For an (associative but not necessarily commutative) ring with - - PowerPoint PPT Presentation

for an associative but not necessarily commutative ring
SMART_READER_LITE
LIVE PREVIEW

For an (associative but not necessarily commutative) ring with - - PowerPoint PPT Presentation

B UCHBERGER T HEORY FOR E FFECTIVE A SSOCIATIVE R INGS T. M., Seven variations on standard bases , (1988) A solution if the ring is a vectorspace over a field A PEL J., Computational ideal theory in finitely generated extension rings , T.C.S. 224


slide-1
SLIDE 1

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

  • T. M., Seven variations on standard bases, (1988)

A solution if the ring is a vectorspace over a field APEL J., Computational ideal theory in finitely generated extension rings, T.C.S. 224 (2000), 1–33 Extension to suitable rings which are algebra over a ring Gateva, Weispfenning and Passau group, Reinert, . . .

slide-2
SLIDE 2

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

  • T. M., Seven variations on standard bases, (1988)

A solution if the ring is a vectorspace over a field APEL J., Computational ideal theory in finitely generated extension rings, T.C.S. 224 (2000), 1–33 Extension to suitable rings which are algebra over a ring Gateva, Weispfenning and Passau group, Reinert, . . .

slide-3
SLIDE 3

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

  • T. M., Seven variations on standard bases, (1988)

A solution if the ring is a vectorspace over a field APEL J., Computational ideal theory in finitely generated extension rings, T.C.S. 224 (2000), 1–33 Extension to suitable rings which are algebra over a ring Gateva, Weispfenning and Passau group, Reinert, . . .

slide-4
SLIDE 4

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

  • T. M., Seven variations on standard bases, (1988)

A solution if the ring is a vectorspace over a field APEL J., Computational ideal theory in finitely generated extension rings, T.C.S. 224 (2000), 1–33 Extension to suitable rings which are algebra over a ring Gateva, Weispfenning and Passau group, Reinert, . . .

slide-5
SLIDE 5

THEOREM For an (associative but not necessarily commutative) ring with identity A, there is a (not necessarily finite nor necessarily countable) set Z and a projection Π : Q := ZZ ։ A so that, denoting I ⊂ Q = ZZ the bilateral ideal I := ker(Π), we have A = Q/I.

slide-6
SLIDE 6

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Let R be an (associative but not necessarily commutative) ring with identity 1R and A another (associative but not necessarily commutative) ring with identity 1A which is a left module on R. We consider A to be effectively given when we are given

  • sets v := {x1, . . . , xj, . . .}, V := {X1, . . . , Xi, . . .}, which are

countable and

  • Z := v ⊔ V = {x1, . . . , xj, . . . , X1, . . . , Xi, . . .};
  • rings R := Zv ⊂ Q := ZZ;
  • projections π : R = Zx1, . . . , xj, . . . ։ R and
  • Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A which satisfies

Π(xj) = π(xj)1A, for each xj ∈ v, so that Π (R) = {r1A : r ∈ R} ⊂ A.

slide-7
SLIDE 7

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Let R be an (associative but not necessarily commutative) ring with identity 1R and A another (associative but not necessarily commutative) ring with identity 1A which is a left module on R. We consider A to be effectively given when we are given

  • sets v := {x1, . . . , xj, . . .}, V := {X1, . . . , Xi, . . .}, which are

countable and

  • Z := v ⊔ V = {x1, . . . , xj, . . . , X1, . . . , Xi, . . .};
  • rings R := Zv ⊂ Q := ZZ;
  • projections π : R = Zx1, . . . , xj, . . . ։ R and
  • Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A which satisfies

Π(xj) = π(xj)1A, for each xj ∈ v, so that Π (R) = {r1A : r ∈ R} ⊂ A.

slide-8
SLIDE 8

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Let R be an (associative but not necessarily commutative) ring with identity 1R and A another (associative but not necessarily commutative) ring with identity 1A which is a left module on R. We consider A to be effectively given when we are given

