• Sonuç bulunamadı

New Sums Identities In Weighted Catalan Triangle With The Powers Of Generalized Fibonacci And Lucas Numbers

N/A
N/A
Protected

Academic year: 2021

Share "New Sums Identities In Weighted Catalan Triangle With The Powers Of Generalized Fibonacci And Lucas Numbers"

Copied!
10
0
0

Yükleniyor.... (view fulltext now)

Tam metin

(1)

TRIANGLE WITH THE POWERS OF GENERALIZED FIBONACCI AND LUCAS NUMBERS

EMRAH KILIÇ AND AYNUR YALÇINER

Abstract. In this paper, we consider a generalized Catalan triangle de…ned by

km n

2n

n k

for positive integer m: Then we compute the weighted half binomial sums with the certain powers of generalized Fibonacci and Lucas numbers of the form

n X k=0 2n n + k km n X r tk;

where Xneither generalized Fibonacci or Lucas numbers, t and r are integers for 1 m 6:After we describe a general methodology to show how to compute the sums for further values of m.

1. Introduction

Shapiro [6] derived the following triangle similar to Pascal’s triangle with entries given by Bn;k = k n 2n n k ;

which called Catalan triangle because the Catalan numbers Cn= n+11 2nn

are the entries in the …rst column.

Shapiro derived sums identities from the Catalan triangle. For example, he gave the following identities:

n X p=1 (Bn;p)2= C2n 1 and n X p=1 Bn;pBn+1;p= C2n:

We also refer to [5] and references therein for other examples.

2000 Mathematics Subject Classi…cation. 11B37.

Key words and phrases. Catalan triangle, sums identites, partial binomial sum, recursions.

(2)

The authors [4] gave also an alternative proof of the identities above and established the following identity:

n

X

p=1

(pBn;p)2= (3n 2) C2(n 1):

In a somewhat di¤erent from the Catalan triangle, K¬l¬ç and Ionascu [2] derived the following result: for any a 2 C f0g ;

n X p=1 2n n + k a k+ a k = 1 an(a + 1) 2n + (n + 1) Cn:

The authors also gave applications to the generalized Fibonacci and Lucas sequences, de…ned by

Un = AUn 1+ Un 2;

Vn = AVn 1+ Vn 2;

where U0= 0; U1= 1; and V0= 2; V1= A; respectively.

The Binet forms are Un=

n n

and Vn = n+ n

where ; = (A p )=2 and = A2+ 4:

For example, we recall one result from [2]:

n X k=0 2n n + k U 2r k = 8 > > > > < > > > > : A2+ 4 r 2r r 22n 2+ rP1 t=0 ( 1)t(n+1) 2r t Vr2nt if r is even, A2+ 4 n r rP1 t=0 ( 1)t(n+1) 2rt Ur2nt if r is odd.

We de…ne a generalized Catalan triangle by taking mth power of

sum-mation index k as follow:

Dn;k(m) =

km

n 2n n k :

When m = 1; the generalized Catalan triangle is reduced to the usual Catalan triangle Bn;p:

In [3], the author considered and computed certain binomial sums weighted by the powers of the summation index.

(3)

In this paper we consider the sums of the forms: for all nonnegative integer m and a 2 Cn f0g S (n; m; a) := n X k=0 2n n + k km n a k+ ( 1)m a k :

The sums S (n; 0; a) were considered and exactly computed in [2]. We …rst exactly compute the sums S (n; 1; a) : Then by using the value of S (n; 1; a) ; we compute S (n; 2; a) : So we will compute S (n; m; a) by us-ing the value of S (n; m 1; a) for the value of m; m = 2; :::; 6: Then we describe a general methodology to compute further values of S (n; m; a) for m > 6. Also we present applications of our results.

2. New Sums Identities From the Catalan Triangle Firstly we compute S (n; 1; a) : Before it we need to evaluate a partial binomial sums by the following lemma. For partial binomial sums, we may refer to [1].

