## On The Power Of Two-Points Based Sampling (1989)

Venue: | Journal of Complexity |

Citations: | 91 - 16 self |

### BibTeX

@ARTICLE{Chor89onthe,

author = {Benny Chor and Oded Goldreich},

title = {On The Power Of Two-Points Based Sampling},

journal = {Journal of Complexity},

year = {1989},

volume = {5},

pages = {96--106}

}

### Years of Citing Articles

### OpenURL

### Abstract

The purpose of this note is to present a new sampling technique and to demonstrate some of its properties. The new technique consists of picking two elements at random, and deterministically generating (from them) a long sequence of pairwise independent elements. The sequence is guarantees to intersect, with high probability, any set of non-negligible density. 1. Introduction In recent years the role of randomness in computation has become more and more dominant. Randomness was used to speed up sequential computations (e.g. primality testing, testing polynomial identities etc.), but its effect on parallel and distributed computation is even more impressive. In either cases the solutions are typically presented such that they are guarateed to produce the desired result with some non-negligible probability. It is implicitly suggested that if a higher degree of confidence is required the algorithm should be run several times, each time using different coin tosses. Since the coin tosses f...

### Citations

1879 | An Introduction to Probability Theory and its Applications - Feller - 1967 |

1766 | How to share a secret - Shamir |

668 | Universal classes of hash functions - Carter, Wegman - 1977 |

604 |
How to generate cryptographically strong sequences of pseudorandom bits
- Blum, Micali
- 1984
(Show Context)
Citation Context ...ur method can be viewed as "expanding randomness" for sampling purposes, without using any unproven assumptions. A much more general method for "expanding randomness" was presented=-= by Blum and Micali [7]-=- and Yao [22]. Under the assumption that 1-1 one-way functions exist it is possible to expand n truly random coin tosses into poly(n) pseudo-random coin tosses that are good with respect to any polyno... |

516 |
Theory and applications of trapdoor functions
- Yao
- 1982
(Show Context)
Citation Context ...n be viewed as "expanding randomness" for sampling purposes, without using any unproven assumptions. A much more general method for "expanding randomness" was presented by Blum and=-= Micali [7] and Yao [22]-=-. Under the assumption that 1-1 one-way functions exist it is possible to expand n truly random coin tosses into poly(n) pseudo-random coin tosses that are good with respect to any polynomial-time alg... |

367 | A Simple Parallel Algorithm for the Maximal Independent Set Problem
- Luby
- 1986
(Show Context)
Citation Context ...in the sequence has to be randomly selected. The latter elements are deterministically computed given the first k. This property has already been demonstrated a useful tool in many applications (e.g. =-=[1,4,5,15,17,20]-=-). Especially inspiring is Luby's methodology of dispancing with randomness in special cases [17]. In Luby's setting the universe size is small but one needs many random elements in it. In case that p... |

217 |
A fast and simple randomized parallel algorithm for the maximal independent set problem
- Alon, Babai, et al.
- 1986
(Show Context)
Citation Context ...in the sequence has to be randomly selected. The latter elements are deterministically computed given the first k. This property has already been demonstrated a useful tool in many applications (e.g. =-=[1,4,5,15,17,20]-=-). Especially inspiring is Luby's methodology of dispancing with randomness in special cases [17]. In Luby's setting the universe size is small but one needs many random elements in it. In case that p... |

153 | The bit extraction problem or t-resilient functions
- Chor, Goldreich, et al.
- 1985
(Show Context)
Citation Context ...se sequences instead of random coin tosses yields the desired result. This technique is extendable to any fixed k (see [4,5]), but does not extend to the case that k grows with the instance size (see =-=[9]-=-). An interesting property of our construction of pairwise independent integers modulo p, is that the sequence of bits obtained by taking the least significant bit of each of the integers is much less... |

147 | A complexity theoretic approach to randomness - Sipser - 1983 |

134 |
RSA and rabin functions: certain parts are as hard as the whole
- Alexi, Chor, et al.
(Show Context)
Citation Context ... possible. (The effect is best examplified by considering L = 2 log p. In this case -- 7 -- there are only O(log 3 p) sequences, while the obvious upper bound is p 2 .) Furthermore, as pointed out in =-=[2]-=-, the sequence of b i 's can be predicted with high probability ( 1 \Gamma L \Gammaffl ), when knowing (1 + 2ffl) \Delta log L bits of x; y. (One needs only to know the least significant bits of both ... |

