International Journal of Analysis and Applications ISSN 2291-8639 Volume 7, Number 2 (2015), 129-144 http://www.etamaths.com ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS N.SHOBANADEVI1,∗ AND J.JAGAN MOHAN2,∗ Abstract. Discrete Mittag - Leffler functions play a major role in the devel- opment of the theory of discrete fractional calculus. In the present article, we analyze qualitative properties of discrete Mittag - Leffler functions and estab- lish sufficient conditions for convergence, oscillation and summability of the infinite series associated with discrete Mittag - Leffler functions. 1. Introduction & Preliminaries Fractional calculus is a mathematical branch investigating the properties of derivatives and integrals of fractional orders. Many scientists have paid lot of attention due to its interesting applications in various fields of science and engi- neering, such as viscoelasticity, diffusion, neurology, control theory and statistics [24]. Like the exponential function in the theory of differential equations, Mittag - Leffler function plays an important role in the theory of fractional differential equations. The definition for one parameter Mittag - Leffler function was given by Gösta Mittag Leffler [22]. Later, Agarwal [1] defined the two parameter Mittag - Leffler function. Definition 1. Let t ∈ R and α,β ∈ R+. The one and two parameter Mittag - Leffler functions are defined by Eα(t) = ∞∑ k=0 tk Γ(αk + 1) ,(1.1) Eα,β(t) = ∞∑ k=0 tk Γ(αk + β) .(1.2) The analogous theory for nabla discrete fractional calculus was initiated by Miller & Ross [21], Gray & Zhang [10] and Atici & Eloe [6], where basic approaches, definitions, and properties of fractional sums and differences were discussed. A series of papers continuing this research has appeared recently [6, 7, 8, 9, 12, 13, 14, 15, 16, 17, 19, 23, 25, 26]. Throughout this article, we shall use the following notations, definitions and known results of nabla discrete fractional calculus [6, 25]. For any a, b ∈ R, Na = {a,a + 1,a + 2, ...........}, Na,b = {a,a + 1,a + 2, ...........,b} where a < b. 2010 Mathematics Subject Classification. 39A10, 39A99. Key words and phrases. fractional order; nabla difference; convergence; oscillation; summabil- ity; periodicity. c©2015 Authors retain the copyrights of their papers, and all open access articles are distributed under the terms of the Creative Commons Attribution License. 129 130 SHOBANADEVI AND MOHAN Definition 2. For any α, t ∈ R, the α rising function is defined by tα = Γ(t + α) Γ(t) , t ∈ R\{......,−2,−1, 0}, 0α = 0. We observe the following properties of rising factorial function. Lemma 1. Assume the following factorial functions are well defined. (1) tα(t + α)β = tα+β. (2) If t ≤ r then tα ≤ rα. (3) If α < t ≤ r then r−α ≤ t−α. Definition 3. Let u : Na → R, α ∈ R+ and choose N ∈ N1 such that N −1 < α < N. (1) (Nabla Difference) The first order backward difference or nabla difference of u is defined by ∇u(t) = u(t) −u(t− 1), t ∈ Na+1, and the Nth - order nabla difference of u is defined recursively by ∇Nu(t) = ∇(∇N−1u(t)), t ∈ Na+N. In addition, we take ∇0 as the identity operator. (2) (Fractional