2013.04.17 Dev Chat

From Eterna Wiki

( Link back to Dev Chat archive )


On Scripting & Puzzle-Solving Bot Competition

 

jeehyung: Hi all [6:01 PM]

angrybirds64: hi [6:01 PM]

Eli Fisker: hi Jee :) [6:01 PM]

jeehyung: welcome to the bi-weekly eterna dev chat : ] [6:01 PM]

angrybirds64: cool [6:01 PM]

starryjess: hi jee [6:02 PM]

jeehyung: hi Eli, angrybirds, starry [6:02 PM]

JR: jee - what did you do with brourd [6:03 PM]

TomoeUzumaki: that is what I want to know [6:03 PM]

TomoeUzumaki: but tell me later, because I have to go for swim practice [6:04 PM]

JR: and how are the switch labs going [6:04 PM]

jeehyung: @JR, Tomoe not sure where Brourd is. I sure miss him. [6:05 PM]

jeehyung: @JR switch lab is going well. I haven't heard any major problem from the synthesis team. [6:05 PM]

JR: and theo - same [6:06 PM]

JR: theo lab [6:06 PM]

starryjess: Brourd won't come back unless we all tell him we miss him [6:06 PM]

starryjess: he's like Spug that way [6:07 PM]

jeehyung: @JR theo is bit delayed as we are having some techical trouble for preparing next puzzle. We need an interface that'll allow adding/removing bases in the game to move on. [6:08 PM]

jeehyung: @starry Spug?? [6:08 PM]

JR: thats a good thing - what is next puzzle? [6:09 PM]

starryjess: oh boy, it's a long story jee [6:09 PM]

jeehyung: @JR It'll be relatively similar puzzle, except that you will be able to shorten/extend the loop you can design [6:09 PM]

jeehyung: I do see a lot of Spug related player puzzles. I'm guestting it's related : P [6:10 PM]

starryjess: Jieux made some player puzzles which combined Eli's Spores and JL's bugs and called them Spugs. Then Tomoe made a comic about it. [6:10 PM]

Eli Fisker: :) [6:10 PM]

starryjess: and Spug was sad in it because no one believed in him [6:11 PM]

JR: site is working fine for me - are there upgrades in the works? [6:11 PM]

jeehyung: I'm looking at the spug comics now : ] [6:11 PM]

jeehyung: @JR nope we don't have updates going on right now. Was someone having trouble with the site? [6:12 PM]

JR: no -just asking [6:12 PM]

Eli Fisker: By the way, Jieux couldn't be here, but I know he is wishing to get numbers on the shadings of the shape data. [6:14 PM]

Eli Fisker: To better distinguish the shades [6:15 PM]

JR: when can we expect switch results? [6:15 PM]

kcabral28: Hi everyone [6:17 PM]

Eli Fisker: hi Kcabral [6:17 PM]

jeehyung: @Eli - yes actually one of our dev is working on it right now. It's close to being done! Hopefully it'll be there next week. [6:18 PM]

jeehyung: @JR we are shooting toward the end of April. [6:19 PM]

jeehyung: hi kcabral28 [6:19 PM]

JR: thx [6:19 PM]

Eli Fisker: Oh, thats really cool, Jee. Thx. Mat will be happy too. [6:19 PM]

JR: do you have any expectations from us concerning how we use scripts? [6:21 PM]

JR: I want to but have not started yet [6:22 PM]

jeehyung: @JR - yes we want to hold a competition where player scripted bots compete against each other and NUPACK/ViennaRNA algorithms in solving player puzzles. [6:23 PM]

JR: cool [6:23 PM]

jeehyung: We actually have all the player scripts solving the player puzzles in the background. We just need a good way to visualize them and make it into a competition [6:23 PM]

JR: nando will be happy [6:23 PM]

Eli Fisker: Vinnie too :) [6:24 PM]

JR: maybe use real people solves as base? [6:26 PM]

JR: I am out of questions so we may want to cut you loose early if no more questions [6:28 PM]

Eli Fisker: Jee, I know it will be useful for an algorithm to be able to solve an structure according to what model it is supposed to use. And the current algorithms can't always. Are these player generated bots going to help an energy model accomplish that? [6:28 PM]

