ON SUMS OF SQUARES OF PELL-LUCAS NUMBERS. Gian Mario Gianella University of Torino, Torino, Italy, Europe.

Transcription

1 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 ON SUMS OF SQUARES OF PELL-LUCAS NUMBERS Zvonko Čerin University of Zagreb, Zagreb, Croatia, Europe cerin@math.hr Gian Mario Gianella University of Torino, Torino, Italy, Europe gianella@dm.unito.it Received: 2/18/06, Revised: 4/10/06, Accepted: 4/24/06, Published: 5/15/06 Abstract In this paper we prove several formulas for sums of squares of even Pell-Lucas numbers, sums of squares of odd Pell-Lucas numbers, and sums of products of even and odd Pell- Lucas numbers. These sums have nice representations as products of appropriate Pell and Pell-Lucas numbers with terms from certain integer sequences. 1. Introduction The Pell and Pell-Lucas sequences P n and Q n are defined by the recurrence relations P 0 = 0, P 1 = 2, P n = 2 P n 1 + P n 2 for n 2, and Q 0 = 2, Q 1 = 2, Q n = 2 Q n 1 + Q n 2 for n 2. In the Sections 2 4 we consider sums of squares of odd and even terms of the Pell-Lucas sequence and sums of their products. These sums have nice representations as products of appropriate Pell and Pell-Lucas numbers. The numbers Q k make the integer sequence A from [8] while the numbers 1 2 P k make A In this paper we shall also need the sequences A001109, A029546, A and A that we shorten to a k, b k, c k and d k with k 0. For the convenience of the reader we shall now explicitly define these sequences.

2 2 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 The first ten terms of the sequence a k are 0, 1, 6, 35, 204, 1189, 6930, 40391, and , it satisfies the recurrence relations a 0 = 0, a 1 = 1 and a n = 6 a n 1 a n 2 for n 2, and it is given by the formula a n = (3+2 2) n (3 2 2) n. 4 2 The first seven terms of the sequence b k are 1, 35, 1190, 40426, , and , it satisfies the recurrence relations b 0 = 1, b 1 = 35, b 2 = 1190 and b n = 35 (b n 1 b n 2 ) b n 3 for n 3, and it is given by the formula b n = (f + F n + + f F n 6), where f ± = 99 ± 70 2 and F ± = 17 ± The first seven terms of the sequence c k are 1, 34, 1155, 39236, , and , it satisfies the recurrence relations c 0 = 1, c 1 = 34 and c n = 34 c n 1 c n 2 for n 2, and it is given by the formula c n = F n+1 n+1 + F F + F. Finally, the first seven terms of the sequence d k are 1, 33, 1121, 38081, , and , it satisfies the recurrence relations d 0 = 1, d 1 = 33 and d n = 34 d n 1 d n 2 for n 2, and it is given by the formula d n = F n F n 6. In the last three sections we look into the alternating sums of squares of odd and even terms of the Pell-Lucas sequence and the alternating sums of products of two consecutive Pell-Lucas numbers. These sums also have nice representations as products of appropriate Pell and Pell-Lucas numbers with terms from the above four integer sequences. These formulas for ordinary sums and for alternating sums have been discovered with the help of a PC computer and all algebraic identities needed for the verification of our theorems can be easily checked in either Derive, Mathematica or Maple V. Running times of all these calculations are in the range of a few seconds. Similar results for Fibonacci and Lucas numbers have recently been discovered by the first author in papers [1], [2], [3], and [4]. They improved some results in [7]. 2. Pell-Lucas even squares The following lemma is needed to accomplish the inductive step in the proof of our first theorem. Lemma 1. For every m 0 and k 0 the following equality holds: 32 a 2 m+1 + Q 2 2k+2m+2 + a m+1 Q 2k Q 2k+2m = a m+2 Q 2k Q 2k+2m+2. (2.1) Proof. Let α = and β = 1 2. Notice that α + β = 2 and α β = 1 so that the real numbers α and β are solutions of the equation x 2 2 x 1 = 0. Since Q j = α j + β j

