## Deterministically Testing Sparse Polynomial Identities of Unbounded Degree (2008)

Citations: | 5 - 0 self |

### BibTeX

@MISC{Bläser08deterministicallytesting,

author = {Markus Bläser and Moritz Hardt and Richard J. Lipton and Nisheeth K. Vishnoi D},

title = {Deterministically Testing Sparse Polynomial Identities of Unbounded Degree },

year = {2008}

}

### OpenURL

### Abstract

We present two deterministic algorithms for the arithmetic circuit identity testing problem. The running time of our algorithms is polynomially bounded in s and m, where s is the size of the circuit and m is an upper bound on the number monomials with non-zero coefficients in its standard representation. The running time of our algorithms also has a logarithmic dependence on the degree of the polynomial but, since a circuit of size s can only compute polynomials of degree at most 2 s, the running time does not depend on its degree. Before this work, all such deterministic algorithms had a polynomial dependence on the degree and therefore an exponential dependence on s. Our first algorithm works over the integers and it requires only black-box access to the given circuit. Though this algorithm is quite simple, the analysis of why it works relies on Linnik’s Theorem, a deep result from number theory about the size of the smallest prime in an arithmetic progression. Our second algorithm, unlike the first, uses elementary arguments and works over any integral domains, but it uses the circuit in a less restricted manner. In both cases the running time has a logarithmic dependence on the largest coefficient of the polynomial.

### Citations

392 | Fast Probabilistic Algorithms for Verification of Polynomial Identities
- Schwartz
- 1980
(Show Context)
Citation Context ... both PSPACE=IP [8] and NP=PCP [9] involved polynomial identity testing as a subroutine. The earliest work in testing polynomial identities is due to DeMillo and Lipton [10], Zippel [11] and Schwartz =-=[12]-=-. These results give efficient randomized algorithms for checking polynomial identities. Indeed, no n-variate polynomial of total degree d can vanish on more than half the points in a combinatorial cu... |

359 | Probabilistic checking of proofs: A new characterization of NP
- Arora, Safra
- 1998
(Show Context)
Citation Context ...esting [6] and primality testing [7] reduce to this problem. Equally many applications have been discovered in complexity theory; to only mention two celebrated results, both PSPACE=IP [8] and NP=PCP =-=[9]-=- involved polynomial identity testing as a subroutine. The earliest work in testing polynomial identities is due to DeMillo and Lipton [10], Zippel [11] and Schwartz [12]. These results give efficient... |

302 | Designing programs that check their work
- Blum, Kannan
- 1995
(Show Context)
Citation Context ...identical to the zero polynomial. Many algorithmic problems such as finding perfect matchings in graphs [2–4], checking the equivalence of read-once branching programs [5], multi-set equality testing =-=[6]-=- and primality testing [7] reduce to this problem. Equally many applications have been discovered in complexity theory; to only mention two celebrated results, both PSPACE=IP [8] and NP=PCP [9] involv... |

235 |
Probabilistic algorithms for sparse polynomials
- Zippel
- 1979
(Show Context)
Citation Context ...elebrated results, both PSPACE=IP [8] and NP=PCP [9] involved polynomial identity testing as a subroutine. The earliest work in testing polynomial identities is due to DeMillo and Lipton [10], Zippel =-=[11]-=- and Schwartz [12]. These results give efficient randomized algorithms for checking polynomial identities. Indeed, no n-variate polynomial of total degree d can vanish on more than half the points in ... |

161 | Matching is as easy as matrix inversion - Mulmuley, Vazirani, et al. - 1987 |

146 |
Modern Computer Algebra
- Gathen, Gerhard
- 2003
(Show Context)
Citation Context ...trassen multiplication. Hence, the entire operation requires Õ(sk) ring operations where s is the size of the given circuit. For details the reader is referred to the book by v. z. Gathen and Gerhard =-=[32]-=-. Our algorithm is stated and analyzed next. It will perform the Kronecker substitution so as to apply to multivariate polynomials as well. Algorithm 2 Input: Arithmetic circuit C on n inputs, paramet... |

