Computer-Implemented Distributed Iteratively Reweighted Least Squares System And Method

摘要

Systems and methods are provided for performing an estimation using an iteratively reweighted least squares technique on a scenario defined by a design matrix, a response variable vector, and a parameter to be estimated. A system and method can be configured to include a root data processor, where a least squares estimate is calculated by the root data processor during each of a plurality of processing iterations. A plurality of node data processors can also be configured to: update values of a weight matrix according to a current state of the least squares estimate, determine a first intermediate value based on the design matrix and the weight matrix, determine a second intermediate value based on the design and weight matrices and the response variable vector. The root data processor calculates an updated least squares estimate based on the intermediate values.