This paper studies an extension of the classical convolution quadrature, a well-known numerical method for calculation of convolution integrals. In contrast to the existing counterpart, which uses the linear multistep formula or Runge-Kutta method, we employ the block generalized Adams method to discretize the underlying initial value problem. Similar to the convolution quadrature method based on the linear multistep formula, the proposed method can also be implemented on an equispaced grid. In addition, the proposed high-order method is as stable as the convolution quadrature based on the Runge-Kutta method, which indicates that it can accurately solve a wide range of problems without becoming unstable. We provide a detailed convergence analysis for the proposed convolution quadrature method and numerically illustrate our theoretical findings for convolution integrals with smooth and weakly singular kernels.

翻译：暂无翻译