## Unprovability of lower bounds on the circuit size in certain fragments of bounded arithmetic (1995)

Venue: | Izvestiya of the R.A.N |

Citations: | 54 - 6 self |

### BibTeX

@ARTICLE{Razborov95unprovabilityof,

author = {Alexander A. Razborov},

title = {Unprovability of lower bounds on the circuit size in certain fragments of bounded arithmetic},

journal = {Izvestiya of the R.A.N},

year = {1995},

volume = {59},

pages = {201--224}

}

### Years of Citing Articles

### OpenURL

### Abstract

To appear in Izvestiya of the RAN Abstract We show that if strong pseudorandom generators exist then the statement "ff encodes a circuit of size n(log * n) for SATISFIABILITY " is not refutable in S22 (ff). For refutation in S12 (ff), this is proven under the weaker assumption of the existence of generators secure against the attack by small depth circuits, and for another system which is strong enough to prove exponential lower bounds for constant-depth circuits, this is shown without using any unproven hardness assumptions. These results can be also viewed as direct corollaries of interpolation-like theorems for certain "split versions " of classical systems of Bounded Arithmetic introduced in this paper.

### Citations

287 |
Algebraic methods in the theory of lower bounds for boolean circuit complexity
- Smolensky
- 1987
(Show Context)
Citation Context ...ns is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. [35, 30, 31, 36, 37]) and in more recent time (=-=[11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]-=-). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which extent the methods developed so far can be useful for attacking central open problems... |

195 |
How easy is local search
- Johnson, Papadimitriou, et al.
- 1988
(Show Context)
Citation Context ....2]: Proposition 2.2. For every (partial) Boolean function f and every d > 0, 2i Cd(Rf ) d -1j <= Sd(f ) <= 2Cd(Rf ). 5s2.3. Polynomial local search problems This concept was originally considered in =-=[13]-=-. We reproduce here the variant of the definition given in [8]. Definition 2.3. A local search problem L consists of a set FL(x) ` N of solutions for every instance x 2 N, an integer-valued cost funct... |

183 |
Separating the polynomial-time hierarchy by oracles (preliminary version
- Yao
- 1985
(Show Context)
Citation Context ...ns is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. [35, 30, 31, 36, 37]) and in more recent time (=-=[11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]-=-). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which extent the methods developed so far can be useful for attacking central open problems... |

182 |
Computational limitations of small-depth circuits
- H˚astad
- 1987
(Show Context)
Citation Context ...ns is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. [35, 30, 31, 36, 37]) and in more recent time (=-=[11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]-=-). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which extent the methods developed so far can be useful for attacking central open problems... |

176 |
The complexity of finite functions
- Boppana, Sipser
- 1990
(Show Context)
Citation Context ...and open problems in Section 8. 2. Background from Complexity Theory In this section we recall necessary definitions and facts from Complexity Theory. 2.1. Boolean Complexity We address the reader to =-=[5]-=- for an excellent treatment of the subject; the sole purpose of this section is to agree upon notation. We denote by Fn the set of all Boolean functions in n variables x1, . . . , xn. Let x1i *) xi an... |

175 | Natural proofs
- Razborov, Rudich
- 1997
(Show Context)
Citation Context ...) such that |s| <= pL(|x|) for all s 2 FL(x). Note that the concept of a polynomial local search (PLS) problem can be relativized in a standard way. 2.4. Natural proofs This concept was introduced in =-=[21]-=-. Let \Gammasand \Lambdasbe complexity classes. Slightly altering the notation from [21], we call a sequence {Cn | n 2 ! } of subsets Cn ` Fn a \Gamma -natural combinatorial property useful against \L... |

149 |
Monotone circuits for connectivity require superlogarithmic depth
- Karchmer, Wigderson
- 1988
(Show Context)
Citation Context |

127 | The monotone circuit complexity of boolean functions
- Alon, Boppana
- 1987
(Show Context)
Citation Context |

122 |
Bounded arithmetic, Bibliopolis
- Buss
- 1986
(Show Context)
Citation Context ...adding to CH0 at most h0 new nodes, we finally compute f by a circuit of size O(size(L)). This completes the proof of Theorem 3.1. 4. Background from Bounded Arithmetic We assume the familiarity with =-=[6]-=- and use the now-standard notation for denoting various hierarchies and fragments of Bounded Arithmetic from that book. We denote by L2 Buss’s first order language which consists of the constant 0, fu... |

119 |
Parity, circuits, and the polynomial time hierarchy
- Furst, Saxe, et al.
- 1984
(Show Context)
Citation Context |

115 |
Feasibly constructive proofs and the propositional calculus
- Cook
- 1975
(Show Context)
Citation Context ... ordinary theories Si2(fl) and T i2(fl). A less trivial example is provided by L = LP V , R = "BASIC2 + \Pi b1-defining axioms for P V -symbols" (see [6, $6.2]), where P V is Cook's equational system =-=[10]-=-. In this case S1R is the theory S12 (LP V ) as defined in [6]. One more example of this sort will be given in Section 6. As we already mentioned, the meaning of this definition is that many (if not a... |

