semigroup.
3. An ~l~or~th~~to determine a mi~irn~i system of generators for the dual of a
;r
g:‘scn scm.~roup.
e would like to thank the referee for his suggestions and comments.
An !jjjitwsotligwr~p is a finitely generated commuta*:ve subsemigroup of N’
(under addition) for some positive integer k. (All sem;groups consider n this
paper arc suppose, 4 to have an identity element, that is to say, y are
ne semigroup and G(S) be the group genen by S.
-ttmi if and only if for all g E G(S) and tt \W
et S be an affine semigroup and defile
L,,.‘(S)
= yp,:s, E S. yI E Q -. t 6 N
Assume that S IS normal. T‘tkes E 15;~(S) n G(S). Then there exists tt E N such
that ttx t S ~e~irn~l~~~till~ Thus tts E S and s E 6’which implies
de~on~i~~~~to~s).
that s c, S. since 5’is normal. Therefore if S is normal. then L.::. (S) n G(S) = S.
The converse is also true. If LLz (S) 17G(S) = S and there exists g E G(S) sue
that tt S for some II E N\(O). then g f- LT..(S) (3 G(S) and this Ieads to
5! E S. have shown the f&owing ch;n-act&zation of normal atfine sernl-
groups.
he second st~~te~ent of the previo
tly retztted with mmtx~l
igroup S is.fir/lif and
e every finitely gener roup is finitely
induces a ring horn0
ip:K[s, . . . I ._q] - K[S], t/9(x,)= .“;, .
ernel is us~l~~i~ydenoh
. . , . (M:.hp)) generates
X”r - /if+ ) generates Is (see 141).
For a given co~~rl~ence 5 on N’,
iti7 = {z E iz”:z = n - h for SCme n s h).
Conversely, for 9 given Z-module H 2 Z’, we define E
-J~/=((Q.~)EN~xN~: c-hrzN].
for this reason (a. h) E mIS,.
not true. As a batter of fact, it is on1
revious resuk
et B
C E Ni” .SUCh
ihQt
ative elcnxms cX a sub
atrix (q . . . q) is
= 5, and ~l~cr~~or~C = 3 x 2 x 5 = 30.
3.2. Cor~lpl~ti~g the set of nonnegative
integer co cients haviftg some of the
dl,.,llxl + . . . + ar, ,,&S~ = 0.
can use the procedure ribed in the previous pxagraph to can
mi a! system of generator , . . . .s,} for the semigroup M’ n F# -‘r.
the ratural projection
71:Nh*zr _ &Jk
n(n I..... (l,,.&_\, . . . . (ih.zr) = (a I..... N!,j.
it ic sb~n in [5] that (n(sl). . . . . n(s,)) generates S.
Once we know how to compute a minim& system of generrators Fx
5’:= M Ti IX’“.we can chu~-k whether K[S] is Gorenstein (in this setting. the ring
X[Sj iS Ll?WilyS Cohen- oschter showed in [I]). The idea is based
on the ~o~~owi$l~ result due to Stanley.
I j.2 in [CI]).
s Iurs trJI irs
S (kx.ze S
this is the
system of generatorsof
I.
3
_.
_3 .
4.
As an ~~p~~i~~tion of the pfocah-e described in the prexd
show flow to compute the solutions with
ne system ofeqrrations with ratiisna! c
form
with .4 an II x
form
e method described in t
of gerzrators of the
et {n,..... n,} bc the ~~entio~ed minimal system of generators. The soiw
zero, and that n, _i
iE (/“t I,.. . .s) and nz E @I.. . . .n,).
always he converse is
ne se~i~ro~~ generatcd by { :O, 11,( 1,
,
ne semigroup. Nevertheless.
since iSp(0. I) + q(l_ ) belongs to &I’for some p. q E Q ‘, then q E & and this
forces p to be a nonnegative integer as well.
L,.
ele
e el
T roposition gives us the iECnhle
. A ~i~i~~~ set t- _generators f& L,+ (Sj Cl GiS) is ci’ntained in rhe set of
n xv whether there exists i., . . . . . E.,.
E 62+ sue at
is occ ws if and only if t
namqptive integer soiutioas of this
not equal results of t L-St
section. It P G(S) if and only if 7
iS @it% it Wily t0 check W ether S is ~~or~~~~ or not.
. Let
s=((2,0,0),(0,2,0).(1,0,1f.(0.1.1),(1,1.1)).
Then G(S) = Z-‘. In order to check whether Lo-(S) n G(S) = S, we must
determine whether all the elements that belong to T, i.e. the elements less or
equal than (2+ 1 + 1,2+ 1 + 1,l + 1 + I> = (4.4.3) that belong to
L,. (S) II G(S) are in S. The element (1, O,O)= I /2(2.0,0) E LuL(S) 71G(S)\S,
and therefore S is not a normal semigroup.
