array}

Their Euler powers are 2,4,6,102,4,6,10, respectively, which are exactly 1-1 in the corresponding prime fields. Hence the required quadratic character in each extension is exactly 1-1.

5. Conclusion

The published theorem handles every odd prime power outside (1), and the tables handle every element of (1). Thus every finite field of odd characteristic has a primitive element gg with g2+4g^2+4 nonsquare. Both Mabilat Conjectures 1 and 2 follow.

0 endorsements
Shivam Patel ·