Loading [MathJax]/extensions/TeX/mathchoice.js
Research article

Permutations involving squares in finite fields

  • Received: 12 August 2021 Revised: 28 December 2021 Accepted: 30 December 2021 Published: 14 April 2022
  • Let p be an odd prime and let Fp be the finite field of p elements. In 2019, Sun studied some permutations involving squares in Fp. In this paper, by the theory of local fields we generalize this topic to Fp2, which gives a partial answer to the question posed by Sun.

    Citation: Hai-Liang Wu, Li-Yuan Wang. Permutations involving squares in finite fields[J]. Electronic Research Archive, 2022, 30(6): 2109-2120. doi: 10.3934/era.2022106

    Related Papers:

    [1] Zehua Wang, Jinrui Guan, Ahmed Zubair . A structure-preserving doubling algorithm for the square root of regular M-matrix. Electronic Research Archive, 2024, 32(9): 5306-5320. doi: 10.3934/era.2024245
    [2] Jiafan Zhang . On the distribution of primitive roots and Lehmer numbers. Electronic Research Archive, 2023, 31(11): 6913-6927. doi: 10.3934/era.2023350
    [3] Xiuyun Guo, Xue Zhang . Determinants and invertibility of circulant matrices. Electronic Research Archive, 2024, 32(7): 4741-4752. doi: 10.3934/era.2024216
    [4] Jorge Garcia Villeda . A computable formula for the class number of the imaginary quadratic field $ \mathbb Q(\sqrt{-p}), \ p = 4n-1 $. Electronic Research Archive, 2021, 29(6): 3853-3865. doi: 10.3934/era.2021065
    [5] Qinlong Chen, Wei Cao . The number of rational points on a class of hypersurfaces in quadratic extensions of finite fields. Electronic Research Archive, 2023, 31(7): 4303-4312. doi: 10.3934/era.2023219
    [6] Fedor Petrov, Zhi-Wei Sun . Proof of some conjectures involving quadratic residues. Electronic Research Archive, 2020, 28(2): 589-597. doi: 10.3934/era.2020031
    [7] Kelvyn Bladen, D. Richard Cutler . Assessing agreement between permutation and dropout variable importance methods for regression and random forest models. Electronic Research Archive, 2024, 32(7): 4495-4514. doi: 10.3934/era.2024203
    [8] Guo-Niu Han . On the existence of permutations conditioned by certain rational functions. Electronic Research Archive, 2020, 28(1): 149-156. doi: 10.3934/era.2020009
    [9] Heesung Shin, Jiang Zeng . More bijections for Entringer and Arnold families. Electronic Research Archive, 2021, 29(2): 2167-2185. doi: 10.3934/era.2020111
    [10] Bin Han . Some multivariate polynomials for doubled permutations. Electronic Research Archive, 2021, 29(2): 1925-1944. doi: 10.3934/era.2020098
  • Let p be an odd prime and let Fp be the finite field of p elements. In 2019, Sun studied some permutations involving squares in Fp. In this paper, by the theory of local fields we generalize this topic to Fp2, which gives a partial answer to the question posed by Sun.



    Permutation is an important mathematical concept. Investigating permutations over finite fields is a classical topic in number theory, combinatorics and finite fields. Let g(x) be a polynomial over a ring R. We say that g(x) is a permutation polynomial if it acts as a permutation of all elements of the ring, i.e., the map

    xg(x)

    is a bijection over R. By the Lagrange interpolation formula it is easy to see that every permutation over a finite field is induced by a permutation polynomial (for the recent progress on permutation polynomial readers may refer to the survey paper [1]).

    Now we introduce some earlier work on this topic. Let p be an odd prime and let aZ with pa. Clearly fa(x)=ax is a permutation polynomial over Z/pZ=Fp. The famous Zolotarev lemma [2] says that the sign of the permutation on Fp induced by fa(x) coincides with the Legendre symbol (ap). This fact provides us with a different proof (see [3,4]) of the law of quadratic reciprocity. Later G. Frobenius [5] generalized Zolotarev's result to the Jacobi symbols. Readers may refer to [6,7] for more related information.

