## A New Efficient Algorithm for Computing Gröbner Bases (F4) (2002)

### Cached

### Download Links

Venue: | IN: ISSAC ’02: PROCEEDINGS OF THE 2002 INTERNATIONAL SYMPOSIUM ON SYMBOLIC AND ALGEBRAIC COMPUTATION |

Citations: | 248 - 53 self |

### BibTeX

@INPROCEEDINGS{Faugère02anew,

author = {Jean-charles Faugère},

title = {A New Efficient Algorithm for Computing Gröbner Bases (F4)},

booktitle = {IN: ISSAC ’02: PROCEEDINGS OF THE 2002 INTERNATIONAL SYMPOSIUM ON SYMBOLIC AND ALGEBRAIC COMPUTATION},

year = {2002},

pages = {75--83},

publisher = {}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper introduces a new efficient algorithm for computing Gröbner bases. To avoid as much as possible intermediate computation, the algorithm computes successive truncated Gröbner bases and it replaces the classical polynomial reduction found in the Buchberger algorithm by the simultaneous reduction of several polynomials. This powerful reduction mechanism is achieved by means of a symbolic precomputation and by extensive use of sparse linear algebra methods. Current techniques in linear algebra used in Computer Algebra are reviewed together with other methods coming from the numerical field. Some previously untractable problems (Cyclic 9) are presented as well as an empirical comparison of a first implementation of this algorithm with other well known programs. This comparison pays careful attention to methodology issues. All the benchmarks and CPU times used in this paper are frequently updated and available on a Web page. Even though the new algorithm does not improve the worst case complexity it is several times faster than previous implementations both for integers and modulo computations.

### Citations

2196 |
The Art of Computer Programming
- Knuth
- 2005
(Show Context)
Citation Context ...� to recover the value © in . In fact the iteration may be stopped when ¦ �� � becomes stable (see [S.R71] for multi modular methods which may be generalized to¥ § adic method). The global complexity =-=[Knu81]-=- of this algorithms is �� £��¤ �� ¨s� Bareiss methods��¤�� � ��¤ �� ¨ ����¤ �� ¨ �� multi modular method �£ ��¤ �� ¨ �£ � ¥ -adic method ��¤ Note that -adic method is also an iterative algorithm (in f... |

500 | Computer Solution of Large Sparse Positive Definite Systems - George, Liu - 1981 |

346 |
Gröbner Bases: an Algorithmic Method in Polynomial Ideal Theory. Recent trends in multidimensional systems theory
- Buchberger
- 1985
(Show Context)
Citation Context ... more powerful. Historically, the Buchberger algorithm was the first algorithm for computing such Gröbner bases. It may eventually be possible to suggest two improvements for the Buchberger algorithm =-=[3, 4, 5]-=-. The first improvement is concerned with strategies: during a Gröbner computation, several choices can be made (select a critical pair, choose a reductor) this aspect of the problem is not directly s... |

271 | Shift register synthesis and BCH decoding - Massey - 1969 |

255 |
An Introduction to Gröbner Bases
- Adams, Loustaunau
- 1994
(Show Context)
Citation Context ... syzygy if v(g) = 0. The so called principal syzygies si,j = fjFi−fiFj are syzygies. The set of all syzygies is a module and abbreviated by Syz (for more information on syzygies we refer to [2] or to =-=[1]-=-). Let PSyz be the module generated by the principal syzygies. For a generic (random) polynomial system (f1, . . . , fm), Syz = PSyz. We can extend the admissible ordering < to P m with the following ... |

202 |
Ein Algorithmus zum Auffinden der Basiselemente des Restklassenringes nach einem nulldimensionalen
- Buchberger
- 1965
(Show Context)
Citation Context ...rithm, one has a lot of choices: select a critical pair in the list of critical pairs. ¦ choose one reductor among a list of reductors when reducing a polynomial by a list of ¦ polynomials. Buchberger=-=[Buc65]-=- proves that these choices are not important for the correctness of the algorithm, but it is well known that these choices are crucial for the total time computation. Moreover the best strategies [Gio... |

179 | Solving sparse linear equations over finite fields - Wiedemann - 1986 |

78 | The multifrontal solution of unsymmetric sets of linear systems - Duff, Reid - 1984 |

