Michael K Gray wrote:Ah, I see now. The Range was assumed to be artificially restricted to x>=0.Spence wrote:The problem I find is that most people do not find stats intuitive so taking shortcuts like this abstracts from what is actually being done and generally more mistakes get made (in my experience).Michael K Gray wrote:Eh???disumbrationist wrote:...the function 1-(1-x)^n can be approximated by n*x if x is near zero and n << 1/x...
I'd put good money on that what you say is not true.
My immediate gut reaction on seeing this supposed equivalence is that it fails spectacularly when x is negative.
(Say x=-0.01 & n=2000; 1 - (-1 + x)^20=-4.39286*10^8 & n*x=-20. Error=-4.39286*10^8. !!!)
Damn fools, take into account Goldmann's coefficient of exponentials, and the minimum possible reciprocate of a polynomial bifurcation can only be a real number greater than pi(^12).
Unless you're stuck in the 18th Century, ll such distributions must be factored in 8 dimensions, and thus the harpsichord's strings are superficially stronger than a spider's leg hair.