Zerg Build Order optimizer. - Page 14
Forum Index > SC2 General |
DarKFoRcE
Germany1215 Posts
| ||
Almania
145 Posts
Even some very simple logic could cut out quite a few million of those BOs. For instance many of them would be make 1-2 hatcheries right away, which simple logic would tell you is not cost efficient. I was only allowing one of each building to be built . But yes, if you're only testing Zerglings, it's pointless testing if a Roach Warren would speed the build up. From the original post (see below) I got the impression the goal was to evaluate every single build order of Zerg ever, and then have a human skim over it and find the ones that really stand out. Which is absurd , I was merely trying to demonstrate that. Wouldn't mapping every possible build up to say 25min of game and then building a DB be more efficient than what you're doing? I don't think theres that many builds specially when you disconsider the obvious silly builds. | ||
Subversion
South Africa3627 Posts
be e-famous! xD | ||
666pool
United States15 Posts
Anyway I did this in a VS computer Very Hard Terran and uploaded a replay for you guys. I rallied the first 7 roaches to their front and attacked, wiping out their marine army with ease. My apologies it's not an actual ladder game. http://replayfu.com/r/jbgnqp | ||
metaldragon
United States251 Posts
Looks like it might be fast enough to shut down a 1/1/1 | ||
Lightspeed
130 Posts
Is transferring workers to gas or to an expansion part of the genome or done when testing? Looking forward to the release. | ||
icezar
Germany240 Posts
Yes it is just a rush, but it is a very good one!!!!! It is unbelievable that this build was done by a program! Please release it a soon a possible or at least give us more build example. What i do is rally all Roach to enemy, after 7th Roach i pool 1 drone off gas and i have just enough to keep making Roaches This is so very hard to hold!!!!!!!! I think the only option is a run-by and then force a base race. | ||
Almania
145 Posts
On October 14 2010 19:43 Lightspeed wrote: Is transferring workers to gas or to an expansion part of the genome or done when testing? Current version doesn't put workers on/take workers off gas, places 3 per extractor. To be fair, most builds do that anyway. Adding commands for "Move 1 worker to gas", "Remove 1 from gas", "Maynard 1 worker to expansion", "Maynard 2 workers to expansion" etc would be a neat improvement though! I wonder if any surprises would come about from that. | ||
tetracycloide
295 Posts
On October 14 2010 13:59 Eknoid4 wrote: that totally wouldn't be the use of this at all. We already have replays that fulfill EXACTLY what you describe for every other game/sport. this only serves to say "what time can i get ___ out by" Replays simply don't work for that purpose in this sport or any other, there are to many other variables for them to do so. It's all done by computers. | ||
Thunderfist
Poland159 Posts
| ||
accaris
98 Posts
So people are clearly reading this thread. | ||
NeoLearner
Belgium1847 Posts
I was just checking out the 7 roach rush and muta builds and it seems the AI always does 10 lord. Is that because you program it with a "fastest" goal? Could we get this thing to solve the 9lord/10lord problem once and for all? If it works, you should team up with an AI programmer. Could make for a cool AI. Rock solid counters. To the people saying: "Well, than you have 7 roaches and he makes marauders and you have no economy and you die". I don't think you quite understood the goal of the program... | ||
icezar
Germany240 Posts
In the 7 Roach build i do like this: go until: 2:10 M:224 G:0 L:3 S:10.0/18 -- BuildRoachWarren but immidiatly after i do Drone OV 5x Roach take a drone off gas 2x Roach ... Keep making roach (and 1 OV if needed) and build a queen when i have 150 spare minerals This way the 7 Roach are at the same time but you have more money. | ||
Marou
Germany1371 Posts
If you release it as a program and not web app please make it crossplatforms (java?) so us, mac users, can enjoy it Imo the web app is just the best : runable from every system and even with your smartphone. And i'm not saying that because i'm a web dev' :D | ||
Baguette
Canada19 Posts
On October 13 2010 13:30 Chill wrote: lol i like how you pretend this will singlehandedly destroy sc2 as an esport or something. It does have the ability to change the early early game quite a lot. Although, after the 15 minute mark, the possibilities grow so much (must faster than say, in Chess) that you would need many supercomputers or distributed computing to calculate any given build, since the permutations would be insane. Unfortunately this is basically an NP hard problem and I like that it is solved using GP. Very cool. | ||
puzzl
United States263 Posts
On October 14 2010 01:43 Lomilar wrote: I'm confused about the extractor trick. Pretty obvious, really: you are using an evolutionary algorithm, which by definition is not optimized except in extremely rare instances. Just like human men have nipples, your evolution will contain confusing, often useless hitches along the way that keep it imperfect. At some point your program may mutate away the extractor trick and realize it is 4 minerals better off from doing so, but with such a small change it is unlikely to do so without many more iterations. I must admit confusion as to why you chose to conquer this in a (notoriously slow and slightly imperfect) evolutionary algorithm rather than going with a fast heuristic approach or a perfect, more robust graph-based approach. With the graph based approach and decent pruning you should be able to "solve" zerg up till at least 10 minutes without too much in the way of difficulty, and in doing so the program would only need to run a single time and could output any and all pertinent scenarios along the way, not just a single end goal. | ||
sylverfyre
United States8298 Posts
| ||
Cyber_Cheese
Australia3615 Posts
infact where the spawning pool starts there is after the overlord pops, and the 13th drone is 24s after the pool starts it might be just as fast to overpool but the 13 pool can do the 7rr with more economy | ||
Cade
Canada1420 Posts
On October 15 2010 07:00 Cyber_Cheese wrote: theres a full 25 seconds in the 7rr build where the pool is done where the roachwarren hasnt started, so its probably better to be doing ~12/13 pool Get us a replay of you get 7 roaches out faster then. | ||
Cyber_Cheese
Australia3615 Posts
On October 15 2010 07:01 Cade wrote: Get us a replay of you get 7 roaches out faster then. its not about having a replay of faster roaches, in fact, it has nothing to do with "faster roaches" im saying that it can be done as fast with better eco edited my post with better reasoning just because its the fastest way doesnt mean it couldnt still be optimized | ||
| ||