Skip to main content
Log in

Solving \(a x^p + b y^p = c z^p\) with abc Containing an Arbitrary Number of Prime Factors

  • Published:
Mediterranean Journal of Mathematics Aims and scope Submit manuscript

Abstract

In this paper we prove new cases of the asymptotic Fermat equation with coefficients. This is done by solving some remarkable S-unit equations and applying a method of Frey–Kraus–Mazur.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

  1. We use the terminology non-trivial local obstructions to distinguish from the ones introduced in Proposition 1.2.

  2. Consider the degree p morphism \(\phi : F_p^{a, b, c} \rightarrow {\mathbb {P}}_1\), \([x:y:z]\mapsto [x:y]\). It is ramified at p points with constant ramification index p.

  3. The set \(F_2^{a, b, c}({\mathbb {Q}})\) is infinite if and only if it is not empty. If \(\mathcal O\in F_3^{a, b, c}({\mathbb {Q}})\) then \((F_3^{a, b, c},\mathcal O)\) is an elliptic curve over \({\mathbb {Q}}\) and \(F_3^{a, b, c}({\mathbb {Q}})\) is a finitely generated group.

  4. Indeed, if \(q\mid x,y\) then \(q^p\mid c\) and \(p\le v_q(c)\le v_q(abc)\).

  5. For example one can take B even and \(A\equiv -1\pmod 4\).

  6. More precisely, \({\bar{\rho }}\) is finite at \(\ell \) if there is a finite flat \({\mathbb {F}}_p\)-vector space scheme H over \({\mathbb {Z}}_\ell \) such that \(H(\bar{{\mathbb {Q}}}_\ell )\) is isomorphic to \({\bar{\rho }}\vert _{G_\ell }\) as \({\mathbb {F}}_p[G_\ell ]\)-modules.

References

  1. Atkin, A.O.L., Li, W.C.W.: Twists of newforms and pseudo-eigenvalues of W-operators. Invent. Math. 48(3), 221–243 (1978)

    Article  MathSciNet  Google Scholar 

  2. Bilu, Y.F., Bugeaud, Y., Mignotte, M.: The Problem of Catalan. Springer, Cham (2014)

    Book  Google Scholar 

  3. Carayol, H.: Sur les représentations galoisiennes modulo l attachées aux formes modulaires. Duke Math. J. 59(3), 785–801 (1989)

    Article  MathSciNet  Google Scholar 

  4. Cohen, H.: Number theory. Vol. II. Analytic and modern tools. Graduate Texts in Mathematics, vol. 240. Springer, New York (2007)

  5. Darmon, H., Diamond, F., Taylor, R.: Fermat’s last theorem. Elliptic curves, modular forms and Fermat’s last theorem. Hong Kong, 1993, pp. 2–140, Int. Press, Cambridge (1993)

  6. Darmon, H., Granville, A.: On the equations \(z^m=F(x, y)\) and \(Ax^p+By^q=Cz^r\). Bull. Lond. Math. Soc. 27(6), 513–543 (1995)

    Article  Google Scholar 

  7. Darmon, H., Merel, L.: Winding quotients and some variants of Fermat’s last theorem. J. Reine Angew. Math. 490, 81–100 (1997)

    MathSciNet  MATH  Google Scholar 

  8. Diamond, F., Kramer, K.: Modularity of a family of elliptic curves. Math. Res. Lett. 2(3), 299–304 (1995)

    Article  MathSciNet  Google Scholar 

  9. Edixhoven, B.: The weight in Serre’s conjectures on modular forms. Invent. Math. 109(3), 563–594 (1992)

    Article  MathSciNet  Google Scholar 

  10. Faltings, G.: Endlichkeitssätze für abelsche Varietäten über Zahlkörpern. Invent. Math. 73(3), 349–366 (1983)

    Article  MathSciNet  Google Scholar 

  11. Freitas, N., Kraus, A.: An application of the symplectic argument to some Fermat-type equations. C. R. Math. Acad. Sci. Paris 354(8), 751–755 (2016)

    Article  MathSciNet  Google Scholar 

  12. Frey, G.: Links between elliptic curves and solutions of \(A-B=C\). J. Indian Math. Soc. (N.S.) 51(1987), 117–145 (1988)

