An Efficient Algorithm for the Compression of Time-Dependent Data from Scientific Simulations

摘要

The efficient storage of simulation data obtained from the solution of time-dependent scientific models represents a significant issue in large-scale computing. One approach to storing this data is to use a mesh independent, or functional, representation of the simulation data. The implementation of such an approach requires an efficient means of computing a functional representation. In this paper, we present a new algorithm for computing this representation that guarantees that a user-specified pointwise error bound is satisfied by the functional approximation. This approach is based on a weighted least squares method to approximate the minimax solution for multi-dimensional simulation data. We present experimental results that demonstrate that this method can be used to efficiently compute the functional representation of large-scale, multi-dimensional data.