Stolarsky mean
Definition
For two positive real numbers x and y , the Stolarsky mean is defined as:
S p ( x , y ) = { x , if x = y , ( x p − y p p ( x − y ) ) 1 / ( p − 1 ) , otherwise .
Derivation
It is derived from the mean value theorem , which states that a secant line , cutting the graph of a differentiable function f at ( x , f ( x ) ) and ( y , f ( y ) ) , has the same slope as a line tangent to the graph at some point ξ in the interval [ x , y ] .
∃ ξ ∈ [ x , y ] f ′ ( ξ ) = f ( x ) − f ( y ) x − y
The Stolarsky mean is obtained by
ξ = [ f ′ ] − 1 ( f ( x ) − f ( y ) x − y )
when choosing f ( x ) = x p .
Special cases
lim p → − ∞ S p ( x , y ) is the minimum .
S − 1 ( x , y ) is the geometric mean .
lim p → 0 S p ( x , y ) is the logarithmic mean . It can be obtained from the mean value theorem by choosing f ( x ) = ln x .
S 1 2 ( x , y ) is the power mean with exponent 1 2 .
lim p → 1 S p ( x , y ) is the identric mean . It can be obtained from the mean value theorem by choosing f ( x ) = x ⋅ ln x .
S 2 ( x , y ) is the arithmetic mean .
S 3 ( x , y ) = Q M ( x , y , G M ( x , y ) ) is a connection to the quadratic mean and the geometric mean .
lim p → ∞ S p ( x , y ) is the maximum .
Generalizations
Additional parameters
The standard two-parameter version extends the one-parameter case with α = p , β = 1 : S α , β ( x , y ) = { x , if x = y , ( β ( x α − y α ) α ( x β − y β ) ) 1 / ( α − β ) , otherwise . (Note that the names of the parameters can vary, and some authors reverse their order.) This extra parameter allows for additional special cases; e.g.:
The r th power mean has α = r and β = 2r .
The harmonic mean has α = -2 and β = -1 .
This can be generalized further to have any number of pairs of parameters a 1 , … , a k , b 1 , … , b k (where Σb = 0 ?) by making the result ( x a 1 − y a 1 a 1 ) b 1 × ⋯ × ( x a k − y a k a k ) b k . The two-parameter case results from a 1 = α , a 2 = β , b 1 = 1 α − β , b 2 = 1 β − α .
More than two variables
There are various different approaches for extending the mean to n variables, such as:
The x-value where the n -1 st derivative equals the divided difference (a generalization of slope) of the values on the p th-degree monic power function : using the mean value theorem for divided differences , S p ( x 1 , … , x n ) = f ( n − 1 ) − 1 ( ( n − 1 ) ! ⋅ f [ x 1 , … , x n ] ) where f ( x ) = x p , or for the two-parameter version, c such that g [ x 1 , … , x n ] f [ x 1 , … , x n ] = g ( n ) ( c ) f ( n ) ( c ) where f ( x ) = x β and g ( x ) = x α , using ln(x ) to replace zero divided by zero if α or β is in ℤ ∩ [ 0 , n ) .
( ∫ ( ∑ i = 1 n u i x i β ) α − β β d 𝐮 ) 1 α − β , where the integral is over a Dirichlet measure , and if it is flat (all concentration parameters equal 1), this is equal to ( ( ∑ i = 1 n x i α + ( n − 2 ) β ∏ j ∈ ℤ ∩ [ 1 , n ] ∖ { i } ( x i β − x j β ) ) ∏ i = 2 n ( i − 1 ) β α + ( i − 2 ) β ) 1 α − β .
Based on a ratio of logarithmic means (for a particular way of extending it): [ ( β / α ) n − 1 ∑ i = 1 n x i α / ∏ j ∈ ℤ ∩ [ 1 , n ] ∖ { i } ( ln ( x i ) − ln ( x j ) ) ∑ i = 1 n x i β / ∏ j ∈ ℤ ∩ [ 1 , n ] ∖ { i } ( ln ( x i ) − ln ( x j ) ) ] 1 α − β .
( ( β α ) 2 ( ∑ i = 1 n p i x i 2 α ) − ∏ i = 1 n x i p i 2 α ( ∑ i = 1 n p i x i 2 β ) − ∏ i = 1 n x i p i 2 β ) 1 2 α − 2 β where the standard two-variable case has p 1 = p 2 = 1/2 .