If you want to ensure that you will get stuck in the local optimum, use gradient descent.

→ View original post on X — @tunguz