  • sets v := {x1, . . . , xj, . . .}, V := {X1, . . . , Xi, . . .}, which are

countable and

  • Z := v ⊔ V = {x1, . . . , xj, . . . , X1, . . . , Xi, . . .};
  • rings R := Zv ⊂ Q := ZZ;
  • projections π : R = Zx1, . . . , xj, . . . ։ R and
  • Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A which satisfies

Π(xj) = π(xj)1A, for each xj ∈ v, so that Π (R) = {r1A : r ∈ R} ⊂ A.

slide-9
SLIDE 9

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Let R be an (associative but not necessarily commutative) ring with identity 1R and A another (associative but not necessarily commutative) ring with identity 1A which is a left module on R. We consider A to be effectively given when we are given

  • sets v := {x1, . . . , xj, . . .}, V := {X1, . . . , Xi, . . .}, which are

countable and

  • Z := v ⊔ V = {x1, . . . , xj, . . . , X1, . . . , Xi, . . .};
  • rings R := Zv ⊂ Q := ZZ;
  • projections π : R = Zx1, . . . , xj, . . . ։ R and
  • Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A which satisfies

Π(xj) = π(xj)1A, for each xj ∈ v, so that Π (R) = {r1A : r ∈ R} ⊂ A.

slide-10
SLIDE 10

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A π : R = Zx1, . . . , xj, . . . ։ R Thus denoting

  • I := ker(Π) ⊂ Q and
  • I := I ∩ R = ker(π) ⊂ R,

we have A = Q/I and R = R/I; moreover we can wlog assume that R ⊂ A. Q as Z-module: using as alphabet V all symbols representing the primes. Further, when considering A as effectively given in this way, we explicitly require that Xixj ≡

i

  • l=1

π(alij)Xl + π(a0ij) mod I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v. If not ZX, Y as left Z[X]-module requires 1 ≥ X ≥ X 2

slide-11
SLIDE 11

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A π : R = Zx1, . . . , xj, . . . ։ R Thus denoting

  • I := ker(Π) ⊂ Q and
  • I := I ∩ R = ker(π) ⊂ R,

we have A = Q/I and R = R/I; moreover we can wlog assume that R ⊂ A. Q as Z-module: using as alphabet V all symbols representing the primes. Further, when considering A as effectively given in this way, we explicitly require that Xixj ≡

i

  • l=1

π(alij)Xl + π(a0ij) mod I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v. If not ZX, Y as left Z[X]-module requires 1 ≥ X ≥ X 2

slide-12
SLIDE 12

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A π : R = Zx1, . . . , xj, . . . ։ R Thus denoting

  • I := ker(Π) ⊂ Q and
  • I := I ∩ R = ker(π) ⊂ R,

we have A = Q/I and R = R/I; moreover we can wlog assume that R ⊂ A. Q as Z-module: using as alphabet V all symbols representing the primes. Further, when considering A as effectively given in this way, we explicitly require that Xixj ≡

i

  • l=1

π(alij)Xl + π(a0ij) mod I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v. If not ZX, Y as left Z[X]-module requires 1 ≥ X ≥ X 2

slide-13
SLIDE 13

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A π : R = Zx1, . . . , xj, . . . ։ R Thus denoting

  • I := ker(Π) ⊂ Q and
  • I := I ∩ R = ker(π) ⊂ R,

we have A = Q/I and R = R/I; moreover we can wlog assume that R ⊂ A. Q as Z-module: using as alphabet V all symbols representing the primes. Further, when considering A as effectively given in this way, we explicitly require that Xixj ≡

i

  • l=1

π(alij)Xl + π(a0ij) mod I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v. If not ZX, Y as left Z[X]-module requires 1 ≥ X ≥ X 2

slide-14
SLIDE 14

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.
slide-15
SLIDE 15

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.
slide-16
SLIDE 16

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.
slide-17
SLIDE 17

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.

