A Computation of Tight Closure in Diagonal Hypersurfaces Anurag K. Singh

advertisement
203, 579]589 Ž1998.
JA977350
JOURNAL OF ALGEBRA
ARTICLE NO.
A Computation of Tight Closure in Diagonal
Hypersurfaces
Anurag K. Singh
Department of Mathematics, Uni¨ ersity of Michigan, East Hall, 525 East Uni¨ ersity A¨ enue,
Ann Arbor, Michigan 48109-1109
Communicated by Craig Huneke
Received August 26, 1997
1. INTRODUCTION
The aim of this paper is to settle a question about the tight closure of
the ideal Ž x 2 , y 2 , z 2 . in the ring R s K w X, Y, Z xrŽ X 3 q Y 3 q Z 3 . where
K is a field of prime characteristic p / 3. ŽLower case letters denote the
images of the corresponding variables.. M. McDermott has studied the
tight closure of various irreducible ideals in R and has established that
xyz g Ž x 2 , y 2 , z 2 .* when p - 200, see wMcx. The general case however
existed as a classic example of the difficulty involved in tight closure
computations, see also wHu, Example 1.2x. We show that xyz g Ž x 2 , y 2 , z 2 .*
in arbitrary prime characteristic p, and furthermore establish that xyz g
Ž x 2 , y 2 , z 2 . F whenever R is not F-pure, i.e., when p ' 2 mod 3. We move
on to generalize these results to the diagonal hypersurfaces R s
K w X 1 , . . . , X n xrŽ X 1n q ??? qX nn ..
These issues relate to the question whether the tight closure I* of an
ideal I agrees with its plus closure, Iqs IRql R, where R is a domain
over a field of characteristic p and Rq is the integral closure of R in an
algebraic closure of its fraction field. In this setting, we may think of the
Frobenius closure of I as I F s IR` l R where R` is the extension of R
obtained by adjoining p e th roots of all nonzero elements of R for e g N.
It is not difficult to see that Iq: I*, and equality in general is a formidable open question. It should be mentioned that in the case when I is an
ideal generated by part of a system of parameters, the equality is a result
E-mail: binku@math.lsa.umich.edu
579
0021-8693r98 $25.00
Copyright Q 1998 by Academic Press
All rights of reproduction in any form reserved.
580
ANURAG K. SINGH
of K. Smith, see wSmx. In the above ring R s K w X, Y, Z xrŽ X 3 q Y 3 q Z 3 .
where K is a field of characteristic p ' 2 mod 3, if one could show that
I* s I F for an ideal I, a consequence of this would be I F : Iq: I* s I F,
by which Iqs I*. McDermott does show that I* s I F for large families of
irreducible ideals and our result xyz g Ž x 2 , y 2 , z 2 . F , we believe, fills in an
interesting remaining case.
2. DEFINITIONS
Our main reference for the theory of tight closure is wHHx. We next
recall some basic definitions.
Let R be a Noetherian ring of characteristic p ) 0. We shall always use
the letter e to denote a variable nonnegative integer, and q to denote the
eth power of p, i.e., q s p e. We shall denote by F, the Frobenius
endomorphism of R, and by F e, its eth iteration, i.e., F e Ž r . s r q. For an
ideal I s Ž x 1 , . . . , x n . : R, we let I w q x s Ž x 1q, . . . , x nq .. Note that F e Ž I . R s
I w q x, where q s p e, as always.
We shall denote by R8 the complement of the union of the minimal
primes of R.
DEFINITION 2.1. A ring R is said to be F-pure if the Frobenius
homomorphism F : M ª M m R F Ž R . is injective for all R-modules M.
For an element x of R and an ideal I, we say that x g I F, the Frobenius
closure of I, if there exists q s p e such that x q g I w q x. A normal domain R
is F-pure if and only if for all ideals I of R, we have I F s I.
We say that x g I*, the tight closure of I, if there exists c g R8 such
that cx q g I w q x for all q s p e 4 0.
It is easily verified that I : I F : I*. Furthermore, I* is always contained in the integral closure of I and is frequently much smaller.
3. PRELIMINARY COMPUTATIONS
We record some determinant computations we shall find useful. Note
that for integers n and m where m G 1, we shall use the notation
ž mn / s
Ž n . Ž n y 1 . ??? Ž n y m q 1 .
.
