ABSTRACT
With the fast development of design of integrated circuit, large-scale control systems and other various applications, the mathematical models in the form of differential-algebraic equations (DAEs) become more complicated. People use computers to generate such models automatically by unified multi-domain langrage e.g. Modelica [11]. To solve these large DAE models, one of the key steps is the differential reduction of the DAE to ODE which can be handled by traditional numerical approaches. The most efficient reduction technology is “the structural analysis method” proposed by Pryce [1]. It is a fixed-point iteration algorithm aiming to get equation differential offset and variable differential offsets. A rough cost estimation for the fixed point algorithm can be found in [3] and it was extended to the fixed point algorithm with parameters for the systems with upper-triangular block structure.
