Geometric and Topological Aspects of

the Representation Theory of Finite

Groups PIMS Summer School and
Workshop July 27 August 5 2016 1st
Edition Jon F. Carlson
Visit to download the full and correct content document:
https://textbookfull.com/product/geometric-and-topological-aspects-of-the-representati
on-theory-of-finite-groups-pims-summer-school-and-workshop-july-27-august-5-2016-
1st-edition-jon-f-carlson/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Geometric Methods in Physics XXXV Workshop and Summer

School Bia■owie■a Poland June 26 July 2 2016 1st
Edition Piotr Kielanowski

https://textbookfull.com/product/geometric-methods-in-physics-
xxxv-workshop-and-summer-school-bialowieza-poland-
june-26-july-2-2016-1st-edition-piotr-kielanowski/

Topological Aspects of Condensed Matter Physics:

Lecture Notes of the Les Houches Summer School: Volume
103, August 2014 1st Edition Claudio Chamon

https://textbookfull.com/product/topological-aspects-of-
condensed-matter-physics-lecture-notes-of-the-les-houches-summer-
school-volume-103-august-2014-1st-edition-claudio-chamon/

Representation Theory of Finite Groups a Guidebook 1st

Edition David A. Craven

https://textbookfull.com/product/representation-theory-of-finite-
groups-a-guidebook-1st-edition-david-a-craven/

Representation Theory of Finite Groups a Guidebook 1st

Edition David A Craven

https://textbookfull.com/product/representation-theory-of-finite-
groups-a-guidebook-1st-edition-david-a-craven-2/
Geometric Methods in Physics XXXVII Workshop and Summer
School Bia■owie■a Poland 2018 Piotr Kielanowski

https://textbookfull.com/product/geometric-methods-in-physics-
xxxvii-workshop-and-summer-school-bialowieza-poland-2018-piotr-
kielanowski/

Theory of Groups and Symmetries Finite groups Lie

groups and Lie Algebras 1st Edition Alexey P. Isaev

https://textbookfull.com/product/theory-of-groups-and-symmetries-
finite-groups-lie-groups-and-lie-algebras-1st-edition-alexey-p-
isaev/

Representation Theory of Finite Monoids 1st Edition

Benjamin Steinberg

https://textbookfull.com/product/representation-theory-of-finite-
monoids-1st-edition-benjamin-steinberg/

A Journey Through Representation Theory From Finite

Groups to Quivers via Algebras Caroline Gruson

https://textbookfull.com/product/a-journey-through-
representation-theory-from-finite-groups-to-quivers-via-algebras-
caroline-gruson/

Complex Semisimple Quantum Groups and Representation

Theory 1st Edition Christian Voigt

https://textbookfull.com/product/complex-semisimple-quantum-
groups-and-representation-theory-1st-edition-christian-voigt/
Springer Proceedings in Mathematics & Statistics

Jon F. Carlson
Srikanth B. Iyengar
Julia Pevtsova Editors

Geometric and
Topological Aspects
of the Representation
Theory of Finite
Groups
PIMS Summer School and Workshop,
July 27–August 5, 2016
Springer Proceedings in Mathematics & Statistics

Volume 242
Springer Proceedings in Mathematics & Statistics
This book series features volumes composed of selected contributions from
workshops and conferences in all areas of current research in mathematics and
statistics, including operation research and optimization. In addition to an overall
evaluation of the interest, scientific quality, and timeliness of each proposal at the
hands of the publisher, individual contributions are all refereed to the high quality
standards of leading journals in the field. Thus, this series provides the research
community with well-edited, authoritative reports on developments in the most
exciting areas of mathematical and statistical research today.

More information about this series at http://www.springer.com/series/10533

Jon F. Carlson Srikanth B. Iyengar
•

Julia Pevtsova
Editors

Geometric and Topological

Aspects of the Representation
Theory of Finite Groups
PIMS Summer School and Workshop,
July 27–August 5, 2016

123
Editors
Jon F. Carlson Julia Pevtsova
Department of Mathematics Department of Mathematics
University of Georgia University of Washington
Athens, GA, USA Seattle, WA, USA

Srikanth B. Iyengar
Department of Mathematics
University of Utah
Salt Lake City, UT, USA

ISSN 2194-1009 ISSN 2194-1017 (electronic)

Springer Proceedings in Mathematics & Statistics
ISBN 978-3-319-94032-8 ISBN 978-3-319-94033-5 (eBook)
https://doi.org/10.1007/978-3-319-94033-5
Library of Congress Control Number: 2018945084

