# Convergence problems of orthogonal series by GyoÌrgy Alexits

By GyoÌrgy Alexits

Similar mathematics_1 books

Identification, Equivalent Models, and Computer Algebra. Statistical Modeling and Decision Science

This paintings offers stipulations for the id of significant suitable periods of versions. Checking those stipulations calls for advanced algebraic computations, that could now be played via laptop. This publication presents suitable algorithms and courses. It incorporates a diskette containing this system

The inverse problem of scattering theory

Half 1. The boundary-value challenge with no singularities --I. specific ideas of the process with no singularities --II. The spectrum and scattering matrix for the boundary-value challenge with no singularities --III. the elemental equation --IV. Parseval's equality --V. The inverse challenge --Part 2.

Additional resources for Convergence problems of orthogonal series

Example text

Z71 lC ι ^ = ΛΓ m V i 2 2 L cfc. Since here m is arbitrarily, this implies that 00 Σ ct^3M2. k=it l + Thus we have established that Σά converges and so our statement is proved. 1, as well as its corollary originate also from him. 4 was proved by KHINTCHINE and KOLMOGOROFF [1]. We shall see in the sequel that these theorems are valid even in a wider range. 7). 4 that the Rademacher series Zcnrn(x) can be the expansion in the functions of the completed system of Rademacher functions of an L-integrable function if and only if f(x) is not merely L-, but also L2-integrable.

CkPk(x)· k=n Now let M denote the maximum of σ(χ); then it follows from Schwarz's inequality (6) that 1 1 lç(x)o%x)ql(x)dxijç{x)pl(x)dx^ 4ü -1 1 -I ^ M \ç(x)o(x)ql{x)dx = M -1 and so a(xo)qn(xo) ^Z\ck\ k+n \Mxo)\ = (m+i)YMO(l) = 0(\). Thus we have established our intermediary theorem. Thus all the cases where a, ß are non-integer numbers are settled. In order to consider also integral values ay ß we use the fact (already mentioned) that for the Legendre polynomials (a = 0,ß = 0) the estimate p f °>(x) = 0(1) holds in (—1, 1).

2 , (proved in the following chapter), we may infer that almost everywhere 1 ^ lim -γ- ]£rk(x) = 0. Cantelli's theorem corresponds to the special case λη = η. The relation (22) is also equivalent to a theorem of BOREL, stating that the ratio of the number of the digits 1 to the number of the digits 0 among the first n digits in the dyadic representation of a number x between 0 and 1 tends to 1 for almost every x, when n —> o°. (*)] =0χ{λ) almost everywhere. This means that the arithmetic means of the sequence obtained from the sequence of digits by replacing the digits 0 by —1 tend to zero almost everywhere.