|
Members in Shoutbox
None.
Shoutbox Search
Shoutbox Commands
/w [name] > Whisper
/r > Reply to last whisper /me > Marks as action Shoutbox Information
Moderators may delete any and all shouts at will.
|
Global Shoutbox
Please log in to shout.
[2017-3-30. : 7:49 pm] Zoan -- Also we are probably mostly just wary because of the shitstorm from SCII[2017-3-30. : 7:46 pm] Zoan -- ButchButch shouted: the new UI is dope, they jus needa fix a few things like demon said. people are so fuckin critical, im so happy sc is getting any attention at all. We even have an opportunity to suggest shit, 1.18 is not terrible imo It needs to look different than the game start lobbies too[2017-3-30. : 7:46 pm]  TiKels -- so if you were able to calculate it in an amount of time represented by a polynomial distribution with 100% accuracy it would be considered a P solved problem?[2017-3-30. : 7:45 pm] Butch -- the new UI is dope, they jus needa fix a few things like demon said. people are so fuckin critical, im so happy sc is getting any attention at all. We even have an opportunity to suggest shit, 1.18 is not terrible imo[2017-3-30. : 7:41 pm] TiKels -- Do you just determine that the number of cities (I assume that's the input?) doesn't leave a solution where the time is listed something like 2^n (where n is number of cities)?[2017-3-30. : 7:40 pm] TiKels -- I get the rough idea, but how do you determine whether something is calculated under "Polynomial time"?[2017-3-30. : 7:39 pm] TiKels -- I feel like you're talking on a level way above my current knowledge haha[2017-3-30. : 7:38 pm] Dem0n -- text box doesn't stay focused so you have to click on it every time you wanna type if you switch windows[2017-3-30. : 7:38 pm] Dem0n -- escape no longer clears your message and instead quits multiplayer, alt+c is no longer "create" and is instead "cancel" (ie quit multiplayer)[2017-3-30. : 7:35 pm] outlawpoet -- that's why we make approximation algorithms that get close enough[2017-3-30. : 7:34 pm] outlawpoet -- AI isn't some panacea, it still relies on CPU time for calculations. An arbitrarily large TSP isn't going to be solvable in polynomial time regardless of the method[2017-3-30. : 7:32 pm] outlawpoet -- TiKelsTiKels shouted: But you would never know if you always got correct answerst with AI. It could give you 10000 optimal answers to 10000 problems, but you'll never know if there is an example of the problem it cant solve in the TSP it's not like an AI is going to have access to different information or better algorithms. What it might do better than humans is choose which algorithm is going to result in the best approximation for any given subset of the data[2017-3-30. : 7:31 pm]  lil-Inferno -- Tbh fam, w/e new UI, people should be mad AF that they removed features and fucced up others LMAO[2017-3-30. : 7:26 pm]  Ahli -- AhliAhli shouted: to detect a structure being under construction, we center a location on that structure type, right? then it centers on it and does not include a certain unit anymore that would be included in the center of the map, right? can we focus on the real problems... that was the right approach to fastest way to detect supply depot construction, right?  [2017-3-30. : 7:23 pm] outlawpoet -- TiKelsTiKels shouted: But you would never know if you always got correct answerst with AI. It could give you 10000 optimal answers to 10000 problems, but you'll never know if there is an example of the problem it cant solve ya, NP-hard problems often take as long to verify as they do to compute[2017-3-30. : 7:22 pm] outlawpoet -- there's a formulation that is NP-complete - "given a length L, decide whether the graph has any tour shorter than L"[2017-3-30. : 7:21 pm] TiKels -- But you would never know if you always got correct answerst with AI. It could give you 10000 optimal answers to 10000 problems, but you'll never know if there is an example of the problem it cant solve[2017-3-30. : 7:21 pm] outlawpoet -- AhliAhli shouted: I would replace the word correct with optimal... correct would be any trip that manages to reach all customers. But you want the optimal plan and not just any... depends on the formulation of the TSP. typically framed as shortest possible distance / least expensive route[2017-3-30. : 7:20 pm] Ahli -- I would replace the word correct with optimal... correct would be any trip that manages to reach all customers. But you want the optimal plan and not just any...[2017-3-30. : 7:19 pm] TiKels -- Outlaw poet wanna take this to a discord or something? You seem pretty knowledgeable[2017-3-30. : 7:19 pm] outlawpoet -- ya, it just takes a long ass time. NP stands for non-deterministic polynomial time[2017-3-30. : 7:16 pm] Dem0n -- TiKelsTiKels shouted: outlawpoet outlawpoet My question is if you designed an algorithm that was able to efficiently solve the traveling salesman problem but were not able to demonstrate that it always got the correct solution (without brute forcing), while being (for lack of a better word) exponentially faster... is that not a solution to an NP problem? Sorry if my understanding of NP is a little elementary If you can't prove 100% correctness, it's not a solution.