This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Diophantine equations of the form Yn=f(X)Y^{n}=f(X) over function fields

Anwesh Ray Department of Mathematics
University of British Columbia
Vancouver BC, Canada V6T 1Z2
anweshray@math.ubc.ca
Abstract.

Let \ell and pp be (not necessarily distinct) prime numbers and FF be a global function field of characteristic \ell with field of constants κ\kappa. Assume that there exists a prime PP_{\infty} of FF which has degree 11, and let 𝒪F\mathcal{O}_{F} be the subring of FF consisting of functions with no poles away from PP_{\infty}. Let f(X)f(X) be a polynomial in XX with coefficients in κ\kappa. We study solutions to diophantine equations of the form Yn=f(X)Y^{n}=f(X) which lie in 𝒪F\mathcal{O}_{F}, and in particular, show that if mm and f(X)f(X) satisfy additional conditions, then there are no non-constant solutions. The results obtained apply to the study of solutions to Yn=f(X)Y^{n}=f(X) in certain rings of integers in p\mathbb{Z}_{p}-extensions of FF known as constant p\mathbb{Z}_{p}-extensions. We prove similar results for solutions in the polynomial ring K[T1,,Tr]K[T_{1},\dots,T_{r}], where KK is any field of characteristic \ell, showing that the only solutions must lie in KK. We apply our methods to study solutions of diophantine equations of the form Yn=i=1d(X+ir)mY^{n}=\sum_{i=1}^{d}(X+ir)^{m}, where m,n,d2m,n,d\geq 2 are integers.

1. Introduction

Let n2n\geq 2 be an integer and κ\kappa be a finite field of characteristic >0\ell>0. Let FF be a global function field with field of constants κ\kappa and assume that there exists a prime PP_{\infty} of FF of degree 11. In other words, we assume that there is a prime PP_{\infty}, which is totally inert in the composite κ¯F\bar{\kappa}\cdot F. The ring of integers 𝒪F\mathcal{O}_{F} shall consist of all functions fFf\in F with no poles away from PP_{\infty}. Given a polynomial f(X)f(X) with coefficients in κ\kappa, we study solutions to the diophantine equation Yn=f(X)Y^{n}=f(X) for which both XX and YY lie in 𝒪F\mathcal{O}_{F}. More precisely, we prove that if certain additional conditions are met, then there are no non-constant solutions, i.e., XX and YY must both belong to the field of constants κ\kappa. Equations of the form Yn=f(X)Y^{n}=f(X) are of significant interest. We shall apply our analysis to study a class of diophantine equations which involve perfect powers in arithmetic progressions. Let m,n,d2m,n,d\geq 2 be integers and let r1r\geq 1, then, there has been significant interest in the classification of integral solutions to the diophantine equation

Yn=(X+r)m+(X+2r)m++(X+dr)m,Y^{n}=(X+r)^{m}+(X+2r)^{m}+\dots+(X+dr)^{m},

cf. for instance, [BGP04, BZ13, Haj15, BPS17, BPSS18, KP18, Pat18, AGP20].

We also explore themes motivated by the Iwasawa theory of function fields. Mazur initiated the Iwasawa theory of elliptic curves over number fields (cf. [Maz72]), which had applications to the study of growth of Mordell–Weil ranks of elliptic curves in certain infinite towers of number fields. One hopes to extend such lines of investigation to curves of higher genus (cf. [Ray22]) and more generally, study the stability and growth of solutions to any diophantine equation in an infinite tower of global fields. In this paper, we shall study certain function field analogues of such questions, however, instead of elliptic curves, we consider the class of diophantine equations of the form Yn=f(X)Y^{n}=f(X), where f(X)f(X) has constant coefficients. Let us explain our results in greater detail. Given any integer nn, there is a unique extension κn/κ\kappa_{n}/\kappa such that Gal(κn/κ)\operatorname{Gal}(\kappa_{n}/\kappa) is isomorphic to /n\mathbb{Z}/n\mathbb{Z}. Given a prime pp, set κn(p)\kappa_{n}^{(p)} to denote κpn\kappa_{p^{n}} and set Fn(p)F_{n}^{(p)} to denote the composite Fκn(p)F\cdot\kappa_{n}^{(p)}. This gives rise to a tower of function field extensions

F=F0(p)F1(p)Fn(p)Fn+1(p).F=F_{0}^{(p)}\subset F_{1}^{(p)}\subset\dots\subset F_{n}^{(p)}\subset F_{n+1}^{(p)}\subset\dots.

Let p\mathbb{Z}_{p} denote the ring of pp-adic integers, i.e., the valuation ring of p\mathbb{Q}_{p}. The constant p\mathbb{Z}_{p}-extension of FF is the infinite union

F(p):=nFn(p).F_{\infty}^{(p)}\mathrel{\mathop{\mathchar 58\relax}}=\bigcup_{n}F_{n}^{(p)}.

