the cutting defined by the cut point x 0 . Thus, V 1 <strong>and</strong> V 2 are nonempty, disjointρ w ∗-open sets such thatF ∗ (z 0 )\{x 0 } = V 1 ∪ V 2 ,<strong>and</strong> V 1 ∪ {x 0 } <strong>and</strong> V 2 ∪ {x 0 } are subcontinua of F ∗ (z 0 ) (see Willard, 1970, section28). Note that because x 0 /∈ Γ(z 0 ),forsomeε 0 > 0, the open ball of radius ε 0centered at x 0 is such that B ε 0(x 0 ) ∩ Γ(z 0 )=∅. With this in mind, define the USCOΓ ∗ (z) :=Γ(z) ∩ F ∗ (z),<strong>and</strong> observe that at z 0 Γ ∗ (z 0 ) is the union of two nonempty closed sets,whereΓ ∗ (z 0 )=F 1 ∪ F 2 ,F 1 ⊂ V 1 <strong>and</strong> F 2 ⊂ V 2 .Because η(z 0 ) ⊂ F ∗ (z 0 ) <strong>and</strong> F ∗ (z 0 )=κ(η(z 0 )) is the unique subcontinuum irreducibleabout η(z 0 ),wehaveatz 0 ,η(z 0 )=n(K(z 0 )) = n 1 (K(z 0 )) ∪ n 2 (K(z 0 )),where n 1 (K(z 0 )) ⊂ F 1 ⊂ V 1 <strong>and</strong> n 2 (K(z 0 )) ⊂ F 2 ⊂ V 2 . By Lemma 11 aboven(K(z 0 )) is minimally essential for N K(z 0 )(K(z 0 )) in S K(z 0 ). Therefore, because eachof the closed sets, n 1 (K(z 0 )) <strong>and</strong> n 2 (K(z 0 )), is a proper subset of n(K(z 0 )), neithern 1 (K(z 0 )) nor n 2 (K(z 0 )) are essential for N K(z 0 )(K(z 0 )) in S K(z 0 ). Therefore, thereare two nonempty, open sets G 1 <strong>and</strong> G 2 withn 1 (K(z 0 )) ⊂ G 1 <strong>and</strong> n 2 (K(z 0 )) ⊂ G 2such that for all δ > 0, there exists Ky Fan sets in S K(z 0 ), E δ1 <strong>and</strong> E δ2 inB hw ∗ ×w ∗ (δ,K(z 0 )) ∩ S K(z 0 ) such thatN(E δ1 ) ∩ G 1 = ∅ <strong>and</strong> N(E δ2 ) ∩ G 2 = ∅.Let U 1 = V 1 ∩ G 1 <strong>and</strong> U 2 = V 2 ∩ G 2 .WehaveU 1 <strong>and</strong> U 2 disjoint open sets suchthat n 1 (K(z 0 )) ⊂ U 1 <strong>and</strong> n 2 (K(z 0 )) ⊂ U 2 <strong>and</strong> for all δ > 0, thereexistE δ1 <strong>and</strong> E δ2in B hw ∗ ×w ∗ (δ,K(z 0 )) ∩ S K(z 0 ) such thatN(E δ1 ) ∩ U 1 = ∅ <strong>and</strong> N(E δ2 ) ∩ U 2 = ∅. (67)Given that the sets N(E δi ) are compact, under [A-1], there exists open sets W 1 <strong>and</strong>W 2 such that for i =1, 2,n i (K(z 0 )) ⊂ W i ⊂ W i ⊂ U i .Thus,wehaveforallδ > 0, E iδ ∈ B hw ∗ ×w ∗ (δ,K(z 0 )) ∩ S K(z 0 ) such thatN(E δ1 ) ∩ W 1 = ∅ <strong>and</strong> N(E δ2 ) ∩ W 2 = ∅. (68)38
Now we have a contradiction: First, because n(K(z 0 )) is a minimal essential setfor N(K(z 0 )) in S K(z 0 ) <strong>and</strong> because n(K(z 0 )) ⊂ £ W 1 ∪ W 2¤ , there exists a positivenumber δ ∗ > 0 such that for all E ∈ B hw ∗ ×w ∗ (δ ∗ ,K(z 0 )) ∩ S K(z 0 ),N(E) ∩ £ W 1 ∪ W 2¤ 6= ∅. (69)But because δ > 0 can be chosen arbitrarily, choosing δ = δ∗ 3, we have by (68) <strong>and</strong>the 3M property, the existence of a Ky Fan setE ∈ B hw ∗ ×w ∗ (3 δ∗ 3 ,K(z0 )) ∩ S K(z 0 ) = B hw ∗ ×w ∗ (δ ∗ ,K(z 0 )) ∩ S K(z 0 )such thathN(E) ∩ W 1 ∪ W 2i = ∅.Thus we must conclude that F ∗ (z) ⊆ N (z) for all z ∈ Z.