This lecture note gives the proof for how deterministic subgradient method converges on non-smooth convex and strongly convex Lipschitz functions.

Is there a non-convex version of this proof?

Like convergence of subgradient method for,

  • non-smooth Lipschitz non-convex functions (like $\frac{1}{1+e^{-\vert x \vert}}$) ?
  • non-smooth non-Lipschitz non-convex functions (like $(1-\vert x \vert)^2$)?

(happy to get references for the the stochastic versions of these too!)

  • $\begingroup$ On non-smooth optimization, better ask at Math SE, I would guess. There's a tag for it there. $\endgroup$ Mar 23, 2021 at 4:35
  • $\begingroup$ Okay. If I dont get answers in a couple of days, I will move it there :D $\endgroup$ Mar 23, 2021 at 4:54


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.