It is easy to see that the Galois group Gal(F(p)/F)\operatorname{Gal}\left(F_{\infty}^{(p)}/F\right) is isomorphic to p\mathbb{Z}_{p}. Let hFh_{F} denote the class number of FF (cf. [Ros02, Chapter 5]). Note that since PP_{\infty} is assumed to have degree 11, it remains inert in Fn(p)F_{n}^{(p)} for all pp. Let 𝒪(p)\mathcal{O}_{\infty}^{(p)} (resp. 𝒪n(p)\mathcal{O}_{n}^{(p)}) be the ring of integers of F(p)F_{\infty}^{(p)} (resp. Fn(p)F_{n}^{(p)}), i.e., the functions fFf\in F_{\infty} with no poles away from PP_{\infty}. We now state the main result.

Theorem (Theorem 3.2).

Let \ell be a prime number and FF be a global function field of characteristic \ell. Let κ\kappa be the field of constants of FF, and let pp and qq be prime numbers that are not necessarily distinct. Assume that qq\neq\ell. Let f(X)f(X) be a polynomial with coefficients in κ\kappa satisfying following conditions.

  1. (1)

    The polynomial f(X)f(X) factorizes into

    f(X)=a0(Xa1)n1(Xat)nt,f(X)=a_{0}(X-a_{1})^{n_{1}}\dots(X-a_{t})^{n_{t}},

    where a0κa_{0}\in\kappa and a1,,ata_{1},\dots,a_{t} are distinct elements in κ\kappa, n1,,ntn_{1},\dots,n_{t} are positive integers and t2t\geq 2.

  2. (2)

    At least two of the exponents nin_{i} are not divisible by qq.

Then, the following assertions hold.

  1. (1)

    Suppose that pp and qq are distinct. Then, for all sufficiently large numbers k>0k>0, the only solutions (X,Y)(X,Y) to Yqk=f(X)Y^{q^{k}}=f(X) that are contained in 𝒪(p)\mathcal{O}_{\infty}^{(p)} are constant.

  2. (2)

    Suppose that phFp\nmid h_{F}, then the only solutions (X,Y)(X,Y) to Yp=f(X)Y^{p}=f(X) that are contained in 𝒪(p)\mathcal{O}_{\infty}^{(p)} are constant.

As a consequence of the above result, we find that for any prime pp, there are only finitely many numbers nn, that are not powers of \ell, for which Yn=f(X)Y^{n}=f(X) has solutions in F(p)F_{\infty}^{(p)}. A more specific criterion applies to any function field FF, cf. Theorem 2.4. The methods used in proving the above result are applied to study another question of independent interest. Let KK be a field of positive characteristic \ell and AA be the polynomial ring K[T1,,Tn]K[T_{1},\dots,T_{n}].

Theorem (Theorem 4.1).

With respect to above notation, let f(X)f(X) be a polynomial with all of its coefficients and roots in KK. Let qq\neq\ell be a prime number and assume that the following conditions are satisfied.

  1. (1)

    f(X)f(X) factorizes into

    f(X)=a0(Xa1)n1(Xat)nt,f(X)=a_{0}(X-a_{1})^{n_{1}}\dots(X-a_{t})^{n_{t}},

    where a0Ka_{0}\in K and a1,,ata_{1},\dots,a_{t} are distinct elements in KK, n1,,ntn_{1},\dots,n_{t} are positive integers and t2t\geq 2.

  2. (2)

    At least two of the exponents nin_{i} are not divisible by qq.

Then, any solution (X,Y)A2(X,Y)\in A^{2} to

(1.1) Yq=f(X)Y^{q}=f(X)

is constant, i.e., XX and YY are both in KK.

It follows from the above result that if n>1n>1 is not a power of \ell, then Xn=f(X)X^{n}=f(X) does not have non-constant solutions in AA.

Organization: Including the introduction, the manuscript consists of 55 sections. In section 2, we prove criteria for the constancy of solutions to Yn=f(X)Y^{n}=f(X) in global function fields FF. The main result proven in section 2 is Theorem 2.4. In section 3, we extend the results in section 2 to prove the constancy of solutions to the above equation in p\mathbb{Z}_{p}-extensions of FF. It is in this section that we prove the main result of the paper, i.e., Theorem 3.2. In section 4, we prove similar results for the polynomial rings over a field. Finally, in section 5, we study the specific case when f(X)=i=1k(X+ir)mf(X)=\sum_{i=1}^{k}(X+ir)^{m}.

2. Constancy of solutions to Yn=f(X)Y^{n}=f(X) in a global function field

In this section, we introduce basic notions and prove results about the solutions to certain diophantine equations over global function fields. Throughout this section, \ell be a prime number and AA be an integral domain of characteristic \ell with field of constants κ\kappa. We introduce the notion of a discrete valuation on AA.

Definition 2.1.

A function d:Ad\mathrel{\mathop{\mathchar 58\relax}}A\rightarrow\mathbb{Z} is said to be a discrete valuation if the following conditions are satisfied.

  1. (1)

    The values taken by dd are non-negative.

  2. (2)

    Let 𝟏\mathbf{1} be the identity element of AA, we have that d(𝟏)=0d(\mathbf{1})=0.

