Optimal Elma player
Moderator: Moporators
Re: Optimal Elma player
Likewise, why would your opinion matter so much? These guys are interested, you are not. Who's gonna forbid them to be interested?
- Grace
- 38mins club
- Posts: 4851
- Joined: 19 Nov 2005, 10:45
- Location: Deep in your Imagination, Twirling your Dreams and Weaving your thoughts.
Re: Optimal Elma player
the model wouldn't even need to generate replays, ramone, just times.
Cyberscore!
___________________________________________________
Targets: 6 Legendary, 23 WC, 20 Pro, 5 Good | AvgTT: 39:59:96
___________________________________________________
Targets: 6 Legendary, 23 WC, 20 Pro, 5 Good | AvgTT: 39:59:96
Re: Optimal Elma player
this is hypocritical. you say zweq's opinion shouldn't matter more than anyone else's(i agree, only the logic of the statement should matter). and then state your own personal opinion as if it should matter, you keep stating that you are not interested in seeing the times, which is a logically worthless argument as to why they shouldn't be found out.ramone wrote:why would zweqs opinion matter more than any elses? I am not at all interested in seeing a list of theoretical times. How could anyone guarantee drives/styles could be kept secret.
This is not intereresting! It sucks.
the only real reason this shouldn't be made is if it will in fact help people cheat. will this be a solo project of dimkadimon, or will he release all of his coding and work so that it may assist somebody in cheating if they so desired? also, is it considered cheating if new styles could be found out through this program? from what i got out of the previous posts, the program doesnt even have the ability to figure out new styles, more so, just to maximize the speed of a particular style already known(perhaps im wrong).
could something be done to give away the fact that somebody is cheating in a replay? nin was able to recognize cheating in a replay without much knowledge of what cheating may look like, maybe somebody couldn't even get away with cheating with this program! my last question though, is does this post even matter, will this optimal player thing ever happen?
God Bless America
Re: Optimal Elma player
In my opinion it's impossible because of a very special factor that takes place in elma: randomness. To prove my point, pick a lev and press gas all the way, without doing anything else (of course, pick a lev in which you'll survive more than 3 seconds by doing this). You'll notice 3-4 different situations may occur if you do it 10 times. On my computer, it does this a lot. I remember trying to make an autolev but couldn't do it because it kept being different all the time. That way, optimal paths cannot be found because they rely on luck.
Re: Optimal Elma player
zebra, commenting on the complexity of the issue (opinion in formation):
zebra wrote:I like the idea of perfect player and would like to see the recs of course but then again if it ruins the WRs and help ppl cheat it's very bad thing.
Re: Optimal Elma player
I also do not like that you are pasting here our private conversation.hexeditor wrote:zebra, commenting on the complexity of the issue (opinion in formation):zebra wrote:I like the idea of perfect player and would like to see the recs of course but then again if it ruins the WRs and help ppl cheat it's very bad thing.
Anyway, that's my opinion and I'm looking forward to what dimkadimon is gonna achieve. I'm not interested in doing that myself. I'd rather play the game as I like it, not trying to 'hack' it. I'm not even sure could I get anything ready so I'd better leave it to someone more experienced.
To get some times, you have to know the replay or the style first. So the question is just about what to publish or what will the program output.SmaXa wrote:the model wouldn't even need to generate replays, ramone, just times.
Please explain Was my explanation wrong?milagros wrote:For this problem you know the mathematical model (set of equations in elma) and there is no uncertainity at all (no need to explain those autolevs now..).
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
Re: Optimal Elma player
Optimal Elma players:
Now seriously. I think that Ramone is on the right way. We should not be able to even know optimal times. It would kill the fun.
Now seriously. I think that Ramone is on the right way. We should not be able to even know optimal times. It would kill the fun.
Lousku wrote:could you mayke shorter sig please mega annoying and also against rules )
Re: Optimal Elma player
Maybe you should learn to read between the lines gimp!gimp wrote:this is hypocritical. you say zweq's opinion shouldn't matter more than anyone else's(i agree, only the logic of the statement should matter). and then state your own personal opinion as if it should matter, you keep stating that you are not interested in seeing the times, which is a logically worthless argument as to why they shouldn't be found out.ramone wrote:why would zweqs opinion matter more than any elses? I am not at all interested in seeing a list of theoretical times. How could anyone guarantee drives/styles could be kept secret.
This is not intereresting! It sucks.
the only real reason this shouldn't be made is if it will in fact help people cheat. will this be a solo project of dimkadimon, or will he release all of his coding and work so that it may assist somebody in cheating if they so desired? also, is it considered cheating if new styles could be found out through this program? from what i got out of the previous posts, the program doesnt even have the ability to figure out new styles, more so, just to maximize the speed of a particular style already known(perhaps im wrong).
could something be done to give away the fact that somebody is cheating in a replay? nin was able to recognize cheating in a replay without much knowledge of what cheating may look like, maybe somebody couldn't even get away with cheating with this program! my last question though, is does this post even matter, will this optimal player thing ever happen?
Of course it would be great fun to know/see best possible times. And then that is done, its over. And I mean OVER! No need to keep playing. It is finding the limits that makes playing fun. I am not interested in seeing these times/recs because they would spoil the (Internal) Elma scene. And possibly spoil future cups and so on. And of course I state my opinion, as do everyone. Never said my opinion were law. I just reacted to hexeditors line: "i wrote to zweq for his opinion, as he's one of the best players around. here it is, folks:"
Of course zweq and everyone is allowed to have whatever opinion they like, as do I!
And about cheating. what you mention about cheating clearly states you know nothign about cheating and the past of the Elasto Mania community, so FUCK OFF!
For fucks sake, is this what you guys really want????
Maybe prog cant be done to find style, maybe this maybe that... there are too damn fucking many maybes. and god damn I get fucking mad. Why did Abula leave. come back and fucking show some balls and Delete this topic.
We (and by that I mean Me, and all guys out there who care about the Elasto Mania scene (not gimp, obviously)) dont want some new hexeditor coming to fuck up a perfectly good game by some hack and then few months later he dont care. Did you even finish all levels hexeditor, can you play this game?
Ok, first of all. At least play til you get under 40 in tt (Uncheated) and then come back and see if you still wanna cheat.
Good damn... I get so fucking mad. I know I dont make the best post... not helping my own cause by all swearing and too lazy to read trough to see I wrote right things and so on.. but damn.. I cant believe ppl just stand by.. its like once (maybe you heard of?) there was this guy hitler, came and fucked up and ppl just stood by. FUCK YOU!
Elasto Mania - ez better
Re: Optimal Elma player
i reserve my post to be at this place. edit:
ramone, i understand what you might feel. no need to bother about cleaning swears; if they're there, they're there.
first of all, gimp is not an authority on the matter. so you were right in not making a fuss about it.
secondo, i think authorities on the matter follow such law: "one point per complete year of playing." so that makes you an authority as big as zweq, and i'm willing to recognize that. like you, i am shocked by the lack of opinions on the matter. and that's precisely why i've took the responsibility to publish zebra's thoughts on the matter (sorry, zebra!)
thirtio, yes, an hitler is always at risks of being elected again.
as i see it, there are only two possible answers to this problem, either a yes or a no. and people like me interested in seeing best recs have nothing to lose, because they mainly hang here just to see recs anyway.
but people that say no, and still want to find styles by themselves (if they still are any big ones left to find), and still have fun in doing so, would be punished by the project, because they know both groups can't be divided. for example, one of the "yes" group would see replays and then attempt to take wrs. and dimkadimon is not going to take part in the debate either. i guess he wouldn't want to harm the scene by doing something we might not want to be done.
as for my own opinion, i am fully aware that it isn't to be taken as law either (as that should be made clear, even if people such as ramone already knows). i've been here for quite some time. i've stopped playing internals six or seven years ago, so i quite don't care. if everyone was like me, and think that internals have nothing left to really give them, not even fun, the debate would not even be happening.
