Generic complexity of undecidable problems

• Authors: Alexei Myasnikov
• Venue: CSR'07 Proceedings of the Second international conference on Computer Science: theory and applications
• Year: 2007

This is an extended abstract of my talk on generic complexity of undecidable problems. It turns out that some classical undecidable problems are, in fact, strongly undecidable, i.e., they are still undecidable on every strongly generic (i.e., "very very large") subset of inputs. For instance, the classical Halting Problem for Turing machines is strongly undecidable. Moreover, we prove an analog of the Rice's theorem for strongly undecidable problems, which provides plenty of examples of strongly undecidable problems. On the other hand, it has been shown recently that many of these classical undecidable problems are easily decidable on some generic (i.e., "very large") subsets of inputs. Altogether, these results lead to an interesting hierarchy of undecidable problems with respect to the size of subsets of inputs where the problems are still undecidable - a frequency analysis of hardness. We construct here some natural super-undecidable problems, i.e., problem which are undecidable on every generic (not only strongly generic) subset of inputs. In particular, there are finitely presented semigroups with super-undecidable word problem. To construct strongly- and super-undecidable problems we introduce a method of generic amplification (an analog of the amplification in complexity theory).