Maximizing Norm over a convex set

[M.T]

Given a convex function f : \mathbb{R}^n \to [0,\infty), and assume that the set {x \in \mathbb{R}^n \mid f(x) \leq 1} is non-empty compact set. 
In YALMIP, my aim is to solve the following non-convex optimization problem:

maxmize ||x||_2
subject to
f(x) \leq 1

For such problem, using BMIBNB is advised since it can handle such optimization problems. When running the code using such solver, quickly a message pops regarding the usage of LMILAB, which was stated that it's not recommended to use such solver.
I must note that I do have MOSEK and GUROBI solvers installed on my machine, as well as CVX. How can use such solvers via BMIBNB?

Please advise.

Thanks in advance,
M.T

[Johan Löfberg]

You ould have to be a tad more explicit in the desciption for any reasonable advice

YALMIP has detected that you have a semidefinite programming problem, and thus selected the only SDP solver you have. Perhaps you did not intend to define a semidefinite constraints, and mistakingly defined a square matrix as symmetric instead of full, or you had a symmetric matrix which you wanted to constrain elementwise but added a psd constraint

https://yalmip.github.io/tutorial/basics/

[M.T]

What i have done, is that used a convex function defined as follows:
- Picked N radom points having d dimensions using rand (in MATLAB): 

P = rand(N,d);

- Defined f to be:

f = @(x)norm(P*x,1);

The problem is that it didnt use SDP, it used fmincon!

[M.T]

Also the constraints were 

Constraints = [f(x) <= 1];

[Johan Löfberg]

you talked about lmilab. YALMIP would never pick lmilab on this model, so you are not telling us everything. 

and if you have a vanilla installation, yalmip would not pick fmincon as upper bound solver in bmibnb, but quadprog by default

>> optimize(norm(x,1) <= 1, -x'*x,sdpsettings('solver','bmibnb'))

* Starting YALMIP global branch & bound.

* Branch-variables : 3

* Upper solver     : QUADPROG

* Lower solver     : LINPROG

* LP solver        : LINPROG

 Node       Upper      Gap(%)       Lower    Open

    1 :   -1.000E+00    66.67     -3.000E+00   2  Improved solution  

    2 :   -1.000E+00    66.67     -3.000E+00   3    

    3 :   -1.000E+00    66.67     -3.000E+00   4    

[Johan Löfberg]

...unless you actually used a sqrtm and formed the norm (bad idea, a quadartic function is much nicer)

>> optimize(norm(x,1) <= 1, -sqrtm(x'*x),sdpsettings('solver','bmibnb'))

* Starting YALMIP global branch & bound.

* Branch-variables : 4

* Upper solver     : fmincon