Nabla Sum) The αth - order fractional nabla sum of u is given by (1.3) ∇−αa u(t) = 1 Γ(α) t∑ s=a+1 (t−ρ(s))α−1u(s), t ∈ Na where ρ(s) = s− 1. Also, we define the trivial sum by ∇−0a u(t) = u(t) for t ∈ Na. (3) (R - L Fractional Nabla Difference) The αth - order Riemann - Liouville type fractional nabla difference of u is given by (1.4) ∇αau(t) = ∇ N [ ∇−(N−α)a u(t) ] , t ∈ Na+N. For α = 0, we set ∇0au(t) = u(t), t ∈ Na. (4) (Caputo Fractional Nabla Difference) The αth - order Caputo type frac- tional nabla difference of u is given by (1.5) ∇αa∗u(t) = ∇ −(N−α) a [ ∇Nu(t) ] , t ∈ Na+N. For α = 0, we set ∇0a∗u(t) = u(t), t ∈ Na. The unified definition for fractional sums and differences is as follows. Definition 4. Let u : Na → R, α ∈ R+ and choose N ∈ N1 such that N −1 < α < N. Then (1) the αth - order nabla fractional sum of u is given by ∇−αa u(t) = 1 Γ(α) t∑ s=a+1 (t−ρ(s))α−1u(s), t ∈ Na. ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 131 (2) the αth - order R - L fractional difference of u is given by (1.6) ∇αau(t) = { 1 Γ(−α) ∑t s=a+1(t−ρ(s)) −α−1u(s), α /∈ N1, ∇Nu(t), α = N ∈ N1, for t ∈ Na+N. Theorem 2. (Power Rule) Let α > 0 and µ > −1. Then, (1) ∇−αa (t−a)µ = Γ(µ+1) Γ(µ+α+1) (t−a)µ+α, t ∈ Na. (2) ∇αa (t−a)µ = Γ(µ+1) Γ(µ−α+1) (t−a) µ−α, t ∈ Na+N . Definition 5. A function u is said to be slowly oscillating if u(t) − u(s) → 0 for any t,s ∈ Na, whenever s →∞, t > s, ts → 1. Definition 6. A function u is said to be T - periodic if u(t + T) = u(t) for all t ∈ Na. The positive integer T is called the period of the function u. Further T is said to be the basic period if there does not exist a smaller period T1 ∈ Z+ such that T1 < T. Definition 7. A continuous and bounded function u is said to be S - asymptotically periodic if there exists T > 0 such that u(t + T)−u(t) → 0 as t →∞. In this case, we say that T is an asymptotic period of u and that u is S - asymptotically T - periodic. 2. Qualitative Properties of Discrete Mittag - Leffler Functions The definitions for one and two parameter discrete Mittag - Leffler functions are given by Atsushi Nagai [2] and Atici & Eloe [8] respectively. Definition 8. Let t ∈ N0, λ ∈ (−1, 1) and α,β ∈ R+. The one and two parameter discrete Mittag - Leffler functions are defined by Fα(λ,t) = ∞∑ k=0 λk tαk Γ(αk + 1) ,(2.1) Fα,β(λ,t α) = ∞∑ k=0 λk tαk Γ(αk + β) .(2.2) Lemma 3. We observe the following properties of (2.1) and (2.2). (1) Fα,1(λ,t α) = Fα(λ,t). (2) F1,1(λ,t 1) = F1(λ,t) = (1 −λ)−t. (3) Fα,β(λ,t α) ≥ 0. (4) Fα,β(λ,t α) ∼ Eα,β(λtα) (t →∞). (5) Fα(λ,t) ≥ [1+(α−1)λ (1−λ)2Γ(t), t ∈ N2. (6) Fα,β(λ,t α) ≤ (1 −λ)−1, 2 ≤ t ≤ β. (7) Fα,β(λ,t α) ≥ 1 (1−λ)Γ(t), 2 ≤ β ≤ t. Proof. The proofs of (1), (2), (3) and (4) follow from (2.1) and (2.2). To prove (5), we consider (2.1). Clearly, Γ(t + αk) Γ(αk + 2) ≥ 1, t ∈ N2. 