  3. (3)

    Given non-zero elements f,gAf,g\in A, we have that d(fg)=d(f)+d(g)d(fg)=d(f)+d(g),

  4. (4)

    d(f+g)max{d(f),d(g)}d(f+g)\leq\operatorname{max}\{d(f),d(g)\},

  5. (5)

    suppose that d(f)<d(g)d(f)<d(g), then d(f+g)=d(f)d(f+g)=d(f).

Let A0A_{0} be the subring of AA consisting of all elements aAa\in A for which d(a)0d(a)\leq 0. Given f,gAf,g\in A, we say that ff divides gg if fh=gfh=g for some hAh\in A. It is clear that if ff divides gg then d(f)d(g)d(f)\leq d(g).

Lemma 2.2.

Let qq be a prime number such that qq\neq\ell, and AA be an integral domain of characteristic \ell equipped with a function dd satisfying the conditions (1) to (5) of Definition 2.1. Let f,g,cAf,g,c\in A satisfy the equation

(2.1) fqgq=c.f^{q}-g^{q}=c.

Then, we have that d(f),d(g)d(c)d(f),d(g)\leq d(c). In particular, ff and gg are contained in A0A_{0} if cc is contained in A0A_{0}.

Proof.

Suppose by way of contradiction that d(f)>d(c)d(f)>d(c), or d(g)>d(c)d(g)>d(c). Assume first that d(f)>d(c)d(f)>d(c). Set e:=(gf)e\mathrel{\mathop{\mathchar 58\relax}}=(g-f), from (2.1), we find that ee divides cc. As a result, d(e)d(c)<d(f)d(e)\leq d(c)<d(f), hence by the property (5), we find that

d(g)=d(f+e)=d(f).d(g)=d\left(f+e\right)=d(f).

Therefore, we have deduced that d(g)>d(c)d(g)>d(c). Rewrite (2.1) as

(g+e)qgq=c,(g+e)^{q}-g^{q}=c,

and expand the left hand side of the above equation via the binomial expansion

qegq1+(q2)e2gq2++eq=c.qeg^{q-1}+{q\choose 2}e^{2}g^{q-2}+\dots+e^{q}=c.

Note that since d(e)<d(g)d(e)<d(g), we find that for all ii such that 2iq2\leq i\leq q,

d((qi)eigqi)<d(qegq1),d\left({q\choose i}e^{i}g^{q-i}\right)<d\left(qeg^{q-1}\right),

and therefore,

d(c)=d((g+e)qgq)=d(qegq1)=(q1)d(g).d(c)=d\left((g+e)^{q}-g^{q}\right)=d\left(qeg^{q-1}\right)=(q-1)d(g).

This implies that d(g)d(c)d(g)\leq d(c), a contradiction. On the other hand, if we assume that d(g)>d(c)d(g)>d(c) (instead of assuming that d(f)>d(c)d(f)>d(c)), the same argument applies. ∎

We shall illustrate the above result in various cases of interest. In this section, we study diophantine equations over global function fields FF. Let \ell be a prime number and denote by 𝔽\mathbb{F}_{\ell} the finite field with \ell elements (i.e. /\mathbb{Z}/\ell\mathbb{Z}). Fix an algebraic closure F¯\bar{F} of FF. Let κ\kappa be the algebraic closure of 𝔽\mathbb{F}_{\ell} in FF, and let κ¯\bar{\kappa} be the algebraic closure of κ\kappa in F¯\bar{F}. Set FF^{\prime} to denote the composite of FF with κ¯\bar{\kappa}.

Following [Ros02, Chapter 5], a prime vv of FF is by definition the maximal ideal of a discrete valuation ring 𝒪vF\mathcal{O}_{v}\subset F with fraction field equal to FF. A divisor of FF is a finite linear combination D=vnvvD=\sum_{v}n_{v}v of primes vv. In the above sum, nvn_{v} are all integers, and the set of primes vv for which nv0n_{v}\neq 0 is referred to as the support of DD. Given a function gFg\in F, denote by div(g)\operatorname{div}(g) the associated principal divisor. Note that any principal divisor has degree 0. Two divisors are considered equivalent if the differ by a principal divisor. The class group of FF is the group of divisor classes of degree 0, and has finite cardinality (cf. [Ros02, Lemma 5.6]). Denote by hFh_{F} the class number, i.e., the number of elements in the class group. Given a natural number NN, denote by hF[N]h_{F}[N] the cardinality of the NN-torsion in class group.

