ebook img

Some $q$-analogs of congruences for central binomial sums PDF

0.13 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Some $q$-analogs of congruences for central binomial sums

SOME q-ANALOGS OF CONGRUENCES FOR CENTRAL BINOMIAL SUMS ROBERTO TAURASO 2 1 Abstract. We establish q-analogs for four congruences involving central binomial coeffi- 0 cients. The q-identitiesnecessary for this purposeare shown viathe q-WZ method. 2 n a J 1. Introduction 0 3 Recently, a number of papers have appeared concerning congruences for central binomial sums (see through the references). Here we would like to draw the attention to one aspect of ] the matter which has been partly neglected so far: q-analogs. In [5, 11], the authors identified T N a first group of such congruences which have a q-counterpart. Among them we mention: for any prime p > 2 . h t ma p−1 2k ≡ p (mod p2), p−1qk 2k ≡ p q⌊p32⌋ (mod [p]q), k 3 k 3 [ Xkp−=10(cid:18) (cid:19) (cid:16) (cid:17) Xkp−=10 (cid:20) (cid:21)q (cid:16) (cid:17) v1 (−1)k 2kk ≡ p5 (mod p), (−1)kq−(k+21) 2kk ≡ p5 q−⌊p54⌋ (mod [p]q), 2 Xk=0 (cid:18) (cid:19) (cid:16) (cid:17) Xk=0 (cid:20) (cid:21)q (cid:16) (cid:17) 15 p−1 21k 2kk ≡ (−1)p−21 (mod p2), p−1 (−qq;kq) 2kk ≡ (−1)p−21q⌊p42⌋ (mod [p]q), 6 k=0 (cid:18) (cid:19) k=1 k(cid:20) (cid:21)q . X X 1 0 where · denotes the Legendre symbol. It has been conjectured in [5] that the first q- · 2 congruence holds modulo [p]2, and we claim that the same can be said for the third one. 1 (cid:0) (cid:1) q However, in this short note, we are not going to refine these q-congruences. Instead, we will : v presentafewmoreexamples of thisphenomena. Moreprecisely weshowthatthecongruences i X r p−1 (1/2)k 2k 3 p−1 (−1/2)k 2k −1 a (1) ≡ − ≡ Q (2) (mod p), p k k 2 k k k=1 (cid:18) (cid:19) k=1 (cid:18) (cid:19) X X p−1 1 2k (2) ≡ 0 (mod p2), k k k=1 (cid:18) (cid:19) X 5 p−1 (−1)k 2k p−1 1 (3) ≡ − (mod p3), 2 k2 k k2 k=1 (cid:18) (cid:19) k=1 X X 2000 Mathematics Subject Classification. 11B65, 11A07 (Primary) 05A10, 05A19, 05A30 (Secondary). Key words and phrases. q-analogs, Gaussian q-binomial coefficients, central binomial coefficients, congruences. 1 2 ROBERTOTAURASO where p > 5 is a prime and Q (2) = (2p−1 − 1)/p is the usual Fermat’s quotient, have as p q-analogs respectively p−1 qk 2k 1 p−1 (−1)k(1+qk +q2k) 2k −1 (4) ≡ − ≡ Q (2;q) (mod [p] ), p q (−q;q) [k] k 2 (−q;q) [k] k k=1 k q(cid:20) (cid:21)q k=1 k q (cid:20) (cid:21)q X X (5) p−1 (1+qk +q2k)q−(k2) 2k ≡ [p]q(p2−1)(1−q)2 (mod [p]2), (1+qk)2[k] k 24 q k=1 q (cid:20) (cid:21)q X (6) p−1 (−1)k(1+3qk +q2k)q−(k2) 