GEORGIAN MATHEMATICAL JOURNAL: Vol. 6, No. 1, 1999, 83-90 ON THE ABSOLUTE SUMMABILITY OF SERIES WITH RESPECT TO BLOCK-ORTHONORMAL SYSTEMS G. NADIBAIDZE Abstract. Theorems determining Weyl’s multipliers for the summability almost everywhere by the |c, 1| method of the series with respect to block-orthonormal systems are proved. In particular, it is stated that if the sequence {ω(n)} is the Weyl multiplier for the summability almost everywhere by the |c, 1| method of all orthogonal series, then there exists a sequence {Nk } such that {ω(n)} will be the Weyl multiplier for the summability almost everywhere by the |c, 1| method of all series with respect to the ∆k -orthonormal systems. The present paper deals with the summability almost everywhere (a.e.) by the |c, α| method of series with respect to block-orthonormal systems. Under the summability by the |c, α| method of the series ∞ X an n=1 is understood the convergence of the series ∞ X (α) (α) σ , n+1 − σn n=1 where σn(α) ∞ 1 X α = α An−k ak An k=1 are the Cesàro (c, α)-means. The problem of the summability a.e. by the |c, α| method of orthogonal series was considered by P.L. Ul’yanov [1]. In particular, he proved that if 1991 Mathematics Subject Classification. 42C20. Key words and phrases. Block-orthonormal systems, Weyl multiplier, series with respect to block-orthonormal systems, summability almost everywhere by the |c, 1| method. 83 c 1999 Plenum Publishing Corporation 1072-947X/99/0100-0083$15.00/0 84 G. NADIBAIDZE the condition ∞ X 1 < ∞, n ω(n) n=1 (1) is fulfilled for a positive nondecreasing sequence {ω(n)}, then the convergence of the series ∞ X an2 ω(n) n=1 guarantees the summability a.e. on (0, 1) by the |c, α| method (α > 12 ) of the series ∞ X an ϕn (x) (2) n=1 for every orthonormal system from L2 (0, 1). If however ∞ X 1 = ∞, n ω(n) n=1 then there exists an even function f (x) ∈ ∩ Lp [0, 2π] such that its Fourier p≥1 series f (x) ∼ ∞ X cn cos nx n=1 converges a.e. on [0, 2π] and for every fixed α > 0 is not |c, α| summable a.e. on [0, 2π] though ∞ X c2n ω(n) < ∞. n=1 Definition 1 (see [2]). Let {Nk } be an increasing sequence of natural numbers, ∆k = (Nk , Nk+1 ], k = 1, 2, . . . , and {ϕn } be a system of functions from L2 (0, 1). The system {ϕn } will be called a ∆k -orthonormal system (∆k -ONS) if: (1) kϕn k2 = 1, n = 1, 2, . . . ; (2) (ϕi , ϕj ) = 0 for i, j ∈ ∆k , i 6= j, k ≥ 1. Definition 2 (see [1]). A positive nondecreasing sequence {ω(n)} will be called the Weyl multiplier for the summability a.e. of series with respect to the ∆k -ONS {ϕn } if the condition ∞ X n=1 a2n ω(n) < ∞ (3) ON THE ABSOLUTE SUMMABILITY OF SERIES 85 guarantees the summability a.e. by the |c, α| method of the corresponding series (2). Below we shall quote the theorem showing that if the sequence {ω(n)} is the Weyl multiplier for the summability a.e. by the |c, 1| method of all orthogonal series (2), then it will be the Weyl multiplier for the summability a.e. by the |c, 1| method of all series (2) with respect to the ∆k -ONS for the increasing sequence of natural numbers {Nk }. Theorem 1. If a positive nondecreasing sequence {ω(n)} is the Weyl multiplier for the summability a.e. by the |c, 1| method of all orthonormal series (2), then there