3.8 Tune fixed steplength for gradient descent Take the cost function g(w) = ww (3.45) where w is an N = 10 dimensional input vector, and g is convex with a single global minimum at w = 0NX1. Code up gradient descent and run it for 100 steps using the initial point w0 = 10 1Nx1, with three steplength values: a = 0.001, a2 = 0.1, and 3 = 1. Produce a cost function history plot to compare the three runs and determine which performs best.