THIRTEEN OPEN

QUESTIONS IN
COMMUTATIVE ALGEBRA

Speaker: Mel Hochster

This is an edited version of the slides used at a talk at a conference in honor of Joseph Lipman on the occasion of his
sixty-fifth birthday at a conference at Purdue University in July, 2004. Each question is followed by some very terse
comments on its status and some references, which are sometimes woven into the comments and sometimes follow
them. The references given are, in most cases, only a small sample of the relevant literature.

1. The Zariski-Lipman conjecture: Let R be a finitely generated C-algebra and P a prime of R. If DerC (R, R)P
is RP -free, then RP is regular.
Comments: RP is normal (Lipman). The hypersurface case is known (Scheja-Storch). The graded case is known
(Hochster). The dimension 2 case implies the general case (Flenner), but is open even for complete intersections in
C4 .
J. Lipman, Free derivation modules on algebraic varieties, Amer. J. Math. 87 (1965) 874–898.
G. Scheja and U. Storch,Über differentielle Abhängigkeit bei idealen analytischer Algebren, Math. Z. 114 (1970)
101–112.
, Differentielle Eigenschaften des Lokaliserungen analytischer Algebren, Math. Ann. 197 (1972) 137–170.
M. Hochster, The Zariski-Lipman conjecture in the graded case, J. of Alg. 47 (1977) 411–424.
H. Flenner, Extendability of differential forms on nonisolated singularities, Invent. Math. 94 (1988) 317–326.

2. Resolution of singularities: Given a reduced and irreducible excellent scheme, or simply a domain finitely
generated over a field, over Z, or over an excellent discrete valuation ring, one wants to find a proper birational map
of a scheme or variety onto it whose local rings are regular. Better yet, if R is a domain, one wants f1 , . . . , fn ∈ R
such that V (f1 , . . . , fn ) is the singular locus and, for all i, R[f1 /fi , f2 /fi , . . . , fn /fi ] is regular. This is a local algebraic
version of blowing up. Iterations of this process have been the main tool in resolving singularities.
Comments: The literature is HUGE. There are a handful of references below. The recent exposition by H. Hauser
listed below contains many more. Hauser calls Zariski “the grandfather” of resolution of singularities. Much notable
work was done by S. S. Abhyankar. Hironaka solved the problem in equal characteristic zero. Lipman gave an optimal
treatment for surfaces. Recent work of De Jong solves the problem in characteristic p and mixed characteristic if the
birationality condition is relaxed and one allows a finite field extension of function fields.
S. S. Abhyankar, Local uniformization of algebraic surfaces over ground fields of characteristic p 6= 0, Ann. Math. 63
(1956) 491–526.
, Resolution of singularities of embedded algebraic surfaces, Academic Press, New York, 1966.
H. Hironaka, Resolution of singularities of an algebraic variety over a field of characteristic zero I, II. Ann. of Math.
(2) 79 (1964) 109–203; 205–326.
J. Lipman, Desingularization of 2-dimensional schemes, Ann. Math. 107 (1978) 151–207.
E. Bierstone and P. D. Milman, Uniformization of analytic spaces, J. Amer. Math. Soc. 2 (1989) 801–836. A. J. de
Jong, Smoothness, semi-stability and alterations, Publ. I.H.E.S. Math. No. 83 (1996) 51–93.
A. J. de Jong, Families of curves and alterations, Ann. Inst. Fourier (Grenoble) 47 (1997) 599–621.
H. Hauser, The Hironaka theorem on resolution of singularities (Or: A proof we always wanted to understand), Bull.
A.M.S. 40 (2003) 323–403.

