Variations on a game of Gale (III): remainder strategies (Q4382474)

[For Parts I and II see J. Symb. Log. 58, No. 3, 1035-1043 (1993; Zbl 0792.03038) and Theor. Comput. Sci. 129, No. 2, 385-396 (1994; Zbl 0809.03042), respectively.] From the text: An infinite set \(X\) is given. D. Gale, in correspondence with J. Mycielski, described the following game in which players ONE and TWO play an inning per positive integer: In the \(n\)th inning ONE chooses a finite subset \(X_n\) of \(X\), and TWO chooses a point \(x_n\) from \((X_1\cup\cdots\cup X_n)\setminus\{x_1,\dots, x_{n- 1}\}\). A play \[ X_1, x_1,\dots, X_n,x_n,\dots \] is won by TWO if \(\bigcup^\infty_{n= 1} X_n= \{x_n: n= 1,2,3,\dots\}\). Gale asked whether TWO could have a winning strategy which depends for each \(n\) on knowledge of only the contents of the set \[ (X_1\cup\cdots\cup X_n)\setminus \{x_1,\dots, x_{n- 1}\}. \] In mathematical terms, is there a function \(F\) defined on the collection \([X]^{<\aleph_0}\) of finite subsets of \(X\) such that: for every sequence \(X_1,x_1,\dots, X_n,x_n,\dots\) where each \(X_n\) is a finite subset of \(X\) and for each \(n\) \[ x_n= F((X_1\cup\cdots\cup X_n)\setminus\{x_1,\dots, x_{n- 1}\}) \] we have \(\bigcup^\infty_{n= 1} X_n= \{x_1,\dots, x_{n- 1},\dots\}\)? We shall call a strategy of this sort a remainder strategy for TWO. If there is some finite subset \(U\) of \(X\) such that \(F(U)\not\in U\), then \(F\) cannot be a winning remainder strategy for TWO, because ONE can defeat \(F\) by choosing \(U\) each inning. So, when studying remainder strategies for TWO we may as well assume that for each finite set \(U\subset X\), \(F(U)\in U\). In Proposition 1.1 of the paper by \textit{K. Ciesielski} and \textit{R. Laver} [Period. Math. Hung. 21, No. 2, 153-158 (1990; Zbl 0666.03038)] it is proved that if \(X\) is a countable set, then the answer is ``yes''. In Theorem 2.1 of the same paper they show that if \(X\) is uncountable then TWO does not have a winning remainder strategy. Their proof reveals a little more: Let us now change the rules of the game such that in each inning ONE is allowed to choose two or fewer (even no) points from \(X\). Let \(\text{SG}(X,2,1)\) denote this game. They prove that if \(X\) is uncountable, then TWO does not have a winning remainder strategy in the game \(\text{SG}(X, 2,1)\). This suggests the following modification of Gale's game: Let \(m\) be a positive integer and let \(f\) be a sequence of positive integers. In the \(n\)th inning ONE chooses a finite subset \(O_n\) of \(X\) and TWO responds with a set \(T_n\) which contains no more than \(m\) elements. Player ONE must further obey the rules that for each \(n\), \hskip 17mm (1) \(O_n\subseteq O_{n+1}\), and \hskip 17mm (2) \(|O_{n+ 1}\setminus O_n|\leq f(n+1)\). A play \((O_1,T_1,\dots, O_n, T_n,\dots)\) is won by TWO if \[ \bigcup^\infty_{n= 1} O_n= \bigcup^\infty_{n= 1} T_n. \] This game is denoted \(\text{SG}(X, f,m)\). If \(f\) is constant and has value \(k\), then the game is denoted \(\text{SG}(X, k,m)\). Under what requirements on \(X\), \(f\) and \(m\) does TWO have a winning remainder strategy? In light of Proposition 1.1 and Theorem 2.1 of Ciesielski and Laver [loc. cit.] our question is a question about uncountable sets. It is easy to see that if for each \(n\), \(f(n)\leq m\), or if for all but finitely many \(n\), \(f(n)< m\), then for any infinite set \(X\) TWO has a winning remainder strategy in the game \(\text{SG}(X, f,m)\). Conjecture 1. For each \(n\) TWO does not have a winning remainder strategy in the game \(\text{SG}(\omega_1,n+ 1,n)\). The purpose of this paper is to give some evidence for this conjecture. First, we verify a few special instances of the conjecture, namely for \(n\) any of 1, 2 or 3 (Theorem 3). Then we show that there is a cardinal number \(\kappa\) such that for each positive integer \(n\), TWO does not have a winning remainder strategy in the game \(\text{SG}(\kappa, n+1,n)\) (Theorem 4). Let us agree that \(\Theta\) denotes the least cardinal with this property. In Theorem 4 we show that \(\Theta\) is no larger than the \(\omega\)th successor of \(2^{\aleph_0}\). In view of Conjecture 1 that \(\Theta= \omega_1\), this upper bound is outlandishly large. In a different direction we show that there exists for each \(n\) an increasing \(f_n\) such that TWO does not have a winning remainder strategy in \(\text{SG}(\omega_1, f_n,n)\) (Theorem 6). Conjecture 1 would have for each \(n\) that for all \(k\), \(f_n(k)= n+1\); from this point of view the \(f_n\) we find is ridiculously fast growing -- it grows exponentially. From this result we then find an even faster growing \(f\) such that for each \(n\), TWO does not have a winning remainder strategy in the game \(\text{SG}(\omega_1, f,n)\) (Theorem 8). If one is willing to strengthen traditional mathematics by adopting additional axioms or if one is willing to weaken the axiom of choice, then the conjecture can be shown to be consistent. We explain this remark in the last section of the paper. These results leave, short of outright proving the conjecture, two obvious tasks: find a better (i.e., lower) upper bound for \(\Theta\), and for each \(n\) find a better (i.e., slower growing) function, say \(g_n\), than \(f_n\) such that TWO does not have a winning remainder strategy in \(\text{SG}(\omega_1, g_n,n)\). It would seem that finding a polynomial \(g_n\) for each \(n\) would already be serious progress.