|
|
Re: My comments on "traditional" algorithms
Posted:
Apr 6, 2000 3:18 PM
|
|
What I should have said before going to the calculator was...
1.5245 is something like 1 and 2/3 percent bigger than 1.5
so if I have 1.5% compounded interest for 20 compunding periods,
how much is that?
old rule of thumb -- from interest rate calculations --
doubling time for k% interest is
about 70/k. so 70/(5/3) = 42.
This means 20 periouds should have about sqrt(2) error....
40%. And indeed, (1.5245)^20 is about 40% greater than 1.5^20.
OK, I could have used logs and the approximation
ln(1+x) ~ x for x near 0.
or the first terms of the taylor series
or the binomial expansion.
Ted Alper
|
|
