optimization with binary parameters

[freezer951000]

Hello,

I have  a pb of linear optimization with binary parameters, how can i use the optimization VI used in labview? for example: id like to minimize this function f(x)=x1-3x2+x3, with constraint: x1..x3 is binary, and x1-3x2+x3>0: it s simple case but i have no idea how to put binary input in this linear optimization VI : 

[Bob_Schor]

What do you mean by "binary input"?  Do you mean that your parameters x1, x2, and x3 are either 0 or 1?  If that is the case, the minimization scheme seems trivial -- compute the function for the eight possible values of x1, x2, x3 (if you create them as bits of an integer, you are testing with the number 0 .. 7), then just see which of the eight values is "optimal" (whatever that means to you).

Bob Schor

[freezer951000]

yes just 0;1, but for my  case i have about 107 parameters; and i want to generate automatically the best combination of 0 and 1 to minimize the cost which is a combination of these parameters

[Bob_Schor]

Whoa -- that involves the corners of a 107-dimension hypercube, a pretty big number!  When you start getting up in these higher-dimensional spaces, the process of minimization becomes much more complicated.  I was just thinking "maybe a version of the Simplex method", but how do you find the "centroid" of, say, a cube (where you are restricted to the corners of the cube)?  You might be able to "random-walk" your way to the minimum, but I wouldn't want to wager more than a few pennies on your chance of success ...

BS

[freezer951000]

Ok so it s impossible, i can mark off the problem, because its a recursive method to calculate the cost, the final cost is the result of assembly of many level of parts, if i optimize assembly parts by assembly parts when i ll sum all assembly part cost , i would have the final cost of the finish product. So how many parameters should i use for simplex method in maximum?