If a commutative semigroup S is given in terms of a Finite presentation, it is
because there exists a set of relators p = {(nl,b,), . . . , (a,,b,)) c N” x IQ”WC
that S is iscmo
If we want t is a normal semigroup, first we must check
whether S is an n order to achieve this, we must recall some
basic concepts.
A commutative semigroup S is said to be cancellative if a + c = b + c im-
plies that 1~= 6, for every a, b? c E S. The semigroup S is said to be reduced if it
has no units. Finally, S is torsion free if na = nb implies u = b, for a. b E S and
0 # n E IV (some authors use the term power-cancellative semigroup to denote
torsion free semigroup). The next result gives us the key to solve the problem of
determining whether a given semigroup is an affine semigroup. It can be found
i&i[g] or in [7]
Thus, we are going to check whether S is reduced, cancellativc and torsion
free.
Hots to decide whew n jnitelv preset ted comnuta tive semigraip is reck& As
before, let CTbe the congruence on N” generated by p. We can assume that
{XU’-XI-l,...,XOl - X”l} is a reduced Grobncr basis of Is (if this is not the
case, we can compute a Grobner basis (see [9,10]) and it is c!ear that the S-
polynomial of two pure binomials is again a pure binomial).
sa at {(~i,br), . . . , (~A) is a canonid system of‘genetxtors of 6.
note by e, the element of Nk which has all its coordinates except the ith
equal to zero and the th coordinate equal to one. Note that if (e;,O) E Q, then
(e;,O) must be in p (t is :oes not hold for systems of generators in general, p
must be a if this is the case,
then the it zero, since p is a
canonical system of generatsrs for
Q-: /Y&p-I eliminating from p
‘COOT ate 0 e rest c#f the elemen
by
d = (((x;, . . . .x~_,‘, o,,. . . . ,,v_,)) E N’-’ x NL-‘:
((.b ..s,_,,Q,_r I....._ U&_l).(yl,.... y_,,o.y I’...., Vk_,)) E 0)
t is not dinicult to prove the next result.
This result enables us to eiiminate the e!ements oft
now on, we will assume that p does noi contain any sue
conditions it is straightforward to prove the folloGng re
details;.
3. Let G beu congruence on N” cd let p = {(al. hi ). . . . . (a,$.b,)) be a
c.monical system qf genercrtors qf G. The fr,liowing conditiom me equiadent:
I. N”/o is reduced.
2. FordiE {I . . . . . .~).b, #O.
J. C. Rosdes. P. A. Garcicc-Sciachcz I Lirtrar Algebra Appl. 286 (1999) 175-186
computing the equations
are nonnegative integers.
these equ;zlims. Once we S’, we can apply the results given in t
:o check whether S’ is normal.
[ij M. Hochst;r, Rings of invariants of tori, Cohen-Macauiay rings generated by monomial . . .md
poiytopes. Ann. Math. 96 (1972) 318-337.
f21 G. Let& Subsemigroups of finitely generated groups with divisor theory. Monatshefre
1 F. Math. 106 (i9SS) 205-210.
[3] L. Ridei. The Theory of Finitely Generated Commutative Se:.koups, Pergamon Pir:s.
Oxford. 1965.
[4] J. Herzog, Generators and rektions of abelian semigroups anu semigroup rings. Manuscrqta
Math. 3 (1970) 175-193.
[5] J.C. Rosaies. P.A. Garcia-Sbnchei, Nonnegative elements of subgroups of Z”, Linear Algebra
Appi. 270 (19!‘5) 351-357.
[6] R. Stanley. Combinatoricc arld Commutative A!gebra. Progress in Mathematics, BirkhCuser,
Boston. Easel. Stuttgdr:, 1983.
[7] F.A. Griliet. The free envelope of a finitely generated commutative semigroup. Trans. Amer.
Math. Sot. 149 (1970) 665 632.
[S] J.C Pnsales, (3&ifinitely generated s::$monoids of &, Semigroup Forun; 50 (!995) 251-262.
!“j P. Buchberger, GrGbner bases: an algorithm method in polynomial ideal theory, in: Y K. Bose
(Cd.), Recent Trendc in Multidimensional System Theory. Reldei, Dordrecht. 1985.
!tO! J.C. Rosaies. Function minimum associated to a congruence on integral n-tuple space,
Sen+rgroup Forum 51 (1995) 87-95.
[l I] J.i. R.lsr:ies, J.M. Urbano-Bianco, A deterministic algorithm to decide if a finitely presented
abeiian monoid is canceilative. Comm. Algebra 24 ( 13)(1996) 42! 74224.