Computable randomness and its properties
Seminar Room 1, Newton Institute
AbstractComputable randomness at first does not seem as natural of a randomness notion as Schnorr and Martin-Löf randomness. However, recently Brattka, Miller, and Nies  have shown that computable randomness is closely linked to differentiability. Why is this so? What are the chances that, say, computable randomness will also be linked to the ergodic theorem? In this talk I will explain how computable randomness is similar to and how it is different from other notions of randomness.
Unlike other notions of randomness, computable randomness is closely linked to the Borel sigma-algebra of a space. This has a number of interesting implications:
- Computable randomness can be extended to other computable probability spaces, but this extension is more complicated to describe .
- Computable randomness is invariant under isomorphisms, but not morphisms (a.e.-computable measure-preserving maps) .
- Computable randomness is connected more with differentiability than with the ergodic theorem.
- Dyadic martingales and martingales whose filtration converges to a "computable" sigma-algebra characterize computable randomness, while more general computable betting strategies do not.
However, this line of research still leaves many open questions about the nature of computable randomness and the nature of randomness in general. I believe the tools used to explore computable randomness may have other applications to algorithmic randomness and computable analysis.
 Vasco Brattka, Joseph S. Miller, and André Nies. "Randomness and differentiability." Submitted.
 Jason Rute. "Computable randomness and betting for computable probability spaces." In preparation.
If it doesn't, something may have gone wrong with our embedded player.
We'll get it fixed as soon as possible.