No label defined (Q4781979)

A typical result in ``statistical group theory'' most likely would say something of type: ``an element of a group \(G\) has a property \(\mathbf P\) with probability \(p\)''. The validity of such a statement does depend on the way one measures subsets in the group \(G\) (especially, when the group \(G\) is infinite). The current paper discusses this at a general methodological level, mostly for free groups. The starting point of the study is the desire to identify measures which can be used in the behavior of generic algorithms on infinite groups. Almost all the measures considered are atomic (which means that every subset in \(G\) is measurable). The first approach used in the paper constructs measures on countable groups based on length or complexity functions. Further consideration leads to Kolmogorov complexity of words as a more adequate concept for the length function on groups. Some related topics are also considered. None of the definitions of measures suggested here is short enough to be included in this review. For the entire collection see [Zbl 0994.00026].