How can calculation this problem

(masoud) #1

How to write this Matlab code

my code:
variable y(n) ;
variable s ;

for n=1:K

for n=1:K



maximize(SUM) ;
subject to
ST <= (1-s);

and error is :
Error using .* (line 262)
Disciplined convex programming error:
Invalid quadratic form(s): not a square.

Error in * (line 36)
z = feval( oper, x, y );

Error in convex (line 42)

(Mark L. Stone) #2

Do you have a proof that the constraint is convex (some kind of second order cone thing?)? I will presume it is non-convex unless you show otherwise, in which case the non-convex designation will be removed (edit: the non-convex designation has now been removed).

Per the CVX Users’ Guide and Why isn’t CVX accepting my model? READ THIS FIRST! , you are not allowed to enter the constraint as you have. And if it is somehow convex, reformulation will be required for entry into CVX.

(masoud) #3

is jointly convex and proof in the paper


a problem (32) is jointly convex y_j * s and s
a problem (32) is jointly convex y_j and s
and finally solved by MOSEK cvx

(masoud) #4


maximize(sum(log(1+M.*y)/log(2))) ;
subject to

sum(((Ek B(n). s+(T-K). y-(sqrt(Ek^2 B(n)^2. s^2-2 (T-K) (Ek B(n)+2).*y(n).*s+(T-K)^2. y(n)^2)))/(2 (T-K)))) <= (1-s);

(masoud) #5


(Mark L. Stone) #6

The paper appears to be , for which there was a later IEEE publication to which I don’t have access.

Even presuming the joint convexity proof (in the transformed variables) is correct, there is no concrete formulation provided in a form suitable for entry into CVX. I suggest you contact the authors and request them to provide you with their CVX code and allow you to post if here for the benefit of forum readers.