3 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 3 and a j = 7α + β α 2j 2 + α + 7β β 2j for every j 0, the difference of the right hand side and the left hand side of the relation (2.1) (after the substitutions β = 1 and α = and the replacement of , α , , , 1 + 2, and with α 6, α 4, α 3, α 2, α and 1 ) is equal α to M, where M = p 16 α 4 0 α 4k+4m + p α 4k with p 0 = p and p ± = 3 α α α 6 ± 14 α 5 5 α 4 3 α 3. Since both polynomials p 0 and p contain α 1 2 as a factor we conclude that M = 0 and the proof is complete. Theorem 1. For every m 0 and k 0 the following equality holds: { m α m + Q 2 Q 4 k, if m = 0; 2k+2i = a m+1 Q 2k Q 2k+2m, if m 1, (2.2) where the sequence α m is defined ( as follows: α 0 = 2, α 1 = 32, and α m α m 1 = 32 a 2 m m ) for m 2 (i. e., α m = 32 ). Proof. When m = 0 we obtain j=1 a2 j Q 2 2k Q 4k 2 = ( α 2 k + β 2 k) 2 α 4 k β 4 k 2 = 2 ( α k) 2 ( β k ) 2 2 = 0. For m 1 the proof is by induction on m. When m = 1 the relation (2.2) is 32 + Q 2 2k + Q 2 2k+2 6 Q 2k Q 2k+2 = 0 which is true since its left hand side is (α 2 2 α 1)(α α 1)(α 4k + α 4k 4 6 α 2 ). Assume that the relation (2.2) is true for m = r. Then α r+1 + r+1 Q 2 2k+2i = 32 a 2 r+1 + α r + Q 2 2k+2r+2 + r Q 2 2k+2i = 32 a 2 r+1 + Q 2 2k+2r+2 + a r+1 Q 2k Q 2k+2r = a r+2 Q 2k Q 2k+2r+2, where the last step uses Lemma 1 for m = r + 1. Hence, (2.2) is true for m = r + 1 and the proof is completed. Remark 1. The following are three additional versions of Theorem 1: For every j 0 and k 1 the following equalities holds: j α j+1 + Q 2 2k+2i = a j+1 Q 2k 2 Q 2k+2j+2, (2.3) j Q 2 2k+2i = γ j + a j+1 Q 2k 1 Q 2k+2j+1, (2.4)

4 4 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 j Q 2 2k+2i = δ j + a j+1 Q 2k+1 Q 2k+2j 1, (2.5) where γ 0 = δ 0 = 8 and γ j γ j 1 = 2 P 2 2 j+1 and δ j δ j 1 = 2 P 2 2 j 1 for every j Pell-Lucas odd squares The initial step in an inductive proof of our second theorem uses the following lemma. Lemma 2. For every k 1 the following identity holds: Q 2 2k+1 Q 2k 1 Q 2k+3 = 32. (3.1) Proof. By the Binet formula Q n = α n + β n so that we have Q 2 2k+1 Q 2k 1 Q 2k+3 = (α 2k+1 + β 2k+1 ) 2 (α 2k 1 + β 2k 1 )(α 2k+3 + β 2k+3 ) = 2 (α β) 2 k+1 (α β) 2 k 1 (α 4 + β 4 ) = 2 + Q 4 = = 32. The following lemma is needed to accomplish the inductive step in the proof of our second theorem. Lemma 3. For every j 0 and k 1 the following equality holds: Q 2 2k+2j+3 + Q 2k 1 (a j+1 Q 2k+2j+3 a j+2 Q 2k+2j+5 ) = 32 a 2 j+2. (3.2) Proof. The difference of the left hand side and the right hand side of (3.2) (after the substitutions β = 1 and α = and the replacement of , , , α , 1 + 2, and with α 6, α 4, α 3, α 2, α and 1 M ) is equal to, where α 16 α 8 with p 0 = p and M = p 0 α 4k+4j + p α 4k p ± = 3 α 14 ± 17 α α 12 ± 14 α 11 5 α 10 3 α 9. Since both polynomials p 0 and p contain α 1 2 as a factor we conclude that M = 0 and the proof is complete. Theorem 2. For every j 0 and k 1 the following equality holds: j Q 2 2k+2i+1 = α j+1 + a j+1 Q 2k 1 Q 2k+2j+3. (3.3)

