11–20 of 51 posts
I remember reading this and thinking "That would be so much simpler to do with a breadth first search". Sure 'genetic' sounds cool, but when the search space is as limited as it is in a build order optimization like this, you should use the right tools for the task.
How would a breadth first search work better? The search space is huge.
This "news" is from 2010.
I remember reading this and thinking "That would be so much simpler to do with a breadth first search". Sure 'genetic' sounds cool, but when the search space is as limited as it is in a build order optimization like this, you should use the right tools for the task.
They also didn't use crossover which I thought was the main advantage of genetic algorithms. How would a breadth first search work better? The search space is huge.
I remember reading this and thinking "That would be so much simpler to do with a breadth first search". Sure 'genetic' sounds cool, but when the search space is as limited as it is in a build order optimization like this, you should use the right tools for the task.
I don't think breadth first search would have been the right approach. The search space is actually quite huge for this. Also, the algorithm wasn't built to create a build that wins the fitness was simply how fast they could produce the army the user inputs. What stands out is if you had asked anyone even a professional player to build this exact army composition (7 roaches) they would almost certainly not have pulle…
Earlier quoted context omitted.
2^10=1024
This assumes e.g. that I can extend my ring finger without also extending the adjacent middle finger. ;)
When I first saw this thread on teamliquid, I thought "Oh man, proggit and HN will love this" so I wrote it up. And sure enough, it was near the top of both, for a day. Next came the random news articles (like this one). And here several years later is an HN repost of a blog article that only exists because HN and proggit upvoted the original blog post. You guys created the article that's being reposted :)
[1] https://news.ycombinator.com/item?id=1856390
From the last time around, the most common criticisms were:
1. Without any form of crossover, it's just hill-climbing/gradient-descent, it's not really a "genetic algorithm".
2. BFS would work better (maybe. IIRC people tried BFS and there's a bunch of subtle things that make the BFS search space huge (e.g. putting workers on and off gas at certain times). Of course since we have a goal state, something like A* might win the day).
Earlier quoted context omitted.
2^10=1024
This assumes e.g. that I can extend my ring finger without also extending the adjacent middle finger. ;)