dual of $min_x||x-2||$ subject to $||x||_* \leq 1$

41 Views Asked by At

I want to find dual problem of

$min_x||x-2||$ subject to $||x||_* \leq 1$

Here $x \in R^n$, $2$ denotes vector of $2s$, $||$ is arbitrary norm and $||_*$ is dual norm.

So I started with this:

$L(x,v) = ||x-2|| + v(||x||_*-1||)$

$g(v) = inf_x( ||x-2|| + v(||x||_*-1||))$

I don't know how to minimize this expression with respect to $x$

Reply to LinAlg's comment:

$min||y||$ subject to

$||x||_* \leq 1$, $y = x-2$

$L(y,v,k) = ||y|| + v(||x||_*-1)+k(y-x+2)$

$g(v,k) = inf_y(||y|| + v(||x||_*-1)+k(y-x+2))$

To minimize function above I need to minimize $||y|| + ky$ where $k \succeq 0$, but it seems to be unbounded below.