Can quad_over_lin solve this problem?

[HaoGong]

How to deal with the subject function X^3/Y^2, where X and Y are variables.

[Mark_L_Stone]

Presuming X >= 0, Y > 0, here is my CVX implementation of the formulation produced by @Michal_Adamaszek earlier in that thread.

How can I write this kind of constraint in cvx

As shown above by @Michal_Adamaszek xy^2\geq z^3 with x,y,z\geq 0 is equivalent to sy\geq z^2,\ xz\geq s^2. The latter can be written in CVX as variables x y z s {z,s,y} == rotated_lorentz(1) {s,x,z} == rotated_lorentz(1) Then use x in place of z^3/y^2

[HaoGong]

Thank you for your help! I have noticed it, but i don’t know how to use it in subject function.

[Mark_L_Stone]

Do you understand it now, or do you know what you need to do?

In that link,
my z is your X
my y is your Y
my x is a new variable which you can name what you want
my s can be your s.

[HaoGong]

Thanks! I know the relationship between the variables in that link and my variables, but i still don’t know how to write the cvx code here,i.e., ‘minimize (X^3/Y^2)’.

[Mark_L_Stone]

In my link, I said to use my x in place of z^3/y^2, so that would be minimize(x). Now just change the variable names to your code.

[HaoGong]

Thanks! I have dealt with this problem.But i have a new question that the first few times the program runs the results are ‘solved’ but suddenly the following results appear: Status: Infeasible
Optimal value (cvx_optval): +Inf. I want to know why this happened.

[HaoGong]

[Mark_L_Stone]

We have no idea what you ran, to include the relation between the various runs you made. Are you using SCA - if so, read my posts on this forum about it.

[HaoGong]

Yes, I used SCA, i want to know which posts may help me. Thanks!

[Mark_L_Stone]

http://ask.cvxr.com/search?q=sca%20order%3Alatest

Note: some of these have the word scalar or scaling in them, and do not pertain to SCA.