Bulletin of Mathematical Analysis and Applications ISSN: 1821-1291, URL: http://www.bmathaa.org Volume 1 Issue 3(2009), Pages 1-9. A STUDY ON ABSOLUTE FACTORS FOR A TRIANGULAR MATRIX (DEDICATED IN OCCASION OF THE 65-YEARS OF PROFESSOR R.K. RAINA) W. T. SULAIMAN Abstract. In this paper a result concerning summability factors theorem for lower triangular matrices is presented. This result generalized and extend the result of Savas [1]. 1. Introduction Let π = (π‘ππ£ ) be a lower ∑ triangular matrix, (π π ) be the sequence of the π-th partial sums of the series ππ , then, we deο¬ne ππ := π ∑ π‘ππ£ π π£ . (1.1) π£=0 A series ∑ ππ is said to be summable β£ π β£π , π ≥ 1, if ∞ ∑ ππ−1 β£ Δππ−1 β£π < ∞. (1.2) π=1 One of the most interesting cases of β£ π β£π -summability is the case when π is chosen to be the Cesaro matrix. That is by putting π‘ππ£ = π΄πΌ−1 Γ(π + πΌ + 1) π−π£ , π£ = 0, 1, ..., π, π΄πΌ , π = 0, 1, ... . π = π΄πΌ Γ(πΌ + 1)Γ(π + 1) π Given any lower triangular matrix π one can associate the matrices π and πˆ, with entries deο¬ned by π‘ππ£ = π ∑ π‘ππ , π πππ π = 0, 1, 2, ..., ˆ π‘ππ£ = π‘ππ£ − π‘π−1,π£ , π=π£ 2000 Mathematics Subject Classiο¬cation. 40F05, 40D25. Key words and phrases. Absolute summability; summability factor; weight function. c β2009 Universiteti i PrishtineΜs, PrishtineΜ, KosoveΜ. Submitted Jun, 2009. Published September, 2009. 1 2 W. T. SULAIMAN respectively. With π π = π‘π := π ∑ π‘ππ£ π π£ = π£=0 ∑π π=0 π ∑ ππ ππ , we deο¬ne and derive the following π‘ππ£ π£=0 ππ := π‘π − π‘π−1 = π ∑ π£ ∑ π=0 π‘ππ ππ ππ − π=0 ππ ππ = π−1 ∑ π ∑ π ∑ ππ ππ π=0 π‘ππ£ = π£=π π‘π−1,π ππ ππ = π ∑ π=0 π ∑ π‘ππ ππ ππ , (1.3) π=0 ˆ π‘ππ ππ ππ , ππ π‘π−1,π = 0. (1.4) π=0 ππ := π’π − π’π−1 = π ∑ π’ ˆππ ππ ππ . (1.5) π=0 We call π a triangle if π is lower triangular and π‘ππ β= 0 for all π. A triangle π΄ is called factorable if its nonzero entries πππ can be written in the form ππ ππ for each π and π. Recall that ˆ π‘ππ = π‘ππ . We also assume that (ππ ) is a positive sequence of numbers such that ππ = π0 + π1 + ... + ππ → ∞, ππ π → ∞. A positive sequence (ππ ) is said to be almost increasing if π΄ππ ≤ ππ ≤ π΅ππ , where (ππ ) is a positive increasing sequence and π΄ and π΅ are positive constants (see [1]). A positive sequence (πΎπ ) is said to be quasi