Xixj ≡

i

  • l=1

π(alij)Xl + π(a0ij) mod I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v

slide-18
SLIDE 18

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.

fij := Xixj −

i

  • l=1

π(alij)Xl − π(a0ij) ∈ I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v

slide-19
SLIDE 19

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.

fij := Xixj −

i

  • l=1

π(alij)Xl − π(a0ij) ∈ I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v Xixj = T(fij) for each Xi ∈ V, xj ∈ v

slide-20
SLIDE 20

EFFECTIVELY GIVEN ASSOCIATIVE RINGS

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I If we fix a term-ordering < on Z we can assume I to be given via its Gr¨

  • bner basis G w.r.t. < and, if < satisfies

Xi > t for each t ∈ v and Xi ∈ V also I is given via its Gr¨

  • bner basis G0 := G ∩ R w.r.t. <.

fij := Xixj −

i

  • l=1

π(alij)Xl − π(a0ij) ∈ I, alij ∈ Zv, ∀Xi ∈ V, xj ∈ v

  • Xixj for each Xi ∈ V, xj ∈ v
  • ⊂ T(I)
slide-21
SLIDE 21

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

Zacharias G., Generalized Gr¨

  • bner bases in commutative

polynomial rings, Bachelor’s thesis, M.I.T. (1978) Effective description of the canonical form of the ring

slide-22
SLIDE 22

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

Zacharias G., Generalized Gr¨

  • bner bases in commutative

polynomial rings, Bachelor’s thesis, M.I.T. (1978) Spear D.A., A constructive approach to commutative ring theory, in Proc. of the 1977 MACSYMA Users’ Conference, NASA CP-2012 (1977), 369–376 Importing a Buchberger Theory from ZZ to A

slide-23
SLIDE 23

BUCHBERGER THEORY FOR EFFECTIVE ASSOCIATIVE RINGS

Zacharias G., Generalized Gr¨

  • bner bases in commutative

polynomial rings, Bachelor’s thesis, M.I.T. (1978) Spear D.A., A constructive approach to commutative ring theory, in Proc. of the 1977 MACSYMA Users’ Conference, NASA CP-2012 (1977), 369–376 M¨

  • ller H.M., On the construction of Gr¨
  • bner bases using

syzygies, J. Symb. Comp. 6 (1988), 345–359 Pritchard F . L., The ideal membership problem in non-commutative polynomial rings,

  • J. Symb. Comp. 22 (1996), 27–48

Lifting Theorem

slide-24
SLIDE 24

ZACHARIAS CANONICAL FORM

The obvious canonical forms of Ac for Zc namely Ac := {r : 0 ≤ r < c} or Ac := {r : 0 < r ≤ c} or Ac := {r : − c

2 < r ≤ c 2} give naturally a computational canonical

form for ZZm For a module I ⊂ ZZm, and each term τ ∈ Z(m) =

  • υei, υ ∈ Z, 1 ≤ i ≤ m
  • ,

considering the principal ideals I(cτ)u := {lc(f ) : f ∈ I, T(f ) = τ} ∪ {0} ⊂ Z, we have ZZm/I ∼ = Zach(ZZm/I) =:

  • τ∈Z(m)

Acτ τ I = I(4X, 2X 2) A = Z ⊕ Z4X ⊕ X 2Z2[X]

slide-25
SLIDE 25

ZACHARIAS CANONICAL FORM

The obvious canonical forms of Ac for Zc namely Ac := {r : 0 ≤ r < c} or Ac := {r : 0 < r ≤ c} or Ac := {r : − c

2 < r ≤ c 2} give naturally a computational canonical

form for ZZm For a module I ⊂ ZZm, and each term τ ∈ Z(m) =

  • υei, υ ∈ Z, 1 ≤ i ≤ m
  • ,

considering the principal ideals I(cτ)u := {lc(f ) : f ∈ I, T(f ) = τ} ∪ {0} ⊂ Z, we have ZZm/I ∼ = Zach(ZZm/I) =:

