## Straight-line programs in geometric elimination theory (1998)

Venue: | J. Pure Appl. Algebra |

Citations: | 59 - 14 self |

### BibTeX

@ARTICLE{Giusti98straight-lineprograms,

author = {M. Giusti and J. Heintz and J. E. Morais and J. Morgenstern L. M. Pardo},

title = {Straight-line programs in geometric elimination theory},

journal = {J. Pure Appl. Algebra},

year = {1998}

}

### Years of Citing Articles

### OpenURL

### Abstract

Dedicated to Volker Strassen for his work on complexity We present a new method for solving symbolically zero–dimensional polynomial equation systems in the affine and toric case. The main feature of our method is the use of problem adapted data structures: arithmetic networks and straight–line programs. For sequential time complexity measured by network size we obtain the following result: it is possible to solve any affine or toric zero–dimensional equation system in non–uniform sequential time which is polynomial in the length of the input description and the “geometric degree ” of the equation system. Here, the input is thought to be given by a straight–line program (or alternatively in sparse representation), and the length of the input is measured by number of variables, degree of equations and size of the program (or sparsity of the equations). The geometric degree of the input system has to be adequately defined. It is always bounded by the algebraic–combinatoric “Bézout number ” of the system which is given by the Hilbert function of a suitable homogeneous ideal. However, in many important cases, the value of the geometric

### Citations

749 | Factoring polynomials with rational coefficients - Lenstra, Lenstra, et al. - 1982 |

228 |
Intersection theory, Ergebnisse der Mathematik und ihrer Grenzgebiete (3
- Fulton
- 1984
(Show Context)
Citation Context ...fi) by n − i generic affine hyperplanes of A n (more generally, we define the affine degree of a closed Zariski subset of A n as the sum of the degrees of its irreducible components. See e.g. [29] or =-=[21]-=- for this notion of degree and its motivation). We denote the affine degree of V (f1,... ,fi) by degV (f1,... ,fi). Let us now define δ := max{degV (f1,... ,fi) ; 1 ≤ i ≤ n} as the (geometric) affine ... |

164 |
The Algebraic Theory of Modular Systems
- Macaulay
- 1916
(Show Context)
Citation Context ...ocally a complete intersection with respect to the isolated points of V . In view of the surprisingly low (linear) regularity bound for homogeneous global complete intersection ideals (see [6], [44], =-=[46]-=-) one is tempted to replace the original equations f1,... ,fn ∈ k[X1,... ,Xn] by new homogeneous ones, G1,... ,Gn ∈ k[X0,... ,Xn], and the affine variety V = V (f1,... ,fn) by the corresponding projec... |

118 | On computing the determinant in small parallel time using a small number of processors - Berkowitz - 1984 |

86 | Pathways to solutions, fixed points and equilibria - Zangwill, Garcia - 1981 |

62 | A probabilistic remark on algebraic program testing
- DeMillo, Lipton
- 1978
(Show Context)
Citation Context ... idea appears implicitly or explicitly in the following representative (although not exhaustive) list of papers, mostly dedicated to probabilistic testing of polynomial identities : [32], [34], [55], =-=[15]-=-, [64]. First applications of this point of view to computer algebra dealt only with elimination of just one variable (absolute primality testing, greatest common divisor computation and factorization... |

60 |
A computational method for diophantine approximation, Algorithms in algebraic geometry and applications
- Krick, Pardo
- 1994
(Show Context)
Citation Context ... following more explicit form : find univariate polynomials q,v1,... ,vn ∈ k[T] such that the set of isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], =-=[40]-=- and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n g... |

55 | Zeros, multiplicities and idempotents for zero-dimensional systems, Algorithms in algebraic geometry and applications
- Alonso, Becker, et al.
- 1996
(Show Context)
Citation Context ...ollowing methods for the construction of a regular sequence of auxiliar homogeneous 2polynomials G1,... ,Gn , from whose zeroes the solutions of the original system f1,...,fn are then extracted (see =-=[1]-=-, [10], [11], [12], [16], [22], [24], [27], [40], [41], [43], [47], [56]). The polynomials G1,... ,Gn are homogeneous in the variables X0,X1,... ,Xn . Moreover they depend on a deformation parameter ε... |

