TVM solve for interest rate, revisited
|
06-19-2022, 06:57 PM
Post: #18
|
|||
|
|||
RE: TVM solve for interest rate, revisited
(06-10-2022 08:25 PM)Albert Chan Wrote: g = n*x / ((1+x)^n-1) = 1 - (n-1)/2*x + (n²-1)/12*x² - (n²-1)/24*x³ + ... We relied on above identity to prove g'' > 0 ⇒ f'' has same sign throughout. It is easy to show with plots (pick any n>1), but harder to proof. (Perhaps I missed something simple ...) I had placed this question of proving it here (response welcome) I managed to proof it, using XCAS. Let R = 1+x > 0, to prove: g = n*(R-1)/(R^n-1) ≥ 1 - (n-1)/2*(R-1) → Or, n ≥ (1 - (n-1)/2*(R-1)) * (R^n-1)/(R-1) // note: multiplied factor > 0 (RHS, R)' = (1 - n*(n+1)/2*R^(n-1) + (n²-1)*R^n - n*(n-1)/2*R^(n+1)) / (R-1)^2 If n > 1, numerator have 3 sign changes, 1 or 3 positive roots (Descartes rules of signs) Taking taylor series, we showed that it had triple equal roots, at x=0 XCAS> M := (1 - n*(n+1)/2*R^(n-1) + (n²-1)*R^n - n*(n-1)/2*R^(n+1)) XCAS> series(M(R=1+x), x, 0, 4, polynom) (n/6-n^3/6)*x^3 + (-n/4+n^2/8+n^3/4-n^4/8)*x^4 2 roots get cancelled by denominator, x^2, we have only 1 extremum, at x=0 Taking limit at x=0, we have LHS = g(0) = 1 RHS = 1 - (n-1)/2*0 = 1 → LHS = RHS For x≠0, two sides don't touch. Example, pick x=∞ LHS = g(∞) = 0 (decay to nothing) RHS = 1 - (n-1)/2*∞ = -∞ → LHS > RHS For x > -1, LHS ≥ RHS QED |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 1 Guest(s)