109 |
Σ 1 1-formulae on finite structures
- Ajtai
- 1983
(Show Context)
Citation Context |

91 | The expressive power of voting polynomials - Aspnes, Beigel, et al. - 1994 |

76 | Monotone Circuits for Matching Require Linear Depth
- Raz, Wigderson
- 1992
(Show Context)
Citation Context |

72 |
Provability of the pigeonhole principle and the existence of infinitely many primes, Journal of Symbolic Logic 53
- Paris, Wilkie, et al.
- 1988
(Show Context)
Citation Context ...the following form: i a >= 2 ^ 8x <= a2(f (x) < a)j oe 9x1, x2 < a2(f (x1) = f (x2) ^ x1 6= x2). Note that, contrary to the common belief, it is open whether T 22 (f ) ` W P HP (f ). But the proof in =-=[16]-=- lets us conclude at least that T 32 (f ) ` W P HP (f ), and this (naturally) extends to showing that T 3R(f ) ` W P HP (f ) for every \Sigma b1-definable f . Now, let A(N, fff) say "fff is an injecti... |

60 |
The gap between monotone and nonmonotone circuit complexity is exponential
- Tardos
- 1988
(Show Context)
Citation Context |

55 | kn application of Boolean complexity to separation problems in bounded arithmetic
- Buss, Krajicek
- 1994
(Show Context)
Citation Context ...d every d > 0, 2i Cd(Rf ) d -1j <= Sd(f ) <= 2Cd(Rf ). 5s2.3. Polynomial local search problems This concept was originally considered in [13]. We reproduce here the variant of the definition given in =-=[8]-=-. Definition 2.3. A local search problem L consists of a set FL(x) ` N of solutions for every instance x 2 N, an integer-valued cost function cL(s, x) and a neighborhood function NL(s, x) such that: a... |

46 | Bounded arithmetic and lower bounds in Boolean complexity
- Razborov
- 1995
(Show Context)
Citation Context ...gical analysis of this situation should start with understanding what is the right "minimal" fragment of ZF C which is really needed for formalizing all these methods, and this question was raised in =-=[19]-=-. It was argued there that the conceivable answer is the second order theory of Bounded Arithmetic V 11 , and no example of a lower bound for explicit function not provable in V 11 has been found sinc... |

41 | On provably disjoint NP-pairs
- Razborov
- 1994
(Show Context)
Citation Context ... author has found a purely complexity characterization of pairs of N P -sets which are provably disjoint in certain fragments of Bounded Arithmetic, including S(S2) + S\Sigma bi - IN D, U 12 and V 12 =-=[20]-=-. This characterization, in particular, gives raise to another and, perhaps, more natural, proof of Theorem 6.4. 10. Acknowledgement I am indebted to Sam Bass, Steven Cook, Mauricio Karchmer, Jan Kraj... |

35 | Witnessing functions in bounded arithmetic and search problems, The
- Chiari, Krajíček
- 1998
(Show Context)
Citation Context ...pact on the future research in this direction is still to be understood. Thus, we confine ourselves here with a simpler combinatorial example which gives a new unexpected proof of a known result from =-=[9]-=- and raises several immediate open questions. Example 1. The proof of the separation theorem works for the monotone case as well. That is to say, if S2R ` (A(N, ff) ^ B(N, fi)) oe 9x(C(N, x, ff) ^ ~D(... |

29 |
RSUV isomorphism
- Takeuti
- 1993
(Show Context)
Citation Context ...at even for the case of V 1 2 , the affirmative answer to a similar question in which we are interested in separating co−NP sets is a straightforward corollary of Proposition 4.2 and RSUV-isomorphism =-=[23, 24, 18]-=-. There are several examples showing that for NP-sets the situation may be different. A couple of them originated from a discussion with Steven Rudich are based upon the lower bound proof for voting p... |

28 | An Equivalence between Second Order Bounded Domain Bounded Arithmetic and First Order Bounded Arithmetic
- Razborov
- 1993
(Show Context)
Citation Context ...malized analogues of Propositions 2.2, 2.1 and Theorem 3.1. In the rest of this section we briefly outline those aspects of this formalization which may appear less obvious. Firstly, we, similarly to =-=[18]-=-, treat V2 simply as a two-sorted first order theory. This allows us to define a language cL2 and the skolemization cV2 of V2 in this language similarly to dL2(fl), dS2(fl). Namely, behind function sy... |

28 | Axiomatizations and conservation results for fragments of bounded arithmetic; in Logic and Computation, proceedings of a Workshop held Carnegie-Mellon
- Buss
- 1987
(Show Context)
Citation Context ... . 11Here Ω is the interpretation of symbols from L \ L2, andM Ω (n) is the result of the computation of M on n when M is fed with the oracle Ω. We also need the following conservation result from =-=[7]-=-: Proposition 4.3. For any strongly regular theory R in a language L ⊇ L2, S 2 R is Σb 2 (L)conservative over T 1 R. functions in terms of PLSFinally, we recall the characterization of Σb 1-defined in... |