    MathSciNet  MATH  Google Scholar 

  13. Halberstadt, E., Kraus, A.: Courbes de Fermat: résultats et problèmes. J. Reine Angew. Math. 548, 167–234 (2002)

    MathSciNet  MATH  Google Scholar 

  14. Kraus, A.: Majorations effectives pour l’équation de Fermat généralisée. Can. J. Math. 49(6), 1139–1161 (1997)

    Article  Google Scholar 

  15. Lang, S.: Integral points on curves. Inst. Hautes Études Sci. Publ. Math. No. 6, pp. 27–43 (1960)

  16. Ljunggren, W.: Zur Theorie der Gleichung \(x^2+1=Dy^4\). Avh. Norske Vid. Akad. Oslo. I. 1942, no. 5–27 (1942)

  17. Martin, G.: Dimensions of the spaces of cusp forms and newforms on \(\Gamma _0(N)\) and \(\Gamma _1(N)\). J. Number Theory 112(2), 298–331 (2005)

    Article  MathSciNet  Google Scholar 

  18. Mazur, B.: Rational isogenies of prime degree. Invent. Math. 44(2), 129–162 (1978)

    Article  MathSciNet  Google Scholar 

  19. Mihăilescu, P.: Primary cyclotomic units and a proof of Catalan’s conjecture. J. Reine Angew. Math. 572, 167–195 (2004)

    MathSciNet  MATH  Google Scholar 

  20. Papadopoulos, I.: Sur la classification de Néron des courbes elliptiques en caractéristique résiduelle 2 et 3. J. Number Theory 44(2), 119–152 (1993)

    Article  MathSciNet  Google Scholar 

  21. Ribenboim, P.: Catalan’s conjecture. Are 8 and 9 the only consecutive powers?. Academic Press, Boston (1994)

    MATH  Google Scholar 

  22. Ribet, K.A.: On modular representations of \(\text{ Gal }(\bar{Q}/ Q)\) arising from modular forms. Invent. Math. 100(2), 431–476 (1990)

    Article  MathSciNet  Google Scholar 

  23. Ribet, K.A.: On the equation \(a^p + 2^\alpha b^p+c^p=0\). Acta Arith. 79(1), 7–16 (1997)

    Article  MathSciNet  Google Scholar 

  24. Schoof, R.: Catalan’s Conjecture. Universitext. Springer, London (2008)

    MATH  Google Scholar 

  25. Serre, J.-P.: Sur les représentations modulaires de degré 2 de \(\text{ Gal } (\bar{{\mathbb{Q}}}/{\mathbb{Q}})\). Duke Math. J. 54(1), 179–230 (1987)

    Article  MathSciNet  Google Scholar 

  26. Silverman, J.H.: The Arithmetic of Elliptic Ccurves, 2nd edn. Graduate Texts in Mathematics, vol. 106. Springer, Dordrecht (2009)

  27. Steiner, R., Tzanakis, N.: Simplifying the solution of Ljunggren’s equation \(X^2+1=2Y^4\). J. Number Theory 37(2), 123–132 (1991)

    Article  MathSciNet  Google Scholar 

  28. Störmer, C.: Solution complète en nombres entiers de l’équation \(m \arctan \frac{1}{x}+n \arctan \frac{1}{y}=k\frac{\pi }{4}\). Bull. Soc. Math. France 27, 160–170 (1899)

    Article  MathSciNet  Google Scholar 

  29. Tate, J.: Algorithm for determining the type of a singular fiber in an elliptic pencil. Modular functions of one variable, IV (Proc. Internat. Summer School, Univ. Antwerp, Antwerp, 1972), pp. 33–52. Lecture Notes in Math., vol. 476. Springer, Berlin (1975)

  30. The LMFDB Collaboration, The L-functions and modular forms database. http://www.lmfdb.org (2013) (Online; accessed 9 Oct 2017)

  31. Ulmer, D.: Conductors of \(\ell \)-adic representations. Proc. Am. Math. Soc. 144(6), 2291–2299 (2016)

    Article  MathSciNet  Google Scholar 

  32. Wiles, A.: Modular elliptic curves and Fermat’s last theorem. Ann. Math. (2) 141(3), 443–551 (1995)

Download references

Acknowledgements

We would like to thank Samuele Anni, Henri Cohen, Nuno Freitas, Roberto Gualdi, Xavier Guitart, Mariagiulia De Maria, Artur Travesa, Carlos de Vera and Gabor Wiese for helpful conversations and comments. The second author is very grateful to Marc Masdeu and Alberto Soto for their help on computational aspects. We would like to thank the anonymous referees for a thorough reading of our paper, and for the numerous helpful suggestions they made to improve the exposition.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Eduardo Soto.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Luis Dieulefait is partially supported by MICINN grant MTM2015-66716-P. Eduardo Soto is partially supported by MICINN grant MTM2016-78623-P.