that being said, the big question is, ramone, what (in your opinion) playing internals really bring to your life? for my part, the game has given me nothing good in the last few years (except some warm relationships). i even know people who has failed serious studies and had to take some university courses back again in order to get their diploma. one thing i'm sure of, is it wouldn't matter to me if it were published, because the game has nothing left to give me. maybe some fun in designing levels, but that's all. i say, publish the optimal internal replays, but not the algorithm or any tool that would permit someone doing so. freeze the wr table. internal are classic, and deserve to be perfectly driven. shame on you, imperfect wr holders. long live the elma-processor.
ramone, i understand what you might feel. no need to bother about cleaning swears; if they're there, they're there.
first of all, gimp is not an authority on the matter. so you were right in not making a fuss about it.
secondo, i think authorities on the matter follow such law: "one point per complete year of playing." so that makes you an authority as big as zweq, and i'm willing to recognize that. like you, i am shocked by the lack of opinions on the matter. and that's precisely why i've took the responsibility to publish zebra's thoughts on the matter (sorry, zebra!)
thirtio, yes, an hitler is always at risks of being elected again.
it's interesting you have been the first one to point out that in such a concise manner. that kinds of sums it up.ramone wrote:Of course it would be great fun to know/see best possible times. And then that is done, its over. And I mean OVER! No need to keep playing. It is finding the limits that makes playing fun. I am not interested in seeing these times/recs because they would spoil the (Internal) Elma scene. And possibly spoil future cups and so on.
as i see it, there are only two possible answers to this problem, either a yes or a no. and people like me interested in seeing best recs have nothing to lose, because they mainly hang here just to see recs anyway.
but people that say no, and still want to find styles by themselves (if they still are any big ones left to find), and still have fun in doing so, would be punished by the project, because they know both groups can't be divided. for example, one of the "yes" group would see replays and then attempt to take wrs. and dimkadimon is not going to take part in the debate either. i guess he wouldn't want to harm the scene by doing something we might not want to be done.
as for my own opinion, i am fully aware that it isn't to be taken as law either (as that should be made clear, even if people such as ramone already knows). i've been here for quite some time. i've stopped playing internals six or seven years ago, so i quite don't care. if everyone was like me, and think that internals have nothing left to really give them, not even fun, the debate would not even be happening.
that being said, the big question is, ramone, what (in your opinion) playing internals really bring to your life? for my part, the game has given me nothing good in the last few years (except some warm relationships). i even know people who has failed serious studies and had to take some university courses back again in order to get their diploma. one thing i'm sure of, is it wouldn't matter to me if it were published, because the game has nothing left to give me. maybe some fun in designing levels, but that's all. i say, publish the optimal internal replays, but not the algorithm or any tool that would permit someone doing so. freeze the wr table. internal are classic, and deserve to be perfectly driven. shame on you, imperfect wr holders. long live the elma-processor.
Last edited by hex on 24 Feb 2009, 00:34, edited 5 times in total.
Re: Optimal Elma player
The only opinion that matters is mine . I don't want to underestimate anyone but I have a feeling there is noone around who can handle patching and machine learning part.
I worked on this thing long time ago but then worked on ballelma/eol instead, it became much bigger project than i thought and i din't have time to get back to ai stuff. I doubt it's going to happen anytime soon (<1 year) so ramone can sleep well.
I would not tell anyone any results or show recs, same as i didn't tell anyone times I got those years ago. (I only told that zamppe pwned both 01 and 02 times)
I don't really play elma anymore so i don't think it affects my hoyling motivation anymore:). I may upload it to some youtube in 5-10 years when the community dies if i ever do it.
I worked on this thing long time ago but then worked on ballelma/eol instead, it became much bigger project than i thought and i din't have time to get back to ai stuff. I doubt it's going to happen anytime soon (<1 year) so ramone can sleep well.
