Reduced-order modeling of large-scale systems by chained aggregation

An approach to model reduction using "chained aggregation" is described. Chained aggregation yields a special system representation, the Generalized Hessenberg Representation. This representation suggests candidate reduced order systems. Manipulations of these candidate systems are presented to obtain reduced-order models. Emphasis is placed on the use of the "restricted QL algorithm" to obtain low-order equivalents.