5 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 5 Proof. The proof is by induction on j. For j = 0 the relation (3.3) is Q 2 2k+1 = 32 + Q 2k 1 Q 2k+3 which is true by the relation (3.1) in Lemma 2. Assume that the relation (3.3) is true for j = r. Then r+1 r Q 2 2k+2i+1 = Q 2 2k+2r+3 + Q 2k+2i+1 = Q 2 2k+2r+3 + α r+1 + a r+1 Q 2k 1 Q 2k+2r+3 = α r a 2 r+2 + a r+2 Q 2k 1 Q 2k+2r+5 = α (r+1)+1 + a r+2 Q 2k 1 Q 2k+2(r+1)+3, where the third step uses Lemma 3. Hence, (3.3) is true also for j = r + 1 and the proof is complete. Another version of Theorem 2 is the following statement: Theorem 3. For every j 0 and k 1 the following equality holds: j β j + Q 2 2k+2i+1 = a j+1 Q 2k 2 Q 2k+2j+4, (3.4) with β 0 = 200 and β j β j 1 = 2 P 2 2j+3 for j 1. Proof. We shall only outline the key steps in an inductive proof leaving the details to the reader because they are analogous to the proof of Theorem 2. The initial step is the equality Q 2 2k+1 = Q 2k 2 Q 2k+4 which holds for every k 1. On the other hand, the inductive step is realized with the following equality: 2 P 2 2r+5 + Q 2 2k+2r+3 + a r+1 Q 2k 2 Q 2k+2r+4 = a r+2 Q 2k 2 Q 2k+2r+6, which holds for every k 1 and every r Pell-Lucas products For the first two steps in a proof by induction of our next theorem we require the following lemma. Lemma 4. For every k 1 the following equalities hold: Q 2k Q 2k+1 = Q 4k (4.1) Q 2k Q 2k+1 + Q 2k+2 Q 2k+3 = 6 Q 2k Q 2k (4.2)

6 6 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 Proof of (4.1). By the Binet formula we have Q 2k Q 2k+1 = ( α 2k + β 2k) ( α 2k+1 + β 2k+1) = α 4k+1 + (α β) 2k (α + β) + β 4k+1 = Q 4k Proof of (4.2). When we apply the Binet formula to the terms in the difference of the left hand side and the right hand side of (4.2) we get 80 + α 4 k+1 + α (α β) 2 k + β (α β) 2 k + β 4 k+1 + α 4 k+5 + α 3 β 2 (α β) 2 k + β 3 α 2 (α β) 2 k + β 4 k+5 6α 4 k+3 6α 3 (α β) 2 k 6β 3 (α β) 2 k 6β 4 k+3. Since α β = 1, this simplifies to p(α) α 4k + p(β) β 4k + q, where p(x) = x 5 6 x 3 + x and q = α 3 β 2 + α 2 β 3 6 α 3 6 β 3 + α + β Now it is easy to check (by the substitutions α = and β = 1 2) that p(α) = 0, p(β) = 0 and q = 0. This implies that the relation (4.2) holds. With the following lemma we shall make the inductive step in the proof of the third theorem. Lemma 5. For every r 1 and k 1 the following equality holds: Q 2k (a r+2 Q 2k+2r+3 a r+1 Q 2k+2r+1 ) Q 2k+2r+2 Q 2k+2r+3 = 2 P 2r+2 P 2r+3. Proof. Let R denote the difference of the left hand side and the right hand side of the above relation. We need to show that R = 0. Let A = α 2k, B = β 2k, U = β 2r, V = β 2r, u = 7α+β 16 and v = α+7β 16. Note that a r+1 is equal to u U + v V. By the Binet formula we get R = (A + B) [ (u α 2 U + v β 2 V )(α 3 A U + β 3 B V ) (u U + v V ) (α A U + β B V )] (α 2 A U + β 2 B V )(α 3 A U + β 3 B V ) [ (α 2 U + β 2 V )(α 3 U + β 3 V ) = l + u A B + (8 u m) ] A2 m U (A B 1) U V + (l v A B n) V 2, where l ± = 4(10 ± 7 2), m = and n = Since A B = 1, u = m 8 and l v = n it follows that R = 0. Theorem 4. For every j 0 and k 1 the following equality holds: { j Q 4 k, if j = 0; α j + Q 2k+2i Q 2k+2i+1 = a j+1 Q 2k Q 2k+2j+1, if j 1, where α 0 = 2, α 1 = 80, and α j+1 α j = 2 P 2j+2 P 2j+3 for j 1. (4.3)