Ž m . Ž m y 1 . ??? Ž 1 .
581
A COMPUTATION OF TIGHT CLOSURE
LEMMA 3.1.
ž
det
ž
n
aqk
n
aqky1
/ ž
/ ž
n
aqky1
n
aqk
ž na /
ž
n
aq1
???
/
???
/
ž
ž
n
a q 2k
/
n
a q 2k y 1
/
???
ž
ž
n
aqk
/
n
nq1
nqk
???
aqk aqk
aqk
.
aqk aqkq1
a q 2k
???
aqk
aqk
aqk
ž
s
???
/
/ž
/ ž /
/ ž /
/ž
Proof. This is evaluated in wMu, p. 682x as well as wRox.
Let F Ž n, a, k . denote the determinant of the matrix
LEMMA 3.2.
¡ žn/
a
ž
M Ž n, a, k . s
nq2
aq1
ž
ž
/
n
aq1
nq2
aq2
???
/
/
???
ž
ž
n
aqk
nq2
aqkq1
???
¢ž
n q 2k
aqk
/ ž
n q 2k
aqkq1
/
???
¦
/
ž
n q 2k
a q 2k
/
.
/§
Then for k G 1 we ha¨ e
F Ž n, a, k .
s n
a
F Ž n q 2, a q 2, k y 1 .
k
k
s Ž s q 2 a y n.
ž / Ł Ł Ž a q r.Ž n y a q r. .
ss1 rs1
Hence
ž na / ž na qq 22 / ??? ž na qq 22 kk /
F Ž n, a, k . s
ž
aqk
k
/ž
aqkq1
a q 2k y 1
???
ky1
1
/ ž
/
/ ž
/ ž
2a y n q k 2a y n q k q 1
2a y n q 2k y 1
???
k
ky1
1
?
.
nyaqk nyaqky1
nyaq1
???
k
ky1
1
ž
/ž
ž
/ž
/
/
582
ANURAG K. SINGH
Proof. We shall perform row operations on M Ž n, a, k . in order to get
zero entries in the first columns from the second row onwards, starting
with the last row and moving up. More precisely, from the Ž r q 1.st row,
subtract the r th row multiplied by Ž naqq2rr .rŽ naqq2rryy12 . starting with r s k,
and continuing until r s 2. The Ž r q 1, s q 1.st entry of the new matrix,
for r G 1, is
ž
n q 2r y
aqrqs
/
s
ž naqq2rr /
ž
n q 2r y 2
aqry1
s Ž s q 2 a y n.
/ž
n q 2r y 2
aqry1qs
/
n q 2r .
Ž a q r.Ž n y a q r. a q r q s
ž
/
We have only one nonzero entry in the first column, namely Ž na . and so we
examine the matrix obtained by deleting the first row and column. Factoring out sŽ s q 2 a y n. from each column for s s 1, . . . , k and 1rŽ a q r .Ž n
y a q r . from each row for r s 1, . . . , k, we see that
det M Ž n, a, k . s n
a
k
k
s Ž s q 2 a y n.
ž / Ł Ł Ž a q r.Ž n y a q r.
ss1 rs1
=det M Ž n q 2, a q 2, k y 1 . .
The required result immediately follows.
LEMMA 3.3. Consider the polynomial ring T s K w A1 , . . . , A m x where Ir, i
denotes the ideal Ir, i s Ž A1i , . . . , Air .T for r F m. Then
a
b
Ž A1 ??? A ry1 . Ž A1 q ??? qA ry1 . g Iry1, aq g q Ž A1 q ??? qA ry1 .
aq g
T
for positi¨ e integers a , b , and g implies
a
b q g y1
g Ir , aq g q Ž A1 q ??? qA r .
Ž A1 ??? A r . Ž A1 q ??? qA r .
aq g
T.
Proof. Consider the binomial expansion of Ž A 1 ??? A r . a Ž A 1
q ??? qA r . bq g y 1 into terms of the form Ž A 1 ??? A ry 1 . a Ž A 1
q ??? qA ry1 . bq gy1yj A ar qj. Such an element is clearly in Ir, aq g whenever
583
A COMPUTATION OF TIGHT CLOSURE
j G g , and so assume g ) j. Now
a
b q g y1yj
Ž A1 ??? A ry1 . Aar qj Ž A1 q ??? qA ry1 .
g Ir , aq g q Aar qj Ž A1 q ??? qA ry1 .