132 SHOBANADEVI AND MOHAN Hence Fα(λ,t) = ∞∑ k=0 λk tαk Γ(αk + 1) ≥ 1 Γ(t) ∞∑ k=0 λk(αk + 1) = [1 + (α− 1)λ (1 −λ)2Γ(t) . Now we consider (2.2) to prove (6) and (7). Clearly, 1 Γ(t) ≤ 1 and Γ(t + αk) Γ(αk + β) ≤ 1, 2 ≤ t ≤ β and Γ(t + αk) Γ(αk + β) ≥ 1, 2 ≤ β ≤ t. Hence Fα,β(λ,t α) = ∞∑ k=0 λk tαk Γ(αk + β) ≤ ∞∑ k=0 λk = 1 1 −λ and Fα,β(λ,t α) = ∞∑ k=0 λk tαk Γ(αk + β) ≥ 1 Γ(t) ∞∑ k=0 λk = 1 (1 −λ)Γ(t) . � Theorem 4. The two parameter discrete Mittag - Leffler function has the following properties. (1) Fα,β(λ,t α) is monotonically increasing on N0. (2) Fα,β(λ,t α) is slowly oscillating on N0. (3) Fα,β(λ,t α) is not a periodic function. (4) Fα,β(λ,t α) is S - asymptotically periodic function on N0,b. Proof. Let t,s ∈ N0 such that t > s. Then t−s = T ∈ Z+. Consider Fα,β(λ,t α) −Fα,β(λ,sα) = ∞∑ k=0 λk Γ(αk + β) [ tαk −sαk ] = ∞∑ k=0 λk Γ(αk + β) [Γ(t + αk) Γ(t) − Γ(s + αk) Γ(s) ] = ∞∑ k=0 λk Γ(αk + β) [Γ(s + T + αk) Γ(s + T) − Γ(s + αk) Γ(s) ] = ∞∑ k=0 λk Γ(αk + β) Γ(s + αk) Γ(s) [(s + T − 1 + kα s + T − 1 )(s + T − 2 + kα s + T − 2 ) ... (s + kα s ) − 1 ] = ∞∑ k=0 λk Γ(αk + β) Γ(s + αk) Γ(s) [( 1 + kα s + T − 1 )( 1 + kα s + T − 2 ) ... ( 1 + kα s ) − 1 ](2.3) > 0. Thus, we have s < t ⇒ Fα,β(λ,sα) < Fα,β(λ,tα). ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 133 Further, letting s →∞ in (2.3), we get Fα,β(λ,t α) −Fα,β(λ,sα) → 0. Hence we have (1) and (2). Let T be any positive integer and consider Fα,β(λ, (t + T) α) = ∞∑ k=0 λk (t + T)αk Γ(αk + β) = ∞∑ k=0 λk Γ(t + T + αk) Γ(t + T)Γ(αk + β) 6= ∞∑ k=0 λk Γ(t + αk) Γ(t)Γ(αk + β) = Fα,β(λ,t α). Thus, Fα,β(λ,t α) is not a T - periodic function. Letting s →∞ in (2.3), we get (2.4) Fα,β(λ, (s + T) α) −Fα,β(λ,sα) → 0. Since Fα,β(λ,t α) is continuous and bounded on N0,b, the proof of (4) is complete. � Lemma 5. Let α, β and γ ∈ R+. The following are valid. (1) ∇Fα(λ,t) = λtα−1Fα,α(λ, (t + α− 1)α), t ∈ N0. (2) ∇ [ tβFα,β+1(λ, (t + β) α) ] = tβ−1Fα,β(λ, (t + β − 1)α), t ∈ N0. (3) ∇−β0 Fα(λ,t) = t βFα,β+1(λ, (t + β) α), t ∈ N0. (4) ∇−γ0 [ tβFα,β+1(λ, (t + β) α) ] = tβ+γFα,β+γ+1(λ, (t + β + γ) α), t ∈ N0. (5) ∇−γ−1 [ (t + 1)β−1Fα,β(λ, (t + β) α) ] = (t + 1)β+γ−1Fα,β+γ(λ, (t + β + γ) α), t ∈ N0. (6) ∇β0∗Fα(λ,t) = λt α−βFα,1(λ, (t + α−β)α), 0 < β < 1, t ∈ N1. (7) ∇γ0 [ tβFα,β+1(λ, (t + β) α) ] = tβ−γFα,β−γ+1(λ, (t + β −γ)α), t ∈ NN . (8) ∇γ−1 [ (t + 1)β−1Fα,β(λ, (t + β) α) ] = (t + 1)β−γ−1Fα,β−γ(λ, (t + β − γ)α), β 6= γ, t ∈ NN−1. (9) ∇β−1 [ (t+1)β−1Fα,β(λ, (t+β) α) ] = λ(t+1)α−1Fα,α(λ, (t+α) α), t ∈ NN−1. 134 SHOBANADEVI AND MOHAN Proof. Consider (1). ∇Fα(λ,t) = ∞∑ k=0 λk Γ(αk + 1) ∇tαk = ∞∑ k=0 λk Γ(αk + 1) [Γ(t + αk) Γ(t) − Γ(t− 1 + αk) Γ(t− 1) ] = ∞∑ k=0 λk Γ(αk + 1) Γ(t− 1 + αk) Γ(t− 1) [t− 1 + αk t− 1 − 1 ] = ∞∑ k=1 λk Γ(αk) Γ(t− 1 + αk) Γ(t) = λ ∞∑ k=0 λk Γ(αk + α) Γ(t− 1 + αk + α) Γ(t) = λ ∞∑ k=0 λk Γ(αk + α) tαk+α−1 = λtα−1 ∞∑ k=0 λk Γ(αk + α) (t + α− 1)αk = λtα−1Fα,α(λ, (t + α− 1)α). Consider (2). ∇ [ tβFα,β+1(λ, (t + β) α) ] = ∞∑ k=0 λk Γ(αk + β + 1) ∇[tβ(t + β)αk] = ∞∑ k=0 λk Γ(αk + β + 1) ∇tαk+β = ∞∑ k=0 λk Γ(αk + β + 1) [Γ(t + αk + β) Γ(t) − Γ(t− 1 + αk + β) Γ(t− 1) ] = ∞∑ k=0 λk Γ(αk + β + 1) Γ(t− 1 + αk + β) Γ(t− 1) [t− 1 + αk + β t− 1 − 1 ] = ∞∑ k=0 λk Γ(αk + β) Γ(t− 1 + αk + β) Γ(t) = ∞∑ k=0 λk Γ(αk + β) tαk+β−1 = tβ−1 ∞∑ k=0 λk Γ(αk + β) (t + β − 1)αk = tβ−1Fα,β(λ, (t + β − 1)α). ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 135 Consider (3). ∇−β0 Fα(λ,t) = ∞∑ k=0 λk Γ(αk + 1) ∇−β0 t αk = ∞∑ k=0 λk Γ(αk + 1) Γ(αk + 1) Γ(αk + β + 1) tβ+αk (using Power Rule) = tβ ∞∑ k=0 λk (t + β)αk Γ(αk + β + 1) (using Lemma 1(1)) = tβFα,β+1(λ, (t + β) α). Consider (4). ∇−γ0 [ tβFα,β+1(λ, (t + β) α) ] = ∞∑ k=0 λk Γ(αk + β + 1) ∇−γ0 [t β(t + β)αk] = ∞∑ k=0 λk Γ(αk + β + 1) ∇−γ0 t αk+β = ∞∑ k=0 λk Γ(αk + β + 1) Γ(αk + β + 1) Γ(αk + β + γ + 1) tαk+β+γ = tβ+γ ∞∑ k=0 λk (t + β + γ)αk Γ(αk + β + γ + 1) (using Lemma 1(1)) = tβ+γFα,β+γ+1(λ, (t + β + γ) α). Consider (5). ∇−γ−1 [ (t + 1)β−1Fα,β(λ, (t + β) α) ] = ∞∑ k=0 λk Γ(αk + β) ∇−γ−1 [(t + 1) β−1(t + β)αk] = ∞∑ k=0 λk Γ(αk + β) ∇−γ−1 (t + 1) αk+β−1 = ∞∑ k=0 λk Γ(αk + β) Γ(αk + β) Γ(αk + β + γ) (t + 1)αk+β+γ−1 = (t + 1)β+γ−1 ∞∑ k=0 λk (t + β + γ)αk Γ(αk + β + γ) = (t + 1)β+γ−1Fα,β+γ(λ, (t + β + γ) α). Consider (6). ∇β0∗Fα(λ,t) = ∇ −(1−β) 0 [ ∇Fα(λ,t) ] (using Definition 3(4)) = ∇−(1−β)0 [ λtα−1Fα,α(λ, (t + α− 1)α) ] (using (1)) = λtα+1−β−1Fα,β+1−β(λ, (t + α + 1 −β − 1)α) (using (4)) = λtα−βFα,1(λ, (t + α−β)α). 136 SHOBANADEVI AND MOHAN (7) and (8) are obtained by replacing γ by −γ in (4) and (5) respectively. Consider (9). ∇β−1 [ (t + 1)β−1Fα,β(λ, (t + β) α) ] = ∞∑ k=0 λk Γ(αk + β) ∇β−1[(t + 1) β−1(t + β)αk] = ∞∑ k=0 λk Γ(αk + β) ∇β−1(t + 1) αk+β−1 = ∞∑ k=1 λk Γ(αk + β) Γ(αk + β) Γ(αk) (t + 1)αk−1 = λ ∞∑ k=0 λk Γ(αk + α) (t + 1)αk+α−1 = λ(t + 1)α−1 ∞∑ k=0 λk (t + α)αk Γ(αk + α) (using Lemma 1(1)) = λ(t + 1)α−1Fα,α(λ, (t + α) α). � Remark 1. From Lemma 5(6), we have (2.5) ∇α0∗Fα(λ,t) = λFα(λ,t), 0 < α < 1, t ∈ N1, implies Fα(λ,t) is an eigenfunction of the operator ∇α0∗. In other words, Fα(λ,t) is a nontrivial solution of the fractional nabla difference equation ∇α0∗u(t) = λu(t), t ∈ N1. Remark 2. From Lemma 5(9), we have ∇α−1 [ (t + 1)α−1Fα,α(λ, (t + α) α) ] = λ(t + 1)α−1Fα,α(λ, (t + α) α), t ∈ N1, implies (t+1)α−1Fα,α(λ, (t+α) α) is an eigenfunction of the operator ∇α−1. That is, (t+ 1)α−1Fα,α(λ, (t+α) α) is the solution of the Riemann - Liouville type fractional nabla difference equation ∇α−1f(t) = λf(t), t ∈ N1. Now, we prove that (t + 1)α−1Fα,α(λ, (t + α) α) is also slowly oscillating on N0 and S - asymptotically periodic on N0,b. For this purpose, let t,s ∈ N0 such that t > s. Then t−s = T ∈ Z+. Now consider ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 137 u(t) −u(s) = (t + 1)α−1Fα,α(λ, (t + α)α) − (s + 1)α−1Fα,α(λ, (t + α)α) = (t + 1)α−1 ∞∑ k=0 λk (t + α)αk Γ(αk + α) − (s + 1)α−1 ∞∑ k=0 λk (s + α)αk Γ(αk + α) = ∞∑ k=0 λk (t + 1)αk+α−1 Γ(αk + α) − ∞∑ k=0 λk (s + 1)αk+α−1 Γ(αk + α) = ∞∑ k=0 λk Γ(αk + α) [ (t + 1)αk+α−1 − (s + 1)αk+α−1 ] = ∞∑ k=0 λk Γ(αk + α) [Γ(t + αk + α) Γ(t + 1) − Γ(s + αk + α) Γ(s + 1) ] = ∞∑ k=0 λk Γ(αk + α) [Γ(s + T + αk + α) Γ(s + T + 1) − Γ(s + αk + α) Γ(s + 1) ] = ∞∑ k=0 λk Γ(αk + α) Γ(s + αk + α) Γ(s + 1) [(s + T − 1 + αk + α s + T ) ... (s + αk + α s + 1 ) − 1 ] = ∞∑ k=0 λk Γ(αk + α) Γ(s + αk + α) Γ(s + 1) [( 1 + αk + α− 1 s + T ) ... ( 1 + αk + α− 1 s + 1 ) − 1 ] . (2.6) Letting s → ∞ in (2.6), we get u(t) − u(s) → 0, i.e., u(s + T) − u(s) → 0. Further, (t + 1)α−1Fα,α(λ, (t + α) α) is continuous and bounded on N0,b. Hence the proof. Finally, we show that (t + 1)α−1Fα,α(λ, (t + α) α) is also not a T - periodic function. Let T be any positive integer and consider u(t + T) = (t + T + 1)α−1Fα,α(λ, (t + T + α) α) = (t + T + 1)α−1 ∞∑ k=0 λk (t + T + α)αk Γ(αk + α) = ∞∑ k=0 λk (t + T + 1)αk+α−1 Γ(αk + α) 138 SHOBANADEVI AND MOHAN = ∞∑ k=0 λk Γ(t + T + αk + α) Γ(t + T + 1)Γ(αk + α) 6= ∞∑ k=0 λk Γ(t + αk + α) Γ(t + 1)Γ(αk + α) = ∞∑ k=0 λk (t + 1)αk+α−1 Γ(αk + α) = (t + 1)α−1 ∞∑ k=0 λk (t + α)αk Γ(αk + α) = (t + 1)α−1Fα,α(λ, (t + α) α) = u(t). 