[2017-3-30. : 7:15 pm] TiKels -- outlawpoetoutlawpoet shouted: I'm not sure I follow your question, TiKels. If you get the correct solution every time, either your heuristic/algorithm is extremely good, extremely lucky, or your search space is tiny outlawpoetoutlawpoet shouted: outlawpoet or you spent a fuckton of CPU time calculating it LOL. solutions are inevitable, mostly it's about time to solve My question is if you designed an algorithm that was able to efficiently solve the traveling salesman problem but were not able to demonstrate that it always got the correct solution (without brute forcing), while being (for lack of a better word) exponentially faster... is that not a solution to an NP problem? Sorry if my understanding of NP is a little elementary[2017-3-30. : 7:14 pm] Dem0n -- l)ark_ssj9kevinl)ark_ssj9kevin shouted: lol there's a fps cap gotta make sure it doesn't get too out of hand. some computers may not be able to handle it![2017-3-30. : 7:14 pm] outlawpoet -- answers to* calculations. I don't care how good your genetic algorithm is, you're not gonna solve a million city TSP without a fuckton of CPU time[2017-3-30. : 7:12 pm] outlawpoet -- TiKelsTiKels shouted: How do AI techniques fit into the P vs NP paradigm? ok I think I understand your angle better now. AI doesn't magically produce answers calculations that can't be done in polynomial time, but it could come up with good heuristics for doing so[2017-3-30. : 6:59 pm] Ahli -- to detect a structure being under construction, we center a location on that structure type, right? then it centers on it and does not include a certain unit anymore that would be included in the center of the map, right?[2017-3-30. : 6:58 pm] outlawpoet -- outlawpoetoutlawpoet shouted: I'm not sure I follow your question, TiKels. If you get the correct solution every time, either your heuristic/algorithm is extremely good, extremely lucky, or your search space is tiny or you spent a fuckton of CPU time calculating it LOL. solutions are inevitable, mostly it's about time to solve[2017-3-30. : 6:58 pm] Azrael -- It's about how much the UI sucks, and the developer basically says "we didnt like the old UI so we made it the way we like it, you better get used to it lol".[2017-3-30. : 6:56 pm] Azrael -- I'm keeping it out of the actual thread, in order to preserve the integrity of the thread, so it can stay on topic (e.g. this UI needs to change), instead of being derailed by silly bickering and how some people choose to present their views.[2017-3-30. : 6:53 pm] Ahli -- https://www.reddit.com/r/starcraft/comments/62fykj/analysis_of_118_exe/ 1.18 contains anticheat and uses opengl 3.3[2017-3-30. : 6:52 pm] Azrael -- I figured you deserved a direct reply, since you had the nerve to name me specifically. Stop acting like MrMeeseeks.[2017-3-30. : 6:51 pm] Azrael -- Trying to be nice to the developers after they effectively said "go away, we're not changing it" is not helpful to anyone, except the developers, who would love for people to stop acting upset so they can go about their business and not worry about it.[2017-3-30. : 6:50 pm] Azrael -- It would seem I'm accurately vocalizing the feelings of many other community members.[2017-3-30. : 6:49 pm] Azrael -- I'm amazed you'd specifically try to state "AZRAEL'S POSTS ARE UNHELPFUL" after seeing how much support they've received.[2017-3-30. : 6:47 pm] Azrael -- There's a reason my posts are heavily upvoted and yours are getting downvoted.[2017-3-30. : 6:47 pm] Azrael -- No one is speaking on your behalf. If you don't like how I present my opinions there, fine. I'm not speaking for you though, so you don't need to try to downplay my posts.[2017-3-30. : 6:47 pm] Azrael -- You're starting community in-fighting for no reason, for the sake of "being nice" to a developer who already said "Nah, it's never coming back".[2017-3-30. : 6:46 pm] Azrael -- Calling out other users by name is immature and obviously not helpful to the community having a united effort.[2017-3-30. : 6:46 pm] outlawpoet -- I'm not sure I follow your question, TiKels. If you get the correct solution every time, either your heuristic/algorithm is extremely good, extremely lucky, or your search space is tiny[2017-3-30. : 6:44 pm] TiKels -- But if an algorithm gets you to the exact solution every time, even though you can't prove it does it always... is it not still a solution?[2017-3-30. : 6:41 pm] TiKels -- Like someone I know used genetic algorithms to get a good solution to the traveling salesman problem, which was able to quickly generate the answer. But you can't really verify it 100% without checking the entire solution space right?[2017-3-30. : 6:13 pm] outlawpoet -- sigsaucysigsaucy shouted: was there anything else i was supposed to get from that screen? everyone can play the campaign right now right? the post was about a typo[2017-3-30. : 6:13 pm] sigsaucy -- was there anything else i was supposed to get from that screen? everyone can play the campaign right now right? |
Members Online: Roy
Roy