7 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 7 Proof. The proof is by induction on j. For j = 0 the relation (4.3) is Q 2k Q 2k+1 2 = Q 4k+1 which is true by (4.1). For j = 1 the relation (4.3) is which is true by (4.2) Q 2k Q 2k+1 + Q 2k+2 Q 2k+3 = 6 Q 2k Q 2k+3 Assume that the relation (4.3) is true for j = r. Then α r+1 + r+1 Q 2k+2i Q 2k+2i+1 = α r + 2 P 2r+2 P 2r+3 + r Q 2k+2i Q 2k+2i+1 + Q 2k+2r+2 Q 2k+2r+3 = a r+1 Q 2k Q 2k+2r P 2r+2 P 2r+3 + Q 2k+2r+2 Q 2k+2r+3 = a r+2 Q 2k Q 2k+2(r+1)+1, where the last step uses Lemma 5. Hence, (4.3) is true also for j = r Alternating Pell-Lucas even squares In this section we look for formulas that give closed forms for alternating sums of squares of Pell-Lucas numbers with even indices. Lemma 6. For every k 0 we have Proof. By the Binet formula we get Q 2 2k = 8 + Q 2k+1 Q 2k 1. (5.1) 8 + Q 2k 1 Q 2k+1 = 8 + (α 2k 1 + β 2k 1 )(α 2k+1 + β 2k+1 ) = 8 + α 4k + β 2 (α β) 2k 1 + α 2 (α β) 2k 1 + β 4k = α 4k β 4k = (α 2k + β 2k ) 2 = Q 2 2k. Lemma 7. For every k 0 we have Q 2 2k Q 2 2k+2 = 32 (1 a k+1 Q 2k ). (5.2) Proof. For real numbers a, b and c let p a b, q a and r a;b c denote αa β a, α a + β a and a α+b β. The α b β b c difference of the left hand side and the right hand side in (5.2) is (q 2k ) 2 (q 2k+2 ) ( r 7;1 16 α 2k + r 1;7 16 β 2k) q 2k. Let α 2k = A and β 2k = B. The above expression reduces to p(α) A 2 + q A B + p(β) B 2 32,

8 8 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 where p(α) = α + 2 β α 4 and q = α + 16 β 2 α 2 β 2. If we replace α and β with and 1 2 we get p(α) = p(β) = 0 and q = 32. Hence, the difference is 32 A B 32 = 0 because A B = 1. Lemma 8. For every k 0 and every r 0 we have 32 c r (1 a k+r+1 Q 2k+2r ) + Q 2 2k+4r+4 Q 2 2k+4r+6 (5.3) = 32 c r+1 (1 a k+r+2 Q 2k+2r+2 ). Proof. Notice that c r = p4 4r+4 for every r 0. The difference of the left hand side and the right hand side in (5.3) is 32 p 4r+4 4 [ 1 ( r 7;1 16 α 2k+2r + r 1;7 16 β 2k+2r) q 2k+2r ] + (q2k+4r+4 ) 2 (q 2k+4r+6 ) 2 32 p 4r+8 4 [ 1 ( r 7;1 16 α 2k+2r+2 + r 1;7 16 β 2k+2r+2) q 2k+2r+2 ]. Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to M α 4 β 4, where M is p(α, β) A 2 U 4 q(α, β) A 2 U 2 V 2 + r(α, β) A B U 3 V + 2 s A B U 2 V 2 where p(β, α)b 2 V 4 + q(β, α)b 2 U 2 V 2 r(β, α)a B U V 3 32(t(α)U 2 t(β)v 2 ), p(α, β) = t(α)(α 4 β 4 α α α 4 β + 14 α + 2 β), q(α, β) = 2 β 4 (7 α + β)(α 4 β 4 1), r(α, β) = 16 α 4 (α + β)(α 6 β 2 1), s = α 4 β 4 (α 4 β 4 )(α 2 β 2 1) and t(α) = α 4 (α 4 1). Replacing α and β with and 1 2 we get M = u + A B U 3 V u A B U V 3 u + U 2 + u V 2, where u ± = 128(140 ± 99 2). Hence, M = 0 so that the difference is zero because A B = 1 and U V = 1. Lemma 9. For every k 0 and every r 0 we have d r Q 2 2k+2r 64 b r 1 Q 2 2k+4r+2 + Q 2 2k+4r+4 = d r+1 Q 2 2k+2r+2 64 b r. Proof. Notice that b r = r 169; α 4r + r 29; β 4r 1 32 and d r = r 5;1 12 α 4r + r 1;5 12 β 4r,