74 | On Wiedemann’s method of solving sparse linear systems - Kaltofen, Saunders - 1991 |

72 | Solving large sparse linear systems over finite fields - LaMacchia, Odlyzko - 1991 |

67 |
A criterion for detecting unnecessary reductions in the construction of Grebner bases," Proceedings of the European Conference on Symbolic and Algebraic Computing, 3-21. (Available as Vol
- Buchberger
- 1979
(Show Context)
Citation Context ...CO LTR 21.024. 1seventually be possible to suggest two improvements: sinces¡ ¢ of the times is spent computing zero it would be useful to have more powerful criterion to remove useless critical pairs =-=[Buc79]-=- (a powerful theoretical criteria exists but it is too costly); this crucial aspect of the problem is not studied in this paper, but is implemented in another algorithm (£ £ [Fau98a]). The second impr... |

64 |
Exact solution of linear equations using p−adic expansions
- Dixon
- 1982
(Show Context)
Citation Context ...th integer coefficients is to use ¥ -adic computations: we choose a prime¥ , and we compute � ¨ �� ©s¡ ¢ ¥ (very often a sparses£ decomposition is more appropriate). Then we define Theorem 3.1 (Dixon =-=[Dix82]-=-) �£ � £� ¢ ��� ¨ ¢ ¢ ���©� © ¨ ¦ �� ¨ � �� © � � �� ¨ � ¢ �� �s¡ ¢ ¥ £¤� � ¦ § ¥� ¤� ¥ ��� � ��� ¥ � � ¦ ¦ �� § ¢ ¨ ¡s¡ ¢ ¥ � � If ¨ is upper bound of the coefficients of � and ¢ it is sufficient to ... |

52 |
A Block Lanczos Algorithm for Finding Dependencies over
- Montgomery
- 1995
(Show Context)
Citation Context ...r context are the following: 16 ©�s3.2.1 Iterative solvers Afters�� � steps we obtain an exact solution (up to rounding errors). This is the case of conjugate gradient method or the Lanczos algorithm =-=[Mon95]-=- since after� steps the result is exact. Another well known iterative algorithm is the Wiedemann algorithm [Wie86, Kal91], which uses the efficient Berlekamp and Massey algorithm [L.69] to recover the... |

48 | Résolution des systèmes d'équations algébriques - Lazard - 1981 |

46 | A sparse matrix library in C++ for high performance architectures - Dongarra, Lumsdaine, et al. - 1994 |

41 |
On an installation of Buchberger's algorithm
- Gebauer, Möller
- 1988
(Show Context)
Citation Context ...e Buchberger Criteria. Since it is not the subject of this paper to improve the Buchberger Criteria we will use a standard implementation of these criteria such as the Gebauer and Moller installation =-=[GM88]-=-: Buchberger Criteria Specification: Input: � ¢ �§¨ � � �§¨ � �¨ £ ¥ £ � � ¡ � ¢ © �� � � © �� �¡ � ¦ ��§ ¨ © �¥� �£���¥ � ¢ © � � ¤ ¥¦ § �� � � �§¥� �� ���¥��§¤ �§ �� ¤��¥ §© � �� �§¤ � § ��� ��� � ¥... |

39 | An algorithmic criterion for the solvability of algebraic systems of equations 251 - Buchberger - 1998 |

39 | Involutive Bases of Polynomial Ideals - Gerdt, Blinkov - 1996 |

27 | Highly parallel sparse triangular solution, in Graph Theory and Sparse Matrix Computations - ALVARADO, POTHEN, et al. - 1993 |

24 | One sugar cube, please” or selection strategies in the Buchberger algorithm
- Giovini, Mora, et al.
- 1991
(Show Context)
Citation Context .... The second improvement is concerned with strategies: during a Gröbner computation, several choices can be made (select a critical pair, choose a reductor) and even if strategies have been proposed (=-=[Gio91]-=- or even [Ger95]) the heuristics which they rely on could not be satisfactorily explained. So it is difficult to be convinced that they are optimal optimizations. Another bad consequence is that it is... |