π½−power increasing sequence if there is a constant πΎ = πΎ(π½, πΎ) ≥ 1 such that πΎππ½ πΎπ ≥ ππ½ πΎπ holds for all π ≥ π ≥ 1. It should be mentioned that every almost increasing sequence is quasi π½−power increasing sequence for any π½ > 0, while the converse need not be true as for example πΎπ = π−π½ , π½ > 0. Here we generalize the quasi π½-power increasing sequence by giving the following. Deο¬nition. A sequence (ππ ) is said to be quasi (π½ − πΎ)−power increasing sequence if there exist πΎ = πΎ(π½, πΎ) ≥ 1 such that πΎππ½ (log π)πΎ ππ ≥ ππ½ (log π)πΎ ππ holds for all π ≥ π ≥ 1, πΎ ≥ 0. Clearly quasi (π½ − 0)−power increasing sequence is the quasi π½− power increasing sequence. The series ∑ ππ is said to be summable β£ π , ππ β£π , π ≥ 1, if ∞ ∑ ππ−1 β£ ππ − ππ−1 β£π < ∞, π=1 where ππ = π 1 ∑ ππ£ π π£ . ππ π£=0 The (πΆ, 1)−mean of the sequence (ππ ) is equal to π (π) = π(π(π)) if (π) limπ→∞ ππ(π) 1 π+1 ∑π = π < ∞. Very recently, Savas [1] proved the following theorem. π£=0 ππ£ , and we said that A STUDY ON ABSOLUTE FACTORS FOR A TRIANGULAR MATRIX 3 Theorem 1.1. Let π΄ be a lower triangular matrix with nonegative entries such that (i) ππ0 = 1, π = 0, 1, ..., (ii) ππ−1,π£ ≥ πππ£ π ππ π ≥ π£ + 1, (iii)ππππ = π(1), 1 = π(ππππ ), ∑π−1 (iv) π£=1 ππ£π£ β£ ˆ ππ,π£+1 β£= π(πππ ). Let ∑ π‘1π denote the π−th (πΆ, 1) mean of (πππ ). If ∞ (v) ∑π£=1 ππ£π£ β£ ππ£ β£π β£ π‘1π£ β£π = π(1), ∞ β£1−π β£ Δππ£ β£π β£ π‘1π£ β£π = π(1), (vi) π£=1 β£ ππ£π£ ∑ then the series ππ ππ is β£ π΄ β£π summable, π ∈ π. The aim of this paper is to give the following three improvements to the previous result (see Theorem 2.): 1. The lower triangular matrix we assumed is not restricted to nonnegative entries. 2. The kind of summability we obtained is β£ π΄, πΏ β£π , πΏ ≥ 0, in which β£ π΄ β£π ≡β£ π΄, 0 β£π , and π is not restricted to integers but π ∈ [1, ∞). 3. An extension is made by assuming further hypothesis. 2. Results. The following is our main result. Theorem 2.1. Let π be a lower triangular matrix, π‘π denote the π−th (πΆ, 1) mean of the sequence (πππ ), and let (ππ ) be a sequence of numbers such that π, π‘π , ππ are all satisfying π β£ π‘ππ β£= π(1), 1 = π(π β£ π‘ππ β£), (2.1) π−1 ∑ β£ π‘π£π£ β£β£ ˆ π‘ππ£ β£= π(β£ π‘ππ β£), (2.2) β£ π‘π£π£ β£β£ ˆ π‘π,π£+1 β£= π(β£ π‘ππ β£), (2.3) π£=1 π−1 ∑ π£=1 ∞ ∑ ππΏπ β£ ˆ π‘ππ£ β£= π(π£ πΏπ ), (2.4) β£ Δπ£ ˆ π‘ππ£ β£= π(β£ π‘ππ β£), (2.5) π=π£+1 π−1 ∑ π£=1 ∞ ∑ ππΏπ β£ Δπ£ ˆ π‘ππ£ β£= π(π£ πΏπ β£ π‘π£π£ β£), (2.6) π=π£+1 ∞ ∑ π£ πΏπ β£ π‘π£π£ β£β£ π‘π£ β£π β£ ππ£ β£π = π(1), (2.7) π£ πΏπ β£ π‘π£ β£π β£ π‘π£π£ β£1−π β£ Δππ£ β£π = π(1), (2.8) π£=1 ∞ ∑ π£=1 4 W. T. SULAIMAN ∑ then the series ππ ππ is summable β£ π, πΏ β£π , π ≥ 1, πΏ ≥ 0. Furthermore if (ππ ) is a quasi (π½−πΎ)−power increasing sequence, 0 < π½ < 1, πΎ ≥ 0, and if (π½π ) is a sequence of numbers satisfying Δππ ≤ π½, (2.9) π½π → 0 ππ π → ∞, (2.10) ∞ ∑ π β£ Δπ½π β£ ππ < ∞, (2.11) π=1 β£ ππ β£ ππ = π(1), (2.12) π ∑ π£ πΏπ−1 β£ π‘π£ β£π = π(ππ ), ππ£π−1 π£=1 (2.13) then the conditions ∑ (2.7) and (2.8) are omitted in order to obtain the β£ π, πΏ β£π −summability of ππ ππ . Proof of Theorem 2.1. We have π ∑ π ∑ ˆ π‘ππ£ ππ£ π£ π£=0 π£=1 ( ) ( ) ( π )( ) π−1 π£ ∑ ∑ ∑ ˆ π‘ππ£ ππ£ π‘ππ ππ πππ Δπ£ = + π£ππ£ π£ π π£=1 π=1 π£=1 ( ) π−1 ∑ 1 1 1 ˆ π+1 ˆ = (π£ + 1)π‘π£ π‘ππ£ ππ£ + Δˆ π‘ππ£ ππ£ + π‘π,π£+1 Δππ£ + π‘π π‘ππ ππ π£(π£ + 1) π£+1 π£+1 π π£=1 ) π−1 ∑ (1 π+1 ˆ ˆ ˆ = π‘π£ π‘ππ£ ππ£ + π‘π£ Δπ£ π‘ππ£ ππ£ + π‘π£ π‘π,π£+1 Δππ£ + π‘π π‘ππ ππ π£ π π£=1 ππ := ˆ π‘ππ£ ππ£ ππ£ = π£ππ£ = ππ1 + ππ2 + ππ3 + ππ4 . In order to prove the Theorem, by Minkowski’s inequality, we have to show that ∞ ∑ π=1 ππΏπ+π−1 β£ πππ β£π < ∞, π = 1, 2, 3, 4. A STUDY ON ABSOLUTE FACTORS FOR A TRIANGULAR MATRIX 5 By Holder’s inequality ∞ ∑ ππΏπ+π−1 β£ ππ1 β£π = π=2 ∞ ∑ ππΏπ+π−1 β£ π=2 ≤ ∞ ∑ π−1 ∑ π£=1 π πΏπ+π−1 π=2 = π(1) π−1 ∑ π£ 1 ˆ π‘π£ π‘ππ£ ππ£ β£π π£ −π β£ π‘π£π£ β£ π β£ π‘π£ β£ β£ ˆ π‘ππ£ β£β£ ππ£ β£ π£=1 ∞ ∑ ∞ ∑ ππΏπ (π β£ π‘ππ β£)π−1 π−1 ∑ = π(1) )π−1 (π−1 ∑ β£ π‘π£π£ β£β£ ˆ π‘ππ£ β£ π£ −π β£ π‘π£π£ β£1−π β£ π‘π£ β£π β£ ˆ π‘ππ£ β£β£ ππ£ β£π π£=1 ∞ ∑ π£ −π β£ π‘π£π£ β£1−π β£ π‘π£ β£π β£ ππ£ β£π π£=1 ∞ ∑ π π£=1 π=2 = π(1) 1−π ππΏπ β£ ˆ π‘ππ£ β£ π=π£+1 π£ πΏπ−π β£ π‘π£π£ β£1−π β£ π‘π£ β£π β£ ππ£ β£π π£=1 = π(1) ∞ ∑ π£ πΏπ β£ π‘π£π£ β£β£ π‘π£ β£π β£ ππ£ β£π = π(1), π£=1 in view of (2.2) and (2.7). ∞ ∑ ππΏπ+π−1 β£ ππ2 β£π = π=2 ∞ ∑ ππΏπ+π−1 β£ π=2 ≤ ∞ ∑ π−1 ∑ π‘π£ Δπ£ ˆ π‘ππ£ ππ£ β£π π£=1 π πΏπ+π−1 π=2 = π(1) π−1 ∑ ∞ ∑ ∞ ∑ π£=1 = π(1) π β£ π‘π£ β£ β£ Δπ£ ˆ π‘ππ£ β£β£ ππ£ β£ (π−1 ∑ π£=1 ∞ ∑ π£=1 π−1 ∑ β£ Δπ£ ˆ π‘ππ£ β£ β£ π‘π£ β£π β£ Δπ£ ˆ π‘ππ£ β£β£ ππ£ β£π π£=1 