Tuning-Free Benchmark Considerations
Thu, 12/08/2016 - 10:41am
For the benchmarks in this category, obviously the type of benchrmark (affine / nonlinear; hybrid / continuous) will affect which tools can run it. Should we be targeting a particular class here? Or maybe these tools should instead use the benchmarks from the other categories, just try to come up with suitable paramters automatically?
Fri, 12/16/2016 - 12:00pm
#1
Benchmarks need to fit the tools
A start would be that everybody who wants to participate in this category proposes a benchmark that fits his tool. Note that this year this category is not about "blind" experiments. It's simply required that your tool can handle the problem without any tweaking of the settings.