Re: Dynamical Systems and Expansion-Contraction
- From: "OsherD" <mdoctorow@xxxxxxxxxxx>
- Date: 28 Apr 2005 12:43:22 -0700
To compare Gumbel's bivariate exponential and Gumbel's bivariate
logistic cdfs, respectively:
1) F(x, y) = 1 - exp(-x) - exp(-y) + exp{-(x + y + axy)}
2) F1(x, y) = (1 + exp(-x) + exp(-y))^(-1)
we can use a few "tricks". Notice that in (1) we have x and y positive
and a in [0, 1], so F(x, y) has the form P(A-->B) = 1 + y - x with x
replaced by exp(-x) + exp(-y) and y replaced by exp{-(x + y + axy)}.
We know that we can make P(A-->B) = 1 + y - x arbitrarily close to 0 by
making x arbitrarily close to 1 and y arbitrarily close to 0 (why?),
under the condition that y < = x, and if we consider the ratio:
3) F(x, y)/F1(x, y)
the negative exponent -1 of F1 becomes a positive exponent 1 so that
the main parentheses term of F1 becomes greater than 1.
Translated into replacing x by exp(-x) + exp(-y) and replacing y by
exp{-(x + y + axy)}, we need to be able to choose exp(-x) + exp(-y)
near 1 and the second exponential near 0. If x and y are near 0.70,
then exp(-x) + exp(-y) is near 1, and if a is near 1 then the second
exponential is near 0.21 which is considerably closer to 0, and 1 times
0.21 is 0.21 < < 1 so the ratio in (3) is considerably less than 1 and
F(x, y) < F1(x, y). Note that a table of positive and negative
exponentials helps to visualize what is happening. In terms of
intervals, "near" above is in terms of "intervals near".
Osher Doctorow
.
- Follow-Ups:
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- References:
- Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Re: Dynamical Systems and Expansion-Contraction
- From: OsherD
- Dynamical Systems and Expansion-Contraction
- Prev by Date: Re: Black Holes
- Next by Date: Re: electron as point particle.
- Previous by thread: Re: Dynamical Systems and Expansion-Contraction
- Next by thread: Re: Dynamical Systems and Expansion-Contraction
- Index(es):
Relevant Pages
|
Loading