3. Convergence & Oscillation In the present section we establish sufficient conditions on convergence and di- vergence of the infinite series (3.1) ∞∑ k=0 λk tαk Γ(αk + β) associated with discrete Mittag - Leffler function. The following theorem discusses the convergence of (3.1) using D’Alembert’s ratio test. Theorem 6. The infinite series (3.1) converges absolutely for each t ∈ Na, λ ∈ R and α,β ∈ R+ such that |λ| < 1. Proof. Consider (3.1). Here ak = λ k t αk Γ(αk + β) = λk Γ(t + αk) Γ(t)Γ(αk + β) = λk 1 Γ(t) Γ(αk + t) Γ(αk + β) . Then ak+1 ak = λ Γ(αk + t + α) Γ(αk + t) Γ(αk + β) Γ(αk + β + α) . As k →∞, ∣∣∣ak+1 ak ∣∣∣ = |λ|(αk)(t+α−t)(αk)(β−(β+α)) = |λ|. Using D’Alembert’s Ratio test, the proof is complete. � Remark 3. Since absolute convergence implies convergence, the infinite series (3.1) converges for |λ| < 1 and diverges for |λ| ≥ 1. Divergent series are often classified further into properly divergent, oscillate finitely and oscillate infinitely series [18]. In the following theorems we discuss these properties for (3.1). Theorem 7. The infinite series (3.1) diverges properly for each t ∈ Na, λ ∈ R and α,β ∈ R+ such that t ≥ β ≥ 1 and λ ≥ 1. Proof. Consider the infinite series (3.1) with λ ≥ 1. Here ak = λ k t αk Γ(αk + β) = λk Γ(t + αk) Γ(t)Γ(αk + β) . ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 139 Let sn be the n-th partial sum of the series (3.1). Then sn = n∑ k=0 ak = n∑ k=0 λk Γ(t + αk) Γ(t)Γ(αk + β) . Since λ ≥ 1 and Γ(t+αk) Γ(αk+β) ≥ 1 for t ≥ β ≥ 1, we have sn ≥ n∑ k=0 1 Γ(t) = (n + 1) Γ(t) and hence lim n→∞ sn = +∞. Thus, the infinite series (3.1) diverges properly for t ≥ β ≥ 1 and λ ≥ 1. � Let λ ∈ R+. Replacing λ by −λ in (3.1), we get an alternating series of the form (3.2) ∞∑ k=1 (−1)k−1λk−1 tαk−α Γ(αk −α + β) . A general criterion for infinite oscillation and a different version of Leibnitz test for alternating series are given in Theorems 8 and 9 respectively. Theorem 8. [4] If A = ∑∞ k=1(−1) k−1ak, where ak > 0 and ak+1 ak → λ > 1 as k →∞, then A oscillates infinitely. Theorem 9. [3] Given an alternating series A = ∑∞ k=1(−1) k−1ak, ak > 0, if ak ak+1 can be expressed in the form (3.3) ak ak+1 = 1 + µ k + O ( 1 kp ) , p > 1 then A is convergent if µ > 0, oscillatory µ ≤ 0. Using Theorems 8 and 9, we now discuss the oscillatory behaviour of (3.2). Theorem 10. The alternating series (3.2) oscillates infinitely for each t ∈ Na, λ ∈ R and α,β ∈ R+ such that λ > 1. Proof. Consider (3.2). Here ak = λ k−1 t αk−α Γ(αk −α + β) = λk−1 Γ(t + αk −α) Γ(t)Γ(αk −α + β) = λk−1 1 Γ(t) Γ(αk + t−α) Γ(αk + β −α) > 0. Then ak+1 ak = λ Γ(αk + t) Γ(αk + t−α) Γ(αk + β −α) Γ(αk + β) . As k →∞, ak+1 ak = λ(αk)(t−(t−α))(αk)((β−α)−β) = λ. Using Theorem 8 the proof is complete. � Theorem 11. The alternating series (3.2) oscillates finitely for each t ∈ Na and α,β ∈ R+. Proof. Consider (3.2) with λ = 1. Here ak = tαk−α Γ(αk −α + β) = Γ(t + αk −α) Γ(t)Γ(αk −α + β) = 1 Γ(t) Γ(αk + t−α) Γ(αk + β −α) > 0. 