    Let k be a positive integer with gcd(k,p1)=1. Then clearly the polynomial gk(x)=xk is a permutation polynomial over Fp. The authors [8] determined the sign of this permutation induced by gk(x) via extending the method of Zolotarev. Moreover, with the tools in group representation theory, Duke and Hopkins [9] generalized this result to finite groups. They also gave the law of quadratic reciprocity on finite groups.

    Recently, Sun [10,11] studied some permutations involving squares in Fp. For example, let p=2n+1 be an odd prime and let b1,,bn be the sequence of all the n quadratic residues among 1,,p1 in ascending order. Then it is easy to see that the sequence

    ¯12,,¯n2, (1.1)

    is a permutation τp of

    ¯b1,,¯bn, (1.2)

    where ¯a denotes the element a mod pZ for each aZ. Sun showed that

    sgn(τp)={1if p3(mod8),(1)(h(p)+1)/2if p7(mod8),

    where h(p) is the class number of Q(p) and sgn(τp) denotes the sign of τp. While studying this topic, Sun and his collaborator [10,12] also determined some products which concerns pth roots of unity. For instance, in the case p3(mod4) Sun [10] obtained

    0<j<k<p/2(ζj2pζk2p)={(p)(p3)/8if 8p3,(1)p+18+h(p)12p(p3)/8iif 8p7. (1.3)

    Later Petrov and Sun [12] showed that if p1(mod8), then

    0<j<k<p/2(ζj2p+ζk2p)=(1)#{1k<p4:(kp)=1}

    and that if p5(mod8), then

    0<j<k<p/2(ζj2p+ζk2p)=(1)#{1k<p4:(kp)=1}εh(p)p,

    where #S denotes the cardinality of a set S and h(p) is the class number of Q(p). These products have close connections with permutations over Fp. Readers may consult [10,12] for details.

    Along this line, the first author [13] determined the sign of τp in the case p1(mod4). Motivated by Sun's work, the first author also studied some permutations on Fp involving primitive roots modulo p. In fact, let gpZ be a primitive root modulo p. Then the sequence

    ¯g2p,¯g4p,,¯gp1p (1.4)

    is a permutation on the sequence (1.2). In [13] the first author gave the sign of this permutation in the case p1(mod4).

    Recently Sun posed the following problem:

    In an arbitrary finite field Fq with 2q, can we get an analogue of the above permutation which involves non-zero squares over Fq?

    In this paper, we mainly generalize the above permutations to Fp2. To do this, we first need to construct two sequences of non-zero squares in Fp2 which are analogues of the sequences (1.1) and (1.4). We now introduce some notations and some basic facts involving local fields.

    Let p=2n+1 be an odd prime, and let ζp21 be a primitive (p21)th root of unity in the algebraic closure ¯Qp of Qp. By [14,p.158 Propositon 7.12] it is easy to see that [Qp(ζp21):Qp]=2 and that the integral closure of Zp in Qp(ζp21) is Zp[ζp21]. Noting that pZp is unramified in Qp(ζp21), we therefore obtain Zp[ζp21]/pZp[ζp21]Fp2. Let Δ3(mod4) be an arbitrary quadratic non-residue modulo p in Z. Then clearly p is inert in the field Q(Δ). Hence Z[Δ]/pZ[Δ]Fp2. Since Qp(ζp21) and Qp(Δ) are both quadratic unramified extensions of Qp, by the local existence theorem (cf. [14,p.321 Theorem 1.4]) we have

    Qp(ζp21)=Qp(Δ).

    By the structure of the unit group of a local field (cf. [14,p.136,Proposition 5.3]) we have

    Zp[ζp21]×=ζp21×(1+pZp[ζp21]),

    where Zp[ζp21]× denotes the group of all invertible elements in Zp[ζp21] and ζp21={ζkp21:kZ}. Hence we can let gZp[ζp21] be a primitive root modulo pZp[ζp21] with gζp21(modpZp[ζp21]). For all xZ[Δ] and yZp[ζp21] we use the symbols ˉx and ˉy to denote the elements x mod pZ[Δ] and y mod pZp[ζp21] respectively.

