How is this algorithms used in CG?

Are there optimizers that are better than gradient descent?

How does this deal with local maxima that gradient ascent can get stuck at, or is it good enough usually?