9 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 9 for every r 0. The difference of the left hand side and the right hand side is ( r 5;1 12 α 4r + r 1;5 12 β 4r) (q 2k+2r ) 2 r 169;29 6 α 4r 4 r 29;169 6 β 4r 4 (q 2k+4r+2 ) 2 + (q 2k+4r+4 ) 2 ( r 5;1 12 α 4r+4 + r 1;5 12 β 4r+4) (q 2k+2r+2 ) 2 + r 169;29 6 α 4r r 29;169 6 β 4r. Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to where M is M, 12 α 4 β 4 p(α, β)a 2 U 4 + q(α, β)(auv ) 2 + r(α, β)abu 3 V 24 s AB(UV ) 2 + with P = (α β) 4, p(β, α)b 2 V 4 + q(β, α)(buv ) 2 + r(β, α)abuv 3 t(α, β)u 2 t(β, α)v 2, p(α, β) = P (α 4 1)(5 α 5 + α 4 β 12 α α + β), q(α, β) = P (P 1)(α + 5 β), r(α, β) = 2 P (α 6 β 2 1)(5 α + β), s = P (P α 2 β 2 ) and t(α, β) = 2 (P β 4 )(169 α + 29 β). Replacing α and β with and 1 2 we get M = u + A B U 3 V + u A B U V 3 u + U 2 u V 2, where u ± = 16(24 ± 17 2). Hence, M = 0 so that the difference is zero because A B = 1 and U V = 1. Theorem 5. a) For every m 0 and k 0 the following equality holds: 2m+1 ( 1) i Q 2 2k+2i = 32 c m (1 a k+m+1 Q 2k+2m ), (5.4) b) For every m 1 and k 0 the following equality holds: 2 m ( 1) i Q 2 2k+2i = d m Q 2 2k+2m 64 b m 1. (5.5) Proof of a). The proof is by induction on m. For m = 0 the relation (5.4) is Q 2 2k Q 2 2k+2 = 32 (1 a k+1 Q 2k ) (i. e., the relation (5.2)) which is true by Lemma 7. Assume that the relation (5.4) is true for m = r. Then 2(r+1)+1 ( 1) i Q 2 2k+2i = 2r+1 ( 1) i Q 2 2k+2i + Q 2 2k+4r+4 Q 2 2k+4r+6 = 32 c r (1 a k+r+1 Q 2k+2r ) + Q 2 2k+4r+4 Q 2 2k+4r+6 = 32 c r+1 (1 a k+(r+1)+1 Q 2k+2(r+1) ),

10 10 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 where the last step uses Lemma 8. completed. Hence, (5.4) is true for m = r + 1 and the proof is Proof of b). The proof is by induction on m. For m = 0 the relation (5.5) is which is true since d 0 = 1 and b 1 = 0. Q 2 2k = d 0 Q 2 2k 64 b 1 Assume that the relation (5.5) is true for m = r. Then 2(r+1) 2r ( 1) i Q 2 2k+2i = ( 1) i Q 2 2k+2i Q 2 2k+4r+2 + Q 2 2k+4r+4 = where the last step uses Lemma 9. completed. d r Q 2 2k+2r 64 b r 1 Q 2 2k+4r+2 + Q 2 2k+4r+6 = d r+1 Q 2 2k+2r+2 64 b r, Hence, (5.5) is true for m = r + 1 and the proof is Lemma 10. For every k 0 we have 6. Alternating Pell-Lucas odd squares Q 2 2k+3 Q 2 2k+1 = 32 a k+1 Q 2k+2. (6.1) Proof. By the Binet formulas the difference of the left hand side and the right hand side in (6.1) is Q 2 2k+3 Q 2 2k+1 32 a k+1 Q 2k+2 = (α 2k+1 + β 2k+1 ) 2 (α 2k+3 + β 2k+3 ) 2 ( 7α + β + 32 α 2k + α + 7β ) β 2k (α 2k+2 + β 2k+2 ), Let α 2k = A and β 2k = B. The above expression reduces to p(α)a 2 q A B + p(β)b 2, where q = 2 α 3 β 3 2 α β 2 α 3 14 β 2 α 14 α 2 β 2 β 3 and p(α) = α 2 (14α α 4 + 2β + 1). It is easy to check (by substitutions α = and β = 1 2) that p(α) = 0, p(β) = 0 and q = 0. This implies that the relation (6.1) is true. Lemma 11. For every k 0 we have 8 + Q 2 2k+1 = Q 2k Q 2k+2. (6.2) Proof. Using the Binet representation of Pell-Lucas numbers we get 8 + Q 2 2k+1 Q 2k Q 2k+2 = 8 + (α 2k+1 + β 2k+1 ) 2 (α 2k + β 2k )(α 2k+2 + β 2k+2 ) = (α β) 2k+1 (α β) 2k (α 2 + β 2 ) = = 0,