53 | Greatest common divisors of polynomials given by straight-line programs
- Kaltofen
- 1988
(Show Context)
Citation Context ... of view to computer algebra dealt only with elimination of just one variable (absolute primality testing, greatest common divisor computation and factorization of multivariate polynomials; see [33], =-=[37]-=-, [38]) and were later extended to multivariate elimination problems by means of “duality techniques” of different type and “Vermeidung von Divisionen” ([19], [24], [26], [40], [41]; see also [31]). S... |

43 | Probabilistic algorithms for deciding equivalence of straight-line programs
- Ibarra, Moran
- 1983
(Show Context)
Citation Context ...s second drawback can be avoided coding rational numbers by straight–line programs. Identity of rational numbers given by such encodings can be detected in random polynomial time by a BPP–test ([54], =-=[34]-=-) or a fixed sample test (as in [32] or [13]) which is due to [28]. Summarizing this we can say that in case Ω := 0Q and k:= 0Q(θ1,... ,θm) with θ1,... ,θm algebraically independent, our algorithms ha... |

37 |
Kähler Differentials
- Kunz
- 1986
(Show Context)
Citation Context ...ule basis of C given by the elements ¯1,ū,...,ū ¯ δ−1 ). Since the k –algebra isomorphism τ is k[X1,... ,Xr]ρ –linear it leaves the traces fixed. We denote them therefore by the same symbol “Tr” (see =-=[42]-=-, Appendix F as reference for traces in noetherian algebras over rings). Let r < j ≤ n. As in the proof of Lemma 1 let us write vj = ∑ a (j) m um with a (j) m ∈ k[X1,... ,Xr]. Observe that ρ ¯ Xj = vj... |

30 | Factorization of polynomials given by straight-line programs
- Kaltofen
- 1989
(Show Context)
Citation Context ...ew to computer algebra dealt only with elimination of just one variable (absolute primality testing, greatest common divisor computation and factorization of multivariate polynomials; see [33], [37], =-=[38]-=-) and were later extended to multivariate elimination problems by means of “duality techniques” of different type and “Vermeidung von Divisionen” ([19], [24], [26], [40], [41]; see also [31]). Straigh... |

30 |
How lower and upper complexity bounds meet in elimination theory
- Pardo
- 1995
(Show Context)
Citation Context ...(for which they seek approximate solutions) is inefficient in terms of bit–complexity if the rational numbers appearing in it are given in binary coding. This fact is due to Liouville’s estimate (see =-=[50]-=-). However this second drawback can be avoided coding rational numbers by straight–line programs. Identity of rational numbers given by such encodings can be detected in random polynomial time by a BP... |

29 | Testing polynomials which are easy to compute - Heintz, Schnorr |

26 |
Subexponential time solving systems of algebraic equations
- Chistov, Grigoriev
- 1983
(Show Context)
Citation Context ...set of isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], =-=[12]-=-, [17], [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input cons... |

24 |
On the intrinsic complexity of elimination theory
- Heintz, Morgenstern
- 1993
(Show Context)
Citation Context ...1,...,fn are generic) our complexity bounds are roughly the same as those obtained by deformation based algorithms. This is not surprising in view of the known lower bounds for elimination problems ( =-=[31]-=-, [57]). An important aspect of elimination procedures consists in the encoding of the polynomials appearing as inputs, outputs or intermediate results of the algorithm. Encoding polynomials ( ) by th... |

24 |
Bounds for traces in complete intersections and degrees in the Nullstellensatz
- Sabia, Solerno
- 1995
(Show Context)
Citation Context ...n(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], [43], [51], =-=[52]-=-), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all irreducible components (and, in particular, the is... |