The field FF^{\prime} is identified with the field of fractions of a projective algebraic curve 𝔛\mathfrak{X} over κ¯\bar{\kappa}. A point w𝔛(κ¯)w\in\mathfrak{X}(\bar{\kappa}) is also referred to a prime of FF^{\prime}, since it corresponds to a valuation ring 𝒪wF\mathcal{O}_{w}\subset F^{\prime} with fraction field FF^{\prime}. Given a prime ww of FF^{\prime} and a prime vv of FF, we say that ww lies above (or divides) vv if the natural inclusion of fields FFF\hookrightarrow F^{\prime} induces an inclusion of valuation rings 𝒪v𝒪w\mathcal{O}_{v}\hookrightarrow\mathcal{O}_{w}. Given a function gFg\in F (resp. gFg\in F^{\prime}), denote by ordv(g)\operatorname{ord}_{v}(g) (resp. ordw(g)\operatorname{ord}_{w}(g)) the order of vanishing of gg at vv (resp. ww). We refer to dv(g):=ordv(g)d_{v}(g)\mathrel{\mathop{\mathchar 58\relax}}=-\operatorname{ord}_{v}(g) (resp. dw(g):=ordw(g)d_{w}(g)\mathrel{\mathop{\mathchar 58\relax}}=-\operatorname{ord}_{w}(g)) the order of the pole of gg at vv (resp. ww). Given a finite and nonempty set of primes SS of FF, the ring of SS-integers 𝒪S\mathcal{O}_{S} consists of all functions gFg\in F such that dv(g)0d_{v}(g)\leq 0 for all primes vSv\notin S. Let S¯\bar{S} be the set of primes of FF^{\prime} that lie above SS. Let AA denote the composite 𝒪Sκ¯\mathcal{O}_{S}\cdot\bar{\kappa}. A function gAg\in A satisfies the property that for all wS¯w\notin\bar{S}, dw(g)0d_{w}(g)\leq 0. According to our conventions, 𝒪F\mathcal{O}_{F} is the ring of SS integers where S:={P}S\mathrel{\mathop{\mathchar 58\relax}}=\{P_{\infty}\}. Since PP_{\infty} is a prime of degree 11, is totally inert in FF^{\prime}. By abuse of notation, the single prime in S¯\bar{S} is also denoted PP_{\infty}.

We list some basic properties of the function dwd_{w} on AA. The following result applies for any ring of S¯\bar{S}-integers.

Lemma 2.3.

Let ff and gg be a functions in AA and ww be a point in 𝔛(κ¯)\mathfrak{X}(\bar{\kappa}). Then, the following assertions hold.

  1. (1)

    Suppose that dw(g)0d_{w}(g)\leq 0 for all wS¯w\in\bar{S}, then, gg is a constant function,

  2. (2)

    dw(fg)=dw(f)+dw(g)d_{w}(fg)=d_{w}(f)+d_{w}(g),

  3. (3)

    suppose that dw(f)>dw(g)d_{w}(f)>d_{w}(g), then, dw(f+g)=dw(f)d_{w}(f+g)=d_{w}(f).

Proof.

Note that since gg is contained in AA, dw(g)0d_{w}(g)\leq 0 for all points wS¯w\notin\bar{S}. Therefore, the assumption that dw(g)0d_{w}(g)\leq 0 implies that gg has no poles, and thus must be a constant function. This proves part (1).

Part (2) clearly follows from the relation ordw(fg)=ordw(f)+ordw(g)\operatorname{ord}_{w}(fg)=\operatorname{ord}_{w}(f)+\operatorname{ord}_{w}(g).

For part (3), we note that f+g=f(1+g/f)f+g=f(1+g/f), and since it is assumed that dw(f)>dw(g)d_{w}(f)>d_{w}(g), it follows that g/fg/f vanishes at ww. As a result, dw(1+g/f)=0d_{w}(1+g/f)=0 and thus,

dw(f+g)=dw(f)+dw(1+f/g)=dw(f),d_{w}(f+g)=d_{w}(f)+d_{w}(1+f/g)=d_{w}(f),

which proves the result. ∎

Recall that PP_{\infty} is a prime of degree 11 and 𝒪F\mathcal{O}_{F} is the associated ring of integers in FF.

Theorem 2.4.

Let \ell be a prime number and FF be a global function field of characteristic \ell. Let 𝒪F\mathcal{O}_{F} be the ring of integers of FF. Denote by κ\kappa the field of constants of FF. Let f(X)f(X) be a polynomial with coefficients in κ\kappa. Let qq\neq\ell be a prime number and let k>0k>0 be the least integer such that hF[qk]=hF[qk1]h_{F}[q^{k}]=h_{F}[q^{k-1}]. Assume that the following conditions are satisfied.

  1. (1)

    f(X)f(X) factorizes into

    f(X)=a0(Xa1)n1(Xat)nt,f(X)=a_{0}(X-a_{1})^{n_{1}}\dots(X-a_{t})^{n_{t}},

    where a0κa_{0}\in\kappa and a1,,ata_{1},\dots,a_{t} are distinct elements in κ\kappa, n1,,ntn_{1},\dots,n_{t} are positive integers and t2t\geq 2.

  2. (2)

    At least two of the exponents nin_{i} are not divisible by qq.

Then, any solution (X,Y)(X,Y) to

(2.2) Yqk=f(X)Y^{q^{k}}=f(X)

for which X,Y𝒪FX,Y\in\mathcal{O}_{F} is constant, i.e., XX and YY are both in κ\kappa.

Proof.

