Analizziamo l'intervallo di dato che la loro media aritmetica (AM) è un piccolo multiplo 1 + δ della loro media geometrica (GM) (con δ ≥ 0 ). Nella domanda, δ ≈ 0,001 ma non sappiamo n .x1≤x2≤⋯≤xn1+δδ≥0δ≈0.001n
Poiché il rapporto di questi mezzi non cambia quando si cambiano le unità di misura, selezionare un'unità per la quale GM è . Pertanto, cerchiamo di massimizzare x n soggetto al vincolo che x 1 + x 2 + ⋯ + x n = n ( 1 + δ ) e x 1 ⋅ x 2 ⋯ x n = 1 .1xnx1+x2+⋯+xn=n(1+δ)x1⋅x2⋯xn=1
x1=x2=⋯=xn−1=x, say, and xn=z≥x. Thus
n(1+δ)=x1+⋯+xn=(n−1)x+z
and
1=x1⋅x2⋯xn=xn−1z.
The solution x is a root between 0 and 1 of
(1−n)xn+n(1+δ)xn−1−1.
It is easily found iteratively. Here are the graphs of the optimal x and z as a function of δ for n=6,20,50,150, left to right:
As soon as n reaches any appreciable size, even a tiny ratio of 1.001 is consistent with one large outlying xn (the upper red curves) and a group of tightly clustered xi (the lower blue curves).
At the other extreme, suppose n=2k is even (for simplicity). The minimum range is achieved when half the xi equal one value x≤1 and the other half equal another value z≥1. Now the solution (which is easily checked) is
xk=1+δ±δ2+2δ−−−−−−√.
For tiny δ, we may ignore the δ2 as an approximation and also approximate the kth root to first order, giving
x≈1+δ−2δ−−√k; z≈1+δ+2δ−−√k.
The range is approximately 32δ−−−√/n.
In this manner we have obtained upper and lower bounds on the possible range of the data. We have learned that they depend heavily on the amount of data n. The upper bound shows the range can be appreciable even for tiny δ, thereby improving our sense of just how close to each other the data points really need to be--and placing a lower limit on their range, too.
Similar analyses, just as easily carried out, can inform you--quantitatively--of how tightly clustered the xi might be in terms of any other measure of spread, such as their variance or coefficient of variation.
x=c(-5,-5,1,2,3,10); prod(x)^(1/length(x))
[1] 3.383363
(mentre la media aritmetica è 1)