2k ≡ −p−1 qk − [p]2q(p4−1)(1−q)4 (mod [p]3). (1+qk)[k]2 k [k]2 240 q k=1 q (cid:20) (cid:21)q k=1 q X X where Q (2;q) = ((−q;q) −1)/[p] . Proofs of (1), (2), (3) can be found in [12, Theorem p p−1 q 3.1] ((2) appeared first in [10]). We are optimistically hopeful that there are plenty of interesting q-analogs to discover. For example, recently in [8], the authors proved that for 0 < q < 1 ∞ (1+2qk)qk2 2k −1 ∞ qk = . [k]2 k [k]2 k=1 q (cid:20) (cid:21)q k=1 q X X By letting q → 1, it gives a well known series identity which then happened to have a congruence version: in [12] we showed that for any prime p > 3 p−1 −1 p−1 1 2k 1 1 ≡ − (mod p3). k2 k 6 k2 k=1 (cid:18) (cid:19) k=1 X X Is there a q-analog for the above congruence? 2. Notations a preliminary results The first two results of this section yield a family of q-analogs of the classical congruence for the harmonic sums: for any prime p > d+2 where d is a positive integer, p−1 1 0 (mod p2) if d is odd, H (d) := ≡ p−1 kd 0 (mod p) if d is even. k=1 (cid:26) X This family depends on two integer parameters a,b and it concerns the sum p−1 qbk [ak]d q k=1 X where 1−qn [n] = = 1+q+···+qn−1. q 1−q Several special cases have already been discussed by numerous authors (see [3, 4, 7, 9]). In particular, K. Dilcher found in [4] a determinant expression in the case when a = 1 and b ∈ {0,1}. We point out that, in this paper, two rational functions in q are congruent modulo [p]r for r ≥ 1 if the numerator of their difference is congruent to 0 modulo [p]r in the q q polynomial ring Z[q] and the denominator is relatively prime to [p] . q SOME q-ANALOGS OF CONGRUENCES FOR CENTRAL BINOMIAL SUMS 3 Theorem 2.1. For any prime p > 2, if a,b,d are integers such that a,d > 0, b ≥ 0 and gcd(a,p) = 1 then p−1 qbk (1−q)d d−1 r +sp d d sp (7) ≡ (−1)dp c 0 − (−1)s (mod [p] ) [ak]d pd s d s 2d q k=1 q s=0 (cid:18) (cid:19) s=0 (cid:18) (cid:19)(cid:18) (cid:19)! X X X where r ≡−b/a (mod p) such that r ∈ {0,1,...,p−1} and 0 0 s r +kp+d−1 d c = (−1)s−k 0 . s d−1 s−k k=0 (cid:18) (cid:19)(cid:18) (cid:19) X Proof. Let q be a p-root of unity such that q 6= 1. Since p−1 p if p | (aj +b), qk(aj+b) = −1+ 0 otherwise, k=1 (cid:26) X it follows that p−1 qbk p−1 j+d−1 = qbk qakjzj (1−qakz)d d−1 k=1 k=1 j≥0(cid:18) (cid:19) X X X p−1 j +d−1 = zj qk(aj+b) d−1 j≥0(cid:18) (cid:19) k=1 X X r +sp+d−1 1 = p 0 zr0+sp− d−1 (1−z)d s≥0(cid:18) (cid:19) X p d−1c zr0+sp 1 = s=0 s − . (1−zp)d (1−z)d P Let z =1+w, then p−1 qbk p ds=−01cs(1+w)r0+sp− 1−(−1+ww)p d = lim (1−qak)d w→0 P (1−(1+w)p)(cid:16)d (cid:17) k=1 X p d−1c r0+sp