    Set ak=k+Δ for 0kp1. Then it is easy to verify that

    {a2kj2:0kp1,1jn}{j2:1jn}

    is a complete system of representatives of

    (Z[Δ]/pZ[Δ])×2:={α2+pZ[Δ]:αZ[Δ]pZ[Δ]}.

    By the isomorphism

    Z[Δ]/pZ[Δ]Zp[ζp21]/pZp[ζp21]

    which sends x mod pZ[Δ] to x mod pZp[ζp21], we can view the sequence

    S:=¯a2012,¯a2022,,¯a20n2,,¯a2p1,,¯a2p1n2,,¯12,,¯n2 (1.5)

    as a permutation πp of the sequence

    S:=¯g2,¯g4,,¯gp21. (1.6)

    Clearly the above two sequences are analogues of the sequences (1.1) and (1.4). We mainly study this permutation in this paper. To state our result, let β0{0,1} be the integer satisfying

    (1)β0(Δ)p12ζp214p21(modpZp[ζp21]). (1.7)

    We also use the symbol sgn(πp) to denote the sign of πp. Now we state the main result of this paper.

    Theorem 1.1.

    sgn(πp)={(1)β0+p+34if p1(mod4),(1)h(p)+12+β0if p3(mod4) and p>3,(1)1+β0if p=3,

    where h(p) is the class number of Q(p).

    The detailed proof of the above theorem will be given in next section.

    Recall that ak=k+Δ for k=0,1,,p1. We begin with several lemmas involving ak. For convenience, we write p=2n+1 and pZ[Δ]=p in this section.

    Lemma 2.1. Let Ap=0kp1ak. Then

    An(n1)p{Δn2(modp)if p1(mod4),(1)n12(modp)if p3(mod4).

    Proof. Since

    0tp1(x+t)xpx(modpZ[x]),

    we have

    An(n1)p=0tp1(Δ+t)n(n1)(2Δ)n(n1)(modp).

    Observing that (Δ)p11(modp), one may get the desired result.

    Lemma 2.2. Let Bp=0kp1(1ap1k). Then

    Bnp1(modp).

    Proof. For each k=0,,p1 we have

    apk=(k+Δ)pk+(Δ)p1ΔkΔ(modp). (2.1)

    Hence we have the following congruences

    Bnp0kp1(1kΔk+Δ)n=2pn(Δ)2n2+n1kn(1k+Δ)n(1pk+Δ)n(2p)1kn(1Δk2)n(modp).

    Noting that

    1kn(xk2)xn1(modpZ[x]), (2.2)

    we obtain

    1kn(1Δk2)n(2p)(modp).

    Hence

    Bnp1(modp).

    Lemma 2.3. Let Cp=0<s<t<p1(t+Δ)(s+Δ). Then

    Cnp(2p)(modp).

    Proof. Clearly we have

    Cp=1s<tn1(t+Δ)(s+Δ)1(pt+Δ)(ps+Δ)×1sn1tn1(pt+Δ)(s+Δ).

    Hence we obtain that Cnp mod p is equal to

    1s<tn(Δt2p)(Δs2p)×1s,tn(1(Δt)(Δ+s))n(modp).

    We first handle the product

    1sn1tn(1(Δt)(Δ+s))n(modp).

    Noting that

    1sn(x+s)1tn(xt)xp11(modpZ[x]),

    we therefore obtain

    1tn(Δt)21sn(Δ+s)(modp).

    Hence

    1sn1tn(1(Δt)(Δ+s))n(2p)n(modp). (2.3)

    We now turn to the product

    1s<tn(Δt2p)(Δs2p).

