Almost fifth powers in arithmetic progression

Almos i h powe s in a i hme ic p og ession L. Hajdu and T. Ko ´acs Uni e si y o Deb ecen, Ins i u e o Ma hema ics and he Numbe Theo y Resea ch G oup o he Hunga ian Academy o Sciences Deb ecen, Hunga y Uni e si y o Deb ecen, Ins i u e o Ma hema ics Deb ecen, Hunga y Abs ac We p o e ha he p oduc o kconsecu i e e ms o a p imi i e a i hme ic p og es- sion is ne e a pe ec i h powe when 3 ≤k≤54. We also p o ide a mo e p ecise s a emen , conce ning he case whe e he p oduc is an ”almos ” i h powe . Ou heo ems yield conside able imp o emen s and ex ensions, in he i h powe case, o ecen esul s due o Gy˝o y, Hajdu and Pin ´e . While he ea lie esul s ha e been p o ed by classical (mainly algeb aic numbe heo e ical) me hods, ou p oo s a e based upon a new ool: we apply genus 2 cu es and he Chabau y me hod (bo h he classical and he ellip ic e ison). Key wo ds: pe ec powe s, a i hme ic p og ession, genus 2 cu es, Chabau y me hod PACS: 11D41, 11B25 1 In oduc ion Conside he Diophan ine equa ion x(x+d). . . (x+ (k−1)d) = byn(1) Email add esses: [email p o ec ed],[email p o ec ed] (L. Hajdu and T. Ko ´acs). 1Resea ch suppo ed in pa by he Hunga ian Academy o Sciences, by he OTKA g an s K67580, K75566, and by he T´ AMOP 4.2.1./B-09/1/KONV-2010- 0007 p ojec . The p ojec is implemen ed h ough he New Hunga y De elopmen Plan, co- inanced by he Eu opean Social Fund and he Eu opean Regional De el- opmen Fund. in non-ze o in ege s x, d, k, b, y, n wi h gcd(x, d) = 1, d≥1, k≥3, n≥2 and P(b)≤k. He e P(u) s ands o he la ges p ime di iso o a non-ze o in ege u, wi h he con en ion P(±1) = 1. The equa ion has a e y ich li e a u e. Fo d= 1 and b= 1, equa ion (1) has been sol ed by E d˝os and Sel idge [9]. This celeb a ed esul can be e o mula ed as ha he p oduc o wo o mo e consecu i e posi i e in ege s is ne e a pe ec powe . The comple e solu ion o (1) in case o d= 1 is due o Sa adha [21] (case k≥4) and Gy˝o y [10] (case k < 4). Fo an o e iew o he huge numbe o ela ed esul s o d > 1 we e e o su ey pape s o Gy˝o y [11], Sho ey [22], [23] and Tijdeman [25]. Now we men ion only esul s which a e closely ela ed o he scope o he p esen pape , ocusing on he comple e solu ion o (1) when he numbe ko e ms is fixed. In case o (k, n) = (3,2) equa ion (1) has infini ely many solu ions, al eady o b= 1 (c. . [25]). Eule (see [8]) p o ed ha (1) has no solu ions wi h b= 1, and (k, n) = (3,3) o (4,2). Obl´a h [18], [19] ob ained simila esul s o (k, n) = (3,4), (3,5) and (5,2). By a conjec u e o E d˝os, equa ion (1) has no solu ions in posi i e in ege s when k > 3 and b= 1. In o he wo ds, he p oduc o kconsecu i e e ms o a p imi i e posi i e a i hme ic p og ession wi h k > 3 is ne e a pe ec powe . By p imi i e a i hme ic p og ession we mean one o he o m x, x +d, . . . , x + (k−1)d, wi h gcd(x, d) = 1. The conjec u e o E d˝os has ecen ly been e ified o ce ain alues o kin a mo e gene al o m; see he pape s [11], [12], [1], [13]. Since now we ocus on he case n= 5, we gi e only he bes known esul o his pa icula exponen . (Though he esul s men