22 | Effective Bézout identities - Berenstein, Yger - 1991 |

22 |
On the efficiency of effective Nullstellensätze
- Giusti, Heintz, et al.
- 1993
(Show Context)
Citation Context ...f V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], =-=[26]-=-, [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all irreducible co... |

19 |
On the computational complexity of polynomials and bilinear mappings. Asurvey.InApplied Algebra, Algebraic Algorithms and Error Correcting Codes
- Heintz
- 1989
(Show Context)
Citation Context ...nly to contain divisions by non–zero elements of k ). We denote the nonscalar size (“nonscalar length” in more traditional terminology) and the nonscalar depth of β by L and ℓ respectively (see [23], =-=[30]-=-, [40], [59] and Section 2 for the notions of straight–line program, arithmetic network and the complexity measures we shall use subsequently). Let us first consider the affine case. Here, we suppose ... |

19 |
On the complexity of zero-dimensional algebraic systems
- Lakshman, Lazard
- 1991
(Show Context)
Citation Context ...v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], =-=[43]-=-, [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all irreducible components (and, in partic... |

18 |
Une approche informatique pour l’approximation diophantienne
- Krick, Pardo
- 1994
(Show Context)
Citation Context ...nomials; see [33], [37], [38]) and were later extended to multivariate elimination problems by means of “duality techniques” of different type and “Vermeidung von Divisionen” ([19], [24], [26], [40], =-=[41]-=-; see also [31]). Straight–line programs and arithmetic circuits have their origins in the design of semi– numerical algorithms and represent a link between numerical analysis and computer science (se... |

17 |
Absolute Primality of Polynomials is Decidable in Random Polynomial Time
- Heintz, Sieveking
- 1981
(Show Context)
Citation Context ... point of view to computer algebra dealt only with elimination of just one variable (absolute primality testing, greatest common divisor computation and factorization of multivariate polynomials; see =-=[33]-=-, [37], [38]) and were later extended to multivariate elimination problems by means of “duality techniques” of different type and “Vermeidung von Divisionen” ([19], [24], [26], [40], [41]; see also [3... |

17 |
Lower bounds for arithmetic networks
- Montaña, Morais, et al.
- 1996
(Show Context)
Citation Context ...de of G by a pair of integers (i,j), where i represents the depth of the node and j is an ordinal number assigned to the node which is given by some ordering of the set of nodes of depth i (see [40], =-=[48]-=- for this type of encoding of arithmetic networks). Let (i,j) be a node of G . If (i,j) has indegree 0 and does not represent an input, the labeling Q assigns to this node a constant from K (which we ... |

14 |
The complexity of partial derivatives, Theoret
- Baur, Strassen
- 1983
(Show Context)
Citation Context ...ynomials depending on the variables Xr+1,... ,Xn ). In order to compute ( ˜ R (l) r+1 ,... , ˜ R (l) n ) we have to evaluate the polynofor 1 ≤ k ≤ i and r < j ≤ n at 3⌈log2δ⌉ + 2 iteration points. By =-=[2]-=- and [40] (see also [49]) this can be done by a division free straight– line program of size O(Llog2δ) ) and depth O(ℓlog2δ). Moreover we have to invert the jacobian matrix at the same 3⌈log2δ⌉ + 2 it... |

14 |
Definability and fast quantifier elimination over algebraically dosed fields, Theoret
- Heintz
- 1983
(Show Context)
Citation Context ...(f1,...,fi) by n − i generic affine hyperplanes of A n (more generally, we define the affine degree of a closed Zariski subset of A n as the sum of the degrees of its irreducible components. See e.g. =-=[29]-=- or [21] for this notion of degree and its motivation). We denote the affine degree of V (f1,... ,fi) by degV (f1,... ,fi). Let us now define δ := max{degV (f1,... ,fi) ; 1 ≤ i ≤ n} as the (geometric)... |

14 | R'esolution des syst`emes d"equations alg'ebriques, Theoretical Computer Science 15 - LAZARD - 1981 |

14 |
Dénominateurs dans le théorème des zéros de
- Philippon
- 1991
(Show Context)
Citation Context ...... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], [43], =-=[51]-=-, [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all irreducible components (and, in particular, ... |

