GUI recipes: Local Wiggle Sequence loses points

Case number:845813-993943
Topic:Game: Tools
Opened by:Madde
Status:Open
Type:Bug
Opened on:Thursday, November 22, 2012 - 15:40
Last modified:Wednesday, April 24, 2013 - 22:40

"Local Wiggle Sequence" in GUI recipes doesn't work as intended anymore. It should only keep score gains (as it did for years) and not losses.

(Thu, 11/22/2012 - 15:40  |  11 comments)


Angus's picture
User offline. Last seen 10 hours 21 min ago. Offline
Joined: 06/04/2008
Groups: Beta Folders

Would be be possible to get complete release notes when there is an update? It's been asked many times before.

tamirh's picture
User offline. Last seen 6 years 10 weeks ago. Offline
Joined: 05/11/2012

Release notes are posted on the front page whenever there's a release. Sometimes there are things we miss, but they should be complete

Angus's picture
User offline. Last seen 10 hours 21 min ago. Offline
Joined: 06/04/2008
Groups: Beta Folders

Well, for example - The GUI local wiggle issue, and the Behavior CI scale change are not noted in the release notes.

It leads one to wonder what else was changed that is not documented.

Madde's picture
User offline. Last seen 10 weeks 4 days ago. Offline
Joined: 05/29/2008
Groups: Void Crushers

*Bump*

Seth describes the function in http://fold.it/portal/node/985980 "The sequential local wiggle in the cookbook will only change the protein if it can improve the score."
So the current behavior is clearly a bug.

tamirh's picture
User offline. Last seen 6 years 10 weeks ago. Offline
Joined: 05/11/2012

Could you give an example of a puzzle+gui script where this happens?

If it's at a point further along than the starting point of the puzzle, you can share a solution with yourself and let us know the title and we will be able to look at it. Thanks

Joined: 09/21/2011
Groups: Void Crushers

Take ANY puzzle and run Co settle script.

Madde's picture
User offline. Last seen 10 weeks 4 days ago. Offline
Joined: 05/29/2008
Groups: Void Crushers
tamirh's picture
User offline. Last seen 6 years 10 weeks ago. Offline
Joined: 05/11/2012

I've been trying these 4 GUI scripts: http://fold.it/portal/recipe/1773 http://fold.it/portal/recipe/1830 http://fold.it/portal/recipe/1831 http://fold.it/portal/recipe/1832

I've tried it on 701, 702, and 703. I reset the puzzle and run the script. I've also tried modifying the protein from the starting pose before running the scripts. I haven't seen a score decrease from any of the local wiggle sections of the scripts.

Are the score decreases you guys are seeing very large or small? I was just doing the check by visually watching the score with the advanced gui to see 3 decimal places. If it's a very small score decrease that's happening I'll add in some code that verifies the score remains the same or increases instead of doing it visually.

Madde's picture
User offline. Last seen 10 weeks 4 days ago. Offline
Joined: 05/29/2008
Groups: Void Crushers

I shared a solution of 703 with myself, it's called "703 GUI test for tamirh"
I ran my "2-6-3 walk" on it and that's the start of the undo graph:










The graph should never go down but it clearly does.
Maybe it happens only in the Windows version of Foldit?

Joined: 09/21/2011
Groups: Void Crushers

This behavior is there since I started folding. I have allways thought it to be normal. Sometimes you loose a lot of points, most of the time small. And it is mostly seen when you use a settler on a pretty stable puzzle where points are hard to get.

Joined: 09/21/2011
Groups: Void Crushers

Because the score can go up-down-up-down-real gain, it might not be a good idea to restore the score at each iteration, but only to restore the best when all requested iterations are done.

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons