An Efficient Algorithm for Computing Parametric Multivariate Polynomial GCD

Transcription

1 An Efficient Algorithm for Computing Parametric Multivariate Polynomial GCD Dong Lu Key Laboratory of Mathematics Mechanization Academy of Mathematics and Systems Science, CAS Joint work with Deepak Kapur, Michael Monagan, Yao Sun and Dingkang Wang July 16-19, 2018, The City University of New York, USA Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 1 / 28

2 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 2 / 28

3 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 3 / 28

4 The Problem Notations k is a field. e 1 := (1, 0) and e 2 := (0, 1). X = {x 1,..., x n } are variables. U = {u 1,..., u m } are parameters. k[u][x ] is the polynomial ring in X. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 4 / 28

5 The Problem Example 1 Let f 1, f 2 C[a][x, y], where { f 1 = ax 3 + (a 3 a + 1)x 2 y + (a 2 + 2)xy 2 + (3a 2 3)y 3, f 2 = ax 3 + (a + 1)x 2 y + 4xy 2 + 3y 3. The results are: when a = 0, gcd(f 1, f 2 ) = y(x + 3y); when a 2 2 = 0, gcd(f 1, f 2 ) = 2x 3 + (a + 2)x 2 y + 4axy 2 + 3ay 3 ; when a(a 2 2) 0, gcd(f 1, f 2 ) = ax 2 + xy + 3y 2. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 5 / 28

6 The Problem Problem: For any given parametric polynomials f 1, f 2,..., f s in k[u][x ], how to divide the parametric space and obtain the corresponding GCD on each branch quickly? Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 6 / 28

7 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 7 / 28

8 Previous Works Non-parametric polynomials case J. Moses and D.Y.Y. Yun: The EZ GCD algorithm. In Proceedings of ACM 73, ACM Press, New York, 1973, (Hensel lifting) R. Zippel: Probabilistic algorithms for sparse polynomials. In Proceedings of EUROSAM 79, Springer-Verlag, 1979, (sparse interpolation) P. Gianni and B. Trager: GCDs and factoring multivariate polynomials using Gröbner bases. In Proceedings of EUROCAL 85, Springer, Berlin, Heidelberg, 1985, (Gröbner basis) T. Sasaki and M. Suzuki: Three new algorithms for multivariate polynomial GCD. Journal of Symbolic Computation, 1992, (Gröbner basis) Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 8 / 28

9 Previous Works Parametric polynomials case S.A. Abramov and K.Y. Kvashenko: On the greatest common divisor of polynomials which depend on a parameter. In Proceedings of ISSAC 1993, (subresultant chain) A. Ayad: Complexity of algorithms for computing greatest common divisors of parametric univariate polynomials. International Journal of Algebra, 2010, (Gaussian elimination) K. Nagasaka: Parametric greatest common divisors using comprehensive Gröbner systems. In Proceedings of ISSAC 2017, (extended the ideas of Gianni and Trager as well as Sasaki and Suzuki) Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 9 / 28

10 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 10 / 28

11 The New Algorithm Non-parametric polynomials case Definition 1 Let I k[x ] be an ideal and g k[x ] \ {0}. The set I : g = {f k[x ] : fg I } is called the quotient ideal (or colon ideal) of I divided by g. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 11 / 28

12 The New Algorithm Non-parametric polynomials case Definition 1 Let I k[x ] be an ideal and g k[x ] \ {0}. The set I : g = {f k[x ] : fg I } is called the quotient ideal (or colon ideal) of I divided by g. Example 2 In k[x, y, z], let I = xy and g = xz. Then I : g = {f : xzf xy } = {f : zf y } = y. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 11 / 28

13 The New Algorithm Lemma 1 Let f 1, f 2 k[x ] \ {0}, then f 1 : f 2 is a principal ideal. Suppose that f 1 : f 2 = f, then we have gcd(f 1, f 2 ) = f 1 f. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 12 / 28

14 The New Algorithm Lemma 1 Let f 1, f 2 k[x ] \ {0}, then f 1 : f 2 is a principal ideal. Suppose that f 1 : f 2 = f, then we have gcd(f 1, f 2 ) = f 1 f. Example 3 Let f 1 = xy and f 2 = xz. According to the Example 2, we have that xy : xz = y. Then gcd(f 1, f 2 ) = f 1 /f = x. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 12 / 28