Lemma 1. For any nonnegative integer t;

t X j=0 2n j 1 j n = 2n 1 t :

Proof. (By induction on t) For t = 0; the claim is obvious. Suppose that the claim is true for k: We show that the claim is true for k + 1: Consider

t+1 X j=0 2n j 1 j n = 2n t + 1 1 t + 1 n + t X j=0 2n j 1 j n ;

which, by the induction hypothesis, equals 2n t + 1 1 t + 1 n + 2n 1 t ;

which, by using the recursion of the binomial coe¢ cient and the property n + 1 k + 1 = n + 1 k + 1 n k ;

(4)

gives us 2n t + 1 2n 1 t 1 t + 1 n + 2n 1 t = 2n 1 t 2n t + 1 1 t + 1 n + 1 = 2n 1 t 2n t + 1 1 = 2n t + 1 2n 1 t = 2n 1 t + 1 ; as claimed.

Now we start with our …rst result. Theorem 1. For n > 0 S (n; 1; a) = n X k=0 2n n k k n a k a k = 1 an (a 1) (a + 1) 2n 1 : Proof. Consider 1 a 1 n X k=0 2n n k k a n+k an k ; which equals n X k=0 2n n k ka n k a2k 1 a 1 = n X k=0 2n n k ka n k 2k X j=0 aj = n X k=0 2n n k k 2k X j=0 an k+j = 2n X t=0 t X j=0 2n j (n j) a t;

which, by Lemma 1, equals

2n X t=0 n 2n 1 t a t= n (a + 1)2n 1 ; which settles the proof.

As a result of Theorem 1, by taking a instead of a, we have the following Corollary: Corollary 1. For n > 0 n X k=0 2n n + k ( 1) k k ak a k = n ( 1) n an (a + 1) (a 1) 2n 1 :

(5)

As a variant of the result of Theorem 1, we have that n X k=0 2n n k k n a n+k an k = (a 1) (a + 1)2n 1 ; which is a polynomial in a: Second we give the result:

Corollary 2. For n > 0 S (n; 2; a) = 1

an(a + 1) 2n 2

n (a + 1)2 2a (2n 1) :

Proof. Consider derivation of the RHS of S (n; 1; a) : d da n X k=0 2n n + k k n a k a k = n X k=0 2n n + k k n d da a k a k = n X k=0 2n n + k k2 n a k 1+ a k 1 = 1 a n X k=0 2n n + k k2 n a k+ a k = 1 aS (n; 2; a) :

On the other hand by taking derivation of the LHS of S (n; 1; a) gives d daS (n; 1; a) = d da 1 an(a 1) (a + 1) 2n 1 = 1 an+1(a + 1) 2n 2 n (a + 1)2 2a (2n 1) : Thus S (n; 2; a) = 1 an(a + 1) 2n 2 n (a + 1)2 2a (2n 1) ; as claimed.

We see that by taking derivation of S (n; 1; a) ; we obtain exact formula for S (n; 2; a) : The process of taking consecutive derivatives could be con-tinued and so we get

S (n; 3; a) = a n(a 1) (a + 1)2n 3hn2(a + 1)2 a (2n 1) (2n 2)i;

S (n; 4; a) = a n(a + 1)2n 4[n3(a + 1)4 2a (2n 1)

(6)

S (n; 5; a) = a n(a 1) (a + 1)2n 5[n4(a + 1)4 a (2n 1) (2n 2) (2n (n 1) + 1) (a + 1)2 (2n 3) (2n 4) a ] and so S (n; 6; a) = a n(a + 1)2n 6[n5(a + 1)6 2a (2n 1) [(a + 1)4 + (n 1) 3n3 3n2+ 4n (a + 1)4 2a (2n 3) (a + 1)2 3n2+ 5 6n 2a 2n2 9n + 10 ii: Generally taking derivative of S (n; m; a) gives a 1S (n; m + 1; a) : Since we can’t …nd an operator or a general recursion rule for them, we couldn’t derive a closed formula for the further values of S (n; m; a) : We leave this problem is an open problem.

3. New Weighted Half Binomial Sums

In this section, we present some applications of our results in order to weighted analogues of the results given [2] including powers of the summa-tion index with the even or odd powers of terms of the generalized binary linear recurrences fUng and fVng whose indices are also in arithmetic

pro-gressions as well as their alternating analogues. We prove the …rst result, others could be similarly derived.