I would not tell anyone any results or show recs, same as i didn't tell anyone times I got those years ago. (I only told that zamppe pwned both 01 and 02 times)
I don't really play elma anymore so i don't think it affects my hoyling motivation anymore:). I may upload it to some youtube in 5-10 years when the community dies if i ever do it.
[carebox]
Re: Optimal Elma player
respects mila. that would be okay for me. as long as i get to see them in my lifetime...milagros wrote:I don't really play elma anymore so i don't think it affects my hoyling motivation anymore:). I may upload it to some youtube in 5-10 years when the community dies if i ever do it.
peace out, foreigner.
Re: Optimal Elma player
I hardly see belma scene dying, not even in 10 years. internal scene? it's dead already
Re: Optimal Elma player
In BruteForce (password breaking program) it may take weeks to solve one password for example MuIkKu-3#
In this Optimal elmanator it would take many weeks to find the best way... Who wants to sacrifice his/her PC for that.
And if you ask me, elma would be dead if we could know those times
In this Optimal elmanator it would take many weeks to find the best way... Who wants to sacrifice his/her PC for that.
And if you ask me, elma would be dead if we could know those times
- Grace
- 38mins club
- Posts: 4851
- Joined: 19 Nov 2005, 10:45
- Location: Deep in your Imagination, Twirling your Dreams and Weaving your thoughts.
Re: Optimal Elma player
Drace - thats why dimkadimon set "Goals" so it wasn't just aimlessly driving.
i think this topic is over, milagros has spoken and said it is possible, we all know it would be good to see, but the scene would die. that pretty much sums it up.
next person to post just got anal raped by jaytea (if the next person to post is jaytea, then they just got anal raped by A.K.B.)
i think this topic is over, milagros has spoken and said it is possible, we all know it would be good to see, but the scene would die. that pretty much sums it up.
next person to post just got anal raped by jaytea (if the next person to post is jaytea, then they just got anal raped by A.K.B.)
Cyberscore!
___________________________________________________
Targets: 6 Legendary, 23 WC, 20 Pro, 5 Good | AvgTT: 39:59:96
___________________________________________________
Targets: 6 Legendary, 23 WC, 20 Pro, 5 Good | AvgTT: 39:59:96
Re: Optimal Elma player
just to be clear, from the start i was interested in the publishing of recs, not of the tool itself. if dimkadimon wants to work on this for his school project, i really wouldn't mind. if he's intelligent enough to design such tool, he's intelligent enough to be sound about it and not to do what the scene doesn't want him to do about it.SmaXa wrote:we all know it would be good to see, but the scene would die.
i can only speak for myself. externals interesting, internals just the same vertexes to be hoyled over and over. just being curious as to why people are still attached to internals after all these years.
still, until unanimity is attained, those recs are best left unpublished.
Re: Optimal Elma player
Programs will always be pwned by zweq. Point made.
Re: Optimal Elma player
I bet Zweq is robot.
Lousku wrote:could you mayke shorter sig please mega annoying and also against rules )
Re: Optimal Elma player
Mats wrote:as for hailing Zweq: this is the CERN:s super computer that once was assigned to beat Zweq in chess. (image taken after opening move by Zweq)
"It's good to be having fun, but it'd be funnier to be good"
Re: Optimal Elma player
zweq is just zweq.exe nice comment puoskari
Team TR
Multi WR in Labyrinth with GRob
Best Internal Total Times, Pipe stats & Pipe archive
World kuski map, World Cup stats
Re: Optimal Elma player
What's the deal with deliberately misquoting people?
Re: Optimal Elma player
but seriously, for a human playing say, table tennis (why not tennis too) to calculate the track of a speeding ball and acting correspondingly is relatively easy (as is controlling the elma bike to produce brutals etc.)
but from a computer it takes loads of arithmetics, dunno how mutch but i bet a lot. so making a program driving internals to the absolute limit can't be easy...one would need really to think up all the "laws" under which the bike would run to make it achieve superhuman performance.
