Suppose U is set of objects, E is a set of {0,1}-valued parameters 6 c! M8 G8 [# l& p- a! @ x ) q5 c- W8 |7 a6 Hfor describing objects in U. For any u in U, define an additive utility $ P9 D" Y) U/ D* L6 G9 T! N9 Y+ k2 z0 r6 l$ S4 G h# H
function f as follows: / M M' e( z3 s3 D
( u: i' P9 N9 V3 Y8 \9 \( l) x
f (u ) e (u ), (对e属于E,e(u)求和) , J8 `6 |7 i# i t i 6 G% ~" K2 u4 b \0 K e E % d5 {, i) L* l$ ` B s: a" I. d. k4 \6 }: V) O% g3 g6 I
where e(u ) 0,1. u is called an optimal solution if it is one of the % U! j3 k+ g5 K2 y
D' Y8 _; F5 {maximum points of function f with respect to normal order. For . N# p( M7 p9 @$ [
% M* n. R, k& N- H! ncertain reasons, some values are missing. It costs if we want to find 3 ]) v4 S; S" q" O 4 k! M) @+ n9 W5 ^out what these values are. We assume that we know nothing about $ u; _ m6 b) u
% E: a0 W0 k7 t/ F: o* X" tthe probability of these values being 0 or 1. So my questions are: 5 Y: A, a; [& I% h) g( a) L8 {
3 A5 H3 l% A& r, W- W5 j
(1.) Which unknown value should we figure out firstly if we want to ( U5 j1 ~' z6 T( E4 u! o* F
' [( {: }/ _4 k
find at least one optimal solution?