exists an increasing sequence of natural numbers {Nk } such that {ω(n)} is the Weyl multiplier for the summability a.e. by the |c, 1| method of all series (2) with respect to the ∆k = (Nk , Nk+1 ]-ONS. Proof. We prove this theorem by the Wang–Ul’yanov’s scheme (see [1]) modifying it accordingly. Let the positive nondecreasing sequence {ω(n)} be the Weyl multiplier for the summability a.e. by the |c, 1| method of all orthogonal series (2). Then condition (1) is fulfilled. As is known (see [1]), for the positive nondecreasing on [n0 , +∞) function ω(x) the series ∞ X m=n0 ∞ X 1 and m ω(m) m=n20 1 √ m ω( m) converge or diverge simultaneously. Therefore, taking into account (1), we have ∞ X 1 √ < ∞. 4 n ω( n) n=1 Then ∞ X R(n) √ < ∞, n ω( 4 n) n=1 where P ∞ R(n) = k=2 ∞ P 1√ 4 k ω( k) k=n+1 (4) 21 1√ 4 k ω( k) 12 . Obviously, R(1) = 1, R(n) < R(n + 1) and limn→∞ R(n) = +∞. Define the sequence k(n) by the recursion formula ( if R(n + 1) ≥ k(n) + 1, k(n) + 1 k(1) = 0, k(n + 1) = n ≥ 1. k(n) if R(n + 1) < k(n) + 1, 86 G. NADIBAIDZE Thus we obtain the nondecreasing sequence of nonnegative integers for which k(n) ≤ R(n), n = 1, 2, . . . . (5) Note that for the sequence k(n) there exists an increasing sequence of natural numbers {Nk } (it is assumed that N0 = 0) which is defined by the formula k(n) = max{k : Nk < n}. Then, taking into account (4) and (5), we find that ∞ X k(n) √ < ∞. n ω( 4 n) n=1 (6) Let {ϕn } be a block-orthonormal system with ∆k = (Nk , Nk+1 ] and condition (3) be fulfilled. Then for the corresponding series (2) we have σn (x) − σn−1 (x) = ∞ X 1 ai (i − 1) ϕi (x), n ≥ 2. n(n − 1) i=1 Denoting by c the absolute positive constants which, generally speaking, may have different values in different inequalities and using (6), we find that ∞ Z1 ∞ Z1 1 X X σn (x) − σn−1 (x)dx ≤ σn (x) − σn−1 (x)2 dx 2 ≤ n=2 0 n=2 ≤c ∞ X 1 2 n n=2 Z1 0 0 n X 2 21 ai (i − 1)ϕi (x) dx ≤ i=1 Z1 NX k(n) ∞ 2 Z1 X 1 ≤c ai (i−1)ϕi (x) + 2 n n=1 i=1 0 0 ≤c Nk(n) X 1 a2i i2 + k(n) 2 n n=1 i=1 ∞ X n X i=Nk(n) +1 n X i=Nk(n) +1 2 12 ai (i−1)ϕi (x) dx ≤ a2i i2 12 ≤ 4 [√ n] ∞ p ∞ n 1 X X X k(n) X 2 2 12 1 2 2 2 c i ai a ≤c ≤ + k(n) i i n2 n2 n=1 n=1 i=1 i=1 √ √ 1 X n ∞ p X 1 k(n)( n 4 n) 2 2 2 2 + + i ai ≤c n2 √ 4 n=1 i=[ n]+1 ON THE ABSOLUTE SUMMABILITY OF SERIES ≤ X ∞ n=1 √ 1 ∞ p X k(n) nω( 4 n)) 2 + √ n2 nω( 4 n)) 12 n=1 1 9 n8 + √ ∞ X ω( 4 n) n3 n=1 n X i=[ n X i=[ √ 4 n]+1 √ 4 i2 ai2 n]+1 i2 ai2 12 ≤ k(n) 12 √ kω( 4 n) n=1 ∞ 12 X 87 ≤ ∞ ∞ ∞ 12 X X 12 1 X 2 2 2 c ω(i) ≤c+c ≤ + c a < ∞, i a ω(i) i i n3 i=1 n=1 i=1 whence by Levy’s theorem ∞ X σn (x) − σn−1 (x) < ∞ a.e. on (0, 1). n=2 The theorem below makes it possible to determine the Weyl multipliers for the summability a.e. of the series (2) with respect to the ∆k -ONS for regularly increasing sequences {Nk }. Theorem 2. Let an increasing sequence of natural numbers {Nk } be given, for which the condition ∞ n X 1 (n → ∞) = O Nk2 Nn2 (7) k=n is fulfilled, and let k(n) = max{k : Nk < n}. If for the positive nondecreasing sequence {ω(n)} condition (1) is fulfilled, then for every ∆k -ONS {ϕn } the condition ∞ X n=1 a2n ω(n) k(n) < ∞ (8) guarantees the summability a.e. by the |c, 1| method of the corresponding series (2). Proof. Let conditions (1), (7) and (8) be fulfilled. Then for the corresponding series (2) we have ∞ Z1 ∞ Z1 1 X X σn (x) − σn−1 (x)dx ≤ σn (x) − σn−1 (x)2 dx 2 ≤ n=2 0 n=2 0 Z1 NX Z1 k(n) 2 1 ≤c ai (i−1)ϕi (x) dx+ 2 n n=1 i=1 ∞ X 0 0 n X i=Nk(n) +1 2 21 ai (i−1)ϕi (x) dx ≤ 88 G. NADIBAIDZE √ 4 [X n] ∞ ∞ p ∞ 21 1 X X X k(n) 1 2 2 2 2 2 ≤c k(n) a ≤ c + i i a i i n2 n2 n=1 n=1 i=1 i=n X 3 ∞ p n X 1 k(n)n 8 2 2 2 + + ≤c i ai n2 √ 4 n=1 i=[ n]+1 √ 1 n ∞ p 1 X k(n) (nω( 4 n)) 2 X 2 2 2 i ≤ + a √ i n2 (nω( 4 n)) 12 √ 4 n=1 i=[ n]+1 ∞ ∞ ∞ 21 X X 12 1 k(n) X 2 2 √ ≤c+c i ω(i) ≤ a i n3 i=1 nω( 4 n) n=1 n=1 ≤c+c ∞ X i2 a2i ω(i) i=1 ∞ X k(n) 12 i=1 n3 = 1 Nk(i)+1 X Nj+1 ∞ ∞ X k(i) X 1 2 X 2 2 + j =c+c i ai ω(i) ≤ n3 n3 n=i i=1 j=k(i)+1 n=Nj +1 X ∞ ∞ X k(i) 1 2 2 ≤c+c i ai ω(i) + (k(i) + 1) + 2 3 i n i=1 n=Nk(i)+1 +1 + ∞ X j=k(i)+1 1 Nj2 12 ∞ X 12 ≤c+c < ∞, i2 ai2 ω(i)k(i) i=1 whence by Levy’s theorem ∞ X σn (x) − σn−1 (x) < ∞ a.e. on (0, 1). n=2 Remark 1. In Theorem 2, the Weyl multipliers defined by conditions (1) and (8) can be assumed to be exact on the set of sequences {Nk } with condition (7) in the sense that if condition (1) is violated, then one can construct a sequence {Nk } for which condition (7) is fulfilled and also there exists a trigonometric series ∞ X bn cos nx, n=1 which is nonsummable by the |c, α| method for almost all x ∈ [0, 2π] (for every fixed α > 0) though ∞ X n=1 b2n ω(n) k(n) < ∞. ON THE ABSOLUTE SUMMABILITY OF SERIES 89 Indeed, let the condition ∞ X 1 =∞ n ω(n) n=1 be fulfilled for the sequence {ω(n)}. We construct an increasing sequence of natural numbers {Nk } in such a way that the condition k=O Nk X 1 β 1 , 0<β≤ , n ω(n) 2 n=1 be fulfilled and the sequence Nkk be increasing. Clearly, condition (7) is fulfilled (see [3], Remark 2). Take k X 1 sk = , k = 1, 2, . . . , n ω(n) n=1 and 1 cm = √ 1 m ω(m) (sm )β+ 2 Then for arbitrary εm = ±1 we have ∞ X (εm cm )2 ω(m)k(m) = m=1 = ∞ X Nk+1 X k=0 m=Nk +1 ≤c ∞ X ∞ X k(m) = 2β+1 m ω(m)(s m) m=1 ∞ X k(m) ≤c 2β+1 m ω(m)(sm ) Nk+1 X k=0 m=Nk +1 Nk+1 X m = 1, 2, . . . . k=0 m=Nk +1 (sNk )β ≤ m ω(m)(sm )2β+1 ∞ X 1 1 ≤ c < ∞. 1+β m ω(m)(sm )1+β m ω(m)(s m) m=1 On the other hand, ∞ X n=0 n+1 2X m=2n +1 c2m 21 ≥ ≥ ∞ X n=0 ∞ X n+1 2X 1 2 (s )1+2β m(ω(m)) m m=2n +1 12 ≥ 1 1 = ∞. n 2 n=1 ω(2 )(s2n ) 21 +β Therefore by Billard’s theorem [1], for almost all choices of εk = ±1 the series ∞ X εm cm cos mx m=1 90 G. NADIBAIDZE is |c, α|-nonsummable (α > 0) at almost every point x ∈ [0, 2π] though ∞ X n=1 where bn = εn cn . b2n ω(n) k(n) < ∞, Remark 2. The above theorems remain also valid for |c, α| methods with cα > 21 . References 1. P. L. Ul’yanov, Solved and unsolved problems of the theory of trigonometric and orthogonal series. (Russian) Uspekhi Mat. Nauk 19(1964), No. 1, 3–69. 2. V. F. Gaposhkin, Series of block-orthogonal and block-independent systems. (Russian) Izv. Vyssh. Uchebn. Zaved. Mat. 1990, No. 5, 12–18; English translation: Soviet Math. (Iz. VUZ), 34(1990), No. 5, 13–20. 3. G. G. Nadibaidze, On some problems connected with series with respect to ∆k -ONS. Bull. Acad. Sci. Georgia 144(1991), No. 2, 233–236. (Received 15.05.1997) Author’s address: J. Javakhishvili Tbilisi State University Faculty of Mechanics and Mathematics 2, University St., Tbilisi 380043 Georgia