140 SHOBANADEVI AND MOHAN Then ak ak+1 = Γ(αk + t−α) Γ(αk + t) Γ(αk + β) Γ(αk + β −α) . For large k, ak ak+1 = (αk)((t−α)−t) [ 1 + O ( 1 αk )] (αk)(β−(β−α)) [ 1 + O ( 1 αk )] = 1 + O ( 1 k2 ) . Using Theorem 9 the proof is complete. � Combining all these results, we have Corollary 1. Let t ∈ Na, λ ∈ R and α,β ∈ R+. The infinite series (3.1) converge, for −1 < λ < 1; diverge properly, for λ ≥ 1 and t ≥ β ≥ 1; oscillate finitely, for λ = −1; oscillate infinitely, for λ < −1. Since the radius of convergence of (3.1) is 1, we have the following result on the uniform convergence of (3.1). Corollary 2. Let t ∈ Na, λ ∈ R and α,β ∈ R+. For any 0 < r < 1, the infinite series (3.1) converges uniformly for each λ ∈ [−r,r]. 4. Preliminaries on Summability The present section contains some basic definitions and results concerning summa- bility theory [11, 20, 3] which will be useful in section 4. The method of convergent series is simply a particular method of associating a definite number called sum denoted by s with the series and using this number in place of the convergent series in calculations. But for the divergent series, this sum does not exist. The problem of divergent series is to associate a number with such a series called Sum denoted by S so that it can be used in place of the divergent series in calculations. Any definite method by which we can associate a Sum with a given divergent series is called the method of summation. The methods of summation are designed primarily for the oscillating series. We now discuss three important summability methods given by Abel, Borel and Le Roy [3, 11, 20]. Let k ∈ N0 and ak,λ ∈ C. Consider a divergent series (4.1) ∞∑ k=0 ak. Abel’s Method: [3, 11, 20] The series (4.1) is said to be Abel - summable (A - summable), if the series (4.2) ∞∑ k=0 akλ k converges in the disk D = {λ : |λ| < 1} and (4.3) lim λ→1− ∞∑ k=0 akλ k = S. ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 141 Then S is called A-sum of the series (4.1) and is denoted by (4.4) ∞∑ k=0 ak = S (A). Borel’s Method: [3, 11, 20] The series (4.1) is said to be Borel - summable (B - summable), if the series (4.5) e−λ ∑ ak λk k! converges to S and (4.6) ∫ ∞ 0 e−λ ∞∑ k=0 ak λk k! dλ = lim λ→∞ ∫ λ 0 e−λ ∞∑ k=0 ak λk k! dλ = S. The complex number S is called B-sum of the series (4.1) and is denoted by (4.7) ∞∑ k=0 ak = S (B). Le Roy’s Method: [3, 11, 20] The series (4.1) is said to be Le Roy - summable (L - summable), if (4.8) lim α→1 lim n→∞ n∑ k=0 ak Γ(αk + 1) Γ(k + 1) = S, 0 < α < 1. Here S is called L-sum of the series (4.1) and is denoted by (4.9) ∞∑ k=0 ak = S (L). Finally we conclude this section with two important theorems on A and L summa- bilities. Theorem 12. [3, 11, 20] Every convergent series is summable (A, L) with Sum equal to sum i.e. Abel and Le Roy methods are regular. Theorem 13. [3, 11, 20] A properly divergent series is not summable (A, L) with finite Sum. 5. Summability of the Infinite Series (3.1) In this section we discuss summability of (3.1) using the results obtained in section 3 and preliminaries described in section 4. The following corollary is a consequence of Theorems 12, 13 and Corollary 1. Corollary 3. For each t ∈ Na and α,β ∈ R+, the infinite series (3.1) is (1) (A, L)- summable for −1 < λ < 1. (2) not (A, L) - summable with finite Sum for λ ≥ 1. (3) uniformly (A, L) - summable for each λ ∈ [−r,r] such that 0 < r < 1. Theorem 14. For each t ∈ Na and α,β ∈ R+, the infinite series (3.1) is A - summable for λ = −1. 