= R[y1 , . . . , ym ] is R ∼
3. Cancellation of indeterminates (Zariski problem): If C[x1 , . . . , xn+m ] ∼ = C[x1 , . . . , xn ]?
Comments: The answer is yes for n = 1 or n = 2 (Fujita, Miyanishi, Sugie: cf. [T. Fujita, On Zariski problem, Proc.
Japan Acad. Ser. A Math. Sci 55 (1979) 106–110], M. Miyanishi and T. Sugie, Affine surfaces containing cylinderlike

1
open sets, J. Math. Kyoto Univ. 20 (1980) 11–42], and [T. Sugie, Characterization of the affine plane and the
affine threespace, in Topological Methods in Algebraic Transformation Groups (editors H. Kraft, T. Petrie, and G. W.
Schwarz), Progress in Math. 80, Birkhäuser, Boston-Basel-Berlin, 1989, 177–190].)
Be wary about cancellation!
Let R, S be C[x, y, z]/ xy − (1 − z 2 ) , C[x, y, z]/ x2 y − (1 − z 2 ) . R ∼
6 S but R[t] ∼



= = S[t] ! (Danielewski surfaces.)
Cf. [W. Danielewski, On the cancellation problem and automorphism groups of affine algebraic varieties, preprint,
Warsaw, 1989], [K.-H. Fieseler, On complex affine surfaces with C+ -action, Comment. Math. Helv. 69 (1994) 5–27].

4. The Jacobian conjecture: If R = C[x1 , . . . , xn ], and f : R → R with f (xi ) = fi for 1 ≤ i ≤ n is such that
det(∂fi /∂xj ) ∈ C − {0}, then f is an automorphism.
Comments: There have been many incorrect proofs: several have been published. Cf. [H. Bass, E. Connell, and
D. Wright, The Jacobian Conjecture: reduction of degree and formal expansion of the inverse, Bull. A.M.S. 7 (1982)
287–330]. The question remains open even if n = 2. A survey is given in [A. van den Essen, Polynomial automorphisms
and the Jacobian conjecture, Algèbre non commutative, groupes quantiques et invariants (Reims, 1995), 55–81, Sémin.
Congr., 2, Soc. Math. France, Paris, 1997.]

5. Set-theoretic definition: Is there a method for determining the least number of generators of an ideal up to
radicals (how many equations are needed to define an algebraic set)?
This seems incredibly difficult. It is not known whether every irreducible curve in A3C is even locally a set-theoretic
complete intersection! It is not known whether, for example, the parametric Moh curve
x = t6 + t31 , y = t8 , z = t10
is a set-theoretic complete intersection at the origin! Cf. [G. Lyubeznik, A survey of problems and results on the
number of defining equations, in Commutative Algebra M.S.R.I. Publ. 15 Springer-Verlag, New York, 1989, 375–390].

6. The strong intersection conjecture: Let (R, m) be a local (Noetherian) ring and let M, N 6= 0 be finitely
generated R-modules such that pdR M < ∞ and M ⊗R N has finite length. Let I = AnnR M . Then dim (N ) ≤
depth I R.
Comments: Cf. [C. Peskine and L. Szpiro, Dimension projective finie et cohomologie locale, Publ. I.H.E.S. 42,
1973].Many other questions considered there have been settled, including conjectures of H. Bass and M. Auslan-
der. In equal characteristic several follow from the existence of big Cohen-Macaulay modules. In mixed characteristic,
others were settled by P. Roberts using intersection theory. M. Auslander’s rigidity conjecture was settled negatively
by R. Heitmann. See also:
M. Hochster, Topics in the homological theory of modules over commutative rings, C.B.M.S. Regional Conf. Ser. in
Math. No. 24, Amer. Math. Soc., Providence, R.I., 1975.
P. Roberts, Le théorème d’intersection, C.R. Acad. Sc. Paris Sér. I 304 (1987) 177–180.
P. Roberts, Multiplicities and Chern classes in local algebra, Cambridge Tracts in Mathematics 133, Cambridge Univ.
Press, Cambridge, England, 1998.
R. Heitmann, A counterexample to the rigidity conjecture for rings, Bull. A.M.S. New Series 29 (1993) 94–97.
7. The Buchsbaum-Eisenbud-Horrocks problem: If R is regular local, dim (R) = n, and `(M ) < ∞, then the
n


j th Betti number of M is ≥ j .