ioned a e alid o any n≥2.) The ollowing s a emen is a combina ion o esul s om [11] (case k= 3), [12] (cases k= 4,5), [1] (cases k= 6,7) and [13] (cases 8 ≤k≤34). Theo em A. The only solu ions o equa ion (1) wi h n= 5,3≤k≤34 and P(b)≤Pk, wi h Pk=                    2,i k= 3,4, 3,i k= 5, 5,i k= 6,7, 7,i 8≤k≤22, k−1 2,i 23 ≤k≤34 a e gi en by (k, d) = (8,1), x ∈ {−10,−9,−8,1,2,3}; (k, d) = (8,2), x ∈ {−9,−7,−5}; 2 (k, d) = (9,1), x ∈ {−10,−9,1,2}; (k, d) = (9,2), x ∈ {−9,−7}; (k, d) = (10,1), x ∈ {−10,1}; (k, d, x) = (10,2,−9). No e ha knowing he alues o k, d and x, all solu ions (x, d, k, b, y, n) o (1) can be easily lis ed. To explain why he case n= 5 in equa ion (1) is special, we need o gi e some insigh in o he me hod o sol ing (1) o fixed k, in he gene al case n≥2. One o he mos impo an ools is he modula me hod, de eloped by Wiles [26]. In [11], [12], [1], [13] all h ee ypes o e na y equa ions (i.e. o signa u es (n, n, 2),(n, n, 3),(n, n, n)) and ela ed esul s o Wiles [26], K aus [16], Da mon and Me el [7], Ribe [20], Benne and Skinne [2], Benne , Va sal and Yazdani [3] and o he s a e used. Howe e , he modula echnique wo ks effec i ely only o ”la ge” exponen s, ypically o n≥7. Thus he ”small” exponen s n= 2,3,5 mus be handled sepa a ely. In ac hese cases a e conside ed in dis inc sec ions, o a e co e ed by sepa a e heo ems in he abo e men ioned pape s. Fu he , he exponen s n= 2,3 has al eady been conside ed in sepa a e pa- pe s. Equa ion (1) wi h n= 2 has a b oad li e a u e in i sel ; see e.g. [15] and he e e ences gi en he e. He e we ocus only on he esolu ion o (1) wi h fixed k. Fo n= 2 and posi i e x, equa ion (1) has been comple ely sol ed (up o a ew excep ional cases) by Hi a a-Kohno, Laish am, Sho ey and Tijdeman [15] o k≤100, and in case o b= 1, e en o k≤109. Thei main ools we e ellip ic cu es and quad a ic esidues. La e , he excep ional emaining cases ha e been handled by Tengely [24], by he help o he Chabau y me hod. A his poin we no e ha we shall e e o he Chabau y me hod equen ly in his pape . Fo he desc ip ion o he me hod, and in pa icula how o use i in he ame o he p og am package Magma [4], we e e o he pape s o B uin [5], [6] and he e e ences gi en he e. When n= 3, wo king mainly wi h cubic esidues, howe e making use o el- lip ic cu es and he Chabau y me hod as well, Hajdu, Tengely and Tijdeman [14] ob ained all solu ions o equa ion (1) wi h k < 32 such ha P(b)≤ki 4≤k≤12 and P(b)< k i k= 3 o k≥13. Fu he , i b= 1 hen hey could sol e (1) o k < 39. The case n= 5 has no ye been closely in es iga ed. In his case (in he abo e men ioned pape s conside ing equa ion (1) o gene al exponen n) mainly classical me hods we e used, due o Di ichle and Lebesgue (see e.g. [13]). Appa en ly, o n= 5 ellip ic cu es a e no applicable. In he p esen pape we show ha in his case he Chabau y me hod (bo h he classical and he ellip ic e sion) can be applied e y efficien ly. As we men ioned, he Chabau y me hod has been al eady used o he cases n= 2,3 in [1], [24], [14]. Howe e , 3 i has been applied only o some pa icula cases and equa ions. To p