We propose an optimized implementation of the Fourier-Motzkin (FM) algorithm for computing all nonnegative, minimal-support invariants of ordinary Petri nets. The FM algorithm is optimized by detecting the parallel place sets created during each of its iteration, and retaining just one member of each set while deleting the rest. The intermediate invariants obtained are then used to generate all minimal-support invariants using an optimized invariant generation process. Experimental results indicate that the proposed algorithm is at least 2.2X faster in execution and requires at least 1.8x less memory than other comparable algorithms
Date of Conference: 10-12 July 2006