11 | Une formule de Jacobi et ses conséquences - Berenstein, Yger - 1991 |

11 |
Évaluation dynamique et clôture algébrique en Axiom
- Duval
- 1995
(Show Context)
Citation Context ...tworks and straight–line programs with parameters in ¯ k (or k ). A similar ground idea may be found in the impressive theoretical and practical work of D. Duval and her school on dynamic evaluation (=-=[18]-=-). Closest to our complexity results comes the numeric method of M. Shub and S. Smale for finding “approximate” solutions of zero–dimensional homogeneous “average” equations systems ([56]). Their meth... |

10 | Borne simplemente exponentielle pour les degrés dans le théorème des zéros sur un corps de charactéristique quelconte - Caniglia, Galligo, et al. - 1988 |

8 |
Smietanski: Sur la complexité du théorème des zéros
- Fitchas, Giusti, et al.
- 1993
(Show Context)
Citation Context ...ted points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], =-=[19]-=-, [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all ir... |

8 | Generic local structure of the morphisms in commutative algebra - Iversen - 1973 |

6 |
The membership problem for unmixed ideals is solvable in single exponential time
- Dickenstein, Fitchas, et al.
- 1991
(Show Context)
Citation Context ...e construction of a regular sequence of auxiliar homogeneous 2polynomials G1,... ,Gn , from whose zeroes the solutions of the original system f1,...,fn are then extracted (see [1], [10], [11], [12], =-=[16]-=-, [22], [24], [27], [40], [41], [43], [47], [56]). The polynomials G1,... ,Gn are homogeneous in the variables X0,X1,... ,Xn . Moreover they depend on a deformation parameter ε and form a regular sequ... |

6 |
A combinatorial proof of the effective Nullstellensatz
- Dubé
- 1993
(Show Context)
Citation Context ... isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], =-=[17]-=-, [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves ... |

6 | Pardo: When polynomial equation systems can be "solved" fast - Giusti, Heintz, et al. - 1995 |

5 | On the Sprindˇzuk–Weissauer approach to universal Hilbert subsets - Fried - 1985 |

4 |
Sur le degré des relations entre polynômes
- Briançon
- 1982
(Show Context)
Citation Context ...,fn forms locally a complete intersection with respect to the isolated points of V . In view of the surprisingly low (linear) regularity bound for homogeneous global complete intersection ideals (see =-=[6]-=-, [44], [46]) one is tempted to replace the original equations f1,... ,fn ∈ k[X1,... ,Xn] by new homogeneous ones, G1,... ,Gn ∈ k[X0,... ,Xn], and the affine variety V = V (f1,... ,fn) by the correspo... |

4 |
Some algebraic and geometric problems
- Canny
- 1988
(Show Context)
Citation Context ...t the set of isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], =-=[10]-=-, [12], [17], [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the inpu... |

4 | Polynomial random test for the equivalence problem of integers given by arithmetic circuits - Hagele, Montana - 1997 |

4 |
How to compute fast a function and all its derivatives, Prépublication No. 49, Université de Nice
- Morgenstern
- 1984
(Show Context)
Citation Context ...he variables Xr+1,... ,Xn ). In order to compute ( ˜ R (l) r+1 ,... , ˜ R (l) n ) we have to evaluate the polynofor 1 ≤ k ≤ i and r < j ≤ n at 3⌈log2δ⌉ + 2 iteration points. By [2] and [40] (see also =-=[49]-=-) this can be done by a division free straight– line program of size O(Llog2δ) ) and depth O(ℓlog2δ). Moreover we have to invert the jacobian matrix at the same 3⌈log2δ⌉ + 2 iteration points. This cos... |