    Let np=#{(x2,y2):1x,yn,x2+y2Δ(modp)}. Then one can easily verify that

    np={n/2if 4p1,(n+1)/2if 4p3. (2.4)

    Let np=#{(x2,y2):1x,yn,x2+Δy2Δ(modp)}. Then

    np={n/2if p1(mod4),(n1)/2if p3(mod4). (2.5)

    By the above we get

    #{(s,t):1s<tn,(Δt2p)(Δs2p)=1}={n24if 4p1,n214if 4p3.

    Therefore we have

    1s<tn(Δt2p)(Δs2p)={(1)n/2if p1(mod4),1if p3(mod4). (2.6)

    Now our desired result follows from (2.3) and (2.6).

    Lemma 2.4. Let Dp=0s<tp1(ap1tap1s). Then Dnp(modp) is equal to

    {(Δ)n2(modp)if p1(mod4),(Δ)n2(1)h(p)+12(2p)(modp)if p3(mod4) and p>3,(Δ)1(modp)if p=3.

    Proof. From (2.1) one may easily verify that Dnp(modp) is equal to

    (tΔt+ΔsΔs+Δ)n0s<tp1(2Δ(ts)(t+Δ)(s+Δ))n(modp).

    We further obtain

    Dnp(2p)n+1(1Δ)n2Cnp0<t<p(1t+Δ)n0<s<t<p(ts)n(modp).

    We first handle the product

    1tp1(1t+Δ)n.

    By (2.2) we have

    1tp1(1t+Δ)n1tn(1Δt2)n(2p)(modp). (2.7)

    We turn to the product

    1s<tp1(ts)n.

    It is clear that

    1s<tp1(ts)n(modp)

    is equal to

    1s<tn(tsp)(s+tp)1sn1tn(1p)(t+sp)(1)n1sn1tn(t+sp)(modp).

    We now divide our proof into the following two cases.

    Case 1. p1(mod4).

    Let 1wn be an arbitrary quadratic non-residue modulo p. Then

    #{(s,t):1s,tn,s+tw(modp)}=w1

    and

    #{(s,t):1s,tn,s+tpw(modp)}=w.

    Hence when p1(mod4) we have

    1sn1tn(t+sp)=(1)#{1wn:(wp)=1}=(1)n/2. (2.8)

    Case 2. p3(mod4).

    Let 1wn be an arbitrary quadratic non-residue modulo p and let 1vn be an arbitrary quadratic residue modulo p. Then

    #{(s,t):1s,tn,s+tw(modp)}=w1

    and

    #{s,t):1s,tn,s+tpv(modp)}=v.

    Hence

    1sn1tn(t+sp)=(1)#{1wn:(wp)=1}(1)p218.

    For each p3(mod4), let h(p) be the class number of Q(p). When p>3, by the class number formula (cf. [15,Chapter 5]) we have

    (2(2p))h(p)=n2#{1wn:(wp)=1}.

    By this one may easily verify that

    #{1wn:(wp)=1}h(p)+12(mod2).

    The readers may also see Mordell's paper [16] for details.

    By the above, we obtain

    1sn1tn(t+sp)={(1)h(p)+12(2p)if p3(mod4) and p>3,1if p=3. (2.9)

    In view of the above, we obtain the desired result.

    Let Φp21(x)Z[x] denote the (p21)th cyclotomic polynomial. We also let

    F(x)=1s<t(p21)/2(x2tx2s),

    and let

    T(x)=(1)p2+78(p212)p214x(p21)4Z[x].

    Let ζ=e2πi/(p21). The following result gives the explict value of F(ζ). As this result is the key element in the proof of our main result, we state this result as an individual theorem.

    Theorem 2.5. Let ζ=e2πi/(p21) be a primitive (p21)th root of unity. Then

    F(ζ)=i(1)p2+78(p212)p214.

    Hence Φp21(x)F(x)T(x) in Z[x].

    Proof. It is sufficient to prove that F(ζ)=T(ζ). We first compute F(ζ)2. We have the following equalities:

    F(ζ)2=1s<tp212(ζ2tζ2s)2=(1)(p21)(p23)81stp212(ζ2tζ2s)=1tp212xp2121xζ2t|x=ζ2t=(p212)p2121tp212ζ2t=1(p212)p212.