Comments: See D. Buchsbaum and D. Eisenbud, Algebra structures for finite free resolutions and some structure
theorems for ideals of codimension 3, Amer. J. Math. 99 (1977) 447–485 and R. Hartshorne, Algebraic vector bundles
on projective spaces: a problem list, Topology 18 (1979) 117–128. This is known if dim (R) ≤ 4. Weaker form: the
sum of the
Betti numbers is at least 2n . Stronger form: the torsion-free rank of a minimal j th module of syzygies
is ≥ n−1
j−1 ,
1 ≤ j
≤ n. Closely related: if `(M ) < ∞ over (R, m), a local ring, and x = x1 , . . . , xn ∈ m, then
` Hj (x; M ) ≥ nj .
Another somewhat related question: if I, J ⊆ R with (R, m) regular local, dim (R) = n, and I + J m-primary
then TorR
1 (R/I, R/J) needs ≥ n − dim (R/I) − dim (R/J) generators. (This is open for K[[x, y, z]] even if I, J are
m-primary: one expects (I ∩ J)/IJ to need at least three generators.)

2
The original problem asks for a lower bound for the dimension of TorR
j (M, K). The second asks for a lower bound for
R R
the dimension of Tor0 (Tor1 (R/I, R/J), K). One may ask, more generally, for lower bounds for the dimensions of more
complicated iterated Tor modules under various conditions on the input modules. Why make it harder? Sometimes,
the “harder” question is easier.
Another closely related question: Let (A, m) be an Artin local ring and let x1 , . . . , xn ∈ m. Is the number of generators
of H1 (x1 , . . . , xn ; A) at least n? This is true if n = 1 and n = 2 (S. Dutta).

8. Positivity of Serre multiplicities: LetP(R, m) be regular
local, dim (R) = n. Let M , N be Noetherian R-
n
modules with `(M ⊗R N ) < ∞. If χ(M, N ) = j=1 ` TorR j (M, N ) then:
(1) dim M + dim N ≤ dim R, (2) χ(M, N ) ≥ 0, and (3) χ(M, N ) > 0 iff equality holds in (1).

Comments: This was proved by Serre if R contains a field or if R b is formal power series over a DVR. Serre proved (1)
in general. Cf. [J.-P. Serre, Algèbre local · Multiplicités, Springer-Verlag Lecture Notes in Math. 11, Springer-Verlag,
New York, 1961. That χ vanishes when it should was proved in [P. Roberts, The vanishing of intersection multiplicities
of perfect complexes, Bull. A.M.S. 13 (1985) 127–130] and [H. Gillet and C. Soulé, K-théorie et nullité des multiplicités
d’intersection, C. R. Acad. Sci. Paris Série I no. 3 t. 300 (1985) 71–74 ]. (2) was proved by O. Gabber: an exposition
is given in [P. Berthelot, Altérations de variétes algébriques [d’après A. J. de Jong], Séminaire BOURBAKI, 48ème
année, no 815, 815-01 – 815-39] using De Jong’s alterations. Positivity of multiplicities remains a frustrating open
question. It is implied by the existence of small Cohen-Macaulay modules, considered in 12. below.

9. Direct summands of regular rings are Cohen-Macaulay: If R is a direct summand as an R-module of a

regular ring S, then must R be Cohen-Macaulay?
This is easy if S is module-finite over R. It is known in equal char.: cf. [M. Hochster and J. L. Roberts, Rings
of invariants of reductive groups acting on regular rings are Cohen-Macaulay, Advances in Math. 13 (1974) 115–
175], [M. Hochster and C. Huneke, Applications of the existence of big Cohen-Macaulay algebras, Advances in Math.
113 (1995), 45–117], and [ , Tight closure in equal characteristic zero, Springer-Verlag, to appear]. The result
for finitely generated C-algebras is sharpened in [J.-F. Boutot, Singularités rationelles et quotients par les groupes
réductifs, Invent. Math. 88 (1987) 65–68]: if S has rational singularities and R is a direct summand, then R has
rational singularities. The question is open in mixed characteristic. It would follow from a solution of the fourth form
of the problem considered in 12. or from a solution of 13.