I never have made programs myself, but i think i handle computer "logic" somewhat and based on that i'd say it's a hell of a job so well done in that field if you get it to work.
in my opinion i never wanna see the results unless the time comes i give shit about internals...though of course i am curious to know how far is it possible to take internals
but from a computer it takes loads of arithmetics, dunno how mutch but i bet a lot. so making a program driving internals to the absolute limit can't be easy...one would need really to think up all the "laws" under which the bike would run to make it achieve superhuman performance.
I never have made programs myself, but i think i handle computer "logic" somewhat and based on that i'd say it's a hell of a job so well done in that field if you get it to work.
in my opinion i never wanna see the results unless the time comes i give shit about internals...though of course i am curious to know how far is it possible to take internals
"It's good to be having fun, but it'd be funnier to be good"
Re: Optimal Elma player
This was said when WR TT was worse than 36.35.XX
Milagros wrote:Theoretical limit might be somewhere around 34 minutes, if there won't be any new version of elma, WRs will go to some 35.40
Zweq wrote:Under 35.30 is possible with current styles, I wouldn't say under 35 is impossible but the interest in the scene towards internals is very low. Might be because of how hard wrs are.
Re: Optimal Elma player
Adding new caption. It's common practice. But I agree, he should have left the quote as it was or have removed text.SveinR wrote:What's the deal with deliberately misquoting people?
Lousku wrote:could you mayke shorter sig please mega annoying and also against rules )
- The_BoneLESS
- 38mins club
- Posts: 4604
- Joined: 7 Sep 2003, 00:30
- Team: HHIT
- Location: Dangerously close to the St-Lawrence River
- Contact:
Re: Optimal Elma player
One main problem (every problem has a solution) i see with this agent in the elasto mania context is the following:
In the example stated in the first post of the topic (the racing game agent), the goal is the same: finishing the track as fast as possible. The main difference between the two games (aside from the fact that in elasto mania, you have to take apples and in addition, in no specific orders) is that, in elma, you can get speed not only by pressing gaz, but with many other tricks (including pushing, bouncing, and a few more). Once again, i don't know too much of how an intelligent agent can learn but, he would sure have to learn all those tricks and how they affect the bike.
The other point i see as problematic is how you can separate the tolerable replays from the untolerable. Let's take the famous stini bounce as an example:
If the agent managed to understand enigma and how to drive the fastest style possible (let's imagine that the actual style is the fastest possible). How could the algorithm know how hard he could bounce? I mean, even if it uses the in-game engine (i don't really know how you would create those replays/calculate these times), how could you seperate bugs from acceptable bounces?
I would sure like to know a little more about mila branch and bound algorithm because, it seems quite interesting to me.
And even though i know some people dislike the idea of a computer finding the best styles in internals, i do believe it would be very interesting simply from the mathematical/logical point of view. And even if someone manages to make that algorithm, i guess it'll take a while before it can realisticly beat the actual WRs.
PS: I'm pretty sure i am mixing up the two concepts (agent/algorithm) but again, i don't have that much knowledge in this area and really, i just wanted to feed the conversation which was slowly drifting towards gibberish.
In the example stated in the first post of the topic (the racing game agent), the goal is the same: finishing the track as fast as possible. The main difference between the two games (aside from the fact that in elasto mania, you have to take apples and in addition, in no specific orders) is that, in elma, you can get speed not only by pressing gaz, but with many other tricks (including pushing, bouncing, and a few more). Once again, i don't know too much of how an intelligent agent can learn but, he would sure have to learn all those tricks and how they affect the bike.
The other point i see as problematic is how you can separate the tolerable replays from the untolerable. Let's take the famous stini bounce as an example:
If the agent managed to understand enigma and how to drive the fastest style possible (let's imagine that the actual style is the fastest possible). How could the algorithm know how hard he could bounce? I mean, even if it uses the in-game engine (i don't really know how you would create those replays/calculate these times), how could you seperate bugs from acceptable bounces?
