Just apply the definitions.
Local truncation error at (t_n,y_n):
|y_(n+1) - y(t_(n+1))|
where y(t_(n+1)) is the exact solution to the problem
dy/dt = f(t,y), y(t_n) = y_n
at
t = t_(n+1)
Global truncation error:
|y_num(b) - y(b)|
where y(b) is the value of the exact solution to the problem
dy/dt = f(t,y), y(a) = y0
and y_num(b) is the approximate solution obtained by the numerical method.
Hint: The exact solution to your problem can easily be obtained by separation of variables.