Theorem 2. Let n 2 N [ f0g and r 2 N: If r is even,

n X k=0 2n n + k k 2U2r tk = n r 2r r 2 2n 2 + nr r 1 X j=0 ( 1)j(tn+1) 2r j V 2n 2 t(r j) nV 2 t(r j) ( 1)jt2(2n 1)

and, if r is odd and t is even,

n X k=0 2n n + k k 2U2r tk = n r 2r r 2 2n 2 + nr r 1 X j=0 ( 1)j 2r j V 2n 2 t(r j) nV 2 t(r j) 2(2n 1) ;

(7)

and, for n > 1; if r and t are odd, n X k=0 2n n + k k 2U2r tk = n n r 1 r 1 X j=0 ( 1)j(tn+1) 2r j U 2n 2 t(r j) n U 2 t(r j) ( 1) j(4n 2) . Proof. Expanding U2r

tk by the binomial formula, consider n X k=0 2n n + k k 2U2r tk = 1 ( )2r n X k=0 2n n + k k 2 2 4 2r X j=0 ( 1)j 2r j (2r j)tk jtk 3 5 : Since = 1, we write n X k=0 2n n + k k 2U2r tk = 1 r n X k=0 2n n + k k 2 ( 1)r(1+tk) 2r r + r 1 X j=0 ( 1)j(1+tk) 2r j ( 2(r j)tk+ 2(r j)tk) ;

which, by changing summation order and Corollary 2, equals n r ( 1)r 2 2r r S(n; 2; ( 1) tr) + n r r 1 X j=0 ( 1)j 2r j S(n; 2; ( 1) jt 2t(r j)) = nr ( 1) r 2 2r r n ( 1)tr+ 1 2n ( 1)trn ( 1)r 2r r (2n 1) ( 1)tr+ 1 2n 2 ( 1)tr(n 1) + r 1 X j=0 ( 1)j 2r j [n( 1) jtn(( 1)jt t(r j)+ ( 1)t(r j) t(r j))2n 2(2n 1)( 1)jt(n 1)(( 1)jt t(r j)+ ( 1)t(r j) t(r j))2n 2] which, by the Binet formulas of fUng and fVng ; gives us the claim.

(8)

Theorem 3. For n > 0 n X k=0 2n n + k ( 1) kkU tk= ( 1)nnUt 8 > < > : n 1U2n 2 t=2 if t 0 (mod 4); Vt=22n 2 if t 2 (mod 4); and, for even t;

n X k=0 2n n + k kUtk= n Ut 8 > < > : Vt=22n 2 if t 0 (mod 4); n 1U2n 2 t=2 if t 2 (mod 4):

Theorem 4. Let n 2 N [ f0g and r 2 N: For even r;

n X k=0 2n n + k k 2V2r tk = n 2r r 2 2n 2 + n r 1 X j=0 ( 1)jtn 2r j V 2n 2 t(r j) nV 2 t(r j)+ ( 1) jt+12(2n 1) : For odd r; (i) For even t;

n X k=0 2n n + k k 2V2r tk = n 2r r 2 2n 2 + n r 1 X j=0 2r j V 2n 2 t(r j) nV 2 t(r j) 2(2n 1) ; n 0

and for odd t;

n X k=0 2n n + k k 2V2r tk = n n 1 r 1 X j=0 2r j ( 1) jnU2n 2 t(r j) n U 2 t(r j) ( 1)j2(2n 1) ; n > 1

Theorem 5. For even t > 0,

n X k=0 2n n + k k 3U tk = nUt 8 > > < > > : Vt=22n 4 n2V2 t=2 (2n 1)(2n 2) if t 0 (mod 4); n 2U2n 4 t=2 n 2 U2 t=2 (2n 1)(2n 2) if t 2 (mod 4);

(9)

and, for all integer t; n X k=0 2n n + k ( 1) kk3U tk= ( 1)nnUt 8 > > < > > : n 2 U2n 4 t=2 n 2 U2 t=2+ (2n 1)(2n 2) if t 0 (mod 4); Vt=22n 4 n2V2 t=2+ (2n 1)(2n 2) if t 2 (mod 4):