131 |
The Book of Prime Number Records
- Ribenboim
(Show Context)
Citation Context ...alue for L is 5.5 due to Heath and Brown. Schinzel, Sierpinski, and Kanold have conjectured the value to be 2. For a detailed discussion of these facts the reader is referred to the book by Ribenboim =-=[30]-=-. We are interested in the value of P (q) where q itself is a prime number. Notice, it is not immediately clear whether or not two distinct primes q1 �= q2, could have P (q1) = P (q2). However, we can... |

122 |
Derandomizing polynomial identity testing means proving circuit lower bounds
- Kabanets, Impagliazzo
(Show Context)
Citation Context ...uire the input degree to be polynomially bounded. When it comes to general circuits, no deterministic sub-exponential time arithmetic circuit identity test is known. In fact, Kabanets and Impagliazzo =-=[28]-=- show that such an algorithm would imply non-trivial circuit lower bounds. Further connections along these lines were shown by Agrawal [27]. 1.2 Our Results We state our two results here informally an... |

86 | A deterministic algorithm for sparse multivariate polynomial interpolation - Ben-Or, Tiwari - 1988 |

58 |
Equivalence of free boolean graphs can be decided probabilistically in polynomial time
- Blum, Chandra, et al.
- 1980
(Show Context)
Citation Context ...given in some representation is identical to the zero polynomial. Many algorithmic problems such as finding perfect matchings in graphs [2–4], checking the equivalence of read-once branching programs =-=[5]-=-, multi-set equality testing [6] and primality testing [7] reduce to this problem. Equally many applications have been discovered in complexity theory; to only mention two celebrated results, both PSP... |

57 | A probabilistic remark on algebraic program testing
- DeMillo, Lipton
- 1978
(Show Context)
Citation Context ...mention two celebrated results, both PSPACE=IP [8] and NP=PCP [9] involved polynomial identity testing as a subroutine. The earliest work in testing polynomial identities is due to DeMillo and Lipton =-=[10]-=-, Zippel [11] and Schwartz [12]. These results give efficient randomized algorithms for checking polynomial identities. Indeed, no n-variate polynomial of total degree d can vanish on more than half t... |

56 | On determinants, matching, and random algorithms - Lovász - 1979 |

52 | Fast parallel algorithms for sparse multivariate polynomial interpolation over finite fields - Grigoriev, Karpinski, et al. - 1990 |

31 | On zero-testing and interpolation of k-sparse multivariate polynomials over finite fields, Theoret - Clausen, Dress, et al. - 1991 |

29 | Randomness efficient identity testing of multivariate polynomials
- Klivans, Spielman
- 2001
(Show Context)
Citation Context ...of nonzero terms and d its degree. There is also some dependence on the characteristic of the underlying field and the bit size of the points queried varies. A survey of these results can be found in =-=[19]-=-. Randomized complexity. A sequence of results [20,13,7,19,21,22] has studied (and determined in several cases) how much randomness is required in testing polynomial identities when only black-box acc... |

29 | Proving lower bounds via pseudo-random generators
- Agrawal
- 2005
(Show Context)
Citation Context ...–26]. We can represent a polynomial with m terms by an arithmetic circuit of size m + 1 and depth 2. Hence, depth-3 circuits can be seen as a generalization of sparse polynomials. Previously, Agrawal =-=[27]-=- gave a “hitting set generator” against the class of depth-2 arithmetic circuits. This result implies a deterministic polynomial identity test for sparse polynomials. However, all of these results req... |

27 |
Primality and identity testing via Chinese Remaindering
- Agrawal, Biswas
- 2003
(Show Context)
Citation Context ...nomial. Many algorithmic problems such as finding perfect matchings in graphs [2–4], checking the equivalence of read-once branching programs [5], multi-set equality testing [6] and primality testing =-=[7]-=- reduce to this problem. Equally many applications have been discovered in complexity theory; to only mention two celebrated results, both PSPACE=IP [8] and NP=PCP [9] involved polynomial identity tes... |

23 | Reducing randomness via irrational numbers
- Chen, Kao
(Show Context)
Citation Context ...me dependence on the characteristic of the underlying field and the bit size of the points queried varies. A survey of these results can be found in [19]. Randomized complexity. A sequence of results =-=[20,13,7,19,21,22]-=- has studied (and determined in several cases) how much randomness is required in testing polynomial identities when only black-box access to the polynomial is granted. Most of these results are not c... |

23 | Blackbox polynomial identity testing for depth 3 circuits - Kayal, Saraf - 2009 |

22 | Randomness-optimal unique element isolation with applications to perfect matching and related problems - Chari, Rohatgi, et al. - 1995 |

21 | Checking Polynomial Identities over any Field: Towards a Derandomization
- Lewin, Vadhan
- 1998
(Show Context)
Citation Context ...f the represented polynomial at specific points. In the case of black-box access, one can show that randomness is necessary for polynomial identity testing (a simple dimension lower bound is given in =-=[13]-=-). When it comes to arithmetic circuits, no such lower bounds are known. However, we only know of deterministic arithmetic circuit identity tests when we have an additional “promise” about the circuit... |

17 | Computational complexity of sparse rational interpolation - Grigoriev, Karpinski, et al. - 1994 |

14 | Pseudorandom generators for low degree polynomials
- Bogdanov
- 2005
(Show Context)
Citation Context ...me dependence on the characteristic of the underlying field and the bit size of the points queried varies. A survey of these results can be found in [19]. Randomized complexity. A sequence of results =-=[20,13,7,19,21,22]-=- has studied (and determined in several cases) how much randomness is required in testing polynomial identities when only black-box access to the polynomial is granted. Most of these results are not c... |

10 |
On some approximation problems concerning sparse polynomials over finite fields
- Karpinski, Shparlinski
- 1996
(Show Context)
Citation Context ...omial in m, n and d. However, as before, the polynomial dependence on d implies an exponential worst-case running time for arithmetic circuits. In the case of finite fields, Karpinski and Shparlinski =-=[23]-=- give a randomized and a deterministic algorithm with runtimes polynomial in m, n and log d. The main issue here is that their deterministic algorithm has a linear dependence on the characteristic of ... |

9 |
Deterministic identity testing for multivariate polynomials
- Lipton, Vishnoi
- 2003
(Show Context)
Citation Context ...endence on the largest coefficient of the polynomial. Key words: theory of computation, polynomial identity testing, arithmetic circuits, derandomization ⋆ This paper builds upon the conference paper =-=[1]-=- by two of the authors. ∗ Corresponding author Email addresses: mblaeser@cs.uni-sb.de (Markus Bläser), mhardt@cs.princeton.edu (Moritz Hardt), rjl@cc.gatech.edu (Richard J. Lipton), nisheeth.vishnoi@g... |

9 | The complexity of sparse polynomial interpolation over finite fields, Applicable Algebra in Engineering, Communication and Computing 5 - Werther - 1994 |

7 |
V.: On the least prime in an arithmetic progression
- Linnik
- 1944
(Show Context)
Citation Context ... order Θ(k log k). Let P (k) denote the smallest prime number in the arithmetic progression {jk + 1 | j ≥ 1}. Linnik’s Theorem gives an unconditional upper bound on P (k). Theorem 6 (Linnik’s Theorem =-=[29]-=-) There is a constant L > 1 (called Linnik’s constant) such that P (k) < k L for every sufficiently large k ≥ k0. The best known value for L is 5.5 due to Heath and Brown. Schinzel, Sierpinski, and Ka... |

7 |
The Brun–Titchmarsh Theorem on average, Analytic Number Theory
- HARMAN
- 1996
(Show Context)
Citation Context ...Remark 12 The analysis of the previous algorithm works even if instead of Linnik’s Theorem we consider a bound on the average size of the smallest prime in an arithmetic progression. Baker and Harman =-=[31]-=- showed that Linnik’s constant is less than 1.93 on average. This leads to improvements by polynomial factors in the runtime of our algorithm. We omit the details. 4 A Simple Modular Arithmetic Approa... |

6 | Locally decodable codes with two queries and polynomial identity testing for depth 3 circuits - Dvir, Shpilka |

6 | Interpolation of depth-3 arithmetic circuits with two multiplication gates - Shpilka - 2007 |

1 | Asymptotically optimal hitting sets against polynomials
- Bläser, Hardt, et al.
- 2008
(Show Context)
Citation Context ...me dependence on the characteristic of the underlying field and the bit size of the points queried varies. A survey of these results can be found in [19]. Randomized complexity. A sequence of results =-=[20,13,7,19,21,22]-=- has studied (and determined in several cases) how much randomness is required in testing polynomial identities when only black-box access to the polynomial is granted. Most of these results are not c... |