10. The direct summand and canonical element conjectures; understanding superheight: Let R be
regular and R ⊆ S module-finite. Is R a direct summand of S?
This is trivial if R ⊇ Q (and then R need only be normal), and was proved in char. p > 0 in [M. Hochster, Contracted
ideals from integral extensions of regular rings, Nagoya Math. J. 51 (1973) 25–43]. It is easy in dimension ≤ 2. It
reduces to the case of formal power series over a DVR in which p generates the maximal ideal. In [R. C. Heitmann, The
direct summand conjecture in dimension three, Annals of Math. (2) 156 (2002) 695–712] the result was finally proved
in mixed characteristic in dimension 3. The direct summand conjecture remains open if dim (R) ≥ 4. An equivalent
conjecture (although it “appears” stronger) is that: if (R, m, K) is local and x = x1 , . . . , xn is a system of parameters,
if one maps the Koszul complex K• = K• (x; R) to a free minimal resolution G• of K beginning with 1R , then the
induced map Kn = R → syz n K is such that the image of 1 is not in (x)syz n K. Letting the system of parameters
n n
vary, one sees that one has a (conjecturally nonzero) canonical element in Hm (syz K), or, as pointed out to me
R n


by Joe Lipman, a (conjecturally, nonzero) canonical element in Torn K, Hm (R) . Using that the direct summand
conjecture implies the canonical element conjecture, one sees that it implies many of the early local homological
conjectures (known from Roberts’ work). Cf. [M. Hochster, Canonical elements in local cohomology modules and the
direct summand conjecture, J. of Algebra 84 (1983) 503–553].
The direct summand conjecture is also equivalent to the statement that if x1 , . . . , xn is a system of parameters of
/ (xt+1
a local ring (R, m), then (x1 · · · xn )t ∈ t+1
1 , . . . , xn )R (the monomial conjecture). Another equivalent Pn statement
is this: if (V, x1 V ) is a complete DVR and T = V [[x2 , ..., xn , y1 , . . . , yn ]], and f = (x1 · · · xn )t − j=1 yj xtj , then, if
R = T /f T , for any map R → S, ht (x1 , . . . , xn )S, if finite, is ≤ n − 1. More generally, one may ask: can one tell what
the largest height a proper expansion of an ideal I of a Noetherian ring R to a Noetherian R-algebra S can achieve?
I.e., what is the superheight of I? The problem remains unsolved even in simple special cases in low dimension.

3
11. Localization of tight closure; tight closure vs. plus closure: If R is an excellent Noetherian domain, char.
p > 0, does the tight closure I ∗ of I commute with localization? Is it the same as the elements in IR+ ∩ I, where R+
is the integral closure of R in an algebraic closure of its fraction field?
e e e e
Comments. Recall that u ∈ I ∗ if for some c 6= 0, cup ∈ I [p ] ∀e  0, where I [p ] = (ip : i ∈ I)R.
IR+ ∩ R ⊆ I ∗ . By [K. E. Smith,Tight closure of parameter ideals, Invent. Math. 115 (1994) 41–60] they are equal if
IRP is generated by part of a system of parameters in RP for all P ⊇ I. Brenner [H. Brenner, Tight closure and plus
closure for cones over elliptic curves, preprint], [ , Tight closure and plus closure in dimension two, preprint] has
done the case of homogeneous ideals in the homogeneous coordinate rings of elliptic curves and of other nonsingular
curves if the base field is the algebraic closure of Z/pZ.
Plus closure commutes with localization, and so the second question would solve the localization problem for tight
closure.

