On uniform definability of types over finite sets (Q2892672)

The paper discusses a new notion of definability of types, called uniform definability of types over finite sets, denoted UDTFS. In detail, a formula \(\phi(x; y)\) (where \(x\), \(y\) are meant as tuples of variables, parameters, respectively) is said to have UDTFS if there exists a formula \(\psi(y; z_0, \dots, z_{k-1})\) such that, for every finite set \(B\) of size \(\geq 2\) consisting of tuples of the same length as \(y\) and for every \(p(x) \in S_\phi (B)\), there are \(c_0, \dots, c_{k-1} \in B\) for which \(\phi(y;c_0, \dots, c_{k-1})\) defines \(p(x)\). A theory is said to have UDTFS if every formula has UDTFS.NEWLINENEWLINEThe paper illustrates some previously known properties of UDTFS. For instance, it explains why both stability and weak o-minimality imply UDTFS and UDTFS implies dependence. Actually it is conjectured that UDTFS characterizes dependence. Then the paper shows that all dp-minimal theories admit UDTFS.