A Taxonomy of Complexity Classes of Functions

This paper comprises a systematic comparison of several complexity classes of functions that are computed nondeterministically in polynomial time or with an oracle in NP. There are three components to this work:*A taxonomy is presented that demonstrates all known inclusion relations of these classes. For (nearly) each inclusion that is not shown to hold, evidence is presented to indicate that the inclusion is false. As an example, consider FewPF, the class of multivalued functions that are nondeterministically computable in polynomial time such that for each x there is a polyomial bound on the number of distinct output values of f(x). We show that FewPF @?"c PF^N^P"t"t. However, we show PF^N^P"t"t @? FewPF if and only if NP = co-NP, and thus PF^N^P"t"t @? FewPF is likely to be false. *Whereas it is known that P^N^P(O(log n)) = P^N^P"t"t @? P^N^P, we show that PF^N^P(O(log n)) = PF^N^P"t"t implies P = FewP and R = NP. Also, we show that PF^N^P"t"t = PF^N^P if and only if P^N^P"t"t = P^N^P. *We show that if every nondeterministic polynomial-time multivalued function has a single-valued nondeterministic refinement (equivalently, if every honest function that is computable in polynomial-time can be inverted by a single-valued nondeterministic function), then there exists a disjoint pair of NP-complete sets such that every separator is NP-hard. The latter is a previously studied open problem that is closely related to investigations on promise problems. This result motivates a study of reductions between partial multivalued functions.

[1]  Yacov Yacobi,et al.  The Complexity of Promise Problems with Applications to Public-Key Cryptography , 1984, Inf. Control..

[2]  Eric Allender,et al.  P-Printable Sets , 1988, SIAM J. Comput..

[3]  Alan L. Selman,et al.  Polynomial Time Enumeration Reducibility , 1978, SIAM J. Comput..

[4]  Mark W. Krentel The Complexity of Optimization Problems , 1988, J. Comput. Syst. Sci..

[5]  Alan L. Selman One-Way Functions in Complexity Theory , 1990, MFCS.

[6]  Ronald V. Book,et al.  Positive Relativizations of Complexity Classes , 1983, SIAM J. Comput..

[7]  Leslie G. Valiant,et al.  NP is as easy as detecting unique solutions , 1985, STOC '85.

[8]  Klaus W. Wagner,et al.  Bounded Query Classes , 1990, SIAM J. Comput..

[9]  Jin-Yi Cai,et al.  On the Power of Parity Polynomial Time , 1989, STACS.

[10]  Timothy J. Long,et al.  Quantitative Relativizations of Complexity Classes , 1984, SIAM J. Comput..

[11]  Seinosuke Toda On the computational power of PP and (+)P , 1989, 30th Annual Symposium on Foundations of Computer Science.

[12]  L. Hemachandra,et al.  Counting in structural complexity theory , 1987 .

[13]  Stephen A. Cook,et al.  The complexity of theorem-proving procedures , 1971, STOC.

[14]  Leslie G. Valiant,et al.  Relative Complexity of Checking and Evaluating , 1976, Inf. Process. Lett..

[15]  Klaus W. Wagner,et al.  The Difference and Truth-Table Hierarchies for NP , 1987, RAIRO Theor. Informatics Appl..

[16]  Mark W. Krentel The complexity of optimization problems , 1986, STOC '86.

[17]  Alan L. Selman,et al.  Complexity Measures for Public-Key Cryptosystems , 1988, SIAM J. Comput..

[18]  Martin Grötschel,et al.  The ellipsoid method and its consequences in combinatorial optimization , 1981, Comb..

[19]  Gerd Wechsung,et al.  On the Power of Probabilistic Polynomial Time:PNP[log] ⊆ PP , 1988 .

[20]  Samuel R. Buss,et al.  On truth-table reducibility to SAT and the difference hierarchy over NP , 1988, [1988] Proceedings. Structure in Complexity Theory Third Annual Conference.

[21]  Jim Kadin,et al.  P^(NP[O(log n)]) and Sparse Turing-Complete Sets for NP , 1989, J. Comput. Syst. Sci..

[22]  R. Beigel NP-hard Sets are P-Superterse Unless R = NP , 1988 .

[23]  Alan L. Selman Natural Self-Reducible Sets , 1988, SIAM J. Comput..

[24]  Lane A. Hemaspaandra,et al.  On the power of probabilistic polynomial time: P/sup NP(log)/ contained in PP , 1989, [1989] Proceedings. Structure in Complexity Theory Fourth Annual Conference.