12. Existence of (big) Cohen-Macaulay (C-M) modules and algebras. Let (R, m, K) be local. We shall say
here that M is a big C-M module or algebra if mM 6= M and every system of parameters R is a regular sequence on
M . We say that M is a (small) C-M module if it is finitely generated.
(1) Does every complete local domain have a small C-M module?
(2) Do there exist big C-M modules?
(3) Do there exist big C-M algebras?
(4) Given a local map of complete local domains R → S do there exist big C-M algebras BR , BS and an R-algebra
map BR → BS (weakly functorial big C-M algebras).
(5) Does a version of tight closure theory exist in mixed characteristic?
Comments: both (2), and (3) reduce to the case of complete (even normal) local domains. Small C-M modules are
known to exist in dimension 2 and for finitely generated N-graded domains R of char. p > 0 with R0 a perfect field if
the ring has an isolated non-C-M point at the origin: this was first observed by R. Hartshorne. The argument is given
in [M. Hochster, Big Cohen-Macaulay modulesand algebras and embeddability in rings of Witt vectors, in Proceedings
of the Queen’s University Commutative Algebra Conference, Queen’s Papers in Pure and Applied Math., No. 42,
1975, 106–195].
Big C-M modules were shown to exist in equal char. (using reduction to char. p) in [ , Topics in the Homological
Theory of Modules over Commutative Rings, Proceedings of the Nebraska Regional C.B.M.S. Conference, (Lincoln,
Nebraska, 1974), A.M.S., Providence, 1975]. (4) was proved in char. p¿0 in [ and C. Huneke, Infinite integral
extensions and big Cohen-Macaulay algebras, Annals of Math. 135 (1992) 53–89]: R+ works, and in char. 0 in the
sequel [ , Applications of the existence of big Cohen-Macaulay algebras, Advances in Math. 113 (1995) 45–117].
But (4) is also true for two rings of mixed char. in dimension ≤ 3. This is shown, making use of Heitmann’s results,
in [M. Hochster, Big Cohen-Macaulay algebras in dimension three via Heitmann’s theorem, J. of Alg. 254 (2002)
395–408]. I have heuristic reasons for believing that the problem of proving (4) in mixed characteristic in general and
the problem of developing an analogue of tight closure in mixed characteristic should be nearly equivalent.

13. The vanishing conjecture for maps of Tor and the strong direct summand conjecture: If A ⊆ R → S
are Noetherian rings with A, S regular and R is module-finite and torsion-free over A, the TorA R
i (M, R) → Tori (M, S)
is 0 for all A-modules M and all i ≥ 1.
Comments: This can be proved in equal characteristic using either tight closure theory or the weakly functorial
existence of big C-M algebras. This powerful result implies both the direct summand conjecture (when S is a field)
and the conjecture that direct summands of regular rings are Cohen-Macaulay (when (A, M, K) is regular local M = K,
and i = 1). Cf. [M. Hochster and C. Huneke, Applications of the existence of big Cohen-Macaulay algebras, Advances
in Math. 113 (1995) 45–117]. f In mixed characteristic, it is a conjecture.
In [N. Ranganathan, Splitting in module-finite extension rings and the vanishing conjecture for maps of Tor, Thesis,
University of Michigan, 2000] the surprising result is obtained that 13. is equivalent to the following strong direct
summand conjecture: let (R, m, K) be a regular local ring in which x ∈ m − m2 , let S be a module finite extension
domain, and let Q be a height one prime of S lying over xR. Then xR ⊆ Q splits as a map of R-modules (which implies
that R splits from S). There is no easy proof known in equal char. 0, and the result is open in mixed characteristic.

4
Here is one not-so-open question: In [Y. Kamoi, K. Kurano, On maps of Grothendieck groups induced by comple-
tion, J. Alg. 254 (2002) 21–43] it’s shown that for a local ring R, G0 (R) → G0 (R) b is injective if R is excellent and
either Henselian, or has an isolated singularity, or is the local ring of an N-graded algebra of finite type over a field.
The authors didn’t know an example of non-injectivity.
Here’s an example: let T = K[x, y, u, v]m and R = T /f T where f = xy − ux2 − yv 2 . f factors in Tb and P = (x, y)R
becomes principal mod either of the the two minimal primes of R. b But [R/(x, y)] 6= 0 (it is 2-torsion) in G0 (R), even if
one maps to RcP . If you are sad that this is not open here’s something to think about: I don’t know a counter-example
if R is normal (conjecture: there is one).

Homework: settle one.

Due: May 18, 2009

•Zariski-Lipman conjecture
•Resolution of singularities
•Cancellation of indeterminates
•Jacobian conjecture
•Set-theoretic definition
•Strong intersection
•Buchsbaum-Eisenbud-Horrocks problem
•Serreconjecture on positivity of multiplicities
•Direct summand of regular is Cohen-Macaulay
•Direct summand conjecture
•Tight closure commutes with localization
•Existence of big and/or small Cohen-Macaulay algebras or modules
•Vanishing of maps of Tor