What is a hard instance of a computational problem?

  • Ker-I Ko
  • Pekka Orponen
  • Uwe Schöning
  • Osamu Watanabe
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 223)


In this paper a measure for the complexity of particular instances with respect to a given decision problem is introduced and investigated. Intuitively, an instance x is considered to be hard for a problem A if every algorithm that decides A and runs "fast" on x needs to look up (a description of) x in a table. A main result states that all problems not in P have infinitely many (polynomially) hard instances. Further, there exist problems in EXPTIME with all their instances being hard. The behavior of hard instances under polynomial reductions and the connections with complexity cores and circuits are studied.


Turing Machine Complexity Core Kolmogorov Complexity Circuit Complexity Hard Instance 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [BS85]
    J.L. Balcázar and U. Schöning, Bi-immune sets for complexity classes, Math. Systems Theory 18 (1985), 1–10.Google Scholar
  2. [ESY85]
    S. Even, A.L. Selman and Y. Yacobi, Hard core theorems for complexity classes, J. Assoc. Comput. Mach. 32 (1985), 205–217.Google Scholar
  3. [Har83a]
    J. Hartmanis, Generalized Kolmogorov complexity and the structure of feasible computations, Proc. 24th FOCS 1983, 439–445.Google Scholar
  4. [Har83b]
    J. Hartmanis, On sparse sets in NP, Information Processing Letters 16 (1983), 55–60.Google Scholar
  5. [HU79]
    J. Hopcroft and J. Ullman, Introduction to Automata, Lanquaqes, and Computation, Addison-Wesley, 1979.Google Scholar
  6. [KL80]
    R. Karp and R. Lipton, Some connections between nonuniform and uniform complexity classes, Proc. 12th ACM Symp. Theory of Computing, 302–309, 1980.Google Scholar
  7. [Ko84]
    K. Ko, A definition of infinite pseudorandom sequences, manuscript, submitted for publication.Google Scholar
  8. [Ko85]
    K. Ko, Non-levelable and immune sets in the accepting density hierarchy for NP, Math. Systems Theory 18 (1985), 189–205.Google Scholar
  9. [Ko165]
    A.N. Kolmogorov, Three approaches to the quantitative definition of information, Prob. Info. Transmission 1 (1965), 1–7.Google Scholar
  10. [Lyn75]
    N. Lynch, On reducibility to complex or sparse sets, J. Assoc. Comput. Mach. 22 (1975), 341–345.Google Scholar
  11. [MM71]
    A.M. Meyer and E.M. McCreight, Computationally complex and pseudo-random zero-one valued functions, in: Z. Kohavi and A.Paz (ed.), Theory of Machines and Computations, Academic Press, 1971.Google Scholar
  12. [Orp84]
    P. Orponen, A classification of complexity core lattices, manuscript, to appear.Google Scholar
  13. [ORS85]
    P. Orponen, D. Russo and U. Schöning, Optimal approximations and polynomially levelable sets, SIAM Journal on Comput., to appear.Google Scholar
  14. [OS84]
    P. Orponen and U. Schöning, The structure of polynomial complexity cores, Proc. 11th Symp. on Math. Found. of Comput. Sci. 1984, Lecture Notes in Computer Science 176, Springer-Verlag, 452–458.Google Scholar
  15. [OS86]
    P. Orponen and U. Schöning, The density and complexity of polynomial cores for intractable sets, Information and Control, to appear.Google Scholar
  16. [Wil83]
    R.E. Wilber, Randomness and the density of hard problems, Proc. 24th FOCS 1983, 335–342.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1986

Authors and Affiliations

  • Ker-I Ko
    • 1
  • Pekka Orponen
    • 2
  • Uwe Schöning
    • 3
  • Osamu Watanabe
    • 4
  1. 1.Dept. of Computer ScienceUniversity of HoustonHoustonU.S.A.
  2. 2.Dept. of Computer ScienceUniversity of HelsinkiHelsinkiFinland
  3. 3.EWH Rheinland-PfalzSeminar für InformatikKoblenzWest Germany
  4. 4.Dept. of Information ScienceTokyo Institute of TechnologyTokyoJapan

Personalised recommendations