Theorem 6. For positive even r;

n X k=0 2n n + k k 4V2r tk = n 2r r 2 2n 3(3n 1) + n r 1 X j=0 ( 1)jtn 2r j V 2n 4 t(r j) h n3Vt(r4 j)+( 1)jt+12(2n 1)(2n2 2n+1)Vt(r2 j)+4(2n 1)(n 1) (2n 3)] ; and, for odd r and even t;

n X k=0 2n n + k k 4V2r tk = n 2r r 2 2n 3(3n 1) + n r 1 X j=0 2r j V 2n 4 t(r j) h n3Vt(r4 j) 2(2n 1)(2n2 2n + 1)Vt(r2 j)+ 4(2n 1)(n 1)(2n 3)i; and, for n > 2 and odd r; t;

n X k=0 2n n + k k 4V2r tk = n n 2 r 1 X j=0 ( 1)jn 2r j U 2n 4 t(r j) h n3 2Ut(r4 j) ( 1)j2(2n 1) (2n2 2n + 1) Ut(r2 j)+ 4(2n 1)(n 1)(2n 3)i: Theorem 7. Let t be a positive even integer. If t 0 (mod 4),

n X k=0 2n n + k k 5U tk= nUtVt=22n 6(n4Vt=24 (2n 1)(2n 2) (2n2 2n + 1)Vt=22 + (2n 1)(2n 2)(2n 3)(2n 4)); and, if t 2 (mod 4); n X k=0 2n n + k k 5U tk= n n 3UtUt=22n 6(n4 2Ut=24 (2n 1)(2n 2) (2n2 2n + 1) Ut=22 + (2n 1)(2n 2)(2n 3)(2n 4): By using the presented results, one can derive many sums formulae sim-ilar to the above sums formulae.

(10)

Acknowledgement 1. Work of A.Y. is supported by the Scienti…c Re-search O¢ ce (BAP) of Selçuk University.

References

[1] R. L. Graham, D. E. Knuth and O. Patashnik, Concrete Mathematics: A Foundation for Computer Science. Addison-Wesley, 1992.

[2] E. K¬l¬ç and E. J. Ionascu, Certain binomial sums with recursive coe¢ cients, The Fibonacci Quarterly 48 (2) (2010), 161-167.

[3] E. K¬l¬ç, N. Ömür and Y.T. Uluta¸s, Some …nite sums involving generalized Fibonacci and Lucas numbers. Discrete Dyn. Nat. Soc. (2011), Art. ID 284261.

[4] J. M. Gutierrez, M. A. Hernandez, P. J. Miana and N. Romero, New identities in the Catalan triangle, J. Math. Anal. Appl. 341 (2008), 52-61.

[5] T. Mansour and Y. Sun, Identities involve Narayana polynomials and Catalan num-bers, Discr. Math. 309 (12) (2009), 4079–4088.

[6] L. W. Shapiro, A Catalan triangle, Discrete Math. 14 (1976), 83-90.

TOBB Economics and Technology University, Mathematics Department 06560 Sogutozu Ankara Turkey

E-mail address : ekilic@etu.edu.tr

Selcuk University, Science Faculty, Department of Mathematics, 42075, Cam-pus, Konya, Turkey

Referanslar

Benzer Belgeler

p 值為 0.093,趨近於 0.05。統計 SIDS 與 non-SIDS 於血液、骨骼肌及心肌 multiple deletion 的比率,兩組之間亦沒有顯著性差異,其中骨骼肌之 p 值為

The Constitutional Court also touched on the issue of applying federal relations pattern between the ex- ecutive and legislative authorities at the level of the constituent entities

Tablo 4.56 Macera Oyunlarını Oynama Sıklığına Göre Bilgisayar Oyunları ve Eğitsel Bilgisayar Oyunları Toplam Puanlarına Ait Kruskal Wallis H Testi Özet Tablosu .... 88

GÜNGÖR GÜNER B E R İL ANILANMERT GÜL DERMAN SEYHUN TOPUZ JALE (NEJDET) ERZEN MERİÇ HIZAL (BAŞOL) NADİDE AKDENİZ GÜLSÜN KARAMUSTAFA AYTAÇ MARMARA KATI TÜLİN

This is because, the division of the attributes of Allah into the attributes of essence and the attributes of action as constructed by Ibādhīyah, was also constructed by

(A) Effect of 20 ␮M palmitate or 20 ␮M 2BrP on the trafficking of GFP-Drd3 cargo vesicles in HEK293 cells cotransfected with GIPC1.. (B) Relative proportion of GFP-Drd3 vesicles

When event studies are carried out by ordinary least squares (OLS) regressions of asset price changes on surprises, as measured by the difference between released values