wd+o(wd)−(−1)d d (−1)s d sp wd+o(wd) = lim s=0 s d s=0 s 2d w→0 (−pw+o(w))d P (cid:0) (cid:1) P (cid:0) (cid:1)(cid:0) (cid:1) d−1 d 1 r +sp d sp = (−1)dp c 0 − (−1)s . pd s d s 2d ! s=0 (cid:18) (cid:19) s=0 (cid:18) (cid:19)(cid:18) (cid:19) X X (cid:3) 4 ROBERTOTAURASO Thefollowing specialcases areworth mentioning. By letting d= 1,2,3 in (7), weobtain these q-congruences which hold modulo [p] : q p−1 qbk p−1 (8) ≡ (1−q) −r , 0 [ak] 2 k=1 q (cid:18) (cid:19) X p−1 qbk (p−1)(p−5) r (p−2−r ) (9) ≡ (1−q)2 − + 0 0 , [ak]2 12 2 k=1 q (cid:18) (cid:19) X p−1 qbk (p−1)(p−3) r (p2−9p+12−3r (p−3)+2r2) (10) ≡ (1−q)3 − − 0 0 0 . [ak]3 8 12 k=1 q (cid:18) (cid:19) X Theorem 2.2. Let b,b,a,d be non-negative integers such that ad = b+b > 0. Then for any prime p > 2 such that gcd(a,p) = 1, p−1 (−1)d−1qbk +qbk p−1 qbk p−1 qbk (11) ≡b(1−q)[p] −ad[p] (mod [p]2) [ak]d q [ak]d q [ak]d+1 q k=1 q k=1 q k=1 q X X X and (12) p−1 (−1)k((−1)dqbk +qbk) p−1 (−1)kqbk p−1 (−1)kqbk ≡ b(1−q)[p] −ad[p] (mod [p]2). [ak]d q [ak]d q [ak]d+1 q q q q k=1 k=1 k=1 X X X Moreover, let b ,b ,b ,b be non-negative integers. 1 1 2 2 If d = b +b > 0 and d = b +b >0 then 1 1 1 2 2 2 (13) qb1j+b2k +(−1)d1+d2qb1j+b2k p−1 qb1j p−1 qb2k p−1 q(b1+b2)k ≡ · − (mod [p] ). [j]d1[k]d2 [j]d1 [k]d2 [k]d1+d2 q 1≤j<k≤p−1 q q k=1 q k=1 q k=1 q X X X X Proof. As regards (11) and (12), it suffices to note that (−1)dqb(p−k) (−1)dqb(p−k)+adk qbp+bk = = [a(p−k)]d ([ap] −[ak] )d [ak]d(1−[ap] /[ak] )d q q q q q q qbk(1−b(1−q)[p] ) [p] q q ≡ 1+d [ak]d [ak] q (cid:18) q(cid:19) qbk b(1−q)[p] qbk ad[p] qbk (14) ≡ − q + q (mod [p]2). [ak]d [ak]d [ak]d+1 q q q q Moreover p−1 qb1j p−1 qb2k p−1 q(b1+b2)k qb1j+b2k qb1j+b2k · − = + [j]d1 [k]d2 [k]d1+d2 [j]d1[k]d2 [j]d1[k]d2 k=1 q k=1 q k=1 q 1≤j<k≤p−1 q q 1≤k<j≤p−1 q q X X X X X and by (14) we get qb1j+b2k qb1(p−j)+b2(p−k) (−1)d1+d2qb1j+b2k = ≡ (mod [p] ). [j]d1[k]d2 [p−j]d1[p−k]d2 [j]d1[k]d2 q 1≤k<j≤p−1 q q 1≤j<k≤p−1 q q 1≤j<k≤p−1 q q X X X Hence the proof of (13) is complete. (cid:3) SOME q-ANALOGS OF CONGRUENCES FOR CENTRAL BINOMIAL SUMS 5 By letting a = 1, d = 1 and b = 0 in (11), and by using (7), we easily find [9, Theorem 1]: for any prime p > 3: p−1 1 1 p−1 1−qk 1 p−1 1+qk (1−q)(p−1) (p2−1)(1−q)2[p] = ≡ + q (mod [p]2). [k] 2 [k] 2 [k] 2 24 q q q q k=1 k=1 k=1 X X X In a similar way, for a = 1, d= 3 and b = 1, (11) yields (15) p−1 qk +q2k p−1 q2k p−1 q2k [p] (1−q)4(p4−1) ≡ (1−q)[p] −3[p] ≡ − q (mod [p]2). [k]3 q [k]3 q [k]4 240 q q q q k=1 k=1 k=1 X X X Inorder to show theq-congruences stated in theintroduction, we needsuitable q-identities. Such identities