Parallel Independence Benchmarking

---

Terminology:

• Jointly epic pairs: For every pair of rules, take all partitions of L1 ⊎ L2 (i.e. all epimorphisms e: L1 ⊎ L2 ↠ G ), and let m1 = e ∘ inj1 and m2 = e ∘ inj2 when both transformation steps can be constructed.
• Fixed instance graph: A fixed instance graph G of significant size was constructed for each grammar, then for each pair of rules all possible pairs of matches m1 : L1 → G and m2 : L2 → G were found such that both transformation steps can be constructed.

---

Benchmarking results for Parallel Independence Algorithms:

• All analysis were executed using a special version of verigraph system available here.


• The instances for benchmarks can be generated running the scripts:
  • "create-pre-inputs-elevator"
  • "create-pre-inputs-guideline1"
  • "create-pre-inputs-guideline2"
  • "create-pre-inputs-guideline3"
• The benchmarks can be executed running the script "run-benchs":
  • For non injective matches, run: "run-benchs <input_file> --all-matches".
  • For Injective Matches, run: "run-benchs <input_file>".
• Reports from analysis are available below and a simple report with the size of each instance of benchmark are available here.