Home Online Manual
Top
Back: Examples of use of Letterplace
Forward: Functionality and release notes of Letterplace
FastBack:
FastForward:
Up: Singular Manual
Top: Singular Manual
Contents: Table of Contents
Index: Index
About: About this document

7.7.2 Example of use of LETTERPLACE over Z

Consider the following paradigmatic example:

 
LIB "freegb.lib";
ring r = integer,(x,y),Dp;
ring R = freeAlgebra(r,5); // length bound is 5
ideal I = 2*x, 3*y;
I = twostd(I);
print(matrix(I)); // pretty prints the generators
==> 3*y,2*x,y*x,x*y

As we can see, over $Z<x,y>$ the ideal $<2x,3y>$ has a finite Groebner basis and indeed

$Z<x,y>/<2x,3y> =$ $Z<x,y>/<2x,3y,yx,xy> =$ $Z<x,y>/<2x,3y,yx-xy,xy>$

and the later is naturally isomorphic to

$Z[x,y]/<2x,3y,xy>$ as a $Z$-algebra.

Now, we analyze the same ideal in the ring with one more variable $z$:

 
LIB "freegb.lib";
ring r = integer,(x,y,z),Dp;
ring R = freeAlgebra(r,5); // length bound is 5
ideal I = 2*x, 3*y;
I = twostd(I);
print(matrix(I)); // pretty prints the generators
==> 3*y,2*x,y*x,x*y,y*z*x,x*z*y,y*z*z*x,x*z*z*y,y*z*z*z*x,x*z*z*z*y

Now we see, that this Groebner basis is potentially infinite and the following argument delivers a proof. Namely, $y z^i x$ and

$x z^i y$ are present in the ideal for all $i>=0$. How can we do this? We wish to express $y*z^i*x$ and

$x*z^i*y$ via the original generators by means of lift:

 
LIB "freegb.lib";
ring r = integer,(x,y,z),Dp;
ring R = freeAlgebra(r,5,2); // length bound is 5, rank of the free bimodule is 2
ideal I = 2*x, 3*y;
matrix T1 = lift(I, ideal(y*z*x,x*z*y));
print(T1);
==> -y*z*ncgen(1),-ncgen(1)*z*y,
==> ncgen(2)*z*x, x*z*ncgen(2)  
-y*z*I[1] + I[2]*z*x; // gives y*z*x
==> y*z*x
matrix T2 = lift(I, ideal(y*z^2*x,x*z^2*y));
print(T2);
==> -y*z*z*ncgen(1),-ncgen(1)*z*z*y,
==> ncgen(2)*z*z*x, x*z*z*ncgen(2)  
-y*z^2*I[1] + I[2]*z^2*x; // gives y*z^2*x
==> y*z*z*x

The columns of matrices, returned by lift, encode the presentation of new elements in terms of generators. From this we conjecture, that in particular

$-y z^i*(2x) + (3y)*z^i x = y z^i x$ holds for all $i>=0$

and indeed, confirm it via a routine computation by hands.

Comparing computations over Q with computations over Z.

In the next example, we first compute over the field of rationals $Q$ and a bit later compare the result with computations over the ring of integers $Z$.

 
LIB "freegb.lib"; // initialization of free algebras
ring r = 0,(z,y,x),Dp; // degree left lex ord on z>y>x
ring R = freeAlgebra(r,7); // length bound is 7
ideal I = y*x - 3*x*y - 3*z, z*x - 2*x*z +y, z*y-y*z-x;
option(redSB); option(redTail); // for minimal reduced GB
option(intStrategy); // avoid divisions by coefficients
ideal J = twostd(I); // compute a two-sided GB of I
J; // prints generators of J
==> J[1]=4*x*y+3*z
==> J[2]=3*x*z-y
==> J[3]=4*y*x-3*z
==> J[4]=2*y*y-3*x*x
==> J[5]=2*y*z+x
==> J[6]=3*z*x+y
==> J[7]=2*z*y-x
==> J[8]=3*z*z-2*x*x
==> J[9]=4*x*x*x+x
LIB "fpadim.lib"; // load the library for K-dimensions
lpMonomialBasis(7,0,J); // all monomials of length up to 7 in Q<x,y,z>/J
==> _[1]=1
==> _[2]=z
==> _[3]=y
==> _[4]=x
==> _[5]=x*x

As we see, we obtain a nice finite Groebner basis J. Moreover, from the form of its leading monomials, we conjecture that

$Q<x,y,z>/J$ is finite dimensional $Q$-vector space. We check it with lpMonomialBasis and obtain an affirmative answer.

Now, for doing similar computations over $Z$ one needs to change only the initialization of the ring, the rest stays the same

 
LIB "freegb.lib"; // initialization of free algebras
ring r = integer,(z,y,x),Dp; // Z and deg left lex ord on z>y>x
ring R = freeAlgebra(r,7); // length bound is 7
ideal I = y*x - 3*x*y - 3*z, z*x - 2*x*z +y, z*y-y*z-x;
option(redSB); option(redTail); // for minimal reduced GB
option(intStrategy); // avoid divisions by coefficients
ideal J = twostd(I); // compute a two-sided GB of I
J; // prints generators of J
==> J[1]=12*x*y+9*z
==> J[2]=9*x*z-3*y
==> J[3]=y*x-3*x*y-3*z
==> J[4]=6*y*y-9*x*x
==> J[5]=6*y*z+3*x
==> J[6]=z*x-2*x*z+y
==> J[7]=z*y-y*z-x
==> J[8]=3*z*z+2*y*y-5*x*x
==> J[9]=6*x*x*x-3*y*z
==> J[10]=4*x*x*y+3*x*z
==> J[11]=3*x*x*z+3*x*y+3*z
==> J[12]=2*x*y*y+75*x*x*x+39*y*z+39*x
==> J[13]=3*x*y*z-3*y*y+6*x*x
==> J[14]=2*y*y*y+x*x*y+3*x*z
==> J[15]=2*x*x*x*x+y*y-x*x
==> J[16]=2*x*x*x*y+3*y*y*z+3*x*y+3*z
==> J[17]=x*x*y*z+x*y*y-x*x*x
==> J[18]=x*y*y*z-y*y*y+x*x*y
==> J[19]=x*x*x*x*x+y*y*y*z+x*x*x
==> J[20]=x*x*x*x*z+x*x*x*y+2*y*y*z+x*x*z+3*x*y+3*z
==> J[21]=x*y*y*y*z-y*y*y*y+x*x*x*x-y*y+x*x
==> J[22]=y*y*y*z*z-x*x*x*x*y
==> J[23]=x*y*y*y*y*z-y*y*y*y*y+x*x*y*y*y
==> J[24]=x*y*y*y*y*y*z-y*y*y*y*y*y+x*x*x*x*y*y+y*y*y*y+x*x*x*x+2*y*y-2*x*x

The output has plenty of elements in each degree (which is the same as length because of the degree ordering), what hints at potentially infinite Groebner basis.

Indeed, one can show that for every $i>=2$ the ideal $J$ contains an element with the leading monomial $x y^i z$.