are not easy to find, but ones they are guessed correctly hopefully they can be proved via the q-WZ method (see for example [6, 13]). For n ≥ k ≥ 0, a pair (F(n,k),G(n,k)) is called q-WZ pair if F(n+1,k)/F(n,k), F(n,k+1)/F(n,k), G(n+1,k)/G(n,k), G(n,k+1)/G(n,k) are all rational functions of qn and qk, and F(n+1,k)−F(n,k) = G(n,k+1)−G(n,k). Let N−1 S(n) = F(n,k) k=0 X then n−1 S(n+1)−S(n) = F(n+1,n)+ (F(n+1,k)−F(n,k)) k=0 X n−1 = F(n+1,n)+ (G(n,k+1)−G(n,k)) k=0 X = F(n+1,n)+G(n,n)−G(n,0) and, by summing over n from 0 to N −1, we get the identity N−1 N−1 N−1 (16) F(N,k) = (F(n+1,n)+G(n,n))− G(n,0) k=0 n=0 n=0 X X X which can be considered as the finite form of [6, Theorem 7.3]. The q-identities we are interested in involves Gaussian q-binomial coefficients n (q;q) (q;q)−1(q;q)−1 if 0 ≤ k ≤n, = n k n−k (cid:20)k(cid:21)q ( 0 otherwise, where (a;q) = n−1(1−aqj) (note that n is a polynomial in q). The next lemma allows n j=0 k q us to reduce a special class of q-binomial coefficients modulo a power of [p] . Q (cid:2) (cid:3) q 6 ROBERTOTAURASO Lemma 2.3. Let p be a prime and let a be a positive integer. For k = 1,...,p−1 we have k (17) apk−1 ≡ (−1)kq−(k+21)1−a[p]q [j1]  (mod [p]2q), (cid:20) (cid:21)q j=1 q X   p−1+k [p] k−1 qj (18) ≡ q 1+[p] (mod [p]3), k [k]  q [j]  q (cid:20) (cid:21)q q j=1 q X   (19) p−1+k p−1 −1 ≡ (−1)kq(k+21)[p]q 1+ [p]q +[p] k−1 1+qj (mod [p]3). k k [k]  [k] q [j]  q (cid:20) (cid:21)q(cid:20) (cid:21)q q q j=1 q X   Proof. Since [ap] ≡ a[p] (mod [p]2), it follows that q q q k k apk−1 = (−1)kq−(k+21) 1− [a[jp]]q ≡ (−1)kq−(k+21)1−a[p]q [j1]  (mod [p]2q). (cid:20) (cid:21)q j=1(cid:18) q (cid:19) j=1 q Y X Moreover   p−1+k [p] k−1 qj[p] [p] k−1 qj = q 1+ q ≡ q 1+[p] (mod [p]3). k [k] [j] [k]  q [j]  q (cid:20) (cid:21)q q j=1(cid:18) q (cid:19) q j=1 q Y X Conguences (17) and (18) easily yield (19).   (cid:3) It should be noted that when p is an odd prime, by using (18) for k = p−1, we recover the q-congruence [3, (3.2)]: (20) p−1 app−−11 ≡ q−(p2)1−a[p]q [j1] ≡ q−(p2) 1− a[p]q(p−21)(1−q) ≡ q(a−1)(p2) (mod [p]2q). (cid:20) (cid:21)q j=1 q (cid:18) (cid:19) X   3. Proof of (4) By [2, (5.17)] (see [5, (4.1)] for a generalization), if n is odd then by n (−1)n−kq(n−2k) n 2k = 0. (−q;q) k k k=0 k (cid:20) (cid:21)q(cid:20) (cid:21)q X Hence for n = p we have that p−1 (−1)k−1q(p−2k)−(p2) p−1 2k = 1 (−q;q)p−1−q−(p2) 2p−1 . (−q;q) [k] k−1 k (−q;q) [p] p−1 k=1 k q (cid:20) (cid:21)q(cid:20) (cid:21)q p−1 q (cid:20) (cid:21)q! X By (17) and (20) we get p−1 qk 2k p−1 q−pk+k 2k (−q;q) −1 p−1 ≡ ≡ ≡Q (2;q) (mod [p] ), p q (−q;q) [k] k (−q;q) [k] k (−q;q) [p] k=1 k q(cid:20) (cid:21)q k=1 k q(cid:20) (cid:21)q p−1 q X X and the first congruence is proved. As regards the second one, we take (−1)k n+k+1 −1 qn+1F(n,k) F(n,k) = and G(n,k) = . (−q;q) [k+1] k+1 1+qn+1 n q(cid:20) (cid:21)q SOME q-ANALOGS OF CONGRUENCES FOR CENTRAL BINOMIAL SUMS 7 This pair can be found in [1][Subsection 2.1] in connection with the irrationality proof of the q-series ∞ (−1)k Ln (2) := q qn−1 k=1 X for |q| 6∈ {0,1}. Hence by (16) we obtain the identity n (−1)k(1+qk +q2k) 2k −1 1 n (−1)k n+k −1 n qk (21) = − (−q;q) [k] k (−q;q) [k] k (−q;q) [k] k=1 k q (cid:20) (cid:21)q n k=1 q (cid:20) (cid:21)q k=1 k q X X X Let n = p−1. Now by (17) and [7, (1.5)] 1 p−1 (−1)k n+k −1 p−1 (−1)k k−1 qj ≡ 1−[p] q (−q;q) [k] k [p]  [j]  p−1 k=1 q (cid:20) (cid:21)q k=1 q j=1 q X X X p−1 k−1 qj p−1q2j−1 ≡ − (−1)k ≡− [j] [2j −1] q q k=1 j=1 j=1 X X X p−1 (p−1)/2 (p−1)(1−q) 1 1 ≡ − + ≡ −Q (2;q) (mod [p] ). p q 2 [j] [2j] q q k=1 k=1 X X By the q-binomial theorem, n k−1 n (x−qj)= xn k k=0(cid:20) (cid:21)qj=0 X Y and for n = p, x = −1 together with (17), we get p−1 q−(k2)(−q;q)k−1 p−1 (−1)k−1(−q;q)k−1 p−1 ≡ = −Q (2;q). p [k] [k] k−1 k=1 q k=1 q (cid:20) (cid:21)q X X Hence (see the dual congruence [7, (5.4)]) p−1 qk p−1 qp−k p−1 q−(k2)(−q;q)k−1 ≡ ≡ ≡ −Q (2;q) (mod [p] ) p q (−q;q) [k] (−q;q) [p−k] [k] k q p−k q q k=1 k=1 k=1 X X X where we used [p−k]q = −q−k[k]q and (−q;q)−p−1k ≡ q−(k2)(−q;q)k−1 (mod [p]q). Therefore, by identity (21), p−1 (−1)k(1+qk +q2k) 2k −1 ≡ −2Q (2;q) (mod [p] ) p q (−q;q) [k] k k=1 k q (cid:20) (cid:21)q X and we are done. 8 ROBERTOTAURASO 4. Proof of (5) Let q−(k+21) n+k+2 n+1 −1 F(n,k) = [k+1] n+1 k+1 q(cid:20) (cid:21)q(cid:20) (cid:21)q and qn+2(1−qk+1)(1−qn+1−k)F(n,k) G(n,k) = − (1+qn+2)(1−qn+2)2 then (16) gives the identity n (1+qk +q2k)q−(k2) 2k n q−(k2) n+k n −1 n qk (22) = − . (1+qk)2[k] k [k] k k [2k] k=1 q (cid:20) (cid:21)q k=1 q (cid:20) (cid:21)q(cid:20) (cid:21)q k=1 q X X X Let n = p−1 then by (19) and (12), we obtain p−1 q−(k2) p−1+k p−1 −1 p−1 (−1)kqk ≡[p] ≡ 0 (mod [p]2). [k] k k q [k]2 q k=1 q (cid:20) (cid:21)q(cid:20) (cid:21)q k=1 q X X Moreover, (11) and (9) implies that p−1 qk p−1 qk [p] (p2−1)(1−q)2 ≡ −[p] ≡ − q (mod [p]2), [2k] q [2k]2 24 q q q k=1 k=1 X X and (5) follows easily from (22). Note that by letting q → 1 in (22) we obtain the identity n n −1 n 3 1 2k 1 n+k n 1 1 = − . 4 k k k k k 2 k k=1 (cid:18) (cid:19) k=1 (cid:18) (cid:19)(cid:18) (cid:19) k=1 X X X which can be exploited to prove an improvement of [12, Theorem 4.2]: p−1 1 2k 8 ≡− H (1)+2p4B (mod p5) p−1 p−5 k k 3 k=1 (cid:18) (cid:19) X for any prime p > 3. 