Mathematics Subject Classification (2010): 20-06, 13-06, 16-06, 55-06

© Springer International Publishing AG 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

This volume is a proceedings of two workshops that were held in the summers of
2015 and 2016. Both events celebrated the mathematical accomplishments of
David J. Benson for the occasion of his 60th birthday. The papers presented at the
meetings were representative of the many mathematical subjects that Dave has
worked on during his distinguished career. The extent of his influence in these
areas, which was clearly on display at both of the meetings, is partly reflected in the
fact that there were two meetings, within the space of a year, organized in Dave’s
honor!
Dave did his graduate study at Cambridge University and wrote his thesis under
the direction of John Thompson. He is well known for establishing new relation-
ships between diverse mathematical areas, particularly connecting group repre-
sentation theory with commutative algebra and homotopy theory. In addition to his
many papers, his conjectures on subjects ranging from branching rules for modules
over symmetric groups to the regularity of cohomology rings of finite groups have
generated much activity. His books, most notably the two-volume survey of group
representations and cohomology, have been standard references for years and an
inspiration for many of the advances in the subject. Dave’s impact on the field has
also been through his numerous collaborations—he has 50 collaborators, according

v
vi Preface

to Math. Sci. Net., including the three editors of this volume. Dave is also much
admired by his colleagues and friends not only for being a fantastic mathematician
but for the whole string of other talents and interests he displays, including juggling,
unicycling, opera singing, and writing a popular book on “Math and Music”.
The first workshop had the title “Groups, Representations and Cohomology” and
was held on June 22–27, 2015 at Sabhal Mòr Ostaig, a school on the Isle of Skye in
Scotland. The organizers were Radha Kessar, Henning Krause, Markus Linckelmann,
Ran Levi, and Peter Symonds. It was preceded by an Anglo-Franco-German
Representation Theory Network Summer School on the subject “Support Varieties in
Modular Representation Theory” held on June 18–22, 2015, organized by Ivo
Dell’Ambrogio, Greg Stevenson, and Jan Stovicek.
The second meeting consisted of a summer school and workshop on the subject
“Geometric Methods in the Representation Theory of Finite Groups”. It took place
at the Pacific Institute for the Mathematical Sciences at the University of British
Columbia in Vancouver, from July 27 to August 5 of 2016. The summer school
featured four series of lectures by Eric Friedlander, Jesper Grodal, Radha Kessar,
and Peter Symonds. The summer school and workshop were organized by the three
editors of these proceedings.
Of the contributors to this volume, Carlson, Erdmann, Greenlees, Iyengar,
Oliver, Pevtsova, and Touzé presented invited lectures at the meeting on the Isle of
Skye. Friedlander, Kessar, and Symonds gave series of several lectures each at the
summer school at PIMS. Their submissions included here are summaries of those
lectures. Avramov, Henke, Huisgen-Zimmermann, Kessar, Krause, Linckelmann,
Lynd, Mathew, Nakano, Sobaje, Stevenson, Webb, and Witherspoon were invited
to speak at the conference in Vancouver.
Partial funding for the participants and speakers at the Vancouver meeting was
also provided by the National Security Agency (under grant NSA-AMS-141009)
and the National Science Foundation (under grant DMS-1624050). The meeting
was also supported by funds from the Sondersforschungsbereich 701 at Bielefeld
Universität. It remains to thank the Pacific Institute for the Mathematical Sciences
for their help in organizing the meeting in Vancouver and this volume.

Athens, GA, USA Jon F. Carlson

Salt Lake City, USA Srikanth B. Iyengar
Seattle, USA Julia Pevtsova
Contents

Restricting Homology to Hypersurfaces . . . . . . . . . . . . . . . . . . . . . . . . . 1

