There he is in all his glory.
Did you know about his brother OrtSol Lore?
Ok, you’ve worked for it, here’s all the third-party benchmarks:
We get this question a lot, and it’s as stupid as the people asking it.
First off, I’m goated. Do I sound like I can be arsed to write case studies that no-one reads?
Second, case studies are for all the n00bz I’m pwning so hard and who lack bragging rights.
If I am destroying n00bz all over the place on problems they all had access to (oftentimes even submitted to the test sets themselves), since before I even wrote a single line of code, what do you think happens on “real” problems?
Nuclear damage is what happens.
Think about that.
Just for jokes, and also to hilariously prove the point above^^, try sorting the MINLPLIB problems by size, and try some of the larger ones with different solvers.
Other solvers’ critical failures might surprise you. In the meantime, here’s our damage:
Problem | Type | Size | Time (seconds) | Notes | Options to reproduce |
junkturn | QCQP | 200,008x140,000 | 60 | Largest problem in MINLPLIB | LOCAL_SEARCH=true, USE_APPROX_EQUALITY_CONSTRAINTS=true |
optcdeg2 | QCQP | 150,002x100,000 | 14 | Second largest problem in MINLPLIB | LOCAL_SEARCH=true |
watercontamination0303 | MBQP | 107,222x108,217 | 7 | Third largest problem in MINLPLIB | NUM_CORES=8 |
watercontamination0202 | MBQP | 106,711x107,209 | 2.6 | Fourth largest problem in MINLPLIB | |
dtoc5 | QCQP | 99,999x49,999 | 10.5 | Fifth largest problem in MINLPLIB | LOCAL_SEARCH=true |
densitymod | MBNLP | 23,529x550 | 5.2 | Largest MINLP in MINLPLIB | LOCAL_SEARCH=true |
transswitch2736spp | MINLP | 23,410x46,821 | 88 | Second largest MINLP in MINLPLIB. Was open until we solved it. | FIRST_FEASIBLE_SOLUTION = true |
transswitch2736spr | MINLP | 23,410x32,845 | 109 | Third largest MINLP in MINLPLIB. Was open until we solved it. | FIRST_FEASIBLE_SOLUTION = true |
powerflow2736spp | NLP | 19,916x46,821 | 22 | Largest NLP in MINLPLIB. | LOCAL_SEARCH = true |