Appendices

Prime Divisors of Cyclotomic Polynomials

In this appendix, we give some lower bounds for the number of prime divisors of \(\ell ^n\pm 1\) for integers \(\ell \ge 3\) and \(n\ge 1\).

Let \(\Phi _n\) be the nth cyclotomic polynomial. A usual description of \(\Phi _n\) is given by the formula

$$\begin{aligned} \Phi _n(X) =\prod _k (X-\zeta _n^k) \end{aligned}$$

where \(\zeta _n = e^{2\pi i/n}\) is a primitive nth root of unity and k ranges over the units of \({\mathbb {Z}}/n{\mathbb {Z}}\). Gauss proved that \(\Phi _n\) is irreducible in \({\mathbb {Z}}[X]\), hence \({\mathbb {Z}}[X]/\Phi _n \simeq {\mathbb {Z}}[\zeta _n] \subseteq {\mathbb {C}}\) is a domain. In particular

$$\begin{aligned} X^n-1 = \prod _{d\mid n} \Phi _d(X) \end{aligned}$$
(3)

is the factorization of \(X^n-1\) in irreducible factors over \({\mathbb {Z}}[X]\). Similarly, write \(n = 2^mn_2\) where \(n_2\) is the largest odd divisor of n. Then

$$\begin{aligned} X^n+1 = \prod _{d\mid n_2} \Phi _{2^{m+1}d} (X) \end{aligned}$$
(4)

since \(X^{2^{m+1}n_2}-1 = (X^n-1) (X^n+1)\).

Let k be a positive integer. The map \({\mathbb {Z}}[X]\rightarrow {\mathbb {Z}}/k{\mathbb {Z}}\), \(X\mapsto \ell \) factors through \({\mathbb {Z}}[\zeta _n] \rightarrow {\mathbb {Z}}/k{\mathbb {Z}}\), \(\zeta _n\mapsto \ell \) if, and only if, \(k\mid \Phi _n(\ell )\).

Lemma A.1

Let \(p\not \mid n\) be a prime and assume that there is a ring homomorphism \(\theta :{\mathbb {Z}}[\zeta _n] \rightarrow {\mathbb {F}}_p\). Then \(\theta (\zeta _n)\) has order n in \({\mathbb {F}}_p^\times \) and \(n\mid p-1\).

Proof

Let \(\alpha = \theta (\zeta _n)\). Then \(\alpha ^n-1 = \prod _d \Phi _d(\alpha )=0\). Notice that \(X^n-1\) is separable over \({\mathbb {F}}_p\) since \(nX^{n-1}\ne 0\) in \({\mathbb {F}}_p[X]\). Hence \(\alpha \) has order n in \({\mathbb {F}}_p^\times \) and the lemma follows. \(\square \)

Lemma A.2

Let p be an odd prime. There is no ring homomorphism \({\mathbb {Z}}[\zeta _p] \rightarrow {\mathbb {Z}}/p^2{\mathbb {Z}}\). There is no ring homomorphism \({\mathbb {Z}}[\zeta _4] \rightarrow {\mathbb {Z}}/4{\mathbb {Z}}\).

Proof

It is enough to prove that \(\Phi _p(X) = \sum _{i=0}^{p-1} X^i\) has no roots in \({\mathbb {Z}}/p^2{\mathbb {Z}}\). The following proof is standard. Assume that there is a root a of \(\Phi _p\) in \({\mathbb {Z}}/p^2{\mathbb {Z}}\). Then \(a = 1 \mod p\), since \(\Phi _p = (X-1)^{p-1}\) in \({\mathbb {F}}_p\). Notice that \(\Phi _p(1 + pb ) = \sum _{i=0}^{p-1} 1 + i pb=p\) in \({\mathbb {Z}}/p^2{\mathbb {Z}}\) for every b. Hence \(\Phi _p(a) = p\) for every \(a\equiv 1\pmod p\).