o e ou esul s we sol e a la ge numbe o genus 2 equa ions by Chabau y me hod, and hen build a kind o sie e sys em based upon hem. 2 New esul s Ou fi s heo em conside ably ex ends Theo em A, in he mos in e es ing case o b= 1 in equa ion (1). We call an a i hme ic p og ession o he o m x, x +d, . . . , x + (k−1)dp imi i e, i gcd (x, d) = 1. Theo em 1 The p oduc o kconsecu i e non-ze o e ms in a p imi i e a i h- me ic p og ession wi h 3≤k≤54 is ne e a fi h powe . In ac Theo em 1 ollows di ec ly om he nex esul . To o mula e i , we need o in oduce a new concep . An a i hme ic p og ession x, x +d, . . . , x + (k−1)dis called i ial i d≤5 and |x+id| ≤ 15 o some i= 0,1, . . . , k −1. Fu he , a solu ion o equa ion (1) is also called i ial, i he e ms x, x + d, . . . , x+(k−1)don he le -hand side o (1) o m a i ial a i hme ic p og es- sion. This concep is needed because o he huge numbe o i ial solu ions; on he o he hand, such solu ions o (1) can be lis ed easily o any fixed k. Theo em 2 Equa ion (1) wi h n= 5,3≤k≤24 and P(b)≤Pkhas he only non i ial solu ions wi h (k, d) = (3,7), x ∈ {−16,−8,−6,2}; (k, d) = (4,7), x ∈ {−16,−15,−12,−9,−6,−5}; (k, d) = (4,11), x ∈ {−27,−6}; (k, d) = (5,7), x ∈ {−16,−12}; (k, d) = (5,11), x ∈ {−36,−32,−12,−8}; (k, d) = (5,13), x ∈ {−40,−27,−25,−12}; (k, d) = (6,7), x ∈ {−32,−25,−10,−3}; (k, d) = (6,9), x ∈ {−25,−20}; (k, d) = (6,13), x ∈ {−40,−25}; (k, d) = (7,7), x ∈ {−39,−32,−27,−22,−20,−15,−10,−3}; (k, d) = (8,7), x ∈ {−39,−27,−22,−10}; (k, d) = (9,7), x ∈ {−39,−34,−32,−24,−22,−17}; (k, d) = (10,7), x ∈ {−39,−24}, 4 whe e he alues o Pka e gi en by k3 4 5 6 7,8 Pk3 5 7 11 13 k9,10,11,12 13,14,15 16,17 18,19,20,21,22,23 24 Pk17 19 23 29 31 Obse e ha Pk> k o k≥4 in Theo em 2, which is a new ea u e abou equa ion (1). As a simple and immedia e co olla y o Theo em 2 we ge he ollowing s a e- men , conce ning he case P(b)≤k. We men ion ha al eady his esul yields conside able imp o emen o Theo em A, in pa icula wi h espec o he bound o P(b). Co olla y 3 Fo n= 5 and 3≤k≤36 all non i ial solu ions o equa ion (1) wi h P(b)≤ka e gi en by (k, d) = (3,7), x ∈ {−16,−8,−6,2}; (k, d) = (5,7), x ∈ {−16,−12}. Ou las heo em p o ides he key o he p oo o Theo em 2 in case o k≥4. I has been p o ed by a kind o sie ing p ocedu e, based upon genus 2 equa- ions and he Chabau y me hod. No e ha ha ing an inc easing a i hme ic p og ession z1< . . . < zl, by symme y we ob ain ha −zl< . . . < −z1is also an inc easing a i hme ic p og ession. Hence dealing wi h such a i hme ic p og essions i is sufficien o gi e only one p og ession om each symme ic pai . Theo em 4 Le 4≤ ≤8and z0< z1< . . . < z −1be a non- i ial p imi i e a i hme ic p og ession. Suppose ha z0=b0x5 0, zi1=bi1x5 i1, zi2=bi2x5 i2, z −1=b −1x5 −1, wi h some indices 0< i1< i2< −1such ha P(b0bi1bi2b −1)≤5. Then he ini ial e m z0and common diffe ence z1−z0o he a i hme ic p og ession z0, . . . , z −1 o he sepa a e alues o = 4, . . . , 8up o symme y is one o = 4 : (−9,7),(−6,7),(−6,11),(−5,7); = 5 : (−32,17),(−25,13),(−20,11),(−16,13),(−12,7),(−12,11),(−12,13), (−10,7),(−8,7),(−8,11),(−4,7),(−3,7),(−1,7),(2,7),(4,7),(4,23); 