Xuan Lin

Hi,Thanks for the reply! I tried to run either:psrecord "gurobi_cl point_mass_04_lnf.mps" --log activity.txt --plot memory_usage.png --interval 1or:psrecord "python3 run_model.py" --log activity.tx...

Hi Riley, I am unsure how many constraints lead to a highly constrained problem, but my model is relatively small (has like 20 binary variables, hence can solve fast) and has more than 700 constrai...

Hi,Thanks! I have another baseline model that is equivalent, but with looser convex relaxations and about 2x more constraints. The solving speed is usually slower. For this case, I got memory usage...

Hi Riley,Yes, running gurobi_cl point_mass_04_lnf.mps works ok and generate the logfile correctly.I figured that if I run:psrecord "gurobi_cl point_mass_04_lnf.mps" --include-children --log activit...

Hi Jaromił, Thank you for the explanation! I have a follow-up question regarding this issue. Are the techniques disabled by activating LazyConstraints inherently incompatible with lazy constraints,...

Hi Riley, Thanks for the answer! I personally haven't used any GPU-based QP solver. I have heard about some research to accelerate QP solvers using GPUs, for example, this one. I think ADMM is know...

Hi Jaromił, Thanks for the answer! I actually didn't know that QPs can be solved with simplex. May I ask a minor question: does the column "unexplored nodes" always goes to 0 as the algorithm termi...

Hi Jaromił,   Thanks for the answer! The column It/Node shows the average number of Simplex iterations performed per node, i.e., the average number of Simplex iteration to solve a problem A to opt...

Hi Riley, Thanks for the answer! I think using simplex instead of concurrent solvers helps to reduce memory usage. I also managed to reduce the model size by improving the formulation. I have a mor...

Hi Riley, Yes, what you said is correct, that is why the code I put here is a simplified model for debugging purposes. I have the complete model (~400 lines) that sets lb=ub to various nonzero valu...