I have some general questions about the barrier algoritm that Gurobi uses to solve a QP.
For which type of problems does the algorithm work the best? Large-scale systems or small-scale systems?
Do dense matrices work the best for the barrier algorithm or sparse matrices? What are the causes of that? Or works a combination better?
For example small, the fastest for small, dense problems and large, sparse problems.
Does the algorithm need a lot of iterations to converge or less iterations, but each iteration is more expensive, in general?
I found a Powerpoint about Gurobi and this was the only thing I found of the above subject, but I don't exactly know what they mean with:
Dozens of expensive iterations: Why many iterations and expensive at both time
Much denser matrices: Why denser matrices? I thought Gurobi could handle sparse matrices better?
Lots of opportunities to exploit parallelism: What is parallelism?
Please sign in to leave a comment.