"There's a problem with this focus movie"

Post here if you are having any kind of problem with the Stardust@home website.

Moderators: Stardust@home Team, DustMods

jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Re: "There's a problem with this focus movie"

Post by jsmaje »

Dan, someone somewhere within the team will have actually made that particular PM: http://aerogel.ssl.berkeley.edu/powermi ... 54&theY=-2, and will know where exactly what and where the supposed 'track' is.
You are better placed than us to get an answer, so can you please do so?

I've had to give up dusting due to the visual obscurity of many PMs, not just to my poorer vision, but not now knowing what to take seriously or discount in the 'real' movies.
Surely the list of duster-clicked vague blobs and dots, mostly artefactual but thought just might be be 'real', must now be overwhelming the team, regardless of how many undergrads they recruit to help spend time reviewing them.

Is this what they really want?
John
DanZ
Site Admin
Posts: 777
Joined: Fri Feb 27, 2009 2:44 pm
Location: Berkeley, CA
Contact:

Re: "There's a problem with this focus movie"

Post by DanZ »

Hi John. Sadly, we have NO undergrads looking at duster clicks, etc. In fact, I'm not sure how it does it, but the old algorithm seems to be enough to satisfy the team for now, and especially with the new foils search, they've identified 50 new interstellar candidates as detailed here (along with plans to acquire new funding). As for the bad tracks, the one you mention certainly has problem coordinates with a -2 Y. Thus the track is probably off the screen and this one should be deleted, just like many others with the same problem. Robert (our tech) was looking into those and finally figured out they weren't being deleted when he thought they were. I will check with him pronto and get a response. Thanks for your patience! Dan
jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Re: "There's a problem with this focus movie"

Post by jsmaje »

Thanks for that Dan. But just what sort of algorithm are you referring to that the team use and are apparently happy with?

I thought calibration tracks were Photoshopped (or whatever) by hand, originally by Anna Butterworth I recall, who would put it within-frame and accurately register the coordinates. If instead it's an automated algorithm that is placing some out-of-frame, then that's plain daft, and surely easily corrected.

The only alternative I can think of is a marking algorithm that accepts clicks within whatever micron radial tolerance, even from a genuine calibration track in an immediately adjacent frame, so classifies the first also as a calibration track.

Just curious.
John
DanZ
Site Admin
Posts: 777
Joined: Fri Feb 27, 2009 2:44 pm
Location: Berkeley, CA
Contact:

Re: "There's a problem with this focus movie"

Post by DanZ »

Hi John. Sorry, I obviously miscommunicated and probably misused algorithm here. What I meant is that the system still identifies those real, non-PM tracks that are getting the most duster clicks, and thus directs the team to where to look for tracks and the best possible chances for candidate particles. But yes, it is true that many of the PM's themselves (track placement within frame and where to click on them) were auto generated by a script that is apparently in need of fine tuning given all these bad coordinates.

Re the PMs with bad coordinates that we thought we were deleting, Robert says he was marking them via an "exclude" field and running a program that would then identify those to be excluded and keep them out of the VM. But now he sees that he did something in error and that's why many are not actually getting deleted, or more accurately, excluded from the VM. He's currently working on a fix and I hope to update you shortly. We're also investigating an entirely new VM system built from scratch and far less prone to error. I imagine that will actually happen and get unveiled sometime in 2017.

More soon!

Dan
jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Re: "There's a problem with this focus movie"

Post by jsmaje »

Thanks again Dan, and credit to Robert for realising his error.
Because I don't find dusting easy now, I'll probably wait until the 2017 re-build before reconsidering.
John
caprarom
Posts: 337
Joined: Thu Aug 02, 2007 7:12 am
Location: Riverview, MI

Re: "There's a problem with this focus movie"

Post by caprarom »

Dan, two comments, possibly for discussion next telecon.
1. Not all the tracks with out of range coordinates need deletion. If they are truly visible, the coordinates just need to be adjusted to where they go off the screen as opposed to where they really terminate beyond the visual field, as Rob has already done for two of them.
2. Your last post makes me wonder how/if the Alpha listing still figures into the team's consideration. Some of the more subtle tracks might not get as many "hits," but can still be highly rated. E.g., 6724440V1 was only recently promoted. It has been flagged just seven times out of 36 views, but is rated 9.5 on the Alpha List.
Mike
Eric Romano
Posts: 8
Joined: Wed Jul 13, 2016 11:25 pm

Re: "There's a problem with this focus movie"

Post by Eric Romano »

Hey all.
I'm new here. Is this forum active?
caprarom
Posts: 337
Joined: Thu Aug 02, 2007 7:12 am
Location: Riverview, MI

Re: "There's a problem with this focus movie"

Post by caprarom »

Hi, Eric. Welcome aboard. The forum is active, but you've joined at a time when there's a bit of a lull awaiting the loading of new movies. Most dusters are probably working more with the foils right now rather than with the aerogel tiles. You can get some status updates from the Third Thursday Telecons posted on the home page (click [View] under "From the desk of --- "). Regards, Mike C.
DanZ
Site Admin
Posts: 777
Joined: Fri Feb 27, 2009 2:44 pm
Location: Berkeley, CA
Contact:

Re: "There's a problem with this focus movie"

Post by DanZ »

caprarom wrote:Dan, two comments, possibly for discussion next telecon.
1. Not all the tracks with out of range coordinates need deletion. If they are truly visible, the coordinates just need to be adjusted to where they go off the screen as opposed to where they really terminate beyond the visual field, as Rob has already done for two of them.
2. Your last post makes me wonder how/if the Alpha listing still figures into the team's consideration. Some of the more subtle tracks might not get as many "hits," but can still be highly rated. E.g., 6724440V1 was only recently promoted. It has been flagged just seven times out of 36 views, but is rated 9.5 on the Alpha List.
Mike
Hi Mike. Great questions! For #1, I agree with you that some of the bad tracks just need the coordinates adjusted. But there was a decision a long long time ago that for time's sake, it was easier to just delete the bad ones - thinking I believe, that a deletion here or there would just be a drop in the bucket and have little impact on the total number of PMs. I would like to bring this up again with the team though and see if they might reconsider, or if a revamped VM in 2017 (which is planned) would completely eliminate this problem, and thus be the main focus going forward. For #2, I know the Alpha list is still one of our most valuable tools we have and no one wants it to go away nor be abandoned by the Red Team. I was just saying that all of these things in tandem certainly keep the machine going. Dan
caprarom
Posts: 337
Joined: Thu Aug 02, 2007 7:12 am
Location: Riverview, MI

Re: "There's a problem with this focus movie"

Post by caprarom »

Thanks, Dan. TTYL in today's telecon.
Post Reply