Since the elements a1,,ata_{1},\dots,a_{t} are distinct elements of κ\kappa, we find that for all i,ji,j such that iji\neq j, (Xai)(Xaj)=ajai(X-a_{i})-(X-a_{j})=a_{j}-a_{i} is a non-zero element of κ\kappa. Therefore, for iji\neq j, the divisors div(Xai)\operatorname{div}(X-a_{i}) and div(Xaj)\operatorname{div}(X-a_{j}) have disjoint supports.

From the equation (5.3), we find that

(2.3) qkdiv(Y)=i=1tnidiv(Xai).q^{k}\operatorname{div}(Y)=\sum_{i=1}^{t}n_{i}\operatorname{div}(X-a_{i}).

Since the divisors div(Xai)\operatorname{div}(X-a_{i}) in the above sum have disjoint supports, and therefore for all ii, there are divisors DiD_{i}^{\prime} such that qkDi=nidiv(Xai)q^{k}D_{i}^{\prime}=n_{i}\operatorname{div}(X-a_{i}). Recall that it is assumed that there are two distinct indices ii and jj such that qniq\nmid n_{i} and qnjq\nmid n_{j}. Without loss of generality, assume that qn1q\nmid n_{1} and qn2q\nmid n_{2}. Therefore, there exist divisors D1D_{1} and D2D_{2} such that qkDi=div(Xai)q^{k}D_{i}=\operatorname{div}(X-a_{i}) for i=1,2i=1,2. The divisor classes [D1][D_{1}] and [D2][D_{2}] in the class group are in the qkq^{k}-torsion subgroup of the class group. Since hF[qk]=hF[qk1]h_{F}[q^{k}]=h_{F}[q^{k-1}], we find that qk1Diq^{k-1}D_{i} is principal for i=1,2i=1,2. Let ff and gg be functions in FF such that

div(f)=qk1D1 and div(g)=qk1D2.\operatorname{div}(f)=q^{k-1}D_{1}\text{ and }\operatorname{div}(g)=q^{k-1}D_{2}.

Thus we find that u1fq=(Xa1)u_{1}f^{q}=(X-a_{1}) and u2gq=(Xa2)u_{2}g^{q}=(X-a_{2}), where u1u_{1} and u2u_{2} are contained in κ\kappa. Note that since (Xai)(X-a_{i}) has no poles away from {P}\{P_{\infty}\}, the same is true for ff and gg, hence, f,g𝒪Ff,g\in\mathcal{O}_{F}. Setting A:=𝒪Fκ¯A\mathrel{\mathop{\mathchar 58\relax}}=\mathcal{O}_{F}\cdot\bar{\kappa}, we may replace ff by u11/qfu_{1}^{1/q}f and gg by u21/qgu_{2}^{1/q}g, and thus assume that fq=(Xa1)f^{q}=(X-a_{1}) and gq=(Xa2)g^{q}=(X-a_{2}) for some elements f,gAf,g\in A. We find that fqgq=a2a1f^{q}-g^{q}=a_{2}-a_{1} is a non-zero element of κ\kappa. The pair (A,dP)(A,d_{P_{\infty}}) satisfies the properties (1) to (5), and therefore dP(f)=dP(g)=0d_{P_{\infty}}(f)=d_{P_{\infty}}(g)=0 by Lemma 2.2. Therefore, by part (1) of Lemma 2.3, we find that ff and gg are both in κ\kappa. Hence, X=fq+a1X=f^{q}+a_{1} is in κ\kappa and thus so is YY. ∎

Remark 2.5.

We make the following observations.

  • Theorem 2.4 above implies that k=1k=1 if qhFq\nmid h_{F}.

  • If the roots aia_{i} are not contained in FF, we may base change FF by an extension κ\kappa^{\prime} of κ\kappa which is generated by the roots aia_{i}.

  • Suppose that f(X)f(X) satisfies the conditions of Theorem 2.4. Then, since qhFq\nmid h_{F} for all but finitely many primes qq, thus Yq=f(X)Y^{q}=f(X) has no non-constant solutions in FF for all but finitely many primes qq. In fact, it is easy to see that Theorem 2.4 implies that Yn=f(X)Y^{n}=f(X) has no non-constant solutions for all but finitely many natural numbers nn.

3. Constancy of solutions to Yn=f(X)Y^{n}=f(X) in the constant p\mathbb{Z}_{p}-extension of a function field

In this section, we apply Theorem 2.4 proven in the previous section, to study questions motivated by Iwasawa theory. Given primes pp and qq (not necessarily distinct) let hn(p,q)h_{n}(p,q) denote #Cl(Fn(p))[q]\#\operatorname{Cl}(F_{n}^{(p)})[q^{\infty}], the cardinality of the qq^{\infty}-torsion in the class group of Fn(p)F_{n}^{(p)}.

Theorem 3.1 (Leitzel, Rosen).

Let pp and qq be (not necessarily distinct) prime numbers and FF be a function field of characteristic \ell. The following assertions hold

  1. (1)

    Suppose that pp and qq are distinct. Then, as nn goes to infinity, the quantity hn(p,q)h_{n}(p,q) is bounded.

  2. (2)

    Suppose that pp does not divide hFh_{F}. Then, hn(p,p)=1h_{n}(p,p)=1 for all nn.