Notice that \(\Phi _4(X)=X^2 + 1\) has no roots in \({\mathbb {Z}}/4{\mathbb {Z}}\). \(\square \)

Lemma A.3

Let \(\ell \ge 3\), \(n\ge 2\) be integers and let p be the largest prime divisor of n, then \(|\Phi _n(\ell )|> p\).

Proof

The Euler’s totient function \(\varphi \),

satisfies that

$$\begin{aligned} p-1\mid \varphi (n). \end{aligned}$$

Hence

$$\begin{aligned} | \Phi _n(\ell ) |= \prod _k |\ell -\zeta _n^k| \ge \prod _k 2 \ge 2^{p-1} \end{aligned}$$

and case \(p\ge 3\) follows.

If \(p=2\) then n is a power of 2, \(n= 2^m\), and

$$\begin{aligned} \Phi _n(\ell ) = \ell ^{2^{m-1}}+1> 2. \end{aligned}$$

\(\square \)

The polynomial \(\Phi _n\) has no real roots for \(n\ge 3\), hence \(|\Phi _n(\ell )| =\Phi _n(\ell )\).

Theorem A.4

Let \(\ell \ge 3, n\ge 3\) be integers. There is a prime divisor p of \(\Phi _n(\ell )\) not dividing 2n. Hence, \(\ell \) has order n in \({\mathbb {F}}_p^\times \).

Proof

Case \(n=2^m\ge 4\).

One has that \(\Phi _{2^m}(X)= X^{2^{m-1}}+ 1\) and \(\Phi _n(\ell ) \ge 10\). If \(4\mid \Phi _n(\ell )\) then \({\mathbb {Z}}[\zeta _n]\rightarrow {\mathbb {Z}}/4{\mathbb {Z}}\), \(\zeta _n\mapsto \ell \) defines a ring homomorphism that restricts to \({\mathbb {Z}}[\zeta _4]\subseteq {\mathbb {Z}}[\zeta _n]\). This contradicts Lemma A.2. Hence either \(\Phi _n(\ell )\) is odd or \(\Phi _n(\ell )/2\ge 5\) is odd.

Case \(p\mid n\), p odd.

Notice that \(\Phi _n(\ell )\) is odd. Indeed, if \(2\mid \Phi _n(\ell )\) then there exists a ring homomorphism

$$\begin{aligned} {\mathbb {Z}}[\zeta _n]\rightarrow {\mathbb {F}}_2 \end{aligned}$$

which induces by restriction a map

$$\begin{aligned} {\mathbb {Z}}[\zeta _p]\rightarrow {\mathbb {F}}_2 \end{aligned}$$

hence \(p\mid 2-1\) by Lemma A.1.

Let us see that either \(\Phi _n(\ell )\) and n are coprime or there is a prime p such that \(\Phi _n(\ell )/p, n\) are coprime. Assume that \(p<q\) are prime divisors of \(\Phi _n(\ell )\) and n. Then there is a ring homomorphism

$$\begin{aligned} {\mathbb {Z}}[\zeta _q] \subseteq {\mathbb {Z}}[\zeta _n] \rightarrow {\mathbb {F}}_p \end{aligned}$$

and \(q\mid p-1\) by Lemma A.1 which contradicts \(p<q\). Hence the greatest common divisor of \(\Phi _n(\ell )\) and n is a possibly trivial power of an odd prime p. If \(p\mid n\) then \(p^2\not \mid \Phi _n(\ell )\) by Lemma A.2. Hence either \(\Phi _n(\ell ), 2n\) are coprime or there is an odd prime divisor p of \(\Phi _n(\ell )\) such that \(\Phi _n(\ell )/p\) and 2n are coprime. In the second case \(\Phi _n(\ell )/p\) is an odd integer \(>1\) by Lemma A.3 and the first part of the theorem follows. The order of \(\ell \) is computed in Lemma A.1. \(\square \)

Corollary A.5

Let \(\ell \ge 3\). Assume that \(n_1, \dots , n_r\) are pairwise different integers \(\ge 3\). Then

$$\begin{aligned} \prod _i \Phi _{n_i}(\ell ) \end{aligned}$$

has at least r odd prime divisors.

