scm007

I have a set of data points, (x, y, z). There are thousands of data points, (probably 1-3 thousand). I need to find a way to branch and bound this data such that a <= x <= b, c <= y <= d, e <= z <= f, in a way that is effective and won't take days to compute. I'm really stuck here as this is the hardest algorithm I've tried to write. I would be VERY grateful.

ANY HELP APPRECIATED,

Steve

pal

Hmm... What's the algorithm called?
Is it a tree?

DaWei

Please note that the forum's rules prohibit cross-posting. Please read them for other facts and restrictions.

__________________
Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code.
Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Arevos

To what end are you using the branch and bound strategy? What end result do you expect?

scm007

Basically, each data point has 3 relevant fields. I need to devise an algorithm such that it will find a solution set of data points (x, y, z) where the mean of each field is within a certain range of a specified value. For instance..

If I had data points (1,4,2), (4,9,8), (6,9,1), and (8,3,6). I would specify that 2<=x<=3, 6<=y<=8, 3<=z<=7. It would select data points (1,4,2) and (4,9,8).

I would also have the algorithm continue to run say 100 extra iterations after it found the first suitable value.

Arevos

Okay, so (1,4,2) and (4,9,8) are chosen because the mean of each individual field is (2.5, 6.5, 5), and the means fall within the parameters supplied - is this correct?

Is there any upper limit on the size of the data set returned? Does it attempt to return the maximum sized set, the first set it finds, or the first n sets, or the best m sets out of n found sets, or what?

crawforddavid2006

You lost me when you said

__________________