22 | User’s guide for the unsymmetric-pattern multifrontal package (UMFPACK
- Davis
- 1993
(Show Context)
Citation Context ... which is predictable. The interest of this method is that both preprocessing may be done simultaneously. A large number of implementation for these methods are available (mainly in C/C++ or Fortran) =-=[Dav95]-=- or even in Matlab. Unfortunately these programs and some algorithms are not very robust: very often the input matrix� must be non singular or square or positive definite. On the other hand, parallel ... |

22 | Ein Algorithmus zum Au nden der Basiselemente des Restklassenringes nach einem nulldimensionalen Polynomideal - Buchberger - 1965 |

18 |
Gaussian elimination and resolution of systems of algebraic equations. EUROCAL
- Lazard, Gröbner-Bases
- 1983
(Show Context)
Citation Context ... total degree (Definition 2.6) are� £ ), then we call ¢ � a (truncated) £ -Gröbner basis of ¢ . The following theorem give a structure to this list when the polynomials are homogenenous. Theorem 2.5 (=-=[Laz83]-=-, [Bec93] p. 471) For homogeneous polynomials� © � � � � � � � , ¢ � is a Gröbner basis “up to degree £ ” that is to say: � §¨ ¦ ¥sis well defined for polynomials � such that � � � �� �s£ . ¦ ¡ ¥ � ¢ ... |

18 | Liu Computer Solution of Large Sparse Positive Definite Systems - George, H - 1981 |

18 | Solving sparse linear equations over - Wiedemann - 1986 |

17 | de Vorst, Parallel sparse LU decomposition on a mesh network of transputers - Stappen, Bisseling, et al. - 1993 |

11 | Fundamental Algorithms
- Bini, Pan
- 1994
(Show Context)
Citation Context ...global complexity for solving the system iss�� �� �� instead ofs��¤� . When the matrix has a regular pattern it is possible to apply even more efficient techniques (for Toeplitz matrices for instance =-=[Bin94]-=-). These methods can not be applied in our case since the pattern of the generated matrices is not regular enough. A significant drawback of these methods is that there is no speedup if we try to solv... |

11 | Distributed solution of sparse linear systems - Heath, Raghavan - 1993 |

10 | C.: Gröbner bases computation using syzygies - Möller, Mora, et al. - 1992 |

9 | The multifrontal solution of unsymmetric sets of linear systems - Du, Reid - 1984 |

7 |
Groebner Bases, a Computationnal Approach to Commutative Algebra. Graduate Texts in Mathematics
- Becker, Weispfenning
- 1993
(Show Context)
Citation Context ...resenting the new algorithm. This section has been divided into several parts: first (2.1), we review the necessary mathematical notations (we make the choice to use the same notations as in the book =-=[Bec93]-=-) and in 2.2 we establish the link between linear algebra (matrices) and polynomial algebra. Then we present (2.3) a basic version of the algorithm without any criteria to eliminate useless pairs. A i... |

7 | Large sparse sets of linear equations - Reid |

6 |
Computing Grobner basis without reduction to zero (F5
- Faugere
- 1998
(Show Context)
Citation Context ...ess critical pairs [Buc79] (a powerful theoretical criteria exists but it is too costly); this crucial aspect of the problem is not studied in this paper, but is implemented in another algorithm (£ £ =-=[Fau98a]-=-). The second improvement is concerned with strategies: during a Gröbner computation, several choices can be made (select a critical pair, choose a reductor) and even if strategies have been proposed ... |

5 |
Buchberger’s algorithm and staggered linear bases
- GEBAUER, MOLLER
- 1986
(Show Context)
Citation Context ... other words, (g1, . . . , gm) is a syzygy. Several papers investigate those issues: Buchberger [4] proposes two criteria to remove a lot of useless critical pairs; staggered linear bases are used in =-=[7]-=-; the idea of [10] is to compute simultaneously a Gröbner basis and a basis of the module of syzygies: a critical pair is not considered if the corresponding syzygy is a linear combination of some ele... |