: Ir , aq g q Ž A1 q ??? qA ry1 , A r .
: Ir , aq g q Ž A1 q ??? qA r .
aq g
aq2 g y1yj
2 a q2 g y1
T
T
T.
4. TIGHT CLOSURE
We now prove the main theorem.
THEOREM 4.1. Let R s K w X 1 , . . . , X n xrŽ X 1n q ??? qX nn . where n G 3
and K is a field of prime characteristic p where p ¦ n. Then
ny 2
g Ž x 1ny1 , . . . , x nny1 . *.
Ž x 1 ??? x n .
Note that there are infinitely many e g N such that p e s q ' 1 mod n.
By wHH, Lemma 8.16x, it suffices to work with powers of p of this form,
and show that for all such q we have
Ž x 1 ??? x n .
Ž ny2 . qq1
g Ž x 1Ž ny1. q , . . . , x Žnny1. q . .
Letting q s nk q 1, it suffices to show
Ž x 1 ??? x n .
Ž ny2 . n k
g Ž x 1Ž ny1. n k , . . . , x Žnny1. n k . .
Let A1 s x 1n, . . . , A n s x nn and note that A1 q ??? qA n s 0. In this notation, we aim to show
Ž A1 ??? A n .
Ž ny2 . k
g Ž AŽ1ny1. k , . . . , AŽnny1. k . .
Our task is then effectively reduced to working in the polynomial ring
K w A1 , . . . , A ny1 x ( K w A1 , . . . , A n xrŽ A1 q ??? qA n . where we need to
show Ž A1 ??? A ny1Ž A1 q ??? qA ny1 ..Ž ny2. k g 2 Iny1, Ž ny1. k q Ž A1 q ??? q
A ny 1 .Ž ny1. k . By repeated use of Lemma 3.3, it suffices to show
Ž A1 A 2 .
Ž ny2 . k
Ž
.
k
Ž A1 q A 2 . g ž AŽ1ny1. k , AŽ2ny1. k , Ž A1 q A 2 . ny1 k / .
We have now reduced our problem to a statement about a polynomial ring
in two variables. The required result follows from the next lemma.
584
ANURAG K. SINGH
LEMMA 4.2. Let K w A, B x be a polynomial ring o¨ er a field K of characteristic p ) 0 and e be a positi¨ e integer such that q s p e ' 1 mod n. If
q s nk q 1, we ha¨ e
Ž A, B . 2 ny3 k : I s Ž AŽ ny1. k , B Ž ny1. k , Ž A q B .
Ž
.
Ž ny1 . k
..
In particular, Ž AB .Ž ny2. k Ž A q B . k g I.
Proof. Note that I contains the following elements: Ž A q B .Ž ny1. k =
A B Ž ny3. k , Ž A q B .Ž ny1. kAky1 B Ž ny3. kq1 , . . . , Ž A q B .Ž ny1. k B Ž ny2. k . We
take the binomial expansions of these elements and consider them modulo
the ideal Ž AŽ ny1. k , B Ž ny1. k .. This shows that the following elements are in
I:
k
žŽ
žŽ
n y 1. k
Ž n y 1 . k AŽ ny2. k B Ž ny1. k ,
AŽ ny1. k B Ž ny2. k q ??? q
k
2k
žŽ
n y 1. k
Ž n y 1 . k AŽ ny2. k B Ž ny1. k .
AŽ ny1. k B Ž ny2. k q ??? q
0
k
/
/
n y 1. k
AŽ ny1. k B Ž ny2. k q ??? q
ky1
???
/
ž
žŽ
/
/
n y 1. k
AŽ ny2. k B Ž ny1. k ,
2k y 1
ž
/
The coefficients of AŽ ny1. k B Ž ny2. k , AŽ ny1. ky1 B Ž ny2. kq1, . . . , AŽ ny2. k B Ž ny1. k
form the matrix
¡ Ž n y 1. k
ž
žŽ
k
n y 1. k
ky1
/ žŽ
/ žŽ
n y 1. k
kq1
n y 1. k
k
/
/
???
???
žŽ
žŽ
???
¢ž
Ž n y 1. k
0
/ ž
Ž n y 1. k
1
/
???
žŽ
n y 1. k
2k
n y 1. k
2k y 1
?
n y 1. k
k
/¦
/
.