β£ π‘π£ β£π β£ ππ£ β£π )π−1 π£=1 ππΏπ (π β£ π‘ππ β£)π−1 π=2 = π(1) π ∞ ∑ ππΏπ β£ Δπ£ ˆ π‘ππ£ β£ π=π£+1 π£ πΏπ β£ π‘π£π£ β£β£ π‘π£ β£π β£ ππ£ β£π = π(1), 6 W. T. SULAIMAN in view of (2.5), (2.6) and (2.7). ∞ ∑ π πΏπ+π−1 π β£ ππ3 β£ = π=2 ∞ ∑ π πΏπ+π−1 β£ π−1 ∑ π=2 ≤ ∞ ∑ π‘π£ ˆ π‘π,π£+1 Δππ£ β£π π£=1 π πΏπ+π−1 π=2 = π(1) π−1 ∑ π β£ π‘π£ β£ β£ π‘π£π£ β£ (π−1 ∑ ∞ ∑ ∞ ∑ ∞ ∑ )π−1 β£ π‘π£π£ β£β£ ˆ π‘π,π£+1 β£ π£=1 ππΏπ (π β£ π‘ππ β£)π−1 π−1 ∑ β£ π‘π£ β£π β£ π‘π£π£ β£1−π β£ ˆ π‘π,π£+1 β£β£ Δππ£ β£π π£=1 β£ π‘π£ β£π β£ π‘π£π£ β£1−π β£ Δππ£ β£π π£=1 = π(1) β£ˆ π‘π,π£+1 β£β£ Δππ£ β£π π£=1 π=2 = π(1) 1−π ∞ ∑ ππΏπ β£ ˆ π‘π,π£+1 β£ π=π£+1 π£ πΏπ β£ π‘π£ β£π β£ π‘π£π£ β£1−π β£ Δππ£ β£π = π(1), π£=1 in view of (2.3), (2.4) and (2.8). ∞ ∑ ππΏπ+π−1 β£ ππ4 β£π = π=2 ∞ ∑ ππΏπ+π−1 β£ π=2 = π(1) = π(1) ∞ ∑ π=1 ∞ ∑ π+1 π‘π π‘ππ ππ β£π π ππΏπ+π−1 β£ π‘π β£π β£ π‘ππ β£π β£ ππ β£π ππΏπ β£ π‘ππ β£β£ π‘π β£π β£ ππ β£π = π(1), π=1 in view of (2.7). In order to complete the proof the rest, it is suο¬cient to show that ∞ ∑ π£ πΏπ β£ π‘π£π£ β£β£ π‘π£ β£π β£ ππ£ β£π = π(1), π£=1 ∞ ∑ π£ πΏπ β£ π‘π£ β£π β£ π‘π£π£ β£1−π β£ Δππ£ β£π = π(1). π£=1 Now, π π ∑ ∑ π£ πΏπ−1 β£ π‘π£ β£π π−1 π£ πΏπ β£ π‘π£π£ β£β£ π‘π£ β£π β£ ππ£ β£π = π(1) ππ£ β£ ππ£ β£π−1 β£ ππ£ β£ π−1 π π£ π£=1 π£=1 π ∑ π£ πΏπ−1 β£ π‘π£ β£π β£ ππ£ β£ ππ£π−1 π£=1 ( π£ ) (π ) π−1 ∑ π£ πΏπ−1 β£ π‘π£ β£π ∑ ∑ ππΏπ−1 β£ π‘π β£π β£ Δ β£ ππ£ β£ β£ +π(1) β£ ππ β£ = π(1) πππ−1 ππ£π−1 π£=1 π=1 π£=1 = π(1) = π(1) π−1 ∑ ππ£ β£ Δππ£ β£ +π(1)ππ β£ ππ β£ π£=1 = π(1) π−1 ∑ π£=1 ππ£ π½π£ + π(1)ππ β£ ππ β£= π(1). A STUDY ON ABSOLUTE FACTORS FOR A TRIANGULAR MATRIX π ∑ π£ πΏπ π 1−π β£ π‘π£ β£ β£ π‘π£π£ β£ π β£ Δππ£ β£ = π(1) π£=1 π ∑ 7 π£ πΏπ+π−1 β£ π‘π£ β£π β£ Δππ£ β£π π£=1 π ∑ π£ πΏπ β£ π‘π£ β£π = π(1) (π£ππ£ π½π£ )π−1 π½π£ π−1 π π£ π£=1 π ∑ π£ πΏπ−1 β£ π‘π£ β£π π£π½π£ ππ£π−1 π£=1 ) (π ) ( π£ π−1 ∑ π£ πΏπ−1 β£ π‘π£ β£π ∑ ∑ ππΏπ−1 β£ π‘π β£π β£ Δ(π£π½π£ ) β£ +π(1) ππ½π = π(1) πππ−1 ππ£π−1 π£=1 π£=1 π=1 = π(1) = π(1) π ∑ ππ£ π½π£ + π(1) π£=1 π ∑ π£ππ£ β£ Δπ½π£ β£ +π(1)πππ π½π π£=1 = π(1). This completes the proof of the Theorem. Recalling Lemma 2.2 and its proof. Lemma 2.2. Let (ππ ) be a quasi (π½ − πΎ)−power increasing sequence, 0 < π½ < 1, πΎ ≥ 0, then the conditions (2.10) and (2.11) implies ∞ ∑ π½π ππ < ∞ (2.14) π=1 ππ½π ππ < ∞. (2.15) Proof. Since