\(L^ 1\)-convergence of double cosine- and Walsh-Fourier series (Q1326658)

From MaRDI portal
scientific article
Language Label Description Also known as
English
\(L^ 1\)-convergence of double cosine- and Walsh-Fourier series
scientific article

    Statements

    \(L^ 1\)-convergence of double cosine- and Walsh-Fourier series (English)
    0 references
    0 references
    21 November 1996
    0 references
    Let \({\displaystyle {\sum^\infty_{j = 0} \sum^\infty_{k = 0}}} v_i v_j a_{jk} \cos jx \cos ky\) be the double Fourier series of an integrable function \(f(x,y)\), which is even in \(x\) and \(y\), on the two-dimensional torus \(T^2 = [-\pi, \pi) \times [-\pi, \pi)\), where \(v_j = {1\over 2}\) if \(j = 0\) and \(v_j = 1\) if \(j \geq 1\), and \[ a_{jk} = {1 \over \pi^2} \int^{\pi}_{-\pi} \int^\pi_{-\pi} f(x,y) \cos jx \cos ky dx dy \quad (j,k \geq 0). \] Let \(S_{mn} (f;x,y) = \sum^m_{j = 0} \sum^n_{k = 0} v_j v_k a_{jk} \cos jx \cos ky\) denote the rectangular partial sum of the above Fourier series. Let \(\Delta_{00} a_{jk} = a_{jk}\) and for \(p \geq 1\) let \(\Delta_{pq} a_{jk} = \Delta_{p-1,q} a_{jk} - \Delta_{p -1,q} a_{j+1,k}\), while for \(q \geq 1\) let \(\Delta_{pq} a_{jk} = \Delta_{p,q-1} a_{jk} - \Delta_{p,q-1} a_{j,k+1}\), where \(j\), \(k\), \(p\), \(q\) are non-negative integers. Let a modified rectangular sum of the Fourier series in question be defined by \[ u_{mn} (f;x,y) = S_{mn} (f;x,y) - a_{m+1, n+1} D_m (x) D_n(y), \] where \(D_m(x) = \sum^m_{j = 0} v_j \cos jx\), \((m,n \geq 0)\) denotes the Dirichlet kernel. The main result, concerning the \(L^1\)-convergence, proved in this paper is the following: Theorem. Let \(f \in L^1(T^2)\), \(f\) is even in \(x\) and \(y\), and \[ \lim_{m,n \to \infty} \log n \left|\sum^m_{j = 0} D_j(x) \Delta_{10} a_{jn} \right|= 0, \;\lim_{m,n \to \infty} \log m \left|\sum^n_{k = 0} D_k(y) \Delta_{01}a_{mk} \right|= 0. \] Then \(\lim_{m,n \to \infty} |u_{mn} (f) - f|= 0\) if and only if \[ \lim_{\lambda \downarrow 1} \limsup_{m,n \to \infty} \Biggl|\sum^m_{j = 0} \sum^{\lambda_n}_{k = n+1} {\lambda_n +1-k \over \lambda_n-n} \;D_j(x) D_k(y) \Delta_{11} a_{jk} \] \[ + \sum^{\lambda_m}_{j = m + 1} \sum^n_{k =0} {\lambda_m +1-j \over \lambda_m - m} D_j(x) D_k(y) \Delta_{11} a_{jk} \] \[ + \sum^{\lambda_m}_{j = m + 1} \sum^n_{k = n + 1} \;{\lambda_m + 1 - j \over \lambda_m -m} {\lambda_n + 1 - k \over \lambda_n -n} D_j(x) D_k(y) \Delta_{11} a_{jk} \Biggr|= 0. \] As an important consequence of this result, we have the following: Corollary: If \(f \in L^1(T^2)\), \(f\) is even in \(x\) and \(y\), and the conditions of the above theorem are satisfied, then \(\lim_{m,n \to \infty} |S_{mn} (f) - f|= 0\) if and only if \(\lim_{m,n \to \infty} a_{mn} \log m \log n = 0\). The above stated main theorem has also been extended to double Walsh-Fourier series.
    0 references
    0 references
    double Fourier series
    0 references
    rectangular partial sum
    0 references
    modified rectangular sum
    0 references
    Dirichlet kernel
    0 references
    \(L^ 1\)-convergence
    0 references
    double Walsh-Fourier series
    0 references

    Identifiers