Ask questions via twitter! Message any question to @answers on twitter. We'll publish the question and send you a reply each time there's a new answer.
Next Question

Answered Question

 
 M¢37  Funded By Mahalo ? |  October 03, 2009 02:27 PM

How did the "towed camera sled" build by Chester Buchanan to locate the USS Scorpion?

Explain the search pattern for finding the USS Scorpion. Why did it take 6 months to find?
Interesting Question?  Yes (0)   No (0)   
RSS
 
 

Best Answer  Chosen by Asker

 
October 03, 2009 08:06 PM
The search pattern used to find the Scorpion was an early application of what's called "Bayesian search theory", which, interestingly, is used today by some email relay software to identify spam, but the general application is to find lost objects.

The first thing you do is formulate some hypothesis as to what could have happened to the lost object. That's important, because some of the hypothesis might involve the object being utterly destroyed, in which case a search is pointless and a probability of zero is assigned.

Next, define an area that you're going to search, and divide that area into a grid.

For each hypothesis, assign to each square in the grid the chance that the lost object would be found in that square of the grid if that hypothesis is true. In the case of looking for a sunk ship, you assign a higher probability of being able to find it in shallow water than in deep water, because it's easier to see things in shallow water, so in the case of a ship, areas of the grid with shallow water get a higher probability if the hypothesis is that it sunk and is still there (i.e. was not hijacked or was not blown to bits so small that it could be swept away by currents or has not slipped into a Bermuda Triangle worm hole etc.).

Now layer the grids for each hypothesis on top of each other and combine their probabilities.

Start searching in the area of the grid that has teh highest probability.

If that area turns up no result, move on to the area of the grid with the next most likely probability, but also do a recaclulation of the combined probabilities of the hypothetical possibility of the object being in each square of the grid by dropping the one you just searched, because if one of the hypothesis was heavily tipping the odds in favor of the area of the grid that you just searched, but was heavily disfavoring an area where the total combined probabilities made an area of the grid, say third, then without that first area's combined probabilities from all hypothesis being included, it might tip the odds so that now the next most likely place to find the lost object is not the one that had been ranked as third most likly, but is now calculated to be the second most likely place to find it, so as you go, the order of the grid areas that are searched might change.

Within each area of the grid, it's a simple sig-zagging back and forth.

Six months might seem like a long time to find the Scorpion, but actually, if people could understand just how vast the possible area was, and if they could understand just how dark it is down there, and how hard it can be to see things, such that there's a limite to how fast one can sweep over an area, they'd realize that six months is pretty darn good.
Asker's Rating:
• Bayesian logic branches are associated with a probability value. So the search used a probability or odd that the item was in a specific location.


Helpful Answer?  (1)   (0)   

Helpful: davepamn

Tip omicron for this answer
Permalink | Report
   Reply  
 
 
 
October 03, 2009 08:22 PM
I wrote a Bayesian expert system in College using a linear algorithm for determining win or loss odds between foot ball teams.

Report
 
 

Answer this Question

How tips and payments work

This question has already been resolved. You may add an answer to it but you will not be eligible to win best answer or any associated tips.

Ask a Question


140 characters left
Top of Page
Buy Mahalo Dollars with Credit Card or PayPal

Top Members

This Week All Time
  • buddawiggi
    buddawiggi
    2nd Degree Black Belt
    26710 Points
    M$774.34 Earned
  • kty2777
    kty2777
    Purple Belt with a Brown Tip
    5216 Points
    M$196.67 Earned
  • opher
    opher
    Purple Belt
    3852 Points
    M$152.92 Earned
   See All
 

Most Popular Tags

mahalo(1567)
iphone(460)
music(458)
google(346)
food(312)
online(288)
beer(277)
money(259)
movies(250)
apple(249)
aotd(235)
health(217)
video(201)
free(200)
dog(200)
   See All
 

Categories

Welcome New Members


 
 
Mahalo Dollars are the currency of Mahalo Answers.

Each Mahalo Dollar costs $1.

Once you earn more than 40 Mahalo Dollars, you can request to be paid via PayPal. Each Mahalo Dollar is currently worth $0.75 when paid out via PayPal. Learn More

 
 

Please log in to use this function.