Eli Fisker: Guess I'm basically asking about the purpose. [6:29 PM]

jeehyung: @Eli - the current competition is the most primitive step : focusing on the computational part. [6:29 PM]

jeehyung: You already know that just designing RNAs that work in the game is easy, but designing RNAs that work in the test tube is hard [6:30 PM]

stlnegril9: Hi Jee - I have a script I'm trying to publish (via "submit Script" button), but it will not. [6:30 PM]

Eli Fisker: very true [6:30 PM]

stlnegril9: should I be able to? [6:30 PM]

stlnegril9: msg is "Evaluation server is not available now." [6:30 PM]

jeehyung: So for the bot, we'll start with (relatively) easier problem : Just solve the puzzle in the computational model. We'll eventually move on to even harder problem. [6:31 PM]

jeehyung: @stlnegril you should be able to. It seems like a problem on our end.... can you save the script content to somewhere else and submit it later? I'm sorry about the trouble. [6:31 PM]

jeehyung: I'll let the script dev inform you when the problem is fixed. [6:32 PM]

stlnegril9: yes, it is saved. thx, will come back to it later [6:32 PM]

kcabral28: @Jee Now that we have the cloudLab up and running I have been having a little trouble analyzing all the synthesis data. [6:32 PM]

Eli Fisker: Ok, Jee. I think it is very cool that players have already made bots that are capable of solving puzzles. They are getting real good already. [6:32 PM]

kcabral28: @Jee I find it is difficult to switch back and forth between solutions to compare them. I was thinking, what if we had an interface that preloaded a number of solutions chosen by the user and a menu that allowed you to quickly switch between solutions. I think it would help people see patterns between the different solutions. [6:32 PM]

Eli Fisker: I agree with Kcabral [6:33 PM]

jeehyung: @Eli I also want to emphasize that even computational solving still is a hot topic among computational biology people : ] Some of them are actually planning to set up a global RNA design competition ..and I hope our player scripts can take part in that. [6:34 PM]

JR: any time frame? [6:34 PM]

Eli Fisker: Wow, that is so cool. Had no idea. Thx for telling [6:34 PM]

jeehyung: @kcabral, Eli. This issue is to be resolved when we move the browsing interface from flash to web version. in the web version you should be able to quickly open/close RNAs without going back & forth from the browsing interface [6:35 PM]

kcabral28: @Jee I don't know what other players are doing to compare synthesis data. With 40+ solutions and multiple rounds for each lab I am starting to feel overwhelmed by all the data and a more efficient way of comparing it all would be beneficial to a lot of people I think.. [6:35 PM]

kcabral28: Oh cool [6:35 PM]

jeehyung: @kcalbral many share the same concern and that's why we want to bring in this new browsing interface [6:35 PM]

jeehyung: hopefully we'll be able to show the prototype soon. It'll be lot easier to browse through solution than it is now [6:36 PM]

Eli Fisker: Jee, thanks for doing that, I will look forward to it. [6:36 PM]

jeehyung: @JR for the competition, we are hoping we'll be ready around Fall. Of course it also depends on when the competition will be actually setup. I think it'll be before 2014, but I'm not entirely sure. [6:37 PM]

JR: thx [6:38 PM]

JR: we will make sure nando is out of bed before then [6:38 PM]

Eli Fisker: hehe [6:38 PM]

Eli Fisker: And Johns puzzle solving script is doing real fine too [6:39 PM]

kcabral28: I'm sure we'll do well [6:39 PM]

JR: agree [6:39 PM]

jeehyung: @Eli - yes everyone was excited : ] [6:39 PM]

Eli Fisker: yep [6:39 PM]

Eli Fisker: hehe, I bet [6:39 PM]

Eli Fisker: We will send an army of bots :) [6:39 PM]

kcabral28: I gotta learn how to script :) [6:40 PM]

jeehyung: We actually have no doubt the players scripts will win. We are more concerned about how we can make the results more interesting to scientific community : ] [6:40 PM]

Eli Fisker: of the sweet puzzle solving kind [6:40 PM]