Proof

Let \(p_i\) be a prime divisor of \(\Phi _{n_i}(\ell )\) coprime to \(2n_i\) as in Theorem A.4. Then \(\ell \) has order \(n_i\) in \({\mathbb {F}}_{p_i}^\times \), thus \(p_i\ne p_j\) for different ij. \(\square \)

For an integer k let \(\omega (k)\) denote the number of prime divisors of k and let \(\sigma (k)\) denote the number of divisors of k.

Corollary A.6

Let \(\ell \ge 3\), \(n\ge 1\) be integers. If \((\ell , n)\ne (3,even)\) then

$$\begin{aligned} \omega (\ell ^n-1)\ge \sigma (n). \end{aligned}$$

Otherwise

$$\begin{aligned} \omega (3^{2t}-1)\ge \sigma (2t)-1. \end{aligned}$$

Proof

Let \(i\in \{1,2\}\) such that \(n \equiv i \pmod 2\). Then

$$\begin{aligned} A:=\prod _{\begin{array}{c} {d\mid n}\\ {d\ge 3} \end{array}}\Phi _d(\ell )=\dfrac{\ell ^n-1}{\ell ^i-1} \end{aligned}$$

has at least \(\sigma (n)-i\) odd prime divisors \(S=\{p_d\}_{d\mid n, d\ge 3}\) as in Theorem A.4. Notice that \(p_d\not \mid \ell ^i-1\) for every \(p_d\in S\). Indeed, if an odd prime p divides \(\ell ^i-1\) then \(\ell \) has order \(\le i\) in \({\mathbb {F}}_{p}^\times \) by Lemma A.1. Thus

$$\begin{aligned} \omega (\ell ^n-1) \ge \sigma (n) - i + \omega (\ell ^i-1). \end{aligned}$$

It is enough to prove that \(\omega (\ell ^i-1) \ge i\) if and only if \((\ell ,i) \ne (3,2)\). If \(i=1\) then \(\ell -1\ge 2\) and \(\omega (\ell -1) \ge 1\). If \(i=2\) then \(gcd(\ell -1,\ell +1)\le 2\). Assume \(\omega (\ell ^2-1) < 2\) then \(\ell -1, \ell +1\) are powers of two. Hence \(\ell =3\). \(\square \)

Corollary A.7

Let \(\ell \ge 3\), \(n\ge 1\) be integers and let \(n_2\) be the largest odd divisor of n. Then

$$\begin{aligned} \omega (\ell ^n+1)\ge \sigma (n_2). \end{aligned}$$

Proof

Let \(n=2^m n_2\) then \(\ell ^n+1 = \prod _{d\mid n_2} \Phi _{2^{m+1} d}(\ell )\) by the polynomial factorization of (4). For every d such that \(2^{m+1} d\ne 2\) consider a prime \(p_d\mid \Phi _{2^{m+1} d}(\ell )\) as in Theorem A.4. If \(m=0\) let \(p_1\) be an arbitrary prime divisor of \(\Phi _{2}(\ell )=\ell +1\). Then \(\prod _{d\mid n_2} p_d\) is a squarefree divisor of \(\ell ^n+1\). \(\square \)

1.1 Catalan Conjecture

One deduces a case of Catalan’s Conjecture.

Theorem A.8

(Partial Catalan’s Conjecture) Let \(\ell \ge 3\) be an integer and assume that

$$\begin{aligned} 2^m - \ell ^n \in \{\pm 1\} \end{aligned}$$

for some integers \(m, n\ge 2\). Then \(m=\ell =3\), \(n=2\).

Proof

Assume that \(2^m = \ell ^n+1\), \(n\ge 2\) and let \(n_2\) be the largest odd divisor of n. Then \(\ell \) is odd and \(\ell ^n+1 \ge 4\). By Corollary A.7 we have that \(1=\omega (\ell ^n +1)\ge \sigma (n_2)\), hence \(n_2=1\) and \(n=2^r\) for some positive r. Since \(2^m=\ell ^{2^r}+1 \equiv 2 \pmod 4\) one has that \(m=1\) and \(2= \ell ^{2^r} + 1\).

Assume that \(2^m = \ell ^n-1\), \(n\ge 2\). If \((\ell ,n)=(3, 2t)\) with t an integer then \(1=\omega (3^{2t}-1) \ge \sigma (2t)-1\) by Corollary A.6. Hence \(t=1\).

