Internat. J. Math. & Math. Sci. VOL. 13 NO. 3 (1990) 501-506 501 CONVEXITY PRESERVING SUMMABILITY MATRICES C.R. SELVARAJ Pennsylvania State University Shenango Valley Campus 147 Shenango Avenue Sharon, PA 16146 (Received February 28, 1989 and in revised form May 27, 1989) ABSTRACT. The maln result of thls paper gives the necessary and sufficient conditions for the Abel matrices to preserve the convexity of sequences. Also, the higher orders of the Cesro method are shown to be convexlty-preservlng matrices. KEY WORDS AND PHRASES. Convex sequences, convexlty-preservlng matrices, domain of a matrix. 1980 AMS SUBJECT CLASSIFICATION CODE. 1. 40 INTRODUCTION. A real sequence (Xn)n= 0 is called convex if Its second order differences A2xn Xn+2 2Xn+l + Xn (1.1) > 0 0,1,2, for n G.H. Toader [I] proved following two theorems characterizing the convex the sequences. THEOREM I.I. If the sequence (xk) is given by k x k then A2Xk (k- i + I) (1.2) bi, bk+ 2. THEOREM 1.2. bk’S [ i=O k) The sequence (x is convex if and only if b k ) 0 for k ) 2 where are given by (1.2). The aim of the next section is to prove the necessary and sufficient conditions for the Abel matrix to preserve the convexity of sequences, using the above theorems. 502 2. C.R. SELVARAJ CONVEXITY PRESERVATION OF ABEL MATRIX. For each sequence (t n) satisfying 0 (Frldy [2, p. 421]) is defined by < t n < and llm n 0 t the Abel matrix A t At In,k] tn(1 tn)k. In [2], Frldy gives the characterization of the domain of the matrix A A sequence x t It follows that all convex I. is in the domain of At if and only if ll/k’llmlxk k sequences need not be in the domain of At An obvious example is the convex sequence k k for k given by x we restrict ourselves to the convex 1,2,3, So, k sequences that are in the domaln of the matrix. x THEOREM 2.1. The Abel matrix A t preserves the convexity of the sequences that are in the domain of A if and only if for all n 0,1,2,..., t (a) 0 and A2yn(1) A2yn(k) (b) 0, for k 2,3,..., 0,I,2,3,..., the sequence {Yn(k)}nO is defined by where for each k (1-t)k n Yn (k) PROOF. t n It is useful to note that A2yn A2yn (0). (2.1) First we prove a general result for a convex sequence in terms of the sequences {Yn(k)}n.0. If x is a convex sequence in the domain of .. . At, then the nth term of the transformed sequence is given by (AtX)n tn tn k=0 b i=0 i tn,k (k- i + l) b I, using (1.2) (22) (I k=i t )k(k n i + I). The change in the order of summation is justified as follows: k k-0 n i-0 (Atx) n + (Ibol )k b O) tn)k + (Ibll (k + I)(I k =0 (1-t) (AtX)n i 0 bl k(1 tn )k, (2.3) for all n, the rlght-hand side (0,I) n Therefore from (2.2), and since x is in the domain of A and t of the equation (2.3) converges. . bl) k =0 n t n -t CONVEXITY PRESERVING SUMMABILITY MATRICES 503 0,1,2,..., It follows that for n A2(AtX)n i=OE b (I i[ -tn+2tn+2 )i 2(I t-n+Itn+l)i -ttn)i ] (I + . Using the equation (2.|), we can write the above equation as follows: A 2(Atx) (b n 0 + b I)A2 (1) + Yn kA2yn(k). (2.4) b kffi2 Now, assume that conditions (a) and (b) given in the theorem are true. equation (2.4) together with Theorem 1.2 implies that A2(AtX)n Then the 0. ) Hence, the sequence At x is convex. Conversely, assume that the matrix At preserves convexity of the sequences in its Then there exists a nonSuppose that the condition (a) fails to hold. L # If 0. O, consider the convex L negative integer N such that do,in. A2yN(1) k) that any sequence (x > It is easy to verify in the domain of A t may be written as in equation (1.2) by taking {-I, -2, -3,...}, which sequence u is k-I b x 0 and 0 b (k E x k k i + if0 l)bl, for k ) (2.5) I. Thus for the above sequence u, b and b k 0 0, for k ) I. Substituting these values of b’s into the equation (2.4) we get A2yN(I) A2(AtU)N -L < O, Atu which contradicts that the transformed sequence is convex. < 0, we repeat a similar argument considering the convex sequence Hence, if the matrix At preserves to get a contradiction. is true. (a) convexity of sequences, then the condition If L v the {1,2,3,...