/§
To show that all monomials of degree Ž2 n y 3. k in A and B are in I, it
suffices to show that this matrix is invertible. Since q s nk q 1 we have
Ž Ž n y 1k. k q r . s Žy1. k Ž 2 k ky r . for 0 F r F k, and so by Lemma 3.1, the deter-
585
A COMPUTATION OF TIGHT CLOSURE
minant of this matrix is
žŽ
n y 1. k
k
k
k
n y 1. k q 1
nk
???
k
k
kq1
2k
???
k
k
/žŽ
ž /ž
s Ž y1 .
k Ž kq1 .
/ ž /
/ ž /
ž /ž / ž /
ž /ž / ž /
2k
k
k
k
2k y 1
k
???
k
k
kq1
2k
???
k
k
s 1.
With this we complete the proof that Ž x 1 ??? x n . ny 2 g Ž x 1ny 1 , . . . , x nny1 .*.
5. FROBENIUS CLOSURE
Let R s K w X 1 , . . . , X n xrŽ X 1n q ??? qX nn . as before, where the characteristic of K is p ¦ n.
LEMMA 5.1. Let R s K w X 1 , . . . , X n xrŽ X 1n q ??? qX nn . where K is a field
of characteristic p. Then R is F-pure if and only if p ' 1 mod n.
Proof. This is Proposition 5.21Žc. of wHRx.
The main result of this section is the following theorem.
THEOREM 5.2. Let R s K w X 1 , . . . , X n xrŽ X 1n q ??? qX nn . where K is a
field of characteristic p. Then
ny2
g Ž x 1ny1 , . . . , x nny1 .
Ž x 1 ??? x n .
F
if and only if p k 1 mod n.
One implication follows from Lemma 5.1, and so we need to consider
the case p k 1 mod n.
The case n s 3 seems to be the most difficult, and we handle that first.
Let R s K w X, Y, Z xrŽ X 3 q Y 3 q Z 3 . where p ' 2 mod 3. We need to
show that xyz g Ž x 2 , y 2 , z 2 . F .
Let A s y 3, B s z 3, and so A q B s yx 3. We first show that when
p s 2, we have xyz g Ž x 2 , y 2 , z 2 . F by establishing that Ž xyz . 8 g
Ž x 2 , y 2 , z 2 .w8x. Note that is suffices to show that Ž xyz . 6 g Ž x 15 , y 15, z 15 ., or
in other words that Ž AB Ž A q B .. 2 g Ž A5, B 5, Ž A q B . 5 ., but this is easily
seen to be true.
586
ANURAG K. SINGH
We may now assume p s 6 m q 5 where m G 0. We shall show that in
this case Ž xyz . p g Ž x 2 , y 2 , z 2 .w px, i.e., that
Ž xyz .
6 mq5
g Ž x 12 mq10 , y 12 mq10 , z 12 mq10 . .
Note that to establish this, it suffices to show
Ž xyz .
6 mq3
g Ž x 12 mq9 , y 12 mq9 , z 12 mq9 . ,
i.e., that Ž ABŽ A q B .. 2 mq1 g Ž A4 mq3, B 4 mq3, Ž A q B . 4 mq3 ..
LEMMA 5.3. Let K w A, B x be a polynomial ring o¨ er a field K of characteristic p s 6 m q 5 where m G 0. Then we ha¨ e
Ž AB Ž A q B . .
2 mq1
g I s Ž A4 mq3 , B 4 mq3 , Ž A q B .
4 mq3
..
Proof. To show that Ž ABŽ A q B .. 2 mq1 g I, we shall show that the
following terms grouped together symmetrically from its binomial expansion,
f 1 s Ž AB .
3 mq1
3m
Ž A q B . , f 3 s Ž AB . Ž A3 q B 3 . , . . . ,
f 2 mq1 s Ž AB .
2 mq1
Ž A2 mq1 q B 2 mq1 . ,
are all in the ideal I. Note that I contains the elements Ž AB . m Ž A q
B . 4 mq3 , Ž AB . my 1 Ž A q B . 4 mq5, . . . , Ž AB .Ž A q B . 6 mq1, Ž A q B . 6 mq3. We
consider the binomial expansions of these elements modulo Ž A4 mq3,
B 4 mq3 ., and get the following elements in I:
ž
ž
ž
4m q 3
4m q 3
4m q 3
f q
f q ??? q
f
,
2m q 2 1
2m q 3 3
3m q 2 2 mq1
/
/
ž
ž
/
/
ž
ž
/
/
4m q 5
4m q 5
4m q 5
f q
f q ??? q
f
,
2m q 3 1
2m q 4 3
3m q 3 2 mq1
???