I would sure like to know a little more about mila branch and bound algorithm because, it seems quite interesting to me.
And even though i know some people dislike the idea of a computer finding the best styles in internals, i do believe it would be very interesting simply from the mathematical/logical point of view. And even if someone manages to make that algorithm, i guess it'll take a while before it can realisticly beat the actual WRs.
PS: I'm pretty sure i am mixing up the two concepts (agent/algorithm) but again, i don't have that much knowledge in this area and really, i just wanted to feed the conversation which was slowly drifting towards gibberish.
Website || TT:38:05:33 || WC5:15th || HHIT for life || 9th world wide ... BAP is next
Re: Optimal Elma player
yes i think its'prwtty much dead , no bad ment. i've had few drinks so sry if thiws seems stupid (as my every post prbly does)The_BoneLESS wrote:i just wanted to feed the conversation which was slowly drifting towards gibberish.
"It's good to be having fun, but it'd be funnier to be good"
-
- Kuski
- Posts: 30
- Joined: 8 May 2005, 00:12
- Location: Adelaide, Australia
- Contact:
Re: Optimal Elma player
Hello all,
I think I've caused some serious confusion here, so let me explain. The agent I want to create is just a computer science exercise, just like people have created chess/backgammon/othello/checkers/poker computer engines. In chess, backgammon and othello programs play stronger than world champion level. In poker they play slightly weaker or on par with world class players. In checkers, the game has been officially solved - this means that the computer has found the optimal line of play, so he is guaranteed to win or draw, but never lose. In all those games the humans are still playing and are still finding the game fascinating. One of the reasons they find it fascinating is because they can learn some strategies from the computer engines. For example, in backgammon a particular line of play was always considered weak, but the computer showed that it is actually good, and now professional humans are using it too. By no means do I want to destroy the fun of Elma, because I know how devastating this will be for everyone, including myself. Elma will remain as it is and all records will belong to humans only.
If I ever create this program, I will never release it to the public. The project will remain closed. I doubt that I will work on this in the near future, at least not for another year, simply because I don't have time. Furthermore, I believe that the agent will not be able to reach world record times, let alone overtake them. Perhaps it might come close on simple levels like Warm Up, but nothing more. The agent is unlikely to find new styles and is unlikely to reproduce bounces. What it should be able to do is make small improvements to existing runs. So the way I see it is you initialize it with your replay and it will try to improve it. If it succeeds, then you can try learning from it. But it might not even succeed at that.
Someone mentioned that this agent cannot possibly work due to the randomness of the game. You see, when we play we somehow manage to adapt to the randomness. Similarly, I expect that the agent can also adapt online, while its playing. It might also need many many runs to get a good time. So I can image running this thing for a whole day/week to get some improvement. Luckily I have access to a cluster of machines at work, so computational power should not be a big problem.
I think I've caused some serious confusion here, so let me explain. The agent I want to create is just a computer science exercise, just like people have created chess/backgammon/othello/checkers/poker computer engines. In chess, backgammon and othello programs play stronger than world champion level. In poker they play slightly weaker or on par with world class players. In checkers, the game has been officially solved - this means that the computer has found the optimal line of play, so he is guaranteed to win or draw, but never lose. In all those games the humans are still playing and are still finding the game fascinating. One of the reasons they find it fascinating is because they can learn some strategies from the computer engines. For example, in backgammon a particular line of play was always considered weak, but the computer showed that it is actually good, and now professional humans are using it too. By no means do I want to destroy the fun of Elma, because I know how devastating this will be for everyone, including myself. Elma will remain as it is and all records will belong to humans only.
If I ever create this program, I will never release it to the public. The project will remain closed. I doubt that I will work on this in the near future, at least not for another year, simply because I don't have time. Furthermore, I believe that the agent will not be able to reach world record times, let alone overtake them. Perhaps it might come close on simple levels like Warm Up, but nothing more. The agent is unlikely to find new styles and is unlikely to reproduce bounces. What it should be able to do is make small improvements to existing runs. So the way I see it is you initialize it with your replay and it will try to improve it. If it succeeds, then you can try learning from it. But it might not even succeed at that.
