What are the chances of finding a particle?

Discuss your experiences with and ideas about Stardust@home here.

Moderators: Stardust@home Team, DustMods

Schlockading
DustMod
Posts: 132
Joined: Thu May 18, 2006 7:20 pm
Location: Mesquite, TX

What are the chances of finding a particle?

Post by Schlockading »

I was wondering what the statistical chances of discovering a particle for a single person are if the focus movies are searched on a regular basis.
Image

Sundevil
Posts: 41
Joined: Wed May 17, 2006 7:19 pm
Location: Ohio, USA

Post by Sundevil »

If they are estimating that there will be 40 to 100 found, and there will be 1.4 million movies... let's just say those aren't the best odds.

Siegfried
Posts: 105
Joined: Wed May 17, 2006 7:18 pm
Location: La Quinta, California
Contact:

Post by Siegfried »

No, I doubt it. I wish I had some cool math to back me up, but I don't.
And God said: E = +mv^2 - Ze^2/r ...and there *WAS* light!

WANTED:
Dead or Alive
^-^
( ^ )
Schroedinger's Cat

Ferrum
Posts: 43
Joined: Thu May 18, 2006 6:59 pm
Location: USA
Contact:

Post by Ferrum »

I'd leave any official response to a moderator, but what exactly do you mean by "a regular basis" schlockading? I'll note that movies that contain fake particles will likely be inserted frequently, and since any movies that have been reported to contain trails will be resubmitted to more searchers, the 1.6 million movie figure will likely increase, although not necessarily substantially. The real difficulty lies in how long the project is expected to last. Let's say 25 weeks (projected end date is in Dec). If one presumes that one can look through 100 movies before it gets slightly monotonous, and sustain this 5 days a week for the duration of the project, an individual user could look through 12,500 movies. 1,600,000 movies/45 particles=1 in 35,555 chance of any movie containing a interstellar particle. Not horrible odds, since this would suggest that as many as 1 in 3 highly motivated participants might have a shot at discovering a particle. Please let me know if anyone observes a flaw in this logic. I'm an accounting major so I hope I'm right :?
The integral sec y dy
From zero to one-sixth of pi
Is the log to base e
Of the square root of three.
Um...times the square root of the fourth power of i.

Siegfried
Posts: 105
Joined: Wed May 17, 2006 7:18 pm
Location: La Quinta, California
Contact:

Post by Siegfried »

Sounds perfectly...er...sound to me! :shock:
And God said: E = +mv^2 - Ze^2/r ...and there *WAS* light!

WANTED:
Dead or Alive
^-^
( ^ )
Schroedinger's Cat

Aquila Hawk
Posts: 66
Joined: Thu May 18, 2006 7:58 pm
Location: Delmar, NY
Contact:

Post by Aquila Hawk »

Siegfried wrote:Sounds perfectly...er...sound to me! :shock:
Actually, and here is that cool math you were looking for, consider that 1 or many dust particles can impact a single section covered by a movie. Therefore, to actually determine an accuate chance of finding a dust particle, you would probably have to look at a poisson distribution. That would give you your chance for finding just about any number of dust particles. The only problem is that we would have to find all the dust particles before we can even generates such a statistic. Let's just for now say that the probability is rediculously low of finding a dust particle, but it will be fun to look anyway.
Everyone talks about SOH CAH TOA, but no one ever talks about CHO SHA CAO.

DustTrailFinder

Post by DustTrailFinder »

If the assumptions are:

1. 40 actual particle trails
2. no one finds more than one in a movie even if more than one exists
3. "discover" means the first report of a trail
4. 1.6 million movies to be initially viewed by 4 people
5. everyone gets an equal chance at being a first, second, third, or fourth viewer

Then I think the chances are essentially 40 in 6,400,000 (1 in 160,000) for each movie viewing by a particular individual. Of course, the chance increases directly by how many movies are viewed by that individual and his efficiency, and indirectly by the number of people who may have viewed the same movies before him/her and their efficiency. In addition, if the movies you look at are part of the second review by 100 people after at least two of the four initial reviewers have already reported a trail, those views will never count as a "discovery," but if only real trails are reviewed, that would be only 16,000 more added to the total of 6,400,000 viewings, an insignificant and statistically irrelevant figure. Chances are that at least a few false trails will be reported by at least two of the first four reviewers, but it would take a rather large number of these to statistically affect the outcome.

Essentially, if it takes 180 days to complete all reviews and an individual does only 10 each day during this period, then I see his chances of discovering a particle at 1,800*40 or 7,200 in 6,400,000, or 1 in 890.

Siegfried
Posts: 105
Joined: Wed May 17, 2006 7:18 pm
Location: La Quinta, California
Contact:

Post by Siegfried »

I'm going to go create a thread on this, but is it possible we will find other particles?
And God said: E = +mv^2 - Ze^2/r ...and there *WAS* light!