142 SHOBANADEVI AND MOHAN Proof. From Corollary 1 we know that (5.1) ∞∑ k=0 (−1)k tαk Γ(αk + β) oscillates finitely. Also, (5.2) ∞∑ k=0 (−1)k tαk Γ(αk + β) λk converges absolutely in the disk D = {λ : |λ| < 1} and (5.3) lim λ→1− ∞∑ k=0 (−1)k tαk Γ(αk + β) λk exists and is finite. So, the infinite series (3.1) is A - summable with finite Sum for λ = −1. � Theorem 15. For each t ∈ Na, 0 < α < 1 and β = 1, the infinite series (3.1) is L - summable for λ < 1. Proof. Consider the alternating series (5.4) ∞∑ k=0 (−1)k(−λ)k tαk Γ(αk + β) with λ ≥ 1. Clearly it oscillates finitely for λ = 1 and oscillate infinitely for λ > 1. Here ak = (−λ)k tαk Γ(αk + 1) . From (4.8), we have S = lim α→1 lim n→∞ n∑ k=0 ak Γ(αk + 1) Γ(k + 1) = lim α→1 lim n→∞ n∑ k=0 ( t + αk − 1 k ) (−λ)k = ( 1 1 + λ )t exists for each λ > −1. Thus (3.1) is L - summable for λ < 1. Hence the proof. � 6. Conclusion Taking α = β = 1 in (3.1), we get (6.1) ∞∑ k=0 λk Γ(t + k) Γ(t)Γ(αk + 1) = ∞∑ k=0 ( t + k − 1 k ) λk = ∞∑ k=0 ( −t k ) λk. We know that, Theorem 16. For each t ∈ Na, the infinite series (6.1) (1) converges to (1 −λ)−t, for −1 < λ < 1; diverges properly, for λ ≥ 1; oscillates finitely, for λ = −1; oscillates infinitely, for λ < −1. (2) is A - summable for −1 ≤ λ < 1 and B - summable for λ < 1 (3) is not (A, B) - summable with finite Sum for λ ≥ 1 (4) is uniformly (A, B) - summable for each λ ∈ [−r,r] such that 0 < r < 1. ANALYSIS OF DISCRETE MITTAG - LEFFLER FUNCTIONS 143 Theorem 16 gives a proper justification to Corollaries 1, 2, 3, Theorems 14 and 15 for α = β = 1. Here we note that Le Roy’s definition coincides with Borel’s, whenever the later is convergent [3, 11, 20]. So one can replace B - Summability by L - Summability in (2), (3) and (4) of Theorem 16. References [1] Agarwal, R.P., A propos d’une note de M.Pierre Humbert, C. R. Acad. Sci., 236 (1953), 2031 - 2032. [2] Atsushi Nagai, Discrete Mittag - Leffler function and its applications, Publ. Res. Inst. Math. Sci., Kyoto Univ., 1302 (2003), 1 - 20. [3] Bromwich, T.J., An Introduction to the Theory of Infinite Series, MacMillan, London, 1908. [4] Butterworth, I.B., Infinite oscillation of alternating series, The Mathematical Gazette, 34 (1950), Number 310, 298 - 300. [5] Elaydi, S., An Introduction to Difference Equations, Undergraduate Texts in Mathematics, 3rd Edition, Springer, New York, 2005. [6] Ferhan M.Atici and Paul W.Eloe, Discrete fractional calculus with the nabla operator, Elec- tron. J. Qual. Theory Differ. Equat., Special Edition I (2009), Number 13, 12 Pages (Electron- ic). [7] Ferhan M.Atici and Paul W.Eloe, Gronwall’s inequality on discrete fractional calculus, Com- puters and Mathematics with Applications, 64 (2012), 3193 - 3200. [8] Ferhan M.Atici and Paul