    Hence F(ζ)=±i(p212)p212. We now compute the argument of F(ζ). Note that for any 1s<t(p21)/2 we have

    ζ2tζ2s=ζt+s(ζtsζ(ts)).

    We therefore obtain

    Arg(ζ2tζ2s)=2πp21(t+s)+π2.

    By this we have

    Arg(F(ζ))=1s<tp212(2πp21(t+s)+π2)=(p21)(p23)π16+2πp211s<tp212(t+s)π2+p218π(mod2πZ).

    Therefore

    F(ζ)=i(1)p2+78(p212)p214=T(ζ).

    This completes the proof.

    Before the proof of our main result, we first observe the following fact. Let S={α1,,αn} be an arbitrary subset of a finite field and let τ be a permutation on S. Then it follows from definition that

    sgn(τ)=1s<tnτ(αt)τ(αs)αtαs.

    Hence

    sgn(πp)=1s<tn¯g2t¯g2sπp(¯g2t)πp(¯g2s).

    The next two propositions handle the numerator and the denominator respectively.

    Proposition 2.6. Set P=pZp[ζp21]. Then

    1s<tp212(g2tg2s)(2p)(2p)p+12gp214(modP). (2.10)

    Proof. Clearly Φp21(x)modpZp[ζp21][x] splits completely in (Zp[ζp21]/P)[x]. As gζp21(modP), by Theorem 2.5 we see that

    1s<tp212(g2tg2s)(modP)

    is equal to

    (2p)(p212)p214gp214(2p)(2p)p+12gp214(modP).

    This completes the proof.

    We now turn to the denominator.

    Proposition 2.7.

    1s<tp212(πp(g2t)πp(g2s))(modp)

    is equal to

    {Δp14(Δ)(p1)24(modp)if p1(mod4),(1)h(p)12(Δ)(p1)24(modp)if p3(mod4) and p>3,(Δ)1(modp)if p=3. (2.11)

    Proof. It is easy to verify that

    1s<tp212(πp(g2t)πp(g2s))(modp)

    is equal to

    An(n1)pBnpDnp1s<tn(t2s2)2(modp).

    By [10,(1.5)] we have

    1s<tn(t2s2)2(1)n+1(modp).

    By the above we obtain that

    1s<tp212(πp(g2t)πp(g2s))(modp)

    is equal to

    {Δp14(Δ)(p1)24(modp)if p1(mod4),(1)h(p)12(Δ)(p1)24(modp)if p3(mod4) and p>3,(Δ)1(modp)if p=3.

    This completes the proof.

    Combining the above two propositions, we now state the proof of our main result.

    Proof of Theorem 1.1. Set Δζαp21(modP) for some αZ. Since (Δ)p11(modP), we obtain

    (p1)αp212(modp21).

    Hence

    αp+12(modp+1).

    Set α=p+12+(p+1)β for some βZ. Then

    (Δ)nζp214p21ζp212βp21(modP).

    By this we obtain

    (1)β(Δ)nζp214p21(modP).

    Hence ββ0(mod2), where β0 is defined as in (1.7). We divide the remaining proof into three cases.

    Case 1. p=3.

    In this case by (2.10) and (2.11) it is easy to see that

    sgn(π3)=(1)1+β0.

    Case 2. p1(mod4).

    By (2.10) and (2.11) we have

    sgn(πp)gp214+p12α+(p1)24α(modP).

    Replacing α by p+12+(p+1)β and noting that gp2121(modP), we obtain that when p1(mod4)

    sgn(πp)=(1)β0+p+34.

    Case 3. p3(mod4) and p>3.

    Similar to the Case 2, we have

    sgn(πp)(2p)gp214(1)h(p)+12g(p1)24α(modP).

    Then via a computation we obtain

    sgn(πp)=(1)h(p)+12+β0.

    In view of the above, we complete the proof.

