Consider a convex function with a unique global minimum. If we initialize gradient descent at a point far from the minimum and use a fixed learning rate, which of the following is most likely to happen as the number of iterations increases?

Question $4$Answer

A$.$

The algorithm will diverge.

B$.$

The algorithm will converge to the global minimum at a constant rate.

C$.$

The algorithm will initially converge quickly, then slow down as it approaches the minimum.

D$.$

The algorithm will oscillate around the minimum without converging.