75 | A Fast Parallel Algorithm for the Maximal Independent Set Problem - RARP, WIGDERSON - 1984 |

47 | On a Set of Almost Deterministic k-Independent Random Variables. The Annals of Probability - Joffe - 1974 |

42 | The complexity of approximate counting - Stockmeyer - 1983 |

38 |
personal communication
- Alon
- 2011
(Show Context)
Citation Context ...ch consisting of L pairwise independent elements. However, k-wise independence may be useful for other purposes. We conclude this section by presenting an implementation (suggested to us by Noga Alon =-=[3]-=-) of the general construction. (The same implemetation was discovered independently by Anderson [5] and Beame [6].) First note that U can be embedded in a finite field whose size is not significantly ... |

21 |
Deterministic simulation of probabilistic constant depth circuits
- AJTAI, WIGDERSON
- 1985
(Show Context)
Citation Context ...in the sequence has to be randomly selected. The latter elements are deterministically computed given the first k. This property has already been demonstrated a useful tool in many applications (e.g. =-=[1,4,5,15,17,20]-=-). Especially inspiring is Luby's methodology of dispancing with randomness in special cases [17]. In Luby's setting the universe size is small but one needs many random elements in it. In case that p... |

4 |
On a sequence of almost deterministic pair-wise independent random variables
- Joffe
- 1971
(Show Context)
Citation Context ... The construction presented in section 3 was first discoved by us in April 1984 [10]. About half a year later, we found out that the construction (but not the application) had been presented by Joffe =-=[12]-=- in 1971 to a Probability Theory conference*. Constructions of pairwise (or k-wise) independent random variables have been used implicitly before in Computer Science works (e.g. [8,18,19,21,16]). Ackn... |

3 |
Private communication
- Beame, Guruswami
- 2003
(Show Context)
Citation Context ...We conclude this section by presenting an implementation (suggested to us by Noga Alon [3]) of the general construction. (The same implemetation was discovered independently by Anderson [5] and Beame =-=[6]-=-.) First note that U can be embedded in a finite field whose size is not significantly -- 6 -- larger than jU j. Thus, without loss of generality, U is a finite field. Let fa 1 ; : : : ; aL g be a set... |

3 |
Provably Good Pattern Generators for a Random Pattern Test
- Spencer
- 1994
(Show Context)
Citation Context |

2 |
Set Splitting", manuscript
- Anderson
- 1985
(Show Context)
Citation Context ...her purposes. We conclude this section by presenting an implementation (suggested to us by Noga Alon [3]) of the general construction. (The same implemetation was discovered independently by Anderson =-=[5]-=- and Beame [6].) First note that U can be embedded in a finite field whose size is not significantly -- 6 -- larger than jU j. Thus, without loss of generality, U is a finite field. Let fa 1 ; : : : ;... |

2 |
Time-Randomness Tradeoff", presented at the AMS conference on probabilistic computational complexity
- Karp, Pippenger
- 1982
(Show Context)
Citation Context ...bility that no run yield the desired result goes exponentially down with the number of runs. This means that we can buy higher degree of confidence in the cost of more coin tosses. Karp and Pippenger =-=[14]-=- have raised the problem of a time-randomness tradeoff in this setting. In particular, can one increase the degree of confidence without increasing the number of coin tosses. They demonstrated an affi... |

2 |
The Complexity of Parallel Computation on Matroids
- Karp, Upfal, et al.
- 1985
(Show Context)
Citation Context |

1 |
RSA/Rabin Least Significant Bits are 2 + poly(log N) Secure", MIT/LCS/TM-260
- Chor, Goldreich
- 1984
(Show Context)
Citation Context ...) The fact that the b i 's can be predicted with very high probability plays a central role in [2]. Chronological Remark The construction presented in section 3 was first discoved by us in April 1984 =-=[10]-=-. About half a year later, we found out that the construction (but not the application) had been presented by Joffe [12] in 1971 to a Probability Theory conference*. Constructions of pairwise (or k-wi... |

1 | Universal Classes of Hash Functions", Jour. of Comp. and Sys - Carter, Wegman - 1979 |