Forums  > Risk Management  > Can I get some advice about optimizing this algorithm?  
Page 1 of 1
Display using:  


Total Posts: 11
Joined: Jan 2016
Posted: 2017-02-17 13:04
Know values: matrix(n x m)—— X,F and Delta; vector—-u;

wanted value: vector—— w;

restrictive condition:

Now shall I set vector — ‘w’ to get the biggest value of R/P?


Total Posts: 1583
Joined: Sep 2006
Posted: 2017-02-17 13:17
Is R correctly specified? You are trying to add a vector to a matrix multiplication?


Total Posts: 11
Joined: Jan 2016
Posted: 2017-02-18 02:01
thanks for your reply;
Yes, I want to get a vector 'w' to make 'R/P' biggest.


Total Posts: 16
Joined: Jul 2009
Posted: 2017-02-21 14:59
Seems like you simply have R = b w and P = sqrt(w^T C w). Their ratio is invariant to re-scaling of w, so you can always choose w so that P = 1. Then you have to maximize R with fixed P which is Lagrange optimization problem: b w + l (w^T C w - 1), where l is Lagrange multiplier (note P^2 to get rid of square root). This is straightforward.
Previous Thread :: Next Thread 
Page 1 of 1