22 | Communication Complexity: A New Approach to Circuit Depth - Karchmer - 1989 |

9 |
A note on a theorem of Razborov
- Barrington
- 1986
(Show Context)
Citation Context |

7 |
Bounded existential induction
- Wilmers
- 1985
(Show Context)
Citation Context ...11 , this is also natural to consider the hierarchy of its subtheories and wonder whether we can do better for them. The strongest theory in this hierarchy to which our method applies is IE1(f ) (see =-=[26]-=- for the definition of IE1), and for this theory we indeed can prove a slightly stronger result. Namely, we may replace t(n) by nk for a fixed constant k > 0 depending only on the quality of the gener... |

6 |
O sloжnosti realizacii line�ino�i funkcii v klasse Π-shem
- Hrapqenko
- 1971
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |

5 |
Niжnie ocenki razmera shem ograniqenno�i glubiny v polnom bazise, soderжawem funkci� logiqeskogo sloжeni
- Razborov
- 1987
(Show Context)
Citation Context |

2 |
Axiomatizations and conservation results for fragments of bounded arithmetic. Contemporary Mathematics
- Bibliopolis, Naples
- 1986
(Show Context)
Citation Context ...adding to CH0 at most h0 new nodes, we finally compute f by a circuit of size O(size(L)). This completes the proof of Theorem 3.1. 4. Background from Bounded Arithmetic We assume the familiarity with =-=[6]-=- and use the now-standard notation for denoting various hierarchies and fragments of Bounded Arithmetic from that book. We denote by L2 Buss's first order language which consists of the constant 0, fu... |

2 |
O minimal~nyh kontaktno-ventil~nyh dvuhpol snikah dl monotonnyh simmetriqeskih funkci$i
- Markov
- 1962
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |

2 |
Ob odno$i bulevsko$i funkcii
- Neqiporuk
- 1966
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60's (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |

2 |
O realizacii line$inyh funkci$i formulami v bazise &; .; \Gamma
- SubbotovskaO
- 1961
(Show Context)
Citation Context |

2 |
Ob odnom metode poluqeniO/ nissnih ocenok slossnosti \Pi -shem. Matem. zametki, t
- Hrapqenko
- 1971
(Show Context)
Citation Context |

2 |
Ob odnom metode poluqeni� niжnih ocenok sloжnosti individualьnyh monotonnyh funkci�i
- Andreev
- 1985
(Show Context)
Citation Context ...ns is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60’s (see e.g. [35, 30, 31, 36, 37]) and in more recent time (=-=[11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]-=-). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which extent the methods developed so far can be useful for attacking central open problems... |

1 |
Si3 and ffiV i2(BD). Archive for Math. Logic
- Takeuti
- 1990
(Show Context)
Citation Context ...at even for the case of V 12 , the affirmative answer to a similar question in which we are interested in separating co-N P sets is a straightforward corollary of Proposition 4.2 and RSUV-isomorphism =-=[23, 24, 18]-=-. There are several examples showing that for N P -sets the situation may be different. A couple of them originated from a discussion with Steven Rudich are based upon the lower bound proof for voting... |

1 |
Ob odnom metode poluqeniO/ nissnih ocenok slossnosti individual~nyh monotonnyh funkci$i
- Andreev
- 1985
(Show Context)
Citation Context |

1 |
Ob odnom metode poluqeniO/ ffffektivnyh nissnih ocenok monotonno$i slossnosti. Algebra i logika
- Andreev
- 1987
(Show Context)
Citation Context |

1 |
Nissnie ocenki monotonno$i slossnosti nekotoryh bulevyh funkci$i
- Razborov
- 1985
(Show Context)
Citation Context |

1 |
O minimalьnyh kontaktno-ventilьnyh dvuhpol�snikah dl� monotonnyh simmetriqeskih funkci�i. InProblemy kibernetiki
- Markov
- 1962
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60’s (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |

1 |
Niжnie ocenki monotonno�i sloжnosti logiqeskogo permanenta
- Razborov
- 1985
(Show Context)
Citation Context ...ns is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60’s (see e.g. [35, 30, 31, 36, 37]) and in more recent time (=-=[11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]-=-). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which extent the methods developed so far can be useful for attacking central open problems... |

1 |
Orealizaciiline�inyh funkci�i formulami v bazise
- Subbotovska�
- 1961
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60’s (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |

1 |
Ob odnom metode poluqeni� niжnih ocenok sloжnosti Π-shem
- Hrapqenko
- 1971
(Show Context)
Citation Context ... complexity of explicitly given Boolean functions is one of the most challenging tasks in the computational complexity. This theory met with a remarkable success at least twice: in the 60’s (see e.g. =-=[35, 30, 31, 36, 37]-=-) and in more recent time ([11, 1, 27, 12, 32, 33, 28, 2, 25, 29, 34, 22, 4, 15, 17]). Both times, however, the period of enthusiasm was followed by understanding that it is not quite clear to which e... |