Accurate Sparse-Matrix Semisymbolic Analysis of Large-Scale RF Circuits

A sophisticated pivoting strategy for the reduction algorithm transforming the general eigenvalue problem to the standard one is presented for both full- and sparse-matrix techniques. The algorithm increases the precision of the semisymbolic analyses especially for the large-scale RF circuits. Furthermore, the accuracy is increased using longer numerical data. First, the long double precision is used. Second, the application of a suitable multiple-precision arithmetic library is suggested. The algorithm is demonstrated in a nonstandard way by estimating the oscillation frequency of a distributed RF oscillator, and by estimating the bandwidth of a distributed microwave amplifier