def Hessian return np matrix 1896 632 -632 -632 316 def LevenbergMarkv

def Hessian(x):
    return np.matrix([[1896*x[0]**2 - 632*x[1] + 4, -632*x[0]],
                     [-632*x[0], 316]])

def LevenbergMarkvardt(x, step = 0, m = 10000, eps = 0.01, M = 1000):
    d = grad(x)
    if np.linalg.norm(d) < eps or step > M:
        return f(x), x, step
    while True:
        d = -(Hessian(x)+m*np.ones([n, n])).I * d
        print(d.shape)
        print(d.transpose())
        new_x = x + d
        print(d, x)
        if f(new_x) < f(x):
            return LevenbergMarkvardt(np.array(new_x)[0], step+1, m / 2.)
        m *= 2