(3) Finally, we will show that F ∗ (z) ∈ M N (z) (C w ∗ f(X)) for all z ∈ Z. Becauseη(z) ⊂ F ∗ (z) for all z ∈ Z <strong>and</strong> η ∈ [N ] ρZ -w ∗, we know that F ∗ (z) ∈ E N (z) (C w ∗ f(X))for all z ∈ Z.Let z 0 be any point in Z <strong>and</strong> Let C 0 ∈ C w ∗ f(X) be a proper subset of F ∗ (z 0 ) <strong>and</strong>consider the nonempty set, F ∗ (z 0 )\C 0 . Because F ∗ (z 0 ) is the unique subcontinuumirreducible about η(z 0 ),η(z 0 ) ∩ [F ∗ (z 0 )\C 0 ] 6= ∅.Let x 0 ∈ η(z 0 )∩[F ∗ (z 0 )\C 0 ] <strong>and</strong> let B ε 0(x 0 ) be an open ball of radius ε 0 > 0 about x 0such that B ε 0(x 0 ) ∩ C 0 = ∅. Therefore, η(z 0 )\B ε 0(x 0 ) ⊂ η(z 0 ) <strong>and</strong> η(z 0 )\B ε 0(x 0 ) ∈P w ∗ f(X). Becauseη(z 0 ) ∈ M N (z 0 )(P w ∗ f(X))there exists some eε > 0 such that for all n, thereexistsz n ∈ B 1 (z 0 ) such thatnBut for some n 0 ,N (z n ) ∩ B w ∗(eε, η(z 0 )\B ε 0(x 0 )) = ∅. (70)N (z 0 ) ∩ B w ∗(eε, η(z 0 )) 6= ∅ for all z 0 ∈ B 1n 0 (z0 ). (71)Because B ε 0(x 0 ) ∩ C 0 = ∅, given (70) <strong>and</strong> (71), we must conclude that for all n, thereexists z n ∈ B 1 (z 0 ) such thatnwhereas, for some n 00 ,N (z n ) ∩ B w ∗(eε,C 0 )=∅,N (z 00 ) ∩ B w ∗(eε,F ∗ (z 0 )) 6= ∅ for all z 00 ∈ B 1 (z0 ).n 00Therefore, for any proper subset C 0 of F ∗ (z 0 ) in C w ∗ f(X), C 0 is not essential forN (z 0 ), implying therefore thatF ∗ (z) ∈ M N (z) (C w ∗ f(X)) for all z ∈ Z.39
- Page 5: cally approximated by continuous fu
- Page 12 and 13: the set x 0 x 1 ∈ C w ∗ f(X) is
- Page 14 and 15: defined on some probability space,
- Page 16 and 17: contains at most one arc (i.e., ¯
- Page 18 and 19: into the collection of Ky Fan sets
- Page 20 and 21: 3.1 Best Response MappingsLetting p
- Page 22 and 23: 4.1 Nikaido-Isoda FunctionsWith eac
- Page 24 and 25: Because Φ(z) × Φ(z) is w ∗ ×
- Page 26 and 27: where for all n, C n ∈ S E n,then
- Page 28 and 29: Definition 4 (The 3M Property - The
- Page 30 and 31: Given (48) and the fact that for n
- Page 32 and 33: We note that if m(z) ∈ P w ∗ f(
- Page 34 and 35: 6.3 AK Convergence of Minimal Nash
- Page 36 and 37: and Gr ρZ ×w ∗f ∗ (·) is the
- Page 38 and 39: In particular, for all n ≥ N 0 an
- Page 42 and 43: According to our main result, under
- Page 44 and 45: By our main approximation result, f
- Page 46 and 47: 9 Appendix 1: USCO FundamentalsIn t
- Page 48 and 49: 9.3 Equi-QuasicontinuityIn order to
- Page 50 and 51: 9.5 Dense SelectionsFor each F ∈
- Page 52 and 53: 10 Appendix 2: The Proof of Lemma 8
- Page 54 and 55: Noting that if E ∈ D eE ,thenn eE
- Page 56 and 57: 12 Appendix 4: The Proof That All K
- Page 58 and 59: Letting E 1 =[E 1 \(X × U 2 )] ∪
- Page 60 and 61: [10] Bryant, V. W. (1970) “The Co
- Page 62: [42] Ward, L. E., Jr. (1958) “A F