4 | The Sparse Matrix Manipulation System User and Reference
- Alvarado
- 1993
(Show Context)
Citation Context ...the input matrix� must be non singular or square or positive definite. On the other hand, parallel algorithms and parallel implementations exist [Van93, Hea, Don, Alv, Pey]. We have modified the smms =-=[Alv93]-=- program in order to work with modulo¥ coefficients in order to evaluate the costs of different algorithms. Solving large linear equations modulo a small prime number has been studied in cryptology [L... |

4 | editors. Graph theory and sparse matrix computation - George, Gilbert, et al. - 1993 |

3 |
Design of nonseparable bidimensional wavelets and filter banks using Grobner bases techniques
- Faug`ere, Martin, et al.
- 1996
(Show Context)
Citation Context ...can be solved, the algorithms are competitive with numerical methods. The main conclusion to be drawn from practice and experience of solving polynomial systems coming from various fields (industrial =-=[Fau98b]-=- problems, pure mathematics [Fro96]) is the following: first of all, even though the computation of a Gröbner basis is a crucial point it must be emphasized that it is only one step in the full solvin... |

3 | Solving systems of algebraic equations - Lazard - 2001 |

3 | Sparse Matrices and their applications - Rose, Willoughby - 1972 |

2 | A.M.: Solving large sparse linear systems over finite fields - LaMacchia, Odlyzko - 1991 |

2 | The Gröbner Basis Algorithm and Subresultant Theory - Mandache - 1994 |

2 | available via anonymous ftp on zariski.harvard.edu - Manual - 1989 |

1 | Homogeneous Parallel Algorithm - Attardi, Traverso - 1996 |

1 |
The Magma Computational Algebra System
- Cannon
- 1998
(Show Context)
Citation Context ...f methodologies in the field of benchmark for linear system solving. We adopt the following points: 19s1. We compare the new algorithm with state of the art Gröbner bases implementation (namely Magma =-=[Can98]-=-, PoSSo/Frisco Gröbner engine [Tra97], Macaulay 2 [Sti89, Gra97], Singular [Gre97], Asir [Tak96], Cocoa, Axiom, Maple [Cha91] and Gb [Faub]). It is also crucial to compare the implementation of the ne... |

1 |
Benchmarks for polynomial solvers. avalaible on the WEB http://posso.lip6.fr/˜jcf/benchs.html
- Faugère
(Show Context)
Citation Context ... high level language/compiler. £ ¤ has been implemented in � and most of the competitors are implemented in C/C++. 2. The list of examples is also a crucial issue: the examples can be easily accessed =-=[Faua]-=- (the web site contains pointers to the Frisco test suite). The list is composed of classical benchmarks (Cyclic � , Katsura � ) but also of industrial examples (coming from signal theory, robotics). ... |

1 |
On line documentation of Gb. avalaible
- Faugère
(Show Context)
Citation Context ...state of the art Gröbner bases implementation (namely Magma [Can98], PoSSo/Frisco Gröbner engine [Tra97], Macaulay 2 [Sti89, Gra97], Singular [Gre97], Asir [Tak96], Cocoa, Axiom, Maple [Cha91] and Gb =-=[Faub]-=-). It is also crucial to compare the implementation of the new algorithm with the Buchberger algorithm implemented by the same person (in this case the author). In our opinion it is also important to ... |

1 |
Singularité résiduelle et Problème du centre
- Fronville
- 1996
(Show Context)
Citation Context ...mpetitive with numerical methods. The main conclusion to be drawn from practice and experience of solving polynomial systems coming from various fields (industrial [Fau98b] problems, pure mathematics =-=[Fro96]-=-) is the following: first of all, even though the computation of a Gröbner basis is a crucial point it must be emphasized that it is only one step in the full solving process (change of ordering, tria... |

1 |
Involutive Polynomial Bases
- Gerdt
- 1995
(Show Context)
Citation Context ...rovement is concerned with strategies: during a Gröbner computation, several choices can be made (select a critical pair, choose a reductor) and even if strategies have been proposed ([Gio91] or even =-=[Ger95]-=-) the heuristics which they rely on could not be satisfactorily explained. So it is difficult to be convinced that they are optimal optimizations. Another bad consequence is that it is very difficult ... |

1 | nouvel algorithme de calcul d’une Base de Gröbner - Un - 1998 |

1 | A note on the relationship between involutive bases and Gröbner bases - Mandache - 1996 |