No label defined (Q4937295)

In the paper by \textit{R. Estrada} [Bol. Soc. Mat. Mex., III. Ser. 3, No.1, 117-133 (1997; Zbl 0899.41019)] the classical Euler-Maclaurin formula \[ \sum^M_{k=N} \varphi (k)= \int^M_N \varphi (x) dx+ {\varphi(N) +\varphi(M) \over 2} +\sum^q_{j=2} c_j( \varphi^{(j-1)} (M)-\varphi^{(j-1)} (N)) +Rq \] was generalized to distributions. Namely, let \(g\) be a distribution with supp\((g) \subset [0,1]\) and \(\varphi\in C^q(N,M)\). Then \[ \sum^{M-1}_{k=N} \langle g(x-k),\varphi (x)\rangle= \int^M_N \varphi (x) dx+ \sum^{q-1}_{j=0} (-1)^j\rho_j( \varphi^{(j)} (M)-\varphi^{(j)} (N))+Rq, \] where \(\rho_j,\;j=0,\dots,q-1\) doesn't depend on \(\varphi\). In the present paper the similar technique is used to get an expression for the sum \[ \sum^{M-1}_{k=N} \sum_{j=P}^{Q-1}\langle g(x-k,y-j),\varphi (x,y)\rangle \] through the relevant integrals of the function \(\varphi(x,y)\). Here \(g(x,y)\) is a distribution in two variables with supp\((g) \subset [0,1]\times [0,1]\). Setting \(g(x,y)=\delta(x-\alpha)\delta(y-\beta)\) and considering the limiting case \(\alpha\to 0+,\;\beta\to 0+\) the authors arrive to a conventional summation formula of Euler-MacLaurin type in two variables which is a bit too long to be presented in our review.