6m q 3
6m q 3
6m q 3
f q
f q ??? q
f
.
3m q 2 1
3m q 3 3
4 m q 2 2 mq1
/
ž
/
ž
/
587
A COMPUTATION OF TIGHT CLOSURE
The coefficients of f 1 , f 3 , . . . , f 2 mq1 arising from these terms form the
matrix
¡ 4m q 3
ž
ž
2m q 2
¢ž
6m q 3
3m q 2
4m q 5
2m q 3
/ ž
/ ž
4m q 3
2m q 3
4m q 5
2m q 4
/
/
???
???
ž
ž
4m q 5
3m q 3
/¦
/
ž
6m q 3
4m q 2
/§
4m q 3
3m q 2
???
/ ž
6m q 3
3m q 3
/
???
.
We need to show that this matrix is invertible, but in the notation of
Lemma 3.2, its determinant is F Ž4 m q 3, 2 m q 2, m. and is easily seen to
be nonzero.
The above lemma completes the case n s 3. We may now assume n G 4
and p s nk q d for 2 F d F n y 1. If k s 0, i.e., 2 F p F n y 1, we have
Ž x 1 ??? x n .
Ž ny2 . p
s y Ž x 1 ??? x ny1 .
Ž ny2 . p
n
x nŽ ny2. pyn Ž x 1n q ??? qx ny1
.
Ž ny1. p
g Ž x 1Ž ny1. p , . . . , x ny1
..
In the remaining case, we have n G 4 and k G 1. To prove that Ž x 1 ???
x n . ny 2 g Ž x 1ny 1, . . . , x nny1 . F , we shall show
Ž x 1 ??? x n .
Ž ny2 . p
g Ž x 1Ž ny1. p , . . . , x nŽ ny1. p . .
This would follow if we could show
Ž x 1 ??? x n .
Ž ny2 . n k
g Ž x 1Ž ny1. n kqn , . . . , x nŽ ny1. n kqn . .
As before, let A1 s x 1n, . . . , A n s x nn. It suffices to show that
Ž A1 ??? A n .
Ž ny2 . k
g Ž AŽ1ny1. kq1 , . . . , AŽnny1. kq1 . .
By Lemma 3.3, this reduces to showing
Ž A1 A 2 .
Ž ny2 . k
k
Ž A1 q A 2 . g I
s AŽ1ny1. kq1 , AŽ2ny1. kq1 , Ž A1 q A 2 .
ž
Ž ny1 . kq1
/.
The only remaining ingredient is the following lemma.
LEMMA 5.4. Let K w A, B x be a polynomial ring o¨ er a field K of characteristic p ) 0 where p s nk q d and where n G 4, k G 1, and 2 F d F n y 1.
588
ANURAG K. SINGH
Then
Ž A, B . 2 ny3 k : I s Ž AŽ ny1. kq1 , B Ž ny1. kq1 , Ž A q B .
Ž
.
Ž ny1 . kq1
..
In particular, Ž AB .Ž ny2. k Ž A q B . k g I.
Proof. Note that I contains the elements Ž A q B .Ž ny1. kq1Ak B Ž ny3. ky1 ,
Ž A q B .Ž ny1. kq1A ky1 B Ž ny3. k , . . . , Ž A q B .Ž ny1. kq1 B Ž ny2. ky1. We take the
binomial expansions of these elements and consider them modulo the
ideal Ž AŽ ny1. kq1, B Ž ny1. kq1 .. This shows that the following elements are in
I:
žŽ
žŽ
žŽ
n y 1. k q 1
Ž n y 1 . k q 1 AŽ ny2. k B Ž ny1. k ,
AŽ ny1. k B Ž ny2. k q ??? q
kq1
2k q 1
/
/
n y 1. k q 1
AŽ ny1. k B Ž ny2. k q ??? q
k
???
ž
žŽ
/
/
n y 1. k q 1
AŽ ny2. k B Ž ny1. k ,
2k
n y 1. k q 1
Ž n y 1 . k q 1 AŽ ny2. k B Ž ny1. k .
