n*******n
发帖数: 407 | 1 Let X be a Poisson variable with parameter λ with a probability mass
function, f(k), where k = 0, 1, 2 … We know the index of log-concavity is
the function rf(k) = f(k)^2/(f(k-1)f(k+1) = (k+1)/k>1. So, Poisson is log-
concave.
Many books have: "The random variable Y is dispersive if, and only if, Y has
a logconcave density."
Poisson variable is discrete. Do we have: A Poisson variable X is dispersive?
Or, more specifically, let X be a Poisson variable with parameter λ1 with a
probability mass function. Let Y be a Poisson variable with parameter λ2
with a probability mass function. Let λ2 > λ1. Is Y more dispersive than X?
Any reference? | n*******n
发帖数: 407 | 2 Dispersive order: Let X and Y be random variables with quantile functions F-
1 and G-1 respectively. If F-1(b)- F-1(a) <= G-1(b)- G-1(a) whenever 0
< 1; then X is said to be smaller than Y in the dispersive order. | T*******x
发帖数: 8565 | 3 从dispersive order的定义看，泊松分布的lambda越大，应该越dispersive。
F-
【在 n*******n 的大作中提到】
: Dispersive order: Let X and Y be random variables with quantile functions F-
: 1 and G-1 respectively. If F-1(b)- F-1(a) <= G-1(b)- G-1(a) whenever 0 : < 1; then X is said to be smaller than Y in the dispersive order.
|
|