W.Eloe, Linear systems of nabla fractional difference equations, Rocky Mountain Journal of Mathematics, 41 (2011), Number 2, 353 - 370. [9] George A.Anastassiou, Nabla discrete fractional calculus and nabla inequalities, Mathematical and Computer Modelling, 51 (2010), 562 - 571. [10] Gray, H.L. and Zhang, N.F., On a new definition of the fractional difference, Mathematics of Computaion, 50 (1988), Number 182, 513 - 529. [11] Hardy, G.H., Divergent Series, Oxford Press, London, 1949. [12] Hein, J., Mc Carthy, S., Gaswick, N., Mc Kain, B. and Spear, K., Laplace transforms for the nabla difference operator, Pan American Mathematical Journal, 21 (2011), Number 3, 79 - 96. [13] Jagan Mohan, J. and Deekshitulu, G.V.S.R., Solutions of nabla fractional difference equations using N - transforms, Commun. Math. Stat., 2 (2014), 1 - 16. [14] Jagan Mohan, J., Deekshitulu, G.V.S.R. and Shobanadevi, N., Stability of nonlinear nabla fractional difference equations using fixed point theorems, Italian Journal of Pure and Applied Mathematics, 32 (2014), 165 - 184. [15] Jagan Mohan, J. and Shobanadevi, N., Stability of linear nabla fractional difference equations, Proceedings of the Jangjeon Mathematical Society, 17 (2014), Number 4, 651 - 657. [16] Jagan Mohan, J., Variation of parameters for nabla fractional difference equations, Novi Sad J. Math., 44 (2014), Number 2, 149 - 159. [17] Jagan Mohan Jonnalagadda, Solutions of perturbed linear nabla fractional difference equa- tions, Differential Equations and Dynamical Systems, 22 (2014), Number 3, 281 - 292. [18] James M.Hylop, Infinite Series, Interscience Publishers, London, 1959. [19] Jan Cermak, Tomas Kisela and Ludek Nechvatal, Stability and asymptotic properties of a linear fractional difference equation, Advances in Difference Equations 2012 (2012), Article ID 122. [20] Lloyd Leroy Smail, Some generalizations in the theory of summable divergent series, Disser- tation, Columbia University, 1913. [21] Miller, K.S. and Ross, B., Fractional difference calculus, Proceedings of the International Symposium on Univalent Functions, Fractional Calculus and Their Applications, 139 - 152, Nihon University, Koriyama, Japan, 1989. [22] Mittag - Leffler, G.M., Sur la nouvelle fonction Eα(x), C. R. Acad. Sci. Paris, 137 (1903), 554 - 558. [23] Nihan Acar, Ferhan M.Atici, Exponential functions of discrete fractional calculus, Applicable Analysis and Discrete Mathematics, 7 (2013), 343 - 353. [24] Podlubny, I., Fractional Differential Equations, Academic Press, San Diego, 1999. [25] Thabet Abdeljawad, Fahd Jarad and Dumitru Baleanu, A semigroup like property for discrete Mittag - Leffler functions, Advances in Difference Equations, 2012 (2012), Article ID 72. 144 SHOBANADEVI AND MOHAN [26] Thabet Abdeljawad and Ferhan M.Atici, On the definitions of nabla fractional operators, Abstract and Applied Analysis, 2012 (2012), Article ID 406757. 1Fluid Dynamics Division, School of Advanced Sciences, VIT University, Vellore - 632014, Tamil Nadu, India 2Department of Mathematics, Birla Institute of Technology and Science Pilani, Hyderabad Campus, Hyderabad - 500078, Telangana, India ∗Corresponding author