  • τ∈Z(m)

Acτ τ I = I(4X, 2X 2) A = Z ⊕ Z4X ⊕ X 2Z2[X]

slide-26
SLIDE 26

ZACHARIAS CANONICAL FORM

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I

  • Xixj for each Xi ∈ V, xj ∈ v
  • ⊂ T(I)

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • A ∼

=

  • ω∈V

 

υ∈v

Acυωυ   ω =:

  • ω∈V

Rωω ⊂ RV ⊂ Q Rω := R/Iω ∼ =

  • υ∈v

Acυωυ ⊂ RV

slide-27
SLIDE 27

ZACHARIAS CANONICAL FORM

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I

  • Xixj for each Xi ∈ V, xj ∈ v
  • ⊂ T(I)

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • A ∼

=

  • ω∈V

 

υ∈v

Acυωυ   ω =:

  • ω∈V

Rωω ⊂ RV ⊂ Q Rω := R/Iω ∼ =

  • υ∈v

Acυωυ ⊂ RV

slide-28
SLIDE 28

ZACHARIAS CANONICAL FORM

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I

  • Xixj for each Xi ∈ V, xj ∈ v
  • ⊂ T(I)

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • A ∼

=

  • ω∈V

 

υ∈v

Acυωυ   ω =:

  • ω∈V

Rωω ⊂ RV ⊂ Q Rω := R/Iω ∼ =

  • υ∈v

Acυωυ ⊂ RV

slide-29
SLIDE 29

ZACHARIAS CANONICAL FORM

Π : Q := Zx1, . . . , xj, . . . , X1, . . . , Xi, . . . ։ A = Q/I π : R = Zx1, . . . , xj, . . . ։ R = R/I

  • Xixj for each Xi ∈ V, xj ∈ v
  • ⊂ T(I)

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • A ∼

=

  • ω∈V

 

υ∈v

Acυωυ   ω =:

  • ω∈V

Rωω ⊂ RV ⊂ Q Rω := R/Iω ∼ =

  • υ∈v

Acυωυ ⊂ RV

slide-30
SLIDE 30

SPEAR’S THEOREM

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • Spear’s intuition that a Buchberger Theory defined in a ring can

be exported to its quotients allow us to impose on A the “natural” Γ-valuation/filtration T(·) : Am → B(m) : f → T(f ) where (Γ, ◦), B ⊂ Γ ⊂ V, is a suibale semigroup.

slide-31
SLIDE 31

SPEAR’S THEOREM

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • T(·) : Am → B(m) : f → T(f )

(Γ, ◦), B ⊂ Γ ⊂ V,

slide-32
SLIDE 32

SPEAR’S THEOREM

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • T(·) : Am → B(m) : f → T(f )

(Γ, ◦), B ⊂ Γ ⊂ V, The associated Γ-graded ring G = G(A) coincides as a set with A and this is sufficient to smoothly export Buchberger test/completion but they don’t conside as rings: the multiplication ⋆ of A does not coincide with the one, ∗, of G

slide-33
SLIDE 33

SPEAR’S THEOREM

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • T(·) : Am → B(m) : f → T(f )

(Γ, ◦), B ⊂ Γ ⊂ V, The associated Γ-graded ring G = G(A) coincides as a set with A and this is sufficient to smoothly export Buchberger test/completion but they don’t conside as rings: the multiplication ⋆ of A does not coincide with the one, ∗, of G For instance, if we consider the Weyl algebra, A = QD, X/I(DX − XD − 1) where G = Q[D, X], D ⋆ X = XD − 1, D ∗ X = XD.

slide-34
SLIDE 34

SPEAR’S THEOREM

B := {ω ∈ V : Iω = R} ⊂

  • υω : υ ∈ v, ω ∈ V
  • T(·) : Am → B(m) : f → T(f )

