EVERY ALGEBRAIC SET IN m-SPACE

IS THE INTERSECTION OF m HYPERSURFACES

David Eisenbud (Waltham) and E.Graham Evans , Jr. (Urbana)*

Abstract

In 1882, Kronecker had announced his theorem based on intersection of hypersurfaces,

Vahlen produced an example which, he claimed, proved the Kronecker’s theorem wrong.
Though the example was wrong but that was remained uncaught until 1942, when Perron
proved the example wrong. In 1961, Kneser observed that the idea behind the example of
Perron was not just for that example but can be generalized. And this entire history helps us
to prove our result.[ CITATION Ser55 \l 16393 ]

Introduction

In this paper, we try to prove that every radical ideal in an n-dimensional space Noetherian
polynomial ring is radical of ideal generated by m elements and try to build same result for
graded rings.

This can easily be seen that this above stated result also strengthens the result (announced
earlier by Kronecker in 1882) that any radical ideal in an m-dimensional Noetherian ring is
radical of an ideal generated by m+1 elements

It is now important to mention two famous problems that remain open in this area. Murthy
has recently shown that if k is a field, then in the ring k[X,Y,Z], any ideal of height 2 which is
locally a complete intersection can be generated by 3 elements.

Murthy also gave an example to show that ideal corresponding even to a non-singular curve
in 3-space may not be generated by 2 elements[ CITATION Mur72 \l 16393 ]. But still there is an
open problem in this field- Is every in 3-space the intersection of 2 hypersurfaces? However,
It can easily be seen that answer to above problem is no in 4-space. The surface S
corresponding to the ideal (x1,x3)∩(x2,x4) ⊆K[x1,x2,…,x4] is not the intersection of 2
hypersurfaces.

AFFINE VARIETY CASE

If I is an ideal of ring R, we will write √ I for the radical of I; that is, √ I is the intersection of
the prime ideals of R which contain I.
Lemma 1. Let R be a Noetherian ring of dimension m, and suppose that R is a polynomial
ring of the form R=L[x] for some ring L. Let J⊆R be an ideal. Then there exist m elements
a1,a2,…,am ∈ J such that √ J =¿ √ ( a 1 , a2 , … , am ). [ CITATION DAV73 \l 16393 ]

Proof: We shall prove the theorem by induction on m, using that dim L=m-1.

Suppose first that m=1, and hence L is Artinian. Let N be nilpotent radical of L, NR is
nilpotent radical in ring R

and set the following notation -

R=¿R/NR=(L/N)[x]
^

^J=¿ (J+NR/NR) ⊆ ^
R.

As every Artinian ring is direct product of Artinian local rings, therefore, L/N is direct
R is clearly a principal ideal ring so there is an element a∈J whose
product of fields, and ^
image a^ ∈ ^J generates ^J . Since NR is nilpotent, it is contained in every prime ideal of R, so
√(a)=√ J , as we have discussed. So, the result is true for m=1.
Now, for m>1. Let P1,P2,…,Pn be minimal primes of L, and let V = L - ¿ i=1 ¿ n Pi . Clearly, V
is a multiplicative closed set being complement of finitely many prime ideals, and LV is 0-
dimesional. Then Rv=Lv[x] is 1-dimensional, so by the case m=1, there is an element a1∈JV
such that √ (a1)= √ J V in Rv. Multiplying by an element of U if necessary, we may assume
that a1∈J. Since J is finitely generated (R and therefore L[x] are Noetherian rings) , there is
an element v∈V such that

v J ⊆√ (a1). (1)

Because,

v∉¿ i=1 ¿ n Pi , dim L/(v)≤ m-2,

and the dimension of R/(v) = L/(v)[x] is ≤ m-1. Let R*=R/(v) and let J*=(J+(v))/(v). By the
induction hypothesis, there exist elements A2,A3,…,Am (m-1 elements) such that

√( A 2 , A 3 ,… , Am)=√ J * (2)

In R*. Let a2,a3,…,am be elements in I which reduced to A2,A3,…,Am modulo (a1).

We need to prove that √ J =¿ √ ( a 1 , a2 , … , a m ).For this we must show that for any prime ideal
P of R, (a1,a2,…,am) ⊆P so that J⊆P . Since (g1) ⊆P, it follows from eqn -1 that either (v) ⊆P or
J⊆P. In the second case, the ideal

P*=P/(v) ⊆ R*
is a prime ideal, so by (2) J*⊆P*. Thus, in the case J+(v) ⊆P+(v)=P. That is J +(v) ⊆P but since
(v) ⊆P so, we have J⊆P and hence in either case J⊆P , hence, we are through.

Result 1. Every Algebraic set in affine m-space is the intersection of m-hypersurfaces.

[ CITATION DAV73 \l 16393 ]

Proof: The proof of above result is nothing but a consequence of above theorem. Replace R
by K[x1,x2,..,xm] with K a field.

Note: We can generalize the same result for projective variety also, only need to define the
graded ring and the same result follow on the same lines.
 References

1. Abhyankar, S.: Algebraic Space Curves ,University of Montreal Lecture Notes(1971). [

CITATION Abh71 \l 16393 ]
2. Eisenbud, D., Evans, E.G.: Generating modules efficiently: Theorems from algebraic
K-theory.[ CITATION Dav73 \l 16393 ]
3. Eisenbud, D., Evans, E.G.: Three conjectures on modules over polynomial rings.
4. Hartshrone , R.: Complete intersection and connectedness. Am .J. Math(1962).
[ CITATION Har62 \l 16393 ]
5. Hartshrone , R.: Cohomological dimension of algebraic varieties (1968).
6. Kronecker, L.: Grundzuge einer arithmetischen Theorie der algebraischen GroBen,
J.Reine Angew. Math. 92, 1-123 (1882).
7. Kneser, M.: Uber die Darstellung algebraischer Raumkurven als Durchschnitte von
Flachen Arch. Math. 11 157-158 (1960).
8. Murthy, M.P.: Generator for certain ideals in regular ring of dimension three. (To
Appear.)[ CITATION Mur72 \l 16393 ]
9. Perron, O.: Uber das Vahlensche Beispiel zu einem Satz von Kronecker. Math. 47,
318-324 (1941).
10. Serre , J.-P.: Faisceaux Algebriques Coherents , Ann of Math. 61, 197-278 (1955).
[ CITATION Ser55 \l 16393 ]
11. Serre , J.-P.: Sur les modules projectifs. Sem. Dudreil-Pisot , 2-13 (1960/61).
12. Vahlen, K.T.: Bemerkung zur vollstandigen Darstellung algebraischer Raumkurven.
J.Reine Angew. Math. 108 ,346-347 (1891).[ CITATION Vah91 \l 16393 ]
13. Van der Waerden, B.L.: Review Zentralblatt fur Math. 24, 276 (1941).

David Eisenbud E. Graham Evans,Jr.

Brandies University University of Illinois at Urbana

Department of Mathematics Department of Mathematics

Waltam, Mass. 02154 Urbana, Ill. 61801

USA USA