2019

monotonic
correspondence
compose
dubious
assume

$$ee$$

optimization over atomic sets

Gauge function: for any compact and convex subset $D$ of a vecter space X, the gange fnction of the convex set is given by

$$Omega_D (X) := inf_t {tvert x belong D}$$

Support function of $D$ is given by

$$% <![CDATA[ Omega_D^*(Y):= sup^{s belongto D} <s,y> %]]>$$

$l1$ norm can take care of sparse data

norm : singular vecter ‘s 1 norm <–> largest eigenvalue

Fundamentals of Unconstrain optimization

Ky Fan-k norm

$$lVert M rVert _k = sum^k_{i=1} sigma_i(M)$$
$$f(M):= lVert M$$