- If I use the normal optimization 40 times and then DCS then the solution of the ACG is not good. It does work if I send a 10 iteration in one shot
If I manually do 10 and then DCS it is still good. Something happens if I optimize normally for too long.
Need to test :
- Does it happens consistently (test with new graphs)
- What conditions prompt this behavior
Results :
####################
It doesn't work for the 6 nodes
Need to test:
- Why ?
####################
Also the inal result looks nice but some corner that should not be fused get fused this way. So it all depend on what is the end goal.