5 = 6 : (−125,61),(−81,17),(−30,31),(−25,8),(−25,11),(−25,13),(−25,17), (−20,9),(−20,13),(−20,19),(−20,29),(−15,7),(−15,11),(−15,13),(−15,23), (−10,7),(−10,11),(−8,7),(−5,7),(−3,7),(−1,11),(−1,13),(1,7),(5,11); = 7 : (−54,19),(−54,29),(−48,23),(−30,11),(−30,13),(−27,17),(−24,13), (−18,7),(−18,11),(−18,13),(−18,19),(−16,11),(−15,7),(−12,7),(−12,11), (−10,7),(−6,7),(−6,11),(−4,9),(−3,13),(−2,7),(−2,17),(2,13),(3,7),(6,7), (8,7),(9,11),(18,7); = 8 : (−405,131),(−125,41),(−100,49),(−32,11),(−27,11),(−27,13), (−25,19),(−24,7),(−16,13),(−10,13),(−9,7),(−5,11),(−4,7),(−2,11), (−1,13),(−1,7),(1,7),(3,11),(4,11),(5,7),(6,17). 3 P elimina ies Be o e gi ing he p oo s o ou esul s, we explain some p inciples and ech- niques which shall be used a he equen ly la e on. We p esen hese ools sepa a ely because in his way he s uc u e o ou p oo s will be mo e ans- pa en . 3.1 Reducing equa ion (1) o a i hme ic p og essions o ”almos ” fi h powe s In a s anda d way, as gcd(x, d) = 1 and n= 5, any solu ion o equa ion (1) can be w i en as x+id =aix5 i(i= 0,1, . . . , k −1) (2) whe e xiis a non-ze o in ege and aiis a fi h powe ee posi i e in ege wi h P(ai)≤k. This obse a ion jus ifies he i le o he pape , as well: he membe s o he a i hme ic p og ession x, x +d, . . . , x + (k−1)da e ”almos ” n- h powe s. 3.2 Lis ing he possible coefficien uples Suppose ha ai1x5 i1< ai2x5 i2<··· < ai x5 i (3) a e (no necessa ily consecu i e) nonze o e ms o a p imi i e a i hme ic p og ession, wi h aijas in (2). In his subsec ion we explain a me hod o lis all he possible coefficien - uples (ai1, ai2, . . . , ai ) co esponding o (3). 6 Obse e ha knowing aijis equi alen o knowing he exponen s νp(aij) o he p imes p≤kin he ac o iza ion o aij. Take an a bi a y p ime p≤k di iding one o he e ms aijx5 ij, and suppose ha ij0is such an index ha νp(aij0x5 ij0)≥νp(aijx5 ij) o all j= 1, . . . , . Since he a i hme ic p og ession is assumed o be p imi i e, one can easily check ha hen o all j= 1, . . . , wi h j=j0we ha e νp(aijx5 ij) = νp(j−j0). As we ha e νp(aij0)<5, we can simply lis all possibili ies o he exponen s o he p ime pin he coefficien s ai1, ai2, . . . , ai . Then combining hese pos- sibili ies o all p imes p≤k, we can lis all he possible coefficien - uples (ai1, ai2, . . . , ai ) which may occu in (3). 3.3 Local es ing o coefficien uples As we will see, some o he coefficien uples lis ed in he p e ious subsec ion in ac canno occu as coefficien s o fi h powe s in a i hme ic p og essions. In many cases his can be shown al eady modulo mwi h some app op ia e choice o m. We shall use he moduli m= 11,25. Le 0 ≤i1< i2<···< i ≤k−1 be indices, and conside a coefficien - uple (ai1, ai2, . . . , ai ), which in ac we would like o exclude - ha is, we would like o show ha no co esponding subsequence ai1x5 i1, . . . , ai x5 i (4) o any app op ia e a i hme ic p og ession exis s. Fo his pu pose, conside (4) modulo m(wi h m= 11 o 25). Obse e ha o ha e such a sequence, we should find app op ia e fi h powe s modulo m. We check all he possibili ies. (Since we