Proof.

For part (1), the result follows from [Ros02, Theorem 11.6]. For function fields of genus 11, the result was proven by Leitzel, cf. [Lei70]. For part (2), the reader is referred to [Ros02, Proposition 11.3]. ∎

Recall notation from the introduction. The prime PP_{\infty} is totally inert in F(p)F_{\infty}^{(p)} for any prime pp. We set 𝒪(p)\mathcal{O}_{\infty}^{(p)} to denote the ring of integers of F(p)F_{\infty}^{(p)}, i.e., the functions with no poles outside {P}\{P_{\infty}\}. The following is the main result of this manuscript.

Theorem 3.2.

Let \ell be a prime number and FF be a global function field with field of constants κ\kappa. Let pp and qq be prime numbers that are not necessarily distinct, and assume that qq\neq\ell. Let f(X)f(X) be a polynomial with coefficients in κ\kappa satisfying the conditions of Theorem 2.4.

Then, the following assertions hold.

  1. (1)

    Suppose that pp and qq are distinct. Then, for all sufficiently large numbers k>0k>0, the only solutions (X,Y)(X,Y) to Yqk=f(X)Y^{q^{k}}=f(X) in 𝒪(p)\mathcal{O}_{\infty}^{(p)} are constant.

  2. (2)

    Suppose that phFp\nmid h_{F}, then the only solutions (X,Y)(X,Y) to Yp=f(X)Y^{p}=f(X) in 𝒪(p)\mathcal{O}_{\infty}^{(p)} are constant.

Proof.

First, we consider the case when p=qp=q. It follows from part (1) of Theorem 3.1 that hn(p,q)h_{n}(p,q) is bounded as nn goes to infinity. Let k>0k>0 be such that qkq^{k} be larger than maxhn(p,q)\operatorname{max}h_{n}(p,q). It follows from Theorem 2.4 that Yqk=f(X)Y^{q^{k}}=f(X) has no non-constant solutions in 𝒪n(p)\mathcal{O}_{n}^{(p)} for all nn, and therefore, no non-constant solutions in 𝒪(p)\mathcal{O}_{\infty}^{(p)}. Hence, there are no non-constant solutions in FF_{\infty} as well.

Next, we consider the case when p=qp=q and that phFp\nmid h_{F}. Note that if phFp\nmid h_{F}, then by part (2) of Theorem 3.1 that hn(p,p)=1h_{n}(p,p)=1 for all nn. It follows from Theorem 2.4 that Yp=f(X)Y^{p}=f(X) has no non-constant solutions in 𝒪n(p)\mathcal{O}_{n}^{(p)} for all nn, and therefore, no non-constant solutions in 𝒪(p)\mathcal{O}_{\infty}^{(p)}. ∎

4. Constancy of solutions to Yn=f(X)Y^{n}=f(X) in a polynomial ring in rr-variables

In this section, we study solutions to equations of the form Yn=f(X)Y^{n}=f(X) in polynomial rings over a field. Let KK be any field of characteristic >0\ell>0 and AA be the polynomial ring K[T1,,Tr]K[T_{1},\dots,T_{r}]. Given a polynomial gg, let di(g)d_{i}(g) be the degree of gg viewed as a polynomial in TiT_{i} over the subring K[T1,,Ti1,Ti+1,,Tr]K[T_{1},\dots,T_{i-1},T_{i+1},\dots,T_{r}]. The pair (A,di)(A,d_{i}) satisfies the conditions (1)–(5) of Definition 2.1. The class group Cl(A)\operatorname{Cl}(A) denotes the group of equivalence classes of Weil divisors. Since AA is a unique factorization domain, we have that Cl(A)=0\operatorname{Cl}(A)=0.

Theorem 4.1.

With respect to above notation, let f(X)f(X) be a polynomial with all of its coefficients in KK. Let qq\neq\ell be a prime number and assume that the following conditions are satisfied.

  1. (1)

    f(X)f(X) factorizes into

    f(X)=a0(Xa1)n1(Xat)nt,f(X)=a_{0}(X-a_{1})^{n_{1}}\dots(X-a_{t})^{n_{t}},

    where a0Ka_{0}\in K and a1,,ata_{1},\dots,a_{t} are distinct elements in KK, n1,,ntn_{1},\dots,n_{t} are positive integers and t2t\geq 2.

  2. (2)

    At least two of the exponents nin_{i} are not divisible by qq.

Then, any solution (X,Y)A2(X,Y)\in A^{2} to

(4.1) Yq=f(X)Y^{q}=f(X)

is constant, i.e., XX and YY are both in KK.

Proof.

