【CVX】SDP and conic form problems

Basic model

LMIs and SDPs with one variable. The generalized eigenvalues of a matrix pair $(A,B)$, where $A,B\in {\mathcal{S}}^n$, are defined as the roots of the polynomial $\det (\lambda B-A)$.
Suppose $B$ is nonsingular, and that $A$ and $B$ can be simltaneously diagonalized by a congruence, i.e., there exists a nonsingular, and that $A$ and $B$ can simultaneously diagonalized by a congruence, i.e, there exists a nonsingular $R\in {\mathbb{R}}^{n\times n}$ such that
$$\begin{gathered} R^{T}AR=diag(a) \\ {R}^{T}BR=diag(b) \end{gathered}$$
where $a,b\in {\mathbb{R}}^n$.
Moreover, a sufficient condition for this to hold is that there exists $t_1,t_2$ such that $t_{1}A+t_{2}B\succ 0$.

Try to find the optimal $t$ that would maximize $c\ast t$ while $A-t\ast B$ is still P.S.D, we could solve the following SDP:
$$\begin{gathered} \min c\ast t \\ s.t.tB\prec A \end{gathered}$$

cvx code

%% generate input data
rng(1);
n = 4;
A = randn(n); A = 0.5*(A'+A);
B = randn(n); B = B'*B;
c = -1;

%% cvx
cvx_begin sdp
    variable t
    minimize c*t
    t*B<=A;
cvx_end

disp('the optimal t obtained is');
disp(t);

yalmip code

%% yalmip
t = sdpvar;
F = [t*B<=A];
obj = c*t;
optimize(F, obj);

disp('the optimal t obtained is');
disp(value(t));

A simple QP with inequalities constraints

Prove that $x^{\ast }=(1,1/2,-1)$ is optimal for the optimization problem
$$\begin{gathered} \min (1/2)x^{T}Px+q^{T}x+r \\ s.t.-1\le x_i\le 1,i=1,2,3 \end{gathered}$$
where
$$P=\left[\begin{array}{ccc}13& 12& -2\\ 12& 17& 6\\ -2& 6& 12\end{array}\right]q=\left[\begin{array}{c}-22.0\\ -14.5\\ 13.0\end{array}\right]r=1$$

cvx code

%% cvx: QP
cvx_begin
    variable x(n)
    minimize ((1/2)*quad_form(x, P)+q'*x+r);
    x>=-1;
    x<=1;
cvx_end

yalmip code

%% yalmip
x = sdpvar(n, 1);
F = [-1<=x<=1];
obj = (1/2)*x'*P*x+q'*x+r
optimize(F, obj);

Reference

Convex Optimiztion S.Boyd Page 203, 215