11 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 11 because α β = 1 and α 2 + β 2 = 6. Hence, 8 + Q 2 2k+1 = Q 2kQ 2k+2. Lemma 12. For all k 0 and r 0 we have L = 0, where L is Q 2 2k+4r+5 Q 2 2k+4r+7 32 c r a k+r+1 Q 2k+2r c r+1 a k+r+2 Q 2k+2r+4. Proof. The expression L is in fact (q 2k+4r+5 ) 2 (q 2k+4r+7 ) 2 32 p 4r+4 4 ( r 7;1 16 α 2k+2r + r 1;7 16 β 2k+2r) q 2k+2r p 4r+8 4 ( r 7;1 16 α 2k+2r+2 + r 1;7 16 β 2k+2r+2) q 2k+2r+4. Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to M α 4 β 4, where M is p(α, β)a 2 U q(α, β)(auv ) r(α, β)abu 3 V with 2 s AB(UV ) 2 p(β, α)b 2 V q(β, α)(buv ) 2 2 r(β, α)abuv 3, p(α, β) = α 6 (α 4 1)(α 4 β 4 α α α 4 β + 14 α + 2 β), q(α, β) = α 2 β 4 ((αβ) 4 1)(7α + β), r(α, β) = α 4 (α + β)(α 6 β 2 1)(α 2 + 6αβ + β 2 ) and s = α 5 β 5 ((αβ) 2 1)(α 4 β 4 ). Replacing α and β with and 1 2 we see easily that all coefficients of the above polynomial vanish so that L = 0. Lemma 13. For every k 0 and every r 0 we have 2(P 2 2r+3 P 2 2r+1) + d r Q 2k+2r Q 2k+2r+2 Q 2 2k+4r+3 + Q 2 2k+4r+5 = d r+1 Q 2k+2r+2 Q 2k+2r+4. Proof. The difference of the left hand side and the right hand side is [ (p ) 2r+3 2 ( ) ] 8 1 p 2r ( r 5;1 12 α 4r + r 1;5 12 β 4r) q 2k+2r q 2k+2r+2 (q 2k+4r+3 ) 2 + (q 2k+4r+5 ) 2 ( r 5;1 12 α 4r+4 + r 1;5 12 β 4r+4) q 2k+2r+2 q 2k+2r+4. Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to where M is (α β) 2 [ p(α, β)a 2 U 4 + q(α, β)(auv ) 2 + r(α, β)abu 3 V 24 s AB(UV ) 2 + p(β, α)b 2 V 4 + q(β, α)(buv ) 2 + r(β, α)abuv 3] with t(α) = α 2 (α 4 1), p(α, β) = t(α)(5 α 5 + α 4 β 12 α α + β), M, 12 (α β) 2 96 [ t(α)u 2 2wUV + t(β)v 2],

12 12 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 q(α, β) = t(αβ)(α + 5 β)/β 2, r(α, β) = (α 2 + β 2 )(α 6 β 2 1)(5 α + β), s = α 3 β 3 ((αβ) 2 1) and w = αβ((αβ) 2 1). Replacing α and β with and 1 2 we get M = u + A B U 3 V + u A B U V 3 u + U 2 u V 2, where u ± = 384(24 ± 17 2). Hence, M = 0 so that the difference is zero since A B = 1 and U V = 1. Theorem 6. a) For every j 0 and k 0 the following equality holds: 2 j+1 ( 1) i Q 2 2k+2i+1 = 32 c j a k+j+1 Q 2k+2j+2, (6.3) b) For every j 1 and k 1 the following equality holds: 2 j 2 P2j ( 1) i Q 2 2k+2i+1 = d j Q 2k+2j Q 2k+2j+2. (6.4) Proof of a). The proof is by induction on j. For j = 0 the relation (6.3) is Q 2 2k+3 Q 2 2k+1 = 32a k+1 Q 2k+2 (i. e., the relation (6.1)) which is true by Lemma 10. Assume that the relation (6.3) is true for j = r. Then 2(r+1)+1 ( 1) i Q 2 2k+2i+1 = 2r+1 ( 1) i Q 2 2k+2i+1 + Q 2 2k+4r+5 Q 2 2k+4r+7 = 32 c r a k+r+1 Q 2k+2r+2 + Q 2 2k+4r+5 Q 2 2k+4r+7 = 32 c r+1 a k+r+2 Q 2k+2r+4, where the last step uses Lemma 12. completed. Hence, (6.3) is true for j = r + 1 and the proof is Proof of b). The proof is again by induction on j. For j = 0 the relation (6.4) is 2 P Q 2 2k+1 = d 0 Q 2k Q 2k+2 which is true by Lemma 11 since d 0 = 1 and P 1 = 2.

