Improved Parallel Composition of Labelled Petri Nets

Abstract

Abstract

STG-Based Resynthesis for Balsa Circuits

“…[15]. There, a construction is defined that transforms an STG into an equivalent one with an injective labelling.…”

“…N 1 can fire an output transition under M 1 , while N 2 does not enable any transition with the same label in A (as above) under M 2 , i.e. the second STG is not able to receive the message sent by the first STG [14,15].…”

“…As argued in the introduction, this composition is free of computation interference; hence, we can apply the methods from [15]: first, we can enforce injectivity; we do not define this here, but it is essentially the inverse of path-splitting defined below. Second, we can apply the optimized parallel composition, which produces fewer places.…”

“…Second, the parallel composition for the cluster might have computation interference; thus, one cannot apply the optimized parallel composition of [15] that produces fewer places and makes transition contractions easier.…”

“…With our translation from a Breeze netlist to Breeze-STGs, we can apply the ideas of [15]: we use an optimized parallel composition, and we can enforce injectivity for the STGs beforehand to avoid complex structures in the initial Balsa-STG N. Additionally, we found that one can achieve better reduction when one afterwards relaxes injectivity in N with shared-path splitting. Since we can prove that N after lambdarization is output-determinate, we can apply all and even more reduction operations than allowed for STG-decomposition (e.g.…”