[Coin-lpsolver] Linear Programming Concept
tunde fabiyi
fabiyitunde at hotmail.com
Mon Apr 28 14:34:07 EDT 2003
maybe you didnt get me right i only wanted to share my idea of what i
developed AND I HAVE NOT REALLY LOOKED AT LP LIKE YOU SAID. for example we
have a linear programme already in standard form and it has 20 variables and
lets say 40 constraints equation .what i am saying is that definitely there
is going to be a particular combination of this variables that will
definitely give us the required result i.e X1 = 1,X2 = 0 ,X3 = 5 ,X4 = 6 ,X5
= 0 .....X20 = 10[lets take for example that this is the optimal solution
for the objective function. now what my model does is that for every of this
variable there exist a maximum value they can be for the variables to be
able to satisfy every of the constraint equation for e.g if we set in
constraint no 1 all other variable in the constraint to zero Apart from X1
we would get the highest possible value for X1 in Constraint 1 which is
equal to the RHS of the equation. now doing this for all the constraints i.e
from 1 to 40 we could get the highest possible value for X1 in all the
constraints equation. you will agree with me that the lowest Value of X1 in
this group would be the highest possible value of X1 in generality. if we do
this for all the variables . is not possible to now have an algorithm that
will try every combination of this this variables and check them against
every constraint equation to see wether the combination will satisfy them
before applying it to the objective function to get and record the objective
value before trying out this combination again.[Note This Algorithm has A
termination Point ,it starts from initializing all variables to Zero and
TRIES EVERY POSSIBLE COMBINATION OF THIS VARIABLES BEARING IN MIND THAT THEY
ALL HAVE A MAXIMUM VALUE]
_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963
More information about the Clp
mailing list