If \((\ell , n)\ne (3, even)\), by Corollary A.6 one has that \(1=\omega (\ell ^n-1) \ge \sigma (n)\). Hence \(n=1\). \(\square \)

This partial result is well known to experts, see [21, B3.3]. See ibid for a complete treatment of Catalan’s conjecture written before Preda Mihăilescu’s proof [19]. See also Bilu–Bugeaud–Mignotte’s book [2] for a minimalistic approach of the proof or Schoof’s book [24] based on two sets of lecture notes by Yuri Bilu.

The Conductor of E[p]

The j-invariant of a Frey curve is given by the formula

$$\begin{aligned} j_E=\dfrac{2^8 (C^2 - AB)^3}{A^2 B^2 C^2}. \end{aligned}$$

Thus one has for the case \((A,B,C) = (ax^p , by^p ,cz^p)\) being pairwise coprime that \(C^2-AB\) and ABC are coprime. Let \(\ell \) be a prime divisor of ABC. Then

$$\begin{aligned} v_\ell (j_E) = 8 v_\ell (2) - 2v_\ell (ABC) \equiv 8v_\ell (2) - 2v_\ell (abc)\pmod p. \end{aligned}$$

Thus \(p\mid v_\ell (j_E)\) if and only if

  • \(\ell \) is odd and \(p\mid v_\ell (abc)\), or

  • \(\ell =2\) and \(v_2(abc) \equiv 4 \pmod p\).

Proposition B.1

Let \(E=E_{A,B,C}\) be the Frey curve as in Theorem 3.4. Let f be a newform in \(S_2(M)\) for some divisor M of \(2^s{\mathrm{rad}}'(abc)\) and let \(\mathfrak p\) be a prime ideal such that

$$\begin{aligned} E[p] \simeq \bar{\rho }_{f,\mathfrak p} \end{aligned}$$

as \({\mathbb {F}}_p[G_{{\mathbb {Q}}}]\)-modules. Then \(M= 2^s {\mathrm{rad}}'(abc)\).

Proof

Let R be the largest (square-free) divisor of \(2^s{\mathrm{rad}}'(abc)\) coprime to 2p. By Tate’s uniformization E[p] is ramified at every prime divisor \(\ell \) of R and so is \(\bar{\rho }_{f,\mathfrak p}\). Thus, \(R\mid M\).

Let \(\ell =2\). If \(s \in \{ 3,5\}\) then Carayol [3] predicts that the lifting \(\rho _{f,\mathfrak p}\) of \(\bar{\rho }_{f,\mathfrak p}\) has conductor exponent s. Thus \(2^s\mid M\). If \(s=0\) then M is odd and so is R. If \(s=1\) then E[p] is ramified at 2 and so is \({\bar{\rho }}_{f,\mathfrak p}\). Hence \(2\mid M\).

One could just avoid case \(p\mid M\) since we will consider big primes p with respect to \({\mathrm{rad}}(abc)\). Still, if \(p\mid {\mathrm{rad}}'(abc)\) then E[p] is not finite at p. That is, \(E[p]\vert _{G_p}\) is reducible and not peu ramifié by [9, Proposition 8.2]. If \(p\not \mid M\) then \(\bar{\rho }_{f,\mathfrak p}\vert _{G_p}\) is either irreducible or reducible and peu ramifié. Thus \(E[p]\vert _{G_p}\not \simeq \bar{\rho }_{f,\mathfrak p}\vert _{G_p}\). This completes the proof. \(\square \)

Mod 24 Exercises

Proof of Lemma 2.13

Let (ABC) be a primitive S-unit point of height \(\ge 3\). Assume \(A=2^r\), \(r\ge 3\). Then \(B+C\equiv 0\pmod 8\) and \(B+C\not \equiv 0\pmod 3\). Hence,

$$\begin{aligned} BC\equiv -1 \pmod 8 \end{aligned}$$

since \(C^{-1} \equiv C \pmod 8\) and

$$\begin{aligned} BC\equiv 1 \pmod 3 \end{aligned}$$

since \(B,C\in \{\pm 1\} \mod 3\). Thus

$$\begin{aligned} \pm q^s \ell ^t = BC \equiv 7 \pmod {24}. \end{aligned}$$
  1. (1)

    By hypothesis \((q,\ell ) \equiv (-5, 5)\) or \((11,-11)\pmod {24}\). Notice that

    $$\begin{aligned} q^s \ell ^t\equiv \pm q^{s+t}\not \equiv \pm 7 \pmod {24}, \end{aligned}$$

    hence A is not a power of two.

