Model reference adaptive control question

Hi, dear friends,

here is a question in model reference adaptive control

plant: y(t) = a*y(t-1)+u(t-1)
model: y_m(t) = a_m*y_m(t-1) + u_m(t-1)
control: u(t) = -theta(t)*y(t) + u_m(t-1)

adaptation:
                       y(t-1)*e(t)
theta(t)=theta(t-1)+ -----------------
                       1+y(t-1)^2
where e(t) is the equation error

the question is to show

       |e(t)|
------------------------ is uniformly bounded and tends to zero
   (1+y(t-1)^2)^(1/2)


I dont have even a clue for this question. Plz give me a hint. Thanks