engineereeyore
Platinum Member
We're not allowed to use the linprog function, so we have to write everything from scratch. We can use lu(), which I do. But I'm confused about something. These are the 5 steps shown for what to do.
1) Calculate the current solution Xb = Yo satisfying BYo = b,
2) Solve lambda(transpose)B = Cb(transpose), and set Rd(transpose) = Cd(transpose) - lambda(transpose)D. If Rd >= 0, optimal solution is found.
3) Determine vector to move into basis B and out of D.
4) Check for unboundedness and determine pivot element.
5) Update B. Return to step 1.
Ok, so hopefully you understand the terminology. The problem I'm having is that the whole purpose is to keep from using an inverse, such as B^-1. So my problem is in step two. How do you solve lambda(transpose)B = Cb(transpose) when you know B and Cb(transpose), don't know lambda, and you aren't supposed to use B inverse?
Can someone explain this to me? Maybe I'm misunderstanding what I'm supposed to do, so if anyone has some knowledge on this, I'd greatly appreciate it.
1) Calculate the current solution Xb = Yo satisfying BYo = b,
2) Solve lambda(transpose)B = Cb(transpose), and set Rd(transpose) = Cd(transpose) - lambda(transpose)D. If Rd >= 0, optimal solution is found.
3) Determine vector to move into basis B and out of D.
4) Check for unboundedness and determine pivot element.
5) Update B. Return to step 1.
Ok, so hopefully you understand the terminology. The problem I'm having is that the whole purpose is to keep from using an inverse, such as B^-1. So my problem is in step two. How do you solve lambda(transpose)B = Cb(transpose) when you know B and Cb(transpose), don't know lambda, and you aren't supposed to use B inverse?
Can someone explain this to me? Maybe I'm misunderstanding what I'm supposed to do, so if anyone has some knowledge on this, I'd greatly appreciate it.