} Next, suppose that the condition (a) non-negative integer J is true, but not such that the sequence {yn(J)}n= 0 (b). Then there exists a is not convex. That is, for some N, A2yN(J) < (2.6) 0. Now, construct a convex sequence x as follows: -(I xj -(I tN+2) j xj_ tN+2 tN+ 1)j tN+l tS)J -(I xj_ 2 tN and choose the other terms of the sequence (x k) in such a way that all k # J 2. By construction, A2xj_2 bj > 0, by (2.6). A2xk bk+2 0 for C. R. SELVARAJ 504 bk’S Using these values of N, together with our assumption (a) in equation (2.4) for n we get A2(Atx)N bjA2yN(j) which again contradicts that Atx < 0, is a convex sequence. This completes the proof. We give below an example of an Abel matrix A t which preserves the convexity of sequences. The Abel matrix A t where the sequence (t EXAMPLE 2.1. is given by + 2’ for n=O,l,2,... ’ n n) We shall show that this matrix A t satisfies the conditions (a) and (b) asserted in Theorem 2.1. Since preserves the convexity of sequences. t Yn (1) t a2yn (1) n + I, n O. In order to see that the condition (b) is satisfied, for each k A2yn(k) k (n + 3) k-1 (n + 4) 2,3,...,we have k (n + 1) + k-1 k-1 (n + 2) (n + 3) 2(n + 2) k It suffices to prove that f(n + I) > f(n) where k (n + 2) k-I (n + 3) f(n) k (n + I) k-I (n + 2) By differentiation it can be easily proved that f(n) is an increasing function. 3. PRESERVING CONVEXITY BY CESARO. In this section we use Theorems I.I and 1.2 to prove that the higher orders of Cesro means preserve introduce In order to prove this result, we Let K represent the set of all real the convexity of sequences. some notations and terminologies. convex sequences. DEFINITION 3.1. For a summabillty matrix A we define the set K(A) as the set of all sequences such that the transformed sequences are convex, i.e., K(A) DEFINITION 3.2. K(A) K}. The matrix A is K-included by the matrix B provided that K(B). DEFINITION 3.3. K(A) {x: Ax c The matrix B is K-stronger than the matrix A provided that K(B). Before we prove the main result we state the following preliminary theorems which are easy to prove. CONVEXITY PRESERVING SUMMABILITY MATRICES 505 Let A be a row-finite matrLx and T be a trLangle (lower triangular -I preserves the convexLty matrix with non-zero dagonal entres). Then the matrix AT THEOREM 3.1. of sequences if and only if T is K-included by A. THEOREM 3.2. Let A be a row-finlte matrix and T be a triangle. Then the matrix AT -! is K-stronger than the identity matrix if and only if A is K-stronger than T. For j e]q, the jth order of the Cesro matrix is given by Hardy [3, pp. 96-98] as (n-k+j+l if k n+j j Cj [n,k] if k 0, > n. In [I], G.H. Toader proved that the Cesro matrix C is K-stronger than the identity matrix. THEOREM For 3.3. each e IN, the matrix -I Cj+ICj preserves the convexity sequences. PROOF. If Cj+I C is represented by it is easy to see that Aj (j+k) ---------, if k O, if k>n. n+j+! j+1 Aj [n,kl Let x be a convex sequence. n, Then n E (Ajx) n Aj[n,k]xk. k=0 A simple calculation shows that (J + n j )Xn (n Writing + j + J + n + (hX)n- (J + ) (Aj X)n-l" (3.1) n (Ajx) n [ i--O (3.2) (n- i + l)e i we can see that the equation (3.1) can be simplified to n + Xn =J Therefore for each n 82 xn In + (J + l)(n- i + I)] i=O e i. 0,I,2,..., [(n-3+J )en+2 nen+ which implies that (n- 3 + J)en+ 2 nen+ + (J + l)A2Xn" (3.3) of 506 C. R. SELVARAJ Since x is a convex sequence, it follows from Theorem 1.2 that A2xn bn+2 0 for n 0, I, Using this fact in the equation (3.3), we obtain en+ 2 0 for n (3.4) 0,I,... Hence, the equation (3.2) together with (3.4) implies that [cf. Theorem 1.2] the transformed sequence REMARK 3.1. Combining the theorems 3.1 and 3.3 we see that K(CI) Since the This completes the proof. is convex. Ajx e: c_ K(C2) c_ K(Cj) K(Cj+ I) preserves the convexity of sequences [I], it follows that Cesro matrix C for each j elNthe jth order of the Cesro matrix Cj {0,I,0,0,...,} which considering the sequence x preserves the convexity. Also, by is not convex, it is easy to prove that A2(AjX)n > 0. Cj+iC Thus the matrix is K-stronger than the identity matrix. Therefore K(C K(C I) 21 K(Cj) K(Cj+ I) Now we state this result in the following theorem. THEOREM 3.4. the sequences. For j ]N, the Jth order Cesro matrix Cj preserves the convexity of Also, Cj+ is K-stronger than Cj, which is K-stronger than the identity matrix. REMARK 3.2. The H6"ider’s summabillty matrix of order Hj (Cl)J Therefore the matrix Hi+ is Rj for J J is given by 1,2 preserves the convexity of the sequences and also the matrix K-stronger than the matrix REFERENCES I. TOADER, G.H., A Hierarchy of Convexity for Sequences, L’anal. Num. et la Th. de l’approx. 12 (1983), 187-192. 2 A Abel Transformations FRIDY, J .., 3. MARDY, G.H., Divergent Series, Oxford University Press, London, 1949. into I Canad. Math. Bull. 25 (1982), 421-427.