5. Proof of (6) By taking (−1)kq−(k+21)(1+qk+1) n+k+2 n+1 −1 F(n,k) = [k+1]2 n+1 k+1 q (cid:20) (cid:21)q(cid:20) (cid:21)q and qn+2(1−qk+1)2(1−qn+1−k)F(n,k) G(n,k) = , (1+qk+1)(1−qn+2)3 (16) yields the identity (23) n (−1)k(1+3qk +q2k)q−(k2) 2k n (−1)k(1+qk)q−(k2) n+k n −1 n qk = − . (1+qk)[k]2 k [k]2 k k [k]2 k=1 q (cid:20) (cid:21)q k=1 q (cid:20) (cid:21)q(cid:20) (cid:21)q k=1 q X X X SOME q-ANALOGS OF CONGRUENCES FOR CENTRAL BINOMIAL SUMS 9 Let n = p−1. Now by (19) p−1 (−1)k(1+qk)q−(k2) p−1+k p−1 −1 p−1 qk +q2k p−1 qk +q2k ≡ [p] +[p]2 [k]2 k k q [k]3 q [k]4 k=1 q (cid:20) (cid:21)q(cid:20) (cid:21)q k=1 q k=1 q X X X (1+qj)(qk +q2k) +[p]2 (mod [p]3). q [j] [k]3 q q q 1≤j<k≤p−1 X Then the q-congruence (6) follows from (23) by using (15) and p−1 qk +q2k (1+qj)(qk +q2k) (1−q)4(p2−1)(p2 −4) ≡ − ≡ (mod [p] ), [k]4 [j] [k]3 360 q q q q k=1 1≤j<k≤p−1 X X which is a straightforward application of (7) and (13). References [1] T.Amdeberhan,D.Zeilberger,q-Ap´eryirrationalityproofsbyq-WZpairs,Adv.inAppl.Math.20(1998), 275–283. [2] G. E. Andrews, Applications of basic hypergeometric functions, SIAMRev.16 (1974), 441–484. [3] G.E.Andrews,q-Analogs ofthebinomialcoefficientcongruences ofBabbage, WolstenholmeandGlaisher, Discrete Math. 204 (1999), 15–25. [4] K.Dilcher,Determinant expressions forq-harmonic congruences and degenerate Bernoulli numbers,Elec- tron. J. Combin. 15 (2008), 1–18. [5] V. J. W. Guo, J. Zeng, Some congruences involving central q-binomial coefficients, Adv. in Appl. Math. 45 (2010), 303–316. [6] M. Mohammed, The q-Markov-WZ method, Ann.Comb. 9 (2005), 205-221. [7] H.Pan, A q-analogue of Lehmer’s congruence, Acta Arith.128 (2007), 303–318. [8] Kh. Hessami Pilehrood, T. Hessami Pilehrood, A q-analogue of the Bailey-Borwein-Bradley identity, J. Symbolic Comput. 46 (2011), 699–711. [9] L.L.Shi,H.Pan,A q-analogue ofWolstenholme’sharmonicseriescongruenc,Amer.Math.Monthly114 (2007), 529–531. [10] Z.W.Sun,R.Tauraso,New congruences forcentral binomialcoefficients,Adv.inAppl.Math.45(2010), 125–148. [11] R.Tauraso, q-Analogs of some congruences involving Catalan numbers, toappearinAdv.inAppl.Math. [12] R.Tauraso,Morecongruences forcentral binomialcoefficients,J.NumberTheory130(2010),2639–2649. [13] D. Zeilberger, Closed form (pun intended!), Contemp. Math. 143 (1993), 579–607. E-mail address: [email protected] Dipartimento di Matematica, Universita` di Roma “Tor Vergata”, via della Ricerca Scien- tifica, 00133 Roma, Italy

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.