    The first author was supported by the National Natural Science Foundation of China (Grant No. 12101321, Grant No. 11971222) and the Natural Science Foundation of the Higher Education Institutions of Jiangsu Province (Grant No. 21KJB110002). The second author was supported by the Natural Science Foundation of the Higher Education Institutions of Jiangsu Province (Grant No. 21KJB110001).

    The authors declare there is no conflicts of interest.



    [1] X.-D. Hou, Permutation polynomials over finite fields-A survey of recent advances, Finite Field Appl., 32 (2015), 82–119. https://doi.org/10.1016/j.ffa.2014.10.001 doi: 10.1016/j.ffa.2014.10.001
    [2] G. Zolotarev, Nouvelle démonstration de la loi de réciprocité de Legendre, Nouvelles Ann. Math., 11 (1872), 354–362.
    [3] M. Riesz, Sur le lemme de Zolotareff et sur la loi de réciprocité des restes quadratiques, Math. Scand., 1 (1953), 159–169. https://doi.org/10.7146/math.scand.a-10376 doi: 10.7146/math.scand.a-10376
    [4] M. Szyjewski, Zolotarev's proof of Gauss reciprocity and Jacobi symbols, Serdica Math. J., 37 (2011), 251–260.
    [5] G. Frobenius, Über das quadratische Reziprozit¨datsgesetz I, Königliche Akademie der Wissenschaften, 1914,335–349.
    [6] A. Brunyate, P. L. Clark, Extending the Zolotarev-Frobenius approach to quadratic reciprocity, Ramanujan J., 37 (2015), 25–50. https://doi.org/10.1007/s11139-014-9635-y doi: 10.1007/s11139-014-9635-y
    [7] R. E. Dressler, E. E. Shult, A simple proof of the Zolotarev-Frobenius theorem, Proc. Amer. Math. Soc., 54 (1976), 53–54. https://doi.org/10.1090/S0002-9939-1976-0389732-8 doi: 10.1090/S0002-9939-1976-0389732-8
    [8] L.-Y. Wang, H.-L. Wu, Applications of Lerch's theorem to permutations of quadratic residues, Bull. Aust. Math. Soc., 100 (2019), 362–371. https://doi.org/10.1017/S000497271900073X doi: 10.1017/S000497271900073X
    [9] W. Duke, K. Hopkins, Quadratic reciprocity in a finite group, Amer. Math. Monthly, 112 (2005), 251–256. https://doi.org/10.1080/00029890.2005.11920190 doi: 10.1080/00029890.2005.11920190
    [10] Z.-W. Sun, Quadratic residues and related permutations and identities, Finite Fields Appl., 59 (2019), 246–283. https://doi.org/10.1016/j.ffa.2019.06.004 doi: 10.1016/j.ffa.2019.06.004
    [11] Z.-W. Sun, On quadratic residues and quartic residues modulo primes, Int. J. Number Theory, 16 (2020), no. 8, 1833–1858. https://doi.org/10.1142/S1793042120500955 doi: 10.1142/S1793042120500955
    [12] F. Petrov, Z.-W. Sun, Proof of some conjecture involving quadratic residues, Electron. Res. Arch., 28 (2020), 589–597. https://doi.org/10.3934/era.2020031 doi: 10.3934/era.2020031
    [13] H.-L. Wu, Quadratic residues and related permuations, Finite Fields Appl., 60 (2019), Article 101576. https://doi.org/10.1016/j.ffa.2019.101576 doi: 10.1016/j.ffa.2019.101576
    [14] J. Neukirch, Algebraic Number Theory, Springer-Verlag Berlin Heidelberg, 1999. https://doi.org/10.1007/978-3-662-03983-0
    [15] Z. I. Borevich, I. R. Shafarevich, Number Theory, Academic Press, 1966.
    [16] L. J. Mordell, The congruence ((p1)/2)!±1(modp), Amer. Math. Monthly, 68 (1961), 145–146. https://doi.org/10.2307/2312481 doi: 10.2307/2312481
  • Reader Comments
  • © 2022 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1180) PDF downloads(64) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog