CM2 April 2022 Q4

Discussion in 'CM2' started by Constantinos Spyrou, Sep 12, 2023.

  1. Constantinos Spyrou

    Constantinos Spyrou Made first post

    Hello

    I kindly ask for an explanation in question 4 on the past paper of April 2022 . When trying to apply Ito's lemma why is the partial derivative with respect to time:
    -a'(T-t)f(t,Xt)?

    If 'a' is a differentiable function with respect to t why is the product rule not being applied here?
     
  2. Steve Hales

    Steve Hales ActEd Tutor Staff Member

    The function f is given as: exp(a(T-t)-x). The partial derivative with respect to t requires the chain rule because we're taking the exponential of a function.
    Let me know how you would implement the product rule.
     
  3. Constantinos Spyrou

    Constantinos Spyrou Made first post

    Thanks for your reply
    f(Xt,t)=exp(a(T-t)-Xt)
    Hence,
    partial(f)/partial(t)=a'(T-t)f(Xt,t) -af(Xt,t)


    Clearly I miss something here, since I don't understand why the chain rule kicks in.
     
  4. Steve Hales

    Steve Hales ActEd Tutor Staff Member

    Here's the chain rule (function-of-a-function rule): if f(t) = g(h(t)), then df/dt = dg/dh * dh/dt.
    In the current question we have h(t) = a(T-t) - x, and g(h(t)) = exp(h(t)). Then:
    partial(f)/partial(t) = dg/dh * dh/dt = f(t) * (-a'(T-t))
     
  5. Sunil Chaudhary

    Sunil Chaudhary Active Member

    Hi Steve,

    Here, I think I am also missing some thing as for me dh(t)/dt is coming as a'(T-t) + a(-1).
    Can you please point out what I am missing here.

    Thanks
     
  6. Steve Hales

    Steve Hales ActEd Tutor Staff Member

    Hi.
    You need the apply the chain rule to a(T-t), but it looks as though you've gone for the product rule instead. The question says that a is a differentiable function. Then, using the chain rule, we have:
    dh(t)/dt = dh(t)/da * da/dt = a'(T-t) * (-1) = -a'(T-t)
     
  7. Sunil Chaudhary

    Sunil Chaudhary Active Member

    So, a(T-t) is a differentiable function of (T-t).

    Further, will it make more sense if I write as below:
    dh(t)/dt = d[a(T-t) - x]/d[a(T-t)] * d[a(T-t)]/d[(T-t)] * d[T-t]/dt = 1 * a'(T-t) * (-1) = -a'(T-t)
     

Share This Page