wo k wi h m= 11 and m= 25, he fi h powe s modulo ma e only {0,±1}and {0,±1,±7}, espec i ely.) Obse e ha by cop imali y, we know ha m|aij1, aij2yields ha m|j1−j2. I we find ha no fi h powe s modulo mexis ha ing also he p e ious p ope y, hen he ac ual coefficien uple (ai1, . . . , ai ) is no alid in he sense ha no unde lying subsequence (4) exis s. We shall illus a e how o use his es la e on. 3.4 Reducing he p oblem o genus 2equa ions We ound wo ways o ge access o genus 2 equa ions. 7 3.4.1 Reduc ion me hod I Suppose ha a0x5 0, a1x5 1, a2x5 2is an a i hme ic p og ession wi h nonze o e ms, and wi h common diffe ence d. Then we ha e (a1x5 1)2−a0x5 0·a2x5 2=d2 which a e he subs i u ions X=−x0x2/x2 1,Y=d/x5 1and A=a0a2,B=a2 1 yields he genus 2 equa ion AX5+B=Y2 in X, Y ∈Q. 3.4.2 Reduc ion me hod II Suppose ha aix5 i, ajx5 j, aux5 u, a x5 a e ou e ms o an a i hme ic p og ession. Then we ha e (j−u)aix5 i+ (u−i)ajx5 j= (j−i)aux5 u and (j− )aix5 i+ ( −i)ajx5 j= (j−i)a x5 . Mul iplying hese iden i ies we ge an equa ion o he o m AX10 +BX5Y5+CY 10 =DZ5,(5) whe e A= (j−u)(j− )a2 i,B= ((j−u)( −i) + (u−i)(j− ))aiaj, C= (u−i)( −i)a2 j,D= (j−i)2aua and X=xi,Y=xj,Z=xux . Then om (5) we can easily ge bo h genus 2 equa ions o e Q A1Z5 1+B1=X2 1and A2Z5 2+B2=X2 2 wi h he no a ion A1= 4AD,B1=B2−4AC,X1= 2AX5/Y 5+B,Z1= Z/Y 2and A2= 4CD,B2=B2−4AC,X2= 2CY 5/X5+B,Z2=Z/X2, espec i ely. The a ional poin s on he genus 2 cu es ob ained by bo h me hods (unde sui able assump ions) can be de e mined by he Chabau y me hod. Then, ollowing he co esponding subs i u ions backwa ds we can de e mine he ac ual membe s o he o iginal a i hme ic p og essions. No e ha in ac in case o k= 3 in he p oo o Theo em 2 we also use genus 1 cu es o e some numbe fields, which can be ea ed by he ellip ic Chabau y 8 me hod. Howe e , since hese a e pa icula cases, we do no include hem in his ”gene al” discussion. 4 P oo s We gi e he p oo s o ou esul s in a specific o de . Fi s we p o e he case k= 3 o Theo em 2. We do so because his esul is needed in he p oo o Theo em 4, which is he nex s ep. The la e esul gi es he key o de i e Theo em 2 o k≥4. Then we con inue by p o ing he cases k≥4 o Theo em 2 and i s co olla y. Finally, we gi e he p oo o Theo em 1, which easily ollows om Theo em 2. In he p oo o case k= 3 o Theo em 2 we shall make use o wo lemmas. The fi s one is due o Benne , B uin, Gy˝o y, Hajdu [1]. Lemma 5 Le Cbe a posi i e in ege wi h P(C)≤5. I he Diophan ine equa ion X5+Y5=CZ5 has solu ions in nonze o cop ime in ege s X, Y and Z, hen C= 2 and X= Y=±1. The second lemma is a esul o K aus [16]. Lemma 6 Le Aand Bbe cop ime posi i e in ege s wi h AB = 2α3β o nonnega i e in ege s αand βwi h α≥4. Then he Diophan ine equa ion AX5+BY 5=Z5 has no solu ions in cop ime nonze o in ege s X, Y and Z. P oo o he case k= 3 o Theo em 2. Fi s lis all he possible coefficien iples (a0, a1, a2) as in (2). This can be done by he me hod explained in Sub- sec ion 3.2. Al oge he we ob ain 182 such iples. Obse e ha a2x5 2, a1x5 1, a0x5 0 is also an a i hme ic p og ession. Hence by symme y i is sufficien o