# Implementing the conjugate gradient descent for the example function

# start simple gradient descent with random guess again
x = np.random.randn(2,1)

# The residual
r = b.T-A @ x
v = r
c = r.T @ r
i = 0
while (i <= IterMax):
    z = A @ v
    t = c/(v.T @ z)
    x += t*v
    r -= t*z;
    d = r.T @ r
    if(np.sqrt(d) < tolerance):
      break
    v = r + (d/c)*v
    c = d
    i+=1

print("Absolute error for the Conjugate gradient solution",np.abs(x-exact))