3 |
Polynomial --time computation of the dimension of components of algebraic varieties in zero--characteristic, Preprint Universit'e Paris Val de Marne
- Chistov
(Show Context)
Citation Context ...thods for the construction of a regular sequence of auxiliar homogeneous 2polynomials G1,... ,Gn , from whose zeroes the solutions of the original system f1,...,fn are then extracted (see [1], [10], =-=[11]-=-, [12], [16], [22], [24], [27], [40], [41], [43], [47], [56]). The polynomials G1,... ,Gn are homogeneous in the variables X0,X1,... ,Xn . Moreover they depend on a deformation parameter ε and form a ... |

3 | Using symmetric functions to describe the solution of a zero– dimensional ideal - Vega, Trujillo - 1995 |

3 |
Systems of algebraic equations solved by means of endomorphisms
- Moeller
- 1993
(Show Context)
Citation Context ...xiliar homogeneous 2polynomials G1,... ,Gn , from whose zeroes the solutions of the original system f1,...,fn are then extracted (see [1], [10], [11], [12], [16], [22], [24], [27], [40], [41], [43], =-=[47]-=-, [56]). The polynomials G1,... ,Gn are homogeneous in the variables X0,X1,... ,Xn . Moreover they depend on a deformation parameter ε and form a regular sequence in k(ε)[X0,X1,... ,Xn]. They define a... |

2 |
Brownawell : Bounds for the degree
- W
- 1987
(Show Context)
Citation Context ...∈ k[T] such that the set of isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], =-=[7]-=-, [8], [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This prepara... |

2 |
A.Galligo and J.Heintz : Some new effectivity bounds
- Caniglia
(Show Context)
Citation Context ...] such that the set of isolated points of V can be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], =-=[8]-=-, [9], [10], [12], [17], [19], [24], [26], [39], [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation ... |

2 |
On the irreducibility of the polynomials P(t m ,Y
- Dèbes
- 1992
(Show Context)
Citation Context ...l is nonuniform by the way we have introduced it. However in view of [28] and the evidence that “efficient” (rather than “effective”) versions of Hilbert’s Irreducibility Theorem must exist (see e.g. =-=[14]-=-, [20], [58] and [62]), we may hope to obtain at least a reasonable randomization (if not uniformization) of the algorithm underlying the next statement which we pronounce without proof. Remark 2 Supp... |

2 |
Sharp Effective Nullstllensatz
- Koll'ar
- 1988
(Show Context)
Citation Context ...n be written as {(v1(t),... ,vn(t)) ; t ∈ A 1 ( ¯ k) , q(t) = 0} (see [24], [40] and Lemma 7 below). As many authors do (see for instance, [3], [4], [7], [8], [9], [10], [12], [17], [19], [24], [26], =-=[39]-=-, [40], [41], [43], [51], [52]), we replace the original input system by n generic k –linear combinations of the equations f1,... ,fs . This preparation of the input conserves all irreducible componen... |

1 |
On real Turing machines that toss coins, manuscript
- Cucker, Karpinski, et al.
- 1995
(Show Context)
Citation Context ...onal numbers by straight–line programs. Identity of rational numbers given by such encodings can be detected in random polynomial time by a BPP–test ([54], [34]) or a fixed sample test (as in [32] or =-=[13]-=-) which is due to [28]. Summarizing this we can say that in case Ω := 0Q and k:= 0Q(θ1,... ,θm) with θ1,... ,θm algebraically independent, our algorithms have a reasonable translation to the (probabil... |

1 |
Ja' : An introduction to parallel algorithms, Addison--Wesley
- Ja'
- 1992
(Show Context)
Citation Context ...utes the rational functions ũ (1),... ,ũ(¯ δ) with a fast and well–parallelizable algorithm for the evaluation of the set of elementary symmetric functions {σm ; 0 ≤ m ≤ ¯ δ} (see [61], Satz 3.1. and =-=[36]-=-, Chapter 8, Exercise 8.15) we obtain an arithmetic circuit γ0 in ¯ k(X1,... ,Xr) which for 0 ≤ m < ¯ δ computes all rational functions (−1) ¯ δ−mσm(ũ (1),... ,ũ(¯ δ) ). The circuit γ0 has size O((i5 ... |