con- side hose 106 iples o which a0≤a2. (I will be clea om ou me hod ha we can do so wi hou loss o gene ali y indeed.) Clea ly, a0x5 0, a1x5 1, a2x5 2is also an a i hme ic p og ession modulo 11 and 25. So we can es he coefficien iples modulo 11 and 25, as explained in Subsec ion 3.3. A e he modulo 11 es we a e le wi h 88 iples; o example (1,1,6) ge s excluded by his me hod. The es modulo 25 excludes 6 mo e iples (e.g. (1,4,3)), and we a e le wi h 82 ones. 9 [6] N. B uin,Diophan ine equa ions o signa u e (n, n, 2), in: Disco e ing ma hema ics wi h Magma, Algo i hms Compu . Ma h. 19 (2006), 63–91. [7] H. Da mon, L. Me el,Winding quo ien s and some a ian s o Fe ma ’s Las Theo em, J. Reine Angew. Ma h. 490 (1997), 81–100. [8] L. E. Dickson,His o y o he heo y o numbe s. Vol. II: Diophan ine analysis, Chelsea Publishing Co., New Yo k 1966. [9] P. E d˝ os, J. L. Sel idge,The p oduc o consecu i e in ege s is ne e a powe , Illinois J. Ma h. 19 (1975), 292-301. [10] K. Gy˝ o y,On he diophan ine equa ion n(n+ 1) . . . (n+k−1) = bxl, Ac a A i h. 83 (1998), 87–92. [11] K. Gy˝ o y,Powe alues o p oduc s o consecu i e in ege s and binomial coefficien s, Numbe Theo y and I s Applica ions, Kluwe Acad. Publ. 1999, 145–156. [12] K. Gy˝ o y, L. Hajdu, N. Sa adha,On he Diophan ine equa ion n(n+ d). . . (n+ (k−1)d) = byl, Canad. Ma h. Bull. 47 (2004), 373-388. Co ec ion: Canad. Ma h. Bull. 48 (2005), 636. [13] K. Gy˝ o y, L. Hajdu, ´ A. Pin ´ e ,Pe ec powe s om p oduc s o consecu i e e ms in a i hme ic p og ession, Composi io Ma h. 145 (2009), 845-864. [14] L. Hajdu, Sz. Tengely, R. Tijdeman,Cubes in p oduc s o e ms in a i hme ic p og ession, Publ. Ma h. Deb ecen 74 (2009), 215–232. [15] N. Hi a a-Kohno, S. Laish am, T. Sho ey, R. Tijdeman,An ex ension o a heo em o Eule , Ac a A i h. 129 (2007), 71–102. [16] A. K aus,Majo a ions effec i es pou l´equa ion de Fe ma g´en´e alis´ee, Canad. J. Ma h. 49 (1997), 1139-1161. [17] A. K aus,On he equa ion xp+yq=z : a su ey, Ramanujan J. 3(1999), 315-333. [18] R. Obl´ a h,¨ Ube das P oduk ¨un au einande olgende Zahlen in eine a i hme ischen Reiche, Publ. Ma h. Deb ecen 1(1950), 222-226. [19] R. Obl´ a h,Eine Beme kung ¨ube P oduk e au einande olgende Zahlen, J. Indian Ma h. Soc. 15 (1951), 135-139. [20] K. Ribe ,On he equa ion ap+ 2αbp+cp= 0, Ac a A i h. 79 (1997), 7–16. [21] N. Sa adha,On pe ec powe s in p oduc s wi h e ms om a i hme ic p og essions, Ac a A i h. 82 (1997), 147–172. [22] T. N. Sho ey,Powe s in a i hme ic p og ession, in: A Pano ama in Numbe Theo y (G. W¨us holz, ed.), Camb idge Uni e si y P ess, Camb idge, 2002, 325- 336. [23] T. N. Sho ey,Powe s in a i hme ic p og ession (II), in: New Aspec s o Analy ic Numbe Theo y, Kyo o 2002, 202-214. 16 [24] Sz. Tengely,No e on he pape ”An ex ension o a heo em o Eule ” by Hi a a-Kohno e al., Ac a A i h. 134 (2008), 329–335. [25] R. Tijdeman,Diophan ine equa ions and diophan ine app oxima ions, in: Numbe Theo y and Applica ions, Kluwe Acad. P ess, 1989, 215-243. [26] A. Wiles,Modula ellip ic cu es and Fe ma ’s Las Theo em, Ann. Ma h. 141 (1995), 443–551. 17