JR: can you give us some parameters [6:40 PM]

Eli Fisker: ah, so you guys are the ones setting up the competition? [6:40 PM]

JR: what they would be looking for? [6:41 PM]

jeehyung: @Eli - oh no, the competition is a separate thing by other people. But we are interested in "investigating" just how our players can beat NUPACK/INFORNA, which were developed by the smartest people in the field. "What made it possible? What was the process? Did ranking/competition help? etc." [6:42 PM]

Eli Fisker: @kcabral, many of us will have to [6:42 PM]

jeehyung: @Eli, kcabral we want to setup a tutorial when we have time...but meanwhile, I recommend Code Academy. That'll get you started on simple javascript (which we use in our interface) [6:43 PM]

Eli Fisker: Jee, ah. So doing good is just a part of the story, you guys want the why too [6:43 PM]

jeehyung: @Eli, doing good will surprise people. Figuring out why so we can do it again, will be ground breaking : ] [6:44 PM]

Eli Fisker: I have made a very very basic intro to scripting, mainly covering how far I got: [6:44 PM]

Eli Fisker: https://docs.google.com/document/d/1VC94wTuRhNd-FN60ueaBc5mUxBsiIy8UzJzQL6ap3CE/edit# [6:44 PM]

Eli Fisker: Jee, I like that perspektive. Should be fun figuring that out. [6:45 PM]

JR: i completed code academy - would recommend [6:45 PM]

jeehyung: oh wow you are always one step ahead of me [6:45 PM]

angrybirds64: hello [6:46 PM]

jeehyung: @Eli - do you mind if we add the link to the intro in our scripting page & documentation? [6:46 PM]

Eli Fisker: Jee, thx. Go ahead. And I will look very much forward to that tutorial you guys are planning [6:46 PM]

Eli Fisker: Jee, one more question. Are there any news on the science paper? [6:50 PM]

jeehyung: Eli - not yet. We are still waiting for their response : ] [6:51 PM]

Eli Fisker: Ah, so it is submitted :) [6:51 PM]

Newy Oudone: Ah... coding... brings back memories. [6:51 PM]

jeehyung: It has been. Now just hoping for the best : ] [6:52 PM]

Eli Fisker: Yep [6:52 PM]

jeehyung: Hi Newy [6:52 PM]

jeehyung: This is a eterna dev chat session. Please throw in any question/suggestion you have about the game : ] [6:53 PM]

angrybirds64: r any of u good freinds of the makers of this games [6:53 PM]

jeehyung: @angrybirds I'm one of the developer : ] [6:53 PM]

angrybirds64: cool u r a genius [6:54 PM]

lroppy: yo [6:55 PM]

JR: study hard and go to college - you can do it too [6:55 PM]

lroppy: just got back from MDs [6:55 PM]

Eli Fisker: hi Iroppy :) [6:55 PM]

lroppy: hi all [6:55 PM]

lroppy: Jee [6:55 PM]

lroppy: any luck slaying some of those bugs i reported? [6:55 PM]

Newy Oudone: You know, I still don't know the point to these puzzles. It seem it isn't enough to just solve the puzzles but rather use as many GU, and as little GC as possible. [6:56 PM]

Newy Oudone: So, I've been challenging myself to do that. o.O [6:57 PM]

Eli Fisker: Newy, good, this will help you with the harder puzzles. In lab things will change [6:57 PM]

Eli Fisker: There it will not be about having most GU's and fewest GC-pairs. [6:58 PM]

jeehyung: @lroppy unfortunately not yet. The dev on the job is still having hard time. I'll go in and look into the problem myself next week once I'm freed from cloud lab works I'm tied to now.. [6:58 PM]

jeehyung: @lroppy Sorry I don't have a good news about this [6:58 PM]

Eli Fisker: But rather have a good balance of basepairs. GU's will be the rarest [6:58 PM]

lroppy: ok thx, Jee [6:58 PM]

jeehyung: I actually have to run now all. Thanks always for the discussion & suggestions. I'll post the chat log in the news item. [6:59 PM]

jeehyung: See you all in next dev chat! [7:00 PM]