(Γ, ◦), B ⊂ Γ ⊂ V, The associated Γ-graded ring G = G(A) coincides as a set with A and this is sufficient to smoothly export Buchberger test/completion but they don’t conside as rings: the multiplication ⋆ of A does not coincide with the one, ∗, of G However an old slogan stated that in order to provide a Buchberger Algorithm on A, one just needs to modify, in the algorithm for G, the multiplication procedure!

slide-35
SLIDE 35

GR ¨

OBNER BASES

f =

s

  • i=1

c(f , ti)ti : c(f , ti) ∈ Z \ {0}, ti ∈ Z, t1 > · · · > ts. T(f ) := t1, lc(f ) := c(f , t1), M(f ) := c(f , t1)t1. Pan: XY = 3X · Y − X · 2Y ∈ I(3X, 2Y ) Let M ⊂ Am be a (left, right, bilateral) A-module. F ⊂ M will be called

slide-36
SLIDE 36

GR ¨

OBNER BASES

f =

s

  • i=1

c(f , ti)ti : c(f , ti) ∈ Z \ {0}, ti ∈ Z, t1 > · · · > ts. T(f ) := t1, lc(f ) := c(f , t1), M(f ) := c(f , t1)t1. Pan: XY = 3X · Y − X · 2Y ∈ I(3X, 2Y ) Let M ⊂ Am be a (left, right, bilateral) A-module. F ⊂ M will be called

slide-37
SLIDE 37

GR ¨

OBNER BASES

f =

s

  • i=1

c(f , ti)ti : c(f , ti) ∈ Z \ {0}, ti ∈ Z, t1 > · · · > ts. T(f ) := t1, lc(f ) := c(f , t1), M(f ) := c(f , t1)t1. Pan: XY = 3X · Y − X · 2Y ∈ I(3X, 2Y ) Let M ⊂ Am be a (left, right, bilateral) A-module. F ⊂ M will be called

slide-38
SLIDE 38

GR ¨

OBNER BASES

f =

s

  • i=1

c(f , ti)ti : c(f , ti) ∈ Z \ {0}, ti ∈ Z, t1 > · · · > ts. T(f ) := t1, lc(f ) := c(f , t1), M(f ) := c(f , t1)t1. Pan: XY = 3X · Y − X · 2Y ∈ I(3X, 2Y ) Let M ⊂ Am be a (left, right, bilateral) A-module. F ⊂ M will be called a (left, right, bilateral) Gr¨

  • bner basis of M if F satisfies the following

condition:

  • for each f ∈ M, there are gi ∈ F,

λi, ρi ∈ B, ai ∈ Rλi \ {0}, bi ∈ Rρi \ {0} such that

  • T(f ) = λi ◦ T(gi) ◦ ρi for all i,
  • M(f ) =

i aiλi ∗ M(gi) ∗ biρi;

slide-39
SLIDE 39

GR ¨

OBNER BASES

f =

s

  • i=1

c(f , ti)ti : c(f , ti) ∈ Z \ {0}, ti ∈ Z, t1 > · · · > ts. T(f ) := t1, lc(f ) := c(f , t1), M(f ) := c(f , t1)t1. Pan: XY = 3X · Y − X · 2Y ∈ I(3X, 2Y ) Let M ⊂ Am be a (left, right, bilateral) A-module. F ⊂ M will be called a (left, right, bilateral) strong Gr¨

  • bner basis of I if F satisfies the

following equivalent conditions

  • for each f ∈ M there is g ∈ F such that M(g) | M(f ),
  • for each f ∈ M there are g ∈ F,

λ, ρ ∈ B, a ∈ Rλ \ {0}, b ∈ Rρ \ {0}, such that T(f ) = λ ◦ T(g) ◦ ρ and M(f ) = aλ ∗ M(g) ∗ bρ.