15 The New Algorithm Problem: How to compute the generator of f 1 : f 2? Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 13 / 28

16 The New Algorithm Problem: How to compute the generator of f 1 : f 2? Original Method Lemma 2 Let f 1, f 2 k[x ] \ {0} and w be a new variable, then wf 1, (w 1)f 2 k[x ] = g for some g k[x ] \ {0}. Moreover, we have f 1 : f 2 = g f 2 and gcd(f 1, f 2 ) = f 1f 2 g. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 13 / 28

17 The New Algorithm New Method Lemma 3 Let f 1, f 2 k[x ] \ {0} and be a monomial order on k[x ] 2 with e 2 e 1. Suppose G is a minimal Gröbner basis of f 1 e 1, f 2 e 1 e 2. Then there is a unique f k[x ] \ {0} such that f e 2 G and f 1 : f 2 = f. Therefore, gcd(f 1, f 2 ) = f 1 f. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 14 / 28

18 The New Algorithm New Method Lemma 3 Let f 1, f 2 k[x ] \ {0} and be a monomial order on k[x ] 2 with e 2 e 1. Suppose G is a minimal Gröbner basis of f 1 e 1, f 2 e 1 e 2. Then there is a unique f k[x ] \ {0} such that f e 2 G and f 1 : f 2 = f. Therefore, gcd(f 1, f 2 ) = f 1 f. Example 4 Let f 1 = xy, f 2 = xz. Given the lexicographic order and extend it to k[x ] 2 in a position over term with e 2 e 1. A minimal Gröbner basis of xy e 1, xz e 1 e 2 is G = {y e 2, xy e 1, xz e 1 e 2 }. So, f = y and xy : xz = y. Moreover, gcd(xy, xz) = f 1 /f = x. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 14 / 28

19 The New Algorithm Parametric polynomials case Theorem 1 Let f 1, f 2 k[u][x ] and be a monomial order on k[x ] 2 with e 2 e 1. Suppose {(A i, G i )} l i=1 is a minimal comprehensive Gröbner system of f 1 e 1, f 2 e 1 e 2. For each branch (A i, G i ), let H i = {f f e 2 G i }. Then we have 1 If H i =, then f 1 = 0 and gcd(f 1, f 2 ) = f 2 on A i. 2 If H i, then H i = {f } and gcd(f 1, f 2 ) = f 1 f on A i. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 15 / 28

20 The New Algorithm Parametric GCD Algorithm Input: f 1, f 2 k[u][x ], a constructible set A k m, and two monomial orders X, U. Output: a comprehensive GCDs: {(A i, h i )} s i=1, where h i = gcd(f 1, f 2) under any specialization from A i and s i=1a i = A. Step 1: compute a minimal comprehensive Gröbner system {(A i, G i )} s i=1 for the module f 1 e 1, f 2 e 1 e 2 on A. Step 2: For every i, let H i = {h h e 2 G i }, then do Step 2.1: if H i is empty, then h i = f 2 on A i and turn to Step 2; otherwise, turn to Step 2.2. Step 2.2: h i = f 1/h on A i. Step 3: return {(A i, h i )} s i=1. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 16 / 28

21 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 17 / 28

22 A Simple Example Example 5 Let f 1, f 2, f 3 C[a, b][x, y, z] be as follows: f 1 = ax 2 + bxy + a 2 xz + abx + abyz + b 2 y, f 2 = ax 2 + bxy + (ab a)xz a 2 x + (b 2 b)yz aby, f 3 = ax 2 + bxy + a 2 xz + (a 2 ab)x + abyz + (ab b 2 )y, Using the new algorithm to compute the GCDs of f 1, f 2, f 3. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 18 / 28

23 A Simple Example Monomial order: X and U are all lexicographic orders with x > y > z and a > b, respectively. We extend X to k[x ] 2 in a position over term with e 2 X e 1. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 19 / 28

