Results 1 
3 of
3
WORKING WITH STRONG REDUCIBILITIES ABOVE TOTALLY ωC.E. DEGREES
"... Abstract. We investigate the connections between the complexity of a c.e. set, as calibrated by its strength as an oracle for Turing computations of functions in the Ershov hierarchy, and how strong reducibilities allows us to compute such sets. For example, we prove that a c.e. degree is totally ω ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
Abstract. We investigate the connections between the complexity of a c.e. set, as calibrated by its strength as an oracle for Turing computations of functions in the Ershov hierarchy, and how strong reducibilities allows us to compute such sets. For example, we prove that a c.e. degree is totally ωc.e. iff every set in it is weak truthtable reducible to a hypersimple, or ranked, set. We also show that a c.e. degree is array computable iff every leftc.e. real of that degree is reducible in a computable Lipschitz way to a random leftc.e. real (an Ωnumber). 1.
Simplicity and Strong Reductions
, 2000
"... A set is called NPsimple if it lies in NP, and its complement is infinite, and does not contain any infinite subsets in NP. Hartmanis, Li and Yesha [HLY86] proved that no set which is hard for NP under manyone (Karp) reductions is NPsimple unless NP # coNP # SUBEXP. However, we can exhibit ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
A set is called NPsimple if it lies in NP, and its complement is infinite, and does not contain any infinite subsets in NP. Hartmanis, Li and Yesha [HLY86] proved that no set which is hard for NP under manyone (Karp) reductions is NPsimple unless NP # coNP # SUBEXP. However, we can exhibit a relativized world in which there is an NPsimple set that is complete under Turing (Cook) reductions, even conjunctive reductions. This raises the questions whether the result by Hartmanis, Li and Yesha generalizes to reductions of intermediate strength. We show that NPsimple sets are not complete for NP under positive bounded truthtable reductions unless UP # SUBEXP. In fact, NPsimple sets cannot be complete for NP under bounded truthtable reductions under the stronger assumption that UP # coUP ## SUBEXP (while there is an oracle relative to which there is an NPsimple set conjuntively complete for NP). We present several other results for di#erent types of reductions, a...
Electronic Colloquium on Computational Complexity, Report No. 71 (2004) Simplicity and Strong Reductions
, 2004
"... A set is called NPsimple if it lies in NP, and its complement is infinite, and does not contain any infinite subsets in NP. Hartmanis, Li and Yesha [HLY86] proved that no set which is hard for NP under manyone (Karp) reductions is NPsimple unless NP ∩ coNP ⊆ SUBEXP. However, we can exhibit a rela ..."
Abstract
 Add to MetaCart
A set is called NPsimple if it lies in NP, and its complement is infinite, and does not contain any infinite subsets in NP. Hartmanis, Li and Yesha [HLY86] proved that no set which is hard for NP under manyone (Karp) reductions is NPsimple unless NP ∩ coNP ⊆ SUBEXP. However, we can exhibit a relativized world in which there is an NPsimple set that is complete under Turing (Cook) reductions, even conjunctive reductions. This raises the questions whether the result by Hartmanis, Li and Yesha generalizes to reductions of intermediate strength. We show that NPsimple sets are not complete for NP under positive bounded truthtable reductions unless UP ⊆ SUBEXP. In fact, NPsimple sets cannot be complete for NP under bounded truthtable reductions under the stronger assumption that UP ∩ coUP ⊆ SUBEXP (while there is an oracle relative to which there is an NPsimple set conjunctively complete for NP). We present several other results for different types of reductions, and show how to prove a similar result for NEXP which does not require any assumptions. We also prove that all NEXPcomplete sets are Plevelable, extending work by Tran [Tra95]. Most of the results are derived by the use of inseparable sets. This technique turns out to be very powerful in the study of truthtable and even (honest) Turing reductions.