Optimization project on the VLSI problem.
- I used a powerful global constraint called DIFFN to ensure non-overlap
- I changed a bit the data structures, Instead of a single [Cx2] array I opted for two single dimension arrays [C][C].
- I found the upper/lower bound limits to be inconsistent, so I changed to different ones.
- I developed a small visualizing program, its useful and It helped understanding some critical behaviour. Its not fully automated but I changed the output from minizinc, so to be easly copied and pasted in the python program.
- I did some statistical comparison between baseline and diffn that we can include in the project report. I included some graphics. Results are as expected, the global constraint is a big improovement.
- I made this repo, if you like github we can work here
- symmetry breaking
file | content |
---|---|
baseline | simplest baseline model |
diffn | global constraint model improovement |
data | contains test data |
display_data | python tool for visualizing results |
optimization_results | contains optimization statistics |