WANTED:
Dead or Alive
^-^
( ^ )
Schroedinger's Cat

Ferrum
Posts: 43
Joined: Thu May 18, 2006 6:59 pm
Location: USA
Contact:

Post by Ferrum »

AH, if i recall correctly from calc class all a poisson distribution can do is to predict the likelyhood of any variation from, in this case, a rate of impact consistent with has been observed, and I don't recall doing any calculations with the poisson formula other than probabilities of specific divergence from certain rates. Since the likelyhood of multiple impacts is statistically unimportant in this case, just rudimentary calculations would probably do just fine.

DTF, excellent point regarding the fact that that only those individuals who discover the particle first will be recognized (in other words only one of the four people to whom the movie has been sent). However 1800*40 doesn't equal 7200.

In any case, can't wait to get started!
The integral sec y dy
From zero to one-sixth of pi
Is the log to base e
Of the square root of three.
Um...times the square root of the fourth power of i.

DustTrailFinder

Post by DustTrailFinder »

Hi Ferrum,

You're right, 1800*40 is 72,000 not 7,200. Thanks.

In the FAQ forum, a moderator has now confirmed that the first to identify a track will be considered the "discoverer."

Additionally, it is now almost impossible to figure any sort of odds because the Berkely scientists are throwing in a wrench in that some movies will contain not actual samples from Star Dust, but samples known to contain actual trails and no trails to rate each viewers efficiency and weed out those who might be speedily clicking through and marking every movie as having a track. If I'm reading the moderators comments correctly, they are going to rely on this methodology and have abandoned the 4 initial views that would then be reviewed by another 100. If all this is true, then the only thing we can really say is that one's chances of being a "discoverer" is directly related to how efficient they are, how many movies they view, and how many actual trails exist.

Bosniac
Posts: 22
Joined: Fri May 19, 2006 4:05 pm
Location: Sara-is-an-ox; B&H

Post by Bosniac »

DustTrailFinder wrote:Hi Ferrum,

You're right, 1800*40 is 72,000 not 7,200. Thanks.

In the FAQ forum, a moderator has now confirmed that the first to identify a track will be considered the "discoverer."

Additionally, it is now almost impossible to figure any sort of odds because the Berkely scientists are throwing in a wrench in that some movies will contain not actual samples from Star Dust, but samples known to contain actual trails and no trails to rate each viewers efficiency and weed out those who might be speedily clicking through and marking every movie as having a track. If I'm reading the moderators comments correctly, they are going to rely on this methodology and have abandoned the 4 initial views that would then be reviewed by another 100. If all this is true, then the only thing we can really say is that one's chances of being a "discoverer" is directly related to how efficient they are, how many movies they view, and how many actual trails exist.
how will they determine if we missed something if they abandone 4 views, could you explain.
You see things; and you say Why? But I dream things that never were; and I say Why not? (B.S.)

DustTrailFinder

Post by DustTrailFinder »

Bosniac

I don't know that they have actually abandoned having four views for each movie. It appears more likely that they have decided to place more confidence in rating each viewers efficiency rather than relying only on reviews by 100 more reviewers.

I think they had a lot more people signing up to assist in this endeavor than they had originally thought would, and have now merely refined their methodology. Additionally, they have found resources to greatly expand their capabililty to support the project, and that has probably added to their ability to interpret and evaluate results.

I'm no expert on any of this, I'm only stating my interpretation, right or wrong, about what I read from the mods here.

Bosniac
Posts: 22
Joined: Fri May 19, 2006 4:05 pm
Location: Sara-is-an-ox; B&H

Post by Bosniac »

Well ok it does sound reasonable. If we are trained enough (they might start testing us right away, like practice... and for fun), we might not make easy assumtions, and then they might go over our potential finds (but here is the person who just finished tutorial session... and still not sure if he gets it).
You see things; and you say Why? But I dream things that never were; and I say Why not? (B.S.)

DustTrailFinder

Post by DustTrailFinder »

Bosniac

I thought the tutorial was pretty much straight forward except for the scratch and another movie that I can't recall at the moment. Although I don't think I'll have much trouble with the scratch, the other one is very troublesome to me, since I really couldn't see the trail in it. However, I would rather find a real trail, than to miss such an obscure trail anyway. I would hope that someone more capable than I am would find it. I think that no matter how many times I review that movie, I would never believe it to be an actual trail.

Bosniac
Posts: 22
Joined: Fri May 19, 2006 4:05 pm
Location: Sara-is-an-ox; B&H

Post by Bosniac »

well same here, although when someone eventually finds one, they will probably post images that we can look at and memorise, compare ect.
You see things; and you say Why? But I dream things that never were; and I say Why not? (B.S.)

Post Reply