Prev: Save a plot in A3 size
Next: Least squares with ill-conditioned mxn matrix
From: Mateusz on 13 Jul 2010 06:55
Hi,

I'm implementing an Call Admission Control algorithm using Markov Decision Processes. In the end I have to maximize the following objective function:

Maximize: sum_x(sum_a(T(x,a)r(x,a)H(x,a))), where H is feasible CAC policy

Subject to:

sum_x(sum_a(T(x,a)H(x,a))) = 1;
sum_a(H(y,a)) = sum_x(sum_a( P(y|x,a) H(x,a)));

H(x,a) > 0, and x,a are all admissable

The question is - How do I define the input to linprog ? I'm aware that I've to find the minimum of the negative objective function (so actually i have the maximum).

I want to know how to define f,A,b. My thought was:
f = [ sum_x(sum_a(T(x,a)r(x,a)))];
A = [sum_x(sum_a(T(x,a))) ; sum_x(sum_a(P(y|x,a)))];
b = [0 1];

I've I'm wrong please correct me.

Best Mateusz
 | 
Pages: 1
Prev: Save a plot in A3 size
Next: Least squares with ill-conditioned mxn matrix