Someone mentioned that this agent cannot possibly work due to the randomness of the game. You see, when we play we somehow manage to adapt to the randomness. Similarly, I expect that the agent can also adapt online, while its playing. It might also need many many runs to get a good time. So I can image running this thing for a whole day/week to get some improvement. Luckily I have access to a cluster of machines at work, so computational power should not be a big problem.
-
- Kuski
- Posts: 30
- Joined: 8 May 2005, 00:12
- Location: Adelaide, Australia
- Contact:
Re: Optimal Elma player
The agent doesn't have the notion of bugs/not bugs. It only cares about one thing: minimize the time to reach a goal. If the time is minimized by having a bounce, then so be it. If this is the case then the action "bounce" will be reinforced and will be more frequent in future runs. If an action does not lead to a fast time (eg dying) then that action will be automatically penalized or "punished", decreasing the frequency of that action in future runs.The_BoneLESS wrote: The other point i see as problematic is how you can separate the tolerable replays from the untolerable. Let's take the famous stini bounce as an example:
If the agent managed to understand enigma and how to drive the fastest style possible (let's imagine that the actual style is the fastest possible). How could the algorithm know how hard he could bounce? I mean, even if it uses the in-game engine (i don't really know how you would create those replays/calculate these times), how could you seperate bugs from acceptable bounces?
I would sure like to know a little more about mila branch and bound algorithm because, it seems quite interesting to me.
And even though i know some people dislike the idea of a computer finding the best styles in internals, i do believe it would be very interesting simply from the mathematical/logical point of view. And even if someone manages to make that algorithm, i guess it'll take a while before it can realisticly beat the actual WRs.
PS: I'm pretty sure i am mixing up the two concepts (agent/algorithm) but again, i don't have that much knowledge in this area and really, i just wanted to feed the conversation which was slowly drifting towards gibberish.
BTW an agent is a computer program that is acting somewhat "independently" to achieve a particular goal in its environment. Its actions are either learned or pre-programmed. An algorithm is just some method. The agent may use many algorithms to achieve its goal. Programs that are not agents can also use many algorithms.
Re: Optimal Elma player
Don't disapoint Boneless he thought it was more like a james bonddimkadimon wrote:BTW an agent is a computer program that is acting somewhat "independently" to achieve a particular goal in its environment. Its actions are either learned or pre-programmed. An algorithm is just some method. The agent may use many algorithms to achieve its goal. Programs that are not agents can also use many algorithms.
- The_BoneLESS
- 38mins club
- Posts: 4604
- Joined: 7 Sep 2003, 00:30
- Team: HHIT
- Location: Dangerously close to the St-Lawrence River
- Contact:
Re: Optimal Elma player
hahaMats wrote:Don't disapoint Boneless he thought it was more like a james bonddimkadimon wrote:BTW an agent is a computer program that is acting somewhat "independently" to achieve a particular goal in its environment. Its actions are either learned or pre-programmed. An algorithm is just some method. The agent may use many algorithms to achieve its goal. Programs that are not agents can also use many algorithms.
but seriously, i get what an agent is, i am simply not familiar with the concept.
Website || TT:38:05:33 || WC5:15th || HHIT for life || 9th world wide ... BAP is next
Re: Optimal Elma player
Im sure that the Optimal Elma Player would do at least 4.32 with hooked bug in warm up, because Vista got 4.33!
Re: Optimal Elma player
4.33 Warm Up?? How?FinMan wrote:Im sure that the Optimal Elma Player would do at least 4.32 with hooked bug in warm up, because Vista got 4.33!
Lousku wrote:could you mayke shorter sig please mega annoying and also against rules )
Re: Optimal Elma player
i wish you good luck for your future project.