Recursive Least Squares Algorithm for Linear Prediction Problems
Sanzheng Qiaq · SIAM Journal on Matrix Analysis and Applications · 1988
A new triangularization technique is presented for solving linear prediction problems. The algorithm is based on the exploitation of the special structure that problems of this type exhibit. The reduced triangular system and the error are computed recursively and the problem is solved when the optimal order has been found. The computational complexity of this algorithm is better than existing methods. In addition, good numerical properties are expected of the method.