slide-40
SLIDE 40

BUCHBERGER THEOREM

THEOREM For any set F ⊂ Am \ {0} the following conditions are equivalent:

  • f ∈ I(F) ⇐

⇒ it has a representation f =

µ

  • i=1

aiλi ⋆ gi ⋆ biρi T(f ) = λ1 ◦T(g1)◦ρ1 and λi ◦T(gi)◦ρi > λi+1 ◦T(gi+1)◦ρi+1∀i;

  • F is a strong Gr¨
  • bner basis of I(F);
slide-41
SLIDE 41

BUCHBERGER THEOREM

THEOREM For any set F ⊂ Am \ {0} the following conditions are equivalent:

  • f ∈ I(F) ⇐

⇒ it has a representation f =

µ

  • i=1

aiλi ⋆ gi ⋆ biρi T(f ) = λ1 ◦ T(g1) ◦ ρ1 and λi ◦ T(gi) ◦ ρi≥ λi+1 ◦ T(gi+1) ◦ ρi+1∀i;

  • F is a weak Gr¨
  • bner basis of I(F);
slide-42
SLIDE 42

BUCHBERGER REDUCTION

In order to perform Buchberger Algorithm we need to solve PROBLEM Given g ∈ Am \ {0} and F ⊂ Am \ {0} decide whether M(g) ∈ M{I2(M{F})} ⊂ Gm in which case return gi ∈ F, λi, ρi ∈ B, ai ∈ Rλi \ {0}, bi ∈ Rρi \ {0} such that

  • T(g) = λi ◦ T(gi) ◦ ρi for all i and
  • M(g) =

i aiλi ∗ M(gi) ∗ biρi.

which is trivial in Q. Then in G performs g → g −

i aiλi ⋆ gi ⋆ biρi

slide-43
SLIDE 43

LIFTING THEOREM

Given a finite set F := {g1, . . . , gu} ⊂ Am, gi = M(gi) − pi =: aiτieιi − pi, we denote M := I(F) and the morphisms sL : Gu → Gm and SL : Au → Am defined as sL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ∗ M(gi), SL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ⋆ gi,

slide-44
SLIDE 44

LIFTING THEOREM

sL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ∗ M(gi), SL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ⋆ gi, A reformulation (and a more efficient procedure) of the classical Buchberger test-completion, which states that “a basis F is Gr¨

  • bner

if and only if each S-polynomial reduces to 0”.

slide-45
SLIDE 45

LIFTING THEOREM

sL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ∗ M(gi), SL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ⋆ gi, A basis F is Gr¨

  • bner if and only if each element u in a minimal basis
  • f the module ker(s) of the syzygies among the leading monomials

M(gi) lifts, via Buchberger reduction of S(u), to a syzygy U ∈ ker(S) among the gi.

slide-46
SLIDE 46

LIFTING THEOREM

sL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ∗ M(gi), SL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ⋆ gi, A basis F is Gr¨

  • bner if and only if each element u in a minimal basis
  • f the module ker(s) of the syzygies among the leading monomials

M(gi) lifts, via Buchberger reduction of S(u), to a syzygy U ∈ ker(S) among the gi.More precisely we have U = S(u) −

µ

  • i=1

aiλi ⋆ ei ⋆ biρi where µ

i=1 aiλi ⋆ gi ⋆ biρi is the Gr¨

  • bner repr. of S(u) mod. F.
slide-47
SLIDE 47

LIFTING THEOREM

sL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ∗ M(gi), SL u

  • i=1
  • ω∈B

aiωω

  • ei
  • :=

u

  • i=1
  • ω∈B

aiωω ⋆ gi, A basis F is Gr¨

  • bner if and only if each element u in a minimal basis
  • f the module ker(s) of the syzygies among the leading monomials

M(gi) lifts, via Buchberger reduction of S(u), to a syzygy U ∈ ker(S) among the gi. As a corollary you get Janet–Schreier Theorem that the lifted ele- ments form a Gr¨

  • bner basis of ker(S)