13 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 13 Assume that the relation (6.4) is true for j = r. Then 2 P 2 2r+3 + 2(r+1) ( 1) i Q 2 2k+2i+1 = 2(P 2 2r+3 P 2 2r+1) ( ) 2r + 2 P2r ( 1) i Q 2 2k+2i+1 Q 2 2k+4r+3 + Q 2 2k+4r+5 = 2(P 2 2r+3 P 2 2r+1) + d r Q 2k+2r Q 2k+2r+2 Q 2 2k+4r+3 + Q 2 2k+4r+5 = d r+1 Q 2k+2r+2 Q 2k+2r+4, where the last step uses Lemma 13. Hence, (6.4) is true for m = r + 1 and the proof is completed. Lemma 14. For every k 0 we have 7. Alternating Pell-Lucas products Q 2k Q 2k+1 Q 2k+2 Q 2k P 2k+3 Q 2k = 80. (7.1) Proof. The difference of the left hand side and the right hand side in (7.1) is q 2k q 2k+1 q 2k+2 q 2k p 2k+3 1 q 2k 80. Let α 2k = A and β 2k = B. The above expression reduces to M α β p(α) A 2 q A B p(β) B 2 80(α β), with p(α) = α(β α 4 α α 2 + α β) and where M is q = (α β)(α 3 β 2 + α 2 β 3 16 α 2 16 α β 16 β 2 α β). If we replace α and β with and 1 2 we get p(α) = p(β) = 0 and q = and 80(α β) = Hence, M = 0 because A B = 1. Lemma 15. For all k 0 and r 0 we have 8 c r+1 (P 2k+2r+5 Q 2k+2r+2 10) + Q 2k+4r+4 Q 2k+4r+5 = 8 c r (P 2k+2r+3 Q 2k+2r 10) + Q 2k+4r+6 Q 2k+4r+7. Proof. The difference L of the left hand side and the right hand side is [ 8 p 4r p 2k+2r+5 1 q 2k+2r+2 10 ] + q 2k+4r+4 q 2k+4r+5 q 2k+4r+6 q 2k+4r+7 8 p 4r+4 4 [ 2 p 2k+2r+3 1 q 2k+2r 10 ] Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to M (α + β)(α β) 2 (α 2 + β 2 ),

14 14 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 where M is p(α, β)a 2 U 4 16 q(α, β)(auv ) r(α, β)abu 3 V s AB(UV ) 2 + with p(β, α)b 2 V 4 16 q(β, α)(buv ) 2 16 r(β, α)abuv 3 t(α)u 2 + t(β)v 2, p(α, β) = α 6 (α 4 1)(α 6 β α 7 + α 3 β 4 α 2 β α ), q(α, β) = α 3 β 4 ((αβ) 4 1), s = α 4 β 4 ((αβ) 2 1)(α 2 + β 2 )(α 2 β 2 ) 2, r(α, β) = α 4 (α β)(α 6 β 2 1)(α 2 + αβ + β 2 ) and t(α) = 80 α 4 (α β) (α 4 1). Replacing α and β with and 1 2 we get M = u + A B U 3 V + u A B U V 3 u + U 2 u V 2, where u ± = 1280(99 ± 70 2). Hence, M = 0 so that the difference is zero since A B = 1 and U V = 1. Lemma 16. For all k 0 and r 0 we have 64 c r + d r Q 2k+2r Q 2k+2r+1 + Q 2k+4r+4 Q 2k+4r+5 = d r+1 Q 2k+2r+2 Q 2k+2r+3 + Q 2k+4r+2 Q 2k+4r+3. Proof. The difference L of the left hand side and the right hand side is 64 p 4r ( r 5;1 12 α 4r + r 1;5 12 β 4r) q 2k+2r q 2k+2r+1 q 2k+4r+2 q 2k+4r+3 + q 2k+4r+4 q 2k+4r+5 ( r 5;1 12 α 4r+4 + r 1;5 12 β 4r+4) q 2k+2r+2 q 2k+2r+3. Let α 2k = A, β 2k = B, α 2r = U and β 2r = V. The above expression reduces to where M is p(α, β)a 2 U 4 q(α, β)(auv ) 2 + r(α, β)abu 3 V + 12 s AB(UV ) 2 with M, 12(α 4 β 4 ) p(β, α)b 2 V 4 q(β, α)(buv ) 2 r(β, α)abuv 3 + t(α)u 2 t(β)v 2, p(α, β) = α (α 4 1)(β 4 α 4 )(5α 5 + α 4 β 12α α + β), s = α 2 β 2 ((αβ) 2 1)(β 4 α 4 )(α + β), r(α, β) = (β 4 α 4 )(α + β)(5α + β)(α 6 β 2 1), q(α, β) = α((αβ) 4 1)(β 4 α 4 )(α + 5β) and t(α) = 768 α 4. Replacing α and β with and 1 2 we get M = u + A B U 3 V + u A B U V 3 + u + U 2 u V 2,