Note that the algebraic closure of KK in AA is equal to KK. We may as well replace KK by its algebraic closure and assume without loss of generality the KK is algebraically closed and that qn1q\nmid n_{1} and qn2q\nmid n_{2}. Since the class group of AA is trivial, the same argument as in the proof of Theorem 2.4 shows that (Xa1)=fq(X-a_{1})=f^{q} and (Xa2)=gq(X-a_{2})=g^{q} for f,gAf,g\in A. Therefore, we find that fqgq=a2a1f^{q}-g^{q}=a_{2}-a_{1}, an element of KK. Lemma 2.2 then implies the di(f)=di(g)=0d_{i}(f)=d_{i}(g)=0 for all ii, hence f,gf,g are both in KK. The result follows from this. ∎

5. Perfect powers that are sums of powers in arithmetic progressions

In this section, we apply the results proven in previous sections to study the solutions of the diophantine equation involving perfect powers in arithmetic progressions

(5.1) Yn=f(X):=(X+r)m+(X+2r)m++(X+dr)m.Y^{n}=f(X)\mathrel{\mathop{\mathchar 58\relax}}=(X+r)^{m}+(X+2r)^{m}+\dots+(X+dr)^{m}.

Here, m,n,r,dm,n,r,d are integers such that m,n,d2m,n,d\geq 2 and r1r\geq 1. Viewing f(X)f(X) as a polynomial with integral coefficients let Δ\Delta denote its discriminant.

Theorem 5.1.

Let FF be a function field with characteristic 5\ell\geq 5 and field of functions κ\kappa. Let qq\neq\ell be a prime. Assume that the following conditions are satisfied.

  1. (1)

    All roots of f(X)f(X) are contained in κ\kappa,

  2. (2)

    r\ell\nmid r,

  3. (3)

    at least one of the following conditions are satisfied:

    1. (a)

      Δ\ell\nmid\Delta,

    2. (b)

      q>mq>m and d0,±1modd\not\equiv 0,\pm 1\mod{\ell}.

Let kk be the minimal value such that hF[qk]=hF[qk1]h_{F}[q^{k}]=h_{F}[q^{k-1}]. Then, there are no non-constant solutions (X,Y)(X,Y) to

Yqk=f(X)=i=1d(X+ir)mY^{q^{k}}=f(X)=\sum_{i=1}^{d}(X+ir)^{m}

in 𝒪F2\mathcal{O}_{F}^{2}.

Proof.

Following notation from the statement of Theorem 2.4, we write

f(X)=a0j=1t(Xai)ni,f(X)=a_{0}\prod_{j=1}^{t}(X-a_{i})^{n_{i}},

where a1,,ata_{1},\dots,a_{t} are all distinct elements of κ\kappa. The result follows from Theorem 2.4 provided the following conditions are satisfied

  1. (1)

    t2t\geq 2.

  2. (2)

    At least two of the exponents nin_{i} are not divisible by qq.

Note that if Δ\ell\nmid\Delta, then all roots of f(X)f(X) are distinct in κ\kappa, hence, ni=1n_{i}=1 for all ii and t=d2t=d\geq 2. In particular, both of the above conditions are satisfied.

On the other hand, assume that q>mq>m. Clearly, all values nin_{i} are less than or equal to degf(X)m\operatorname{deg}f(X)\leq m, and since q>mq>m, it follows that qniq\nmid n_{i} for all ii. It suffices to check that t2t\geq 2 if d0,±1modd\not\equiv 0,\pm 1\mod{\ell}. Suppose not, then f(X)f(X) is of the form d(X+a)md(X+a)^{m}, for some aκa\in\kappa.

Expanding f(X)=i=1d(X+ir)mf(X)=\sum_{i=1}^{d}(X+ir)^{m}, one obtains

i=1d(X+ir)m=i=1dj=0m(mj)ijrjXmj=j=0m(mj)rj(i=1dij)Xmj=dXm+mr(d(d+1)2)Xm1+(m2)r2(d(d+1)(2d+1)6)Xm2+.\begin{split}&\sum_{i=1}^{d}(X+ir)^{m}\\ =&\sum_{i=1}^{d}\sum_{j=0}^{m}{m\choose j}i^{j}r^{j}X^{m-j}\\ =&\sum_{j=0}^{m}{m\choose j}r^{j}\left(\sum_{i=1}^{d}i^{j}\right)X^{m-j}\\ =&dX^{m}+mr\left(\frac{d(d+1)}{2}\right)X^{m-1}+{m\choose 2}r^{2}\left(\frac{d(d+1)(2d+1)}{6}\right)X^{m-2}+\dots.\end{split}

Since f(X)=d(X+a)mf(X)=d(X+a)^{m} and d\ell\nmid d, we find that

aj=rj(1di=1dij)a^{j}=r^{j}\left(\frac{1}{d}\sum_{i=1}^{d}i^{j}\right)

for all values of jj. In particular, we find that

a=r((d+1)2) and a2=r2((d+1)(2d+1)6).a=r\left(\frac{(d+1)}{2}\right)\text{ and }a^{2}=r^{2}\left(\frac{(d+1)(2d+1)}{6}\right).

We thus arrive at the relation

(5.2) a2=r2((d+1)2)2=r2((d+1)(2d+1)6).a^{2}=r^{2}\left(\frac{(d+1)}{2}\right)^{2}=r^{2}\left(\frac{(d+1)(2d+1)}{6}\right).