π½π → 0, then Δπ½π → 0, and hence ∞ ∞ ∞ ∞ π£ ∑ ∑ ∑ ∑ ∑ π½π ππ ≤ ππ β£ Δπ½π β£= β£ Δπ½π£ β£ ππ π=1 π£=π π=1 = ∞ ∑ β£ Δπ½π£ β£ π£=1 = π(1) π£=1 π£ ∑ π=1 ππ½ (log π)πΎ ππ π−π½ (log π)−πΎ π=1 ∞ ∑ β£ Δπ½π£ β£ π£ π½ (log π£)πΎ ππ£ π£=1 = π(1) ∞ ∑ β£ Δπ½π£ β£ π£ π½ (log π£)πΎ ππ£ π£=1 = π(1) ∞ ∑ π£ ∑ π=1 π£ ∑ π−π½ (log π)−πΎ ππ (log π)−πΎ π−π½−π , 0 < π < π½ + π < 1 π=1 β£ Δπ½π£ β£ π£ π½ (log π£)πΎ ππ£ π£ π (log π£)−πΎ π£=1 = π(1) ∞ ∑ π£=1 = π(1) = π(1) ∞ ∑ π£=1 ∞ ∑ π£=1 π£ ∑ π=1 β£ Δπ½π£ β£ π£ π½+π ππ£ ∫ π£ π₯−π½−π ππ₯ 0 β£ Δπ½π£ β£ π£ π½+π ππ£ π£ 1−π½−π π£ β£ Δπ½π£ β£ ππ£ = π(1), π−π½−π 8 W. T. SULAIMAN in view of (2.11). ∞ ∑ ππ½π ππ = πππ Δπ½π£ ≤ πππ π=π£ ∞ ∑ β£ Δπ½π£ β£ π=π£ = π1−π½ (log π)−πΎ ππ½ (log π)πΎ ππ ∞ ∑ β£ Δπ½π£ β£ π=π£ ≤π 1−π½ −πΎ (log π) ∞ ∑ π£ π½ (log π£)πΎ ππ£ β£ Δπ½π£ β£ π=π£ ≤ = ∞ ∑ π=π£ ∞ ∑ π£ 1−π½ (log π£)−πΎ ππ£ π£ π½ (log π£)πΎ ππ£ β£ Δπ½π£ β£ π£ππ£ β£ Δπ½π£ β£< ∞, π=π£ in view of (2.11). β‘ ∑π Corollary 2.3. Let (ππ ) be a positive sequence such that ππ = π£=0 ππ£ → ∞, π‘π denote the π−th (πΆ, 1) mean of the sequence (πππ ), and let (ππ ) be a sequence of numbers such that ππ , π‘π , ππ are all satisfying πππ = π(ππ ), ππ = π(πππ ), ∞ ∑ ∞ ∑ π£ πΏπ π£=1 ∞ ∑ π£=1 π£ πΏπ ππ = π(π£ πΏπ /ππ£−1 ), ππ ππ−1 (2.17) ππ£ β£ π‘π£ β£π β£ ππ£ β£π = π(1), ππ£ (2.18) ππΏπ π=π£+1 ( ππ£ ππ£ (2.16) )π−1 β£ π‘π£ β£π β£ Δππ£ β£π = π(1), (2.19) ∑ then the series ππ ππ is summable β£ π , ππ β£π , π ≥ 1, πΏ ≥ 0. Furthermore if (ππ ) is a quasi (π½−πΎ)−power increasing sequence, 0 < π½ < 1, πΎ ≥ 0, and if (π½π ) is a sequence of numbers satisfying (2.9)-(2.13), then the conditions ∑ (2.17) are (2.18) omitted in order to obtain the β£ π , ππ β£π −summability of ππ ππ . Proof. The proof follows from Theorem 2.1 by putting π ≡ (π , ππ ), that is π‘ππ£ = ππ ππ£−1 ππ ππ£ ππ£ ˆ , π‘ππ£ = , Δπ£ ˆ π‘ππ£ = − . ππ ππ ππ−1 ππ ππ−1 β‘ References [1] E. Savas, A study on absolute summability factors for a triangular matrix, Math. Ineq. Appl. 12 (2009) 41–46. [2] W. T. Sulaiman, A study relation between two summability methods, Proc. Amer. Math. Soc. 115 (1992) 303–312. A STUDY ON ABSOLUTE FACTORS FOR A TRIANGULAR MATRIX 9 Waad T. Sulaiman Department of Computer Engineering , College of Engineering, University of Mosul, Iraq E-mail address: waadsulaiman@hotmail.com