Karni, Edi (contributor) - 2005
0 ⇔
X
s∈E
π(s | E)u
s
(x
s
) ≥
X
s∈E
0
π(s | E)u
s
(y
s
),
where, for all B ∈ E, π(s | B)=π(s)/
P
t∈B
π(t) is the …
<y
E
⇔
X
s∈E
π(s | E)u
s
(x
s
) ≥
X
s∈E
π(s | E)u
s
(y
s
).
Hence the probability measure π has the interpretation of a … Bayesian prior
and, sincex
E
<y
E
ifandonlyif
P
s∈E
π(s | E)u
s
(x
s
) ≥
P
s∈E
π(s | E)u
s
(y
s
),
π(·|E) is the posterior …