15 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 15 where u ± = 768(17 ± 12 2). Hence, M = 0 so that the difference is zero since A B = 1 and U V = 1. Theorem 7. a) For every j 0 and k 0 the following equality holds: 2 j+1 ( 1) i Q 2k+2i Q 2k+2i+1 = 8 c j (P 2k+2j+3 Q 2k+2j 10), (7.2) b) For every j 0 and k 0 the following equality holds: 2 j β j + ( 1) i Q 2k+2i Q 2k+2i+1 = d j Q 2k+2j Q 2k+2j+1, (7.3) where the sequence β j is determined by the conditions: β 0 = 0 and β j+1 = β j + 64 c j for every j 0. Proof of a). The proof is by induction on j. For j = 0 the relation (7.2) is Q 2k Q 2k+1 Q 2k+2 Q 2k+3 = 8 P 2k+3 Q 2k + 80 (i. e., the relation (7.1)) which is true by Lemma 14. Assume that the relation (7.2) is true for j = r. Then 2(r+1)+1 ( 1) i Q 2k+2i Q 2k+2i+1 = 2r+1 ( 1) i Q 2k+2i Q 2k+2i+1 + Q 2k+4r+4 Q 2k+4r+5 Q 2k+4r+6 Q 2k+4r+7 = 8c r (P 2k+2r+3 Q 2k+2r 10) + Q 2k+4r+4 Q 2k+4r+5 Q 2k+4r+6 Q 2k+4r+7 = 8c r+1 (P 2k+2r+5 Q 2k+2r+2 10), where the last step uses Lemma 15. completed. Hence, (7.2) is true for j = r + 1 and the proof is Proof of b). The proof is once again by induction on j. For j = 0 the relation (7.3) is β 0 + Q 2k Q 2k+1 = d 0 Q 2k Q 2k+1 which is true since d 0 = 1 and β 0 = 0. Assume that the relation (7.3) is true for j = r. Then β r+1 + 2(r+1) 2r ( 1) i Q 2k+2i Q 2k+2i+1 = β r+1 β r + β r + ( 1) i Q 2k+2i Q 2k+2i+1 Q 2k+4r+2 Q 2k+4r+3 + Q 2k+4r+4 Q 2k+4r+5 = 64 c r + d r Q 2k+2r Q 2k+2r+1 Q 2k+4r+2 Q 2k+4r+3 + Q 2k+4r+4 Q 2k+4r+5 where the last step uses Lemma 16. completed. = d r+1 Q 2k+2r+2 Q 2k+2r+3, Hence, (7.3) is true for j = r + 1 and the proof is

16 16 INTEGERS: ELECTRONIC JOURNAL OF COMBINATORIAL NUMBER THEORY 6 (2006), #A15 We mention also the following equalities that have been discovered while attempting to prove Theorem 7. Theorem 8. For every k 0 we have Q 2k Q 2k+1 = Q 2k+3 Q 2k 2 80 = 2 (P 2k+2 P 2k 1 6). References [1] Z. Čerin, Properties of odd and even terms of the Fibonacci sequence, Demonstratio Mathematica, 39 (1) (2006), [2] Z. Čerin, On sums of squares of odd and even terms of the Lucas sequence, Proccedings of the 11th Fibonacci Conference, (to appear). [3] Z. Čerin, Some alternating sums of Lucas numbers, Central European Journal of Mathematics 3 (1) (2005), [4] Z. Čerin, Alternating Sums of Fibonacci Products, (preprint). [5] V. E. Hoggatt, Jr., Fibonacci and Lucas numbers, The Fibonacci Association, Santa Clara, [6] R. Knott, Fibonacci numbers and the Golden Section, [7] V. Rajesh and G. Leversha, Some properties of odd terms of the Fibonacci sequence, Mathematical Gazette, 88 (2004), [8] N. J. A. Sloane, On-Line Encyclopedia of Integer Sequences, [9] S. Vajda, Fibonacci and Lucas numbers, and the Golden Section: Theory and Applications, Halsted Press, Chichester (1989).