    Assume that

    $$\begin{aligned} 0\equiv 2^r q^{s} = \ell ^t + \varepsilon \equiv (- 3)^t +\varepsilon \pmod 8 \end{aligned}$$

    for some \(\varepsilon \in \{\pm 1\}\). Then \(\varepsilon =-1\) and t is even. Proposition 2.6 implies

    $$\begin{aligned} (q,\ell ) \in \{(3,5),(5,3),(3,7),(3,17)\}. \end{aligned}$$

    Condition \(q\equiv -\ell \pmod {24}\) leads to a contradiction. Similarly, \(2^r \ell ^t = q^r+\varepsilon \) has no solution.

  2. (2)

    Assume that \((2^r,- q^s \ell ^t, \varepsilon )\) is an S-point for some unit \(\varepsilon \). Then \(- \varepsilon q^s \ell ^t\equiv 7\pmod {24}\). Thus st are odd and \(\varepsilon = -1\). That is

    $$\begin{aligned} 2^r = q^s \ell ^t + 1\equiv -1\pmod 3, \end{aligned}$$

    hence r is odd, \(r=2f+1\). Thus, 2 is a square in \({\mathbb {F}}_q\), i.e. \(q\equiv \pm 1\pmod 8\). Indeed

    $$\begin{aligned} \genfrac(){}0{1}{q}=\genfrac(){}0{2}{q}^r = \genfrac(){}0{2}{q}. \end{aligned}$$

    Assume that \(2^r +(-1)^{a}q^{s} +(-1)^{b}\ell ^{t}=0\). Then

    $$\begin{aligned} (-1)^{a+b} q^s \ell ^t \equiv 7\pmod {24}. \end{aligned}$$

    Hence ab have same parity and st are odd. Thus

    $$\begin{aligned} 2^r = q^s + \ell ^t\equiv 1\pmod 3 \end{aligned}$$

    and r is even. Thus q is a square in \({\mathbb {F}}_\ell \).

    Assume that \((2^r q^s , -\ell ^t , \varepsilon )\) is an S-point. Then \(\ell ^t \equiv \varepsilon \pmod 8\) and hence t is even and \(\varepsilon =1\).

    Assume that \((2^r \ell ^t , -q^s , \varepsilon )\) is an S-point. Then \(\varepsilon =1\) and s is even. By Proposition 2.6\(q\in \{ 3,5,7,17\}\), hence

    $$\begin{aligned} q\not \equiv 11 \pmod {24}. \end{aligned}$$
  3. (3)

    By hypothesis

    $$\begin{aligned} \ell \equiv -1 \pmod {24} \end{aligned}$$

    and \(q\equiv \pm 5\) or \(\pm 11 \pmod {24}\) since \(q \ge 5\). Thus \(q^s \ell ^t\not \equiv \pm 7 \pmod {24}\) and A is not a power of two.

    Assume that \(2^r q^{s} = \ell ^{t} +1\). Then t is either 1 or an odd prime by Lemma 2.8. Case \(t=1\) implies \(\ell \equiv -1 \pmod q\). Case t odd prime implies \(\ell \) Mersenne hence

    $$\begin{aligned} \ell \equiv 0, 1\pmod 3. \end{aligned}$$

    Assume that \(2^r q^{s}=\ell ^{t} -1\). Hence t is even and Proposition 2.6 implies \(\ell \in \{3,5,7, 17\}\), then \(\ell \not \equiv -1\pmod {24}\). Similarly, case \(2^r \ell ^s = q^t\pm 1\) is not allowed by Lemma 2.6.

\(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Dieulefait, L., Soto, E. Solving \(a x^p + b y^p = c z^p\) with abc Containing an Arbitrary Number of Prime Factors. Mediterr. J. Math. 18, 48 (2021). https://doi.org/10.1007/s00009-020-01678-1

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s00009-020-01678-1

Mathematics Subject Classification

Navigation