Fast algorithms for Calderón-Zygmund singular integral operators (Q1917556)

The paper is devoted to the construction of a fast algorithm for the numerical evaluation of the Calderón-Zygmund singular integral \[ Tf= \int K(x,y)f(y)dy, \] where \(x=(x_1,\dots,x_n)\), \(y=(y_1,\dots,y_n)\), the operator \(T:L^2(\mathbb{R}^n)\to L^2(\mathbb{R}^n)\) is continuous, and the kernel \(K(x,y)\) satisfies several estimates. The author considers an increasing sequence \(\{V_j\}_{j\in \mathbb{Z}}\) of closed subspaces of \(L^2(\mathbb{R}^n)\) such that \(\bigcap^\infty_{-\infty} V_j=\{0\}\), \(\bigcup^\infty_{-\infty} V_j\) is dense in \(L^2(\mathbb{R}^n)\), \(f(x)\in V_j\Leftrightarrow f(2x)\in V_{j+1}\), \(W_j\) be the orthogonal complement to \(V_j\) inside \(V_{j+1}\). Let \(\phi_{j,k}(\psi_{j,k})k\in \mathbb{Z}^n\) be an orthogonal basis for \(V_j(W_j)\). Let \(P_j:L^2(\mathbb{R}^n)\to V_j\) be the orthogonal projector and \(Q_j=P_{j+1}-P_j\). The author's nonstandard decomposition of \(T\) is \[ T=\sum^\infty_{-\infty} P_jTQ_j+\sum^\infty_{-\infty} Q_j TP_j+\sum^\infty_{-\infty} Q_j TQ_j, \] where the matrix of \(\sum^\infty_{-\infty} P_j TQ_j\) is lower diagonal, the matrix of \(\sum^\infty_{-\infty} Q_j TP_j\) is upper diagonal, the matrix of \(\sum^\infty_{-\infty} Q_j TQ_j\) is diagonal. The operator \(T_m\) is defined as \((T_m\phi_{j,k},\psi_{j,l})=(T\phi_{j,k},\psi_{j,l})\), whenever \(1\leq|k-l|\leq m\) and \((T_m\phi_{j,k},\psi_{j,l})=0\) if \(|k-l|> m\). The author proves that \[ |T-T_m|\asymp cm^{-\gamma}\sqrt{\log m}. \]