The relation holds in /\mathbb{Z}/\ell\mathbb{Z}. Since r\ell\nmid r and d1modd\nmid-1\mod{\ell} by assumption, we find that the above relation (5.2) gives us

(d+1)4=2d+16.\frac{(d+1)}{4}=\frac{2d+1}{6}.

This is not possible since it is assumed that d1modd\not\equiv 1\mod{\ell}. Thus, Theorem 2.4 applies to give the result. ∎

Theorem 5.2.

Let FF be a function field with characteristic 5\ell\geq 5 and field of functions κ\kappa. Let qq\neq\ell be a prime. Assume that the following conditions are satisfied.

  1. (1)

    All roots of f(X)f(X) are contained in κ\kappa,

  2. (2)

    r\ell\nmid r,

  3. (3)

    at least one of the following conditions are satisfied:

    1. (a)

      Δ\ell\nmid\Delta,

    2. (b)

      q>mq>m and d0,±1modd\not\equiv 0,\pm 1\mod{\ell}.

Let pp be any prime number. The following assertions hold.

  1. (1)

    Suppose that pqp\neq q. Then for all large enough values of k>0k>0, there are no non-constant solutions to Yqk=f(X)Y^{q^{k}}=f(X) in 𝒪(p)\mathcal{O}_{\infty}^{(p)}.

  2. (2)

    Suppose that p=qp=q and phFp\nmid h_{F}. Then, there are no non-constant solutions to Yp=f(X)Y^{p}=f(X) in 𝒪(p)\mathcal{O}_{\infty}^{(p)}.

Proof.

It follows from the proof of Theorem 5.1 that the conditions of Theorem 3.2 are satisfied, and thus the result follows. ∎

Theorem 5.3.

Let KK be a field of characteristic 5\ell\geq 5 and let AA be the polynomial ring K[T1,,Tr]K[T_{1},\dots,T_{r}]. Let qq\neq\ell be a prime. Assume that the following conditions are satisfied.

  1. (1)

    All roots of f(X)f(X) are contained in KK,

  2. (2)

    r\ell\nmid r,

  3. (3)

    at least one of the following conditions are satisfied:

    1. (a)

      Δ\ell\nmid\Delta,

    2. (b)

      q>mq>m and d0,±1modd\not\equiv 0,\pm 1\mod{\ell}.

Then, any solution (X,Y)A2(X,Y)\in A^{2} to

(5.3) Yq=f(X)Y^{q}=f(X)

is constant, i.e., XX and YY are both in KK.

Proof.

It follows from the proof of Theorem 5.1 that the conditions of Theorem 4.1 are satisfied. The result thus follows from Theorem 4.1. ∎

References

  • [AGP20] Alejandro Argáez-García and Vandita Patel. On perfect powers that are sums of cubes of a seven term arithmetic progression. Journal of Number Theory, 214:440–451, 2020.
  • [BGP04] Michael A Bennett, Kálmán Gyory, and Ákos Pintér. On the diophantine equation 1k+2k++xk=yn1^{k}+2^{k}+\dots+x^{k}=y^{n}. Compos. Math, 140(6):1417–1431, 2004.
  • [BPS17] Michael A Bennett, Vandita Patel, and Samir Siksek. Perfect powers that are sums of consecutive cubes. Mathematika, 63(1):230–249, 2017.
  • [BPSS18] Attila Bérczes, István Pink, Gamze Savaş, and Gökhan Soydan. On the diophantine equation (x+1)k+(x+2)k++(2x)k=yn(x+1)^{k}+(x+2)^{k}+...+(2x)^{k}=y^{n}. Journal of Number Theory, 183:326–351, 2018.
  • [BZ13] Meng Bai and Zhongfeng Zhang. On the diophantine equation (x+1)2+(x+2)2++(x+d)2=yn(x+1)^{2}+(x+2)^{2}+\dots+(x+d)^{2}=y^{n}. Functiones et Approximatio Commentarii Mathematici, 49(1):73–77, 2013.
  • [Haj15] L Hajdu. On a conjecture of schaffer concerning the equation 1k+ +xk= yn. Journal of Number Theory, 155:129–138, 2015.
  • [KP18] Angelos Koutsianas and Vandita Patel. Perfect powers that are sums of squares in a three term arithmetic progression. International Journal of Number Theory, 14(10):2729–2735, 2018.
  • [Lei70] James RC Leitzel. Class number in constant extensions of elliptic function fields. Proceedings of the American Mathematical Society, 25(1):183–188, 1970.
  • [Maz72] Barry Mazur. Rational points of abelian varieties with values in towers of number fields. Inventiones mathematicae, 18(3):183–266, 1972.
  • [Pat18] Vandita Patel. Perfect powers that are sums of consecutive squares. Comptes Rendus Mathématiques de l’Académie des Sciences, 40(2):33–38, 2018.
  • [Ray22] Anwesh Ray. Rational points on algebraic curves in infinite towers of number fields. The Ramanujan Journal, pages 1–16, 2022.
  • [Ros02] Michael Rosen. Number theory in function fields, volume 210. Springer Science & Business Media, 2002.