24 A Simple Example Monomial order: X and U are all lexicographic orders with x > y > z and a > b, respectively. We extend X to k[x ] 2 in a position over term with e 2 X e 1. Step 1: Compute a minimal comprehensive Gröbner system (CGS) G 0 of f 1 e 1, f 2 e 1 e 2, and get six branches {(A i, G i )} 6 i=1. The first branch of G 0 is (A 1, G 1 ), where A 1 = V( 0 ) \ V( a 3 a 2 b + a 2 ), G 1 = {f 1 e 1, (x + az + b) e 2, ((a 2 ab + a)xz+ (a 2 + ab)x + (ab b 2 + b)yz + (ab + b 2 )y) e 1 + e 2 }. Then, H 1 = {x + az + b (x + az + b) e 2 G 1 } and the GCD of f 1 and f 2 on A 1 is h 1 = f 1 /(x + az + b) = ax + by. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 19 / 28

25 A Simple Example Similarly, the GCDs of f 1 and f 2 on other five branches are: (A 2, h 2 ) = (V( a b + 1 ) \ V( 2b 2 3b + 1 ), (b 1)x + by), (A 3, h 3 ) = (V( a, b 1 ), y), (A 4, h 4 ) = (V( 2a + 1, 2b 1 ) \ V( b 1 ), 1 2 x xy xz 1 4 x 1 4 yz y), (A 5, h 5 ) = (V( a, b ), 0), (A 6, h 6 ) = (V( a ) \ V( ab 3 ab 2 b 4 + 2b 3 b 2 ), by). Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 20 / 28

26 A Simple Example Step 2: For A 1, compute the GCD of h 1 and f 3. A minimal CGS G 1 of h 1 e 1, f 3 e 1 e 2 on A 1 has one branch: (A 1, G 11 ) = (V( 0 ) \ V( a 3 a 2 b + a 2 ), {e 2, h 1 e 1 }). Then H 11 = {1} and the GCD of h 1 and f 3 on A 1 is h 11 = h 1 /1 = ax + by. So, the GCD of f 1, f 2, f 3 on A 1 is gcd(f 1, f 2, f 3 ) = ax + by. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 21 / 28

27 A Simple Example Similarly, we can compute the GCD of h i and f 3 on A i, where i = 2,..., 6. Then the parametric GCDs of {f 1, f 2, f 3 } are (V( 0 ) \ V( a 3 a 2 b + a 2 ), ax + by), (V( a b + 1 ) \ V( 2b 2 3b + 1 ), (b 1)x + by), (V( a, b 1 ), y), (V( 2a + 1, 2b 1 ) \ V( b 1 ), x + y), (V( a, b ), 0), (V( a ) \ V( ab 3 ab 2 b 4 + 2b 3 b 2 ), by). Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 22 / 28

28 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 23 / 28

29 Implementation The new algorithm and two algorithms proposed by Nagasaka have been implemented in the Singular. More comparative examples can be generated by the codes at: The experimental data was obtained on a Core i GHz with 4GB Memory running Windows 7. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 24 / 28

30 Implementation Table: Timings(sec.) Ex. New NGT NSS > 1h > 1h > 1h > 1h > 1h > 1h > 1h > 1h NGT: K. Nagasaka, P. Gianni and B. Trager. NSS: K. Nagasaka, T. Sasaki and M. Suzuki. K. Nagasaka: Parametric greatest common divisors using comprehensive Gröbner systems, in Proceedings of ISSAC 2017, July 25-28, Kaiserslautern Germany. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 25 / 28

31 Outline 1 The Problem 2 Previous Works 3 The New Algorithm 4 A Simple Example 5 Implementation 6 Conclusions Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 26 / 28

32 Conclusions 1 We compute the GCDs of parametric polynomials by using quotient ideal and comprehensive Gröbner system of a module. 2 Our method does not need to compute the primitive part of parametric polynomials with respect to main variable. 3 Without any knowledge of f 1 or f 2 being zero or not on different branch, the GCDs of f 1 and f 2 can be obtained from the minimal comprehensive Gröbner system of f 1 e 1, f 2 e 1 e 2. 4 When we compute the GCDs of more than two parametric polynomials, the results can be obtained iteratively. Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 27 / 28

33 Thanks for your attention! Dong Lu (KLMM, AMSS, CAS) Algorithm for Computing Parametric GCD July 16-19, CUNY, New York 28 / 28