AŽ ny1. k B Ž ny2. k q ??? q
1
kq1
/
ž
/
The coefficients of AŽ ny1. k B Ž ny2. k , AŽ ny1. ky1 B Ž ny2. kq1, . . . , AŽ ny2. k B Ž ny1. k
form the matrix
¡ Ž n y 1. k q 1
ž
žŽ
¢ž Ž
kq1
n y 1. k q 1
k
n y 1. k q 1
1
/ žŽ
/ žŽ
/ žŽ
n y 1. k q 1
kq2
n y 1. k q 1
kq1
n y 1. k q 1
2
/
/
/
???
???
???
???
žŽ
žŽ
žŽ
n y 1. k q 1
2k q 1
n y 1. k q 1
2k
n y 1. k q 1
kq1
/¦
/
/§
.
To show that all monomials of degree Ž2 n y 3. k in A and B are in I, it
suffices to show that this matrix is invertible. The determinant of this
matrix is
žŽ
n y 1. k q 1
kq1
kq1
kq1
ž
n y 1. k q 2
nk q 1
???
kq1
kq1
kq2
2k q 1
???
kq1
kq1
/žŽ
/ž
/ ž
/ ž
/
/
which is easily seen to be nonzero since the characteristic of the field is
p s nk q d where 2 F d F n y 1.
A COMPUTATION OF TIGHT CLOSURE
589
Remark 5.5. It is worth noting that xyz g Ž x 2 , y 2 , z 2 .* in the ring
R s K w X, Y, Z xrŽ X 3 q Y 3 q Z 3 . is, in a certain sense, unexplained. Under mild hypotheses on a ring, tight closure has a ‘‘colon-capturing’’
property: for x 1 , . . . , x n part of a system of parameters for an excellent
local Žor graded. equidimensional ring A, we have Ž x 1 , . . . , x ny1 .: A x n :
Ž x 1 , . . . , x ny1 .* and various instances of elements being in the tight closure
of ideals are easily seen to arise from this colon-capturing property.
To illustrate our point, we recall from wHo, Example 5.7x how z 2 g Ž x, y .*
in the ring R above is seen to arise from colon-capturing. Consider the
Segre product T s RaS where S s K wU, V x. Then the elements x¨ y yu,
xu and y¨ form a system of parameters for the ring T. This ring is not
Cohen]Macaulay as seen from the relation on the parameters
2
2
Ž zu . Ž z¨ . Ž x¨ y yu . s Ž z¨ . Ž xu . y Ž zu . Ž y¨ . .
The colon-capturing property of tight closure shows
Ž zu . Ž z¨ . g Ž xu, y¨ . :T Ž x¨ y yu . : Ž xu, y¨ . *.
There is a retraction R m K S ª R under which U ¬ 1 and V ¬ 1. This
gives us a retraction from T ª R which, when applied to Ž zu.Ž z¨ . g
Ž xu, y¨ .*, shows z 2 g Ž x, y .* in R.
REFERENCES
wHHx M. Hochster and C. Huneke, Tight closure, invariant theory, and the Briançon]Skoda
theorem, J. Amer. Math. Soc. 3 Ž1990., 31]116.
wHox M. Hochster, Tight closure in equal characteristic, big Cohen]Macaulay algebras, and
solid closure, Contemp. Math. 159 Ž1994., 173]196.
wHRx M. Hochster and J. Roberts, The purity of the Frobenius and local cohomology, Ad¨ .
in Math. 21 Ž1976., 117]172.
wHux C. Huneke, Tight closure, parameter ideals and geometry, in ‘‘Commutative Algebra,’’
Birkhauser,
Basel, in press.
¨
wMcx M. McDermott, ‘‘Tight Closure, Plus Closure and Frobenius Closure in Cubical
Cones,’’ Thesis, Univ. of Michigan, 1996.
wMux T. Muir, ‘‘A Treatise on the Theory of Determinants’’ Žrevised and enlarged by W. H.
Metzler., Longmans, Green, New YorkrLondon, 1933.
wRox P. Roberts, A computation of local cohomology, Contemp. Math. 159 Ž1994., 351]356.
wSmx K. E. Smith, Tight closure of parameter ideals, In¨ ent. Math. 115 Ž1994., 41]60.
Download