Hacker News new | ask | show | jobs
by Epa095 899 days ago
A bit disappointed that there were no reimplementation and real benchmarking happening.
2 comments

Next article apparently. I was very much looking forward to that, too.

> In our next article, we'll look to compile crosswords using a MILP model.

and

> In the next article, we attempt to formulate and solve a MILP to compile crossword puzzles. This is still a difficult problem, though given the improvement in computer and solver speed, there is some basis for optimism that the task is now possible.

Exactly. In particular here I suspect the perf is dramatically worse than 1800s/20e9=90ns. Constant factors are real-world annoyance :)