How can we declare initial feasible points while solving an optimization problem in cvx?
Starting (initial) values can not be used by CVX to solve a problem. I’m not aware of any primal-dual solvers which can use starting values (that includes all the solvers which CVX can call). If you know of any primal-dual solvers which can make use of starting values, please let us know.
If you have starting values which you think might help a solution, such as solving a QP by an active set method, you’ll have to use a different tool (for instance, you can call cplexqp directly under MATLAB and provide starting values for the primal variables) - whether those will help, especially for convex QPs, is another matter.