Luchezar L. Avramov and Srikanth B. Iyengar
Thick Subcategories of the Relative Stable Category . . . . . . . . . . . . . . . 25
Jon F. Carlson
Nilpotent Elements in Hochschild Cohomology . . . . . . . . . . . . . . . . . . . 51
Karin Erdmann
Rational Cohomology and Supports for Linear Algebraic Groups . . . . . 67
Eric M. Friedlander
Anderson and Gorenstein Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
J. P. C. Greenlees and V. Stojanoska
Understanding Finite Dimensional Representations Generically . . . . . . . 131
K. R. Goodearl and B. Huisgen-Zimmermann
Descent of Equivalences and Character Bijections . . . . . . . . . . . . . . . . . 181
Radha Kessar and Markus Linckelmann
Length Categories of Infinite Height . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Henning Krause and Dieter Vossieck
On Automorphisms and Focal Subgroups of Blocks . . . . . . . . . . . . . . . 235
Markus Linckelmann
Extensions of the Benson-Solomon Fusion Systems . . . . . . . . . . . . . . . . 251
Ellen Henke and Justin Lynd
Examples of Descent up to Nilpotence . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Akhil Mathew
Globally Irreducible Weyl Modules for Quantum Groups . . . . . . . . . . . 313
Skip Garibaldi, Robert M. Guralnick and Daniel K. Nakano

vii
viii Contents

A Remark on the Construction of Centric Linking Systems . . . . . . . . . . 327

B. Oliver
Varieties of Elementary Subalgebras of Maximal Dimension
for Modular Lie Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Julia Pevtsova and Jim Stark
Varieties Related to the Problem of Lifting Gr-Modules to G . . . . . . . . . 377
Paul Sobaje
Complete Boolean Algebras are Bousfield Lattices . . . . . . . . . . . . . . . . 393
Greg Stevenson
Endotrivial Modules for Infinite Groups . . . . . . . . . . . . . . . . . . . . . . . . 407
Peter Symonds
Cohomology of Algebraic Groups with Coefficients in Twisted
Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
Antoine Touzé
Bilinear Forms on Grothendieck Groups of Triangulated
Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Peter Webb
Varieties for Modules of Finite Dimensional Hopf Algebras . . . . . . . . . . 481
Sarah Witherspoon
Restricting Homology to Hypersurfaces

Luchezar L. Avramov and Srikanth B. Iyengar

To Dave Benson on the occasion of his 60th birthday

2010 Mathematics Subject Classification 13D07 (primary)

16E45, 13D02, 13D40 (secondary)
Introduction
This work concerns an analogue for commutative rings of Carlson’s theory of rank
varieties for elementary abelian groups [8]. It takes the following form: given a
(noetherian, commutative) local ring R that is a quotient of a local ring P, so that
R := P/I for an ideal I , the goal is to study an R-module M by its restrictions to
hypersurfaces P/( f ) where f ∈ I is a regular element (that is to say, not a zero
divisor). The rationale is that homological algebra over such hypersurfaces is well
understood, especially when P is a regular ring; then one can, for example, take
recourse to the theory of matrix factorizations initiated by Eisenbud [9].
In this endeavour, a basic question is how the properties of M change as we vary
the element f . The results below, contained in Theorem 2.1, addresses this issue.
If f , g in I are regular elements with f − g in nI , where n is the maximal ideal
of P, then for any R-module M there are isomorphisms of k-vector spaces

Partly supported by NSF grants DMS-1103176 (LLA) and DMS-1700985 (SBI). We are grateful
to Chris Drupieski for comments on earlier version of this manuscript.

L. L. Avramov
Department of Mathematics, University of Nebraska, Lincoln, NE 68588, U.S.A.
e-mail: avramov@math.unl.edu
S. B. Iyengar (B)
Department of Mathematics, University of Utah, Salt Lake City, UT 84112, U.S.A.
e-mail: iyengar@math.utah.edu

© Springer International Publishing AG 2018 1

J. F. Carlson et al. (eds.), Geometric and Topological Aspects of the Representation
Theory of Finite Groups, Springer Proceedings in Mathematics & Statistics 242,
https://doi.org/10.1007/978-3-319-94033-5_1
2 L. L. Avramov and S. B. Iyengar

(k, M ) ∼
P/(f ) P/(g)
Tor i = Tor i (k, M ) for each i ≥ 0.

It follows that when M is finitely generated, its projective dimension as a module

over P/(f ) and over P/(g) are simultaneously finite. This latter result was proved
by Avramov [2] when P is regular and I is generated by a regular sequence; it is
part of the theory of cohomological support varieties for modules over complete
intersections [4]. Jorgensen [12] generalized this to any ideal I in a domain P.
The isomorphism above yields more: the Betti numbers of M over P/(f ) and over
P/(g) are equal. Moreover, the statement carries over to the context of graded rings
and implies that the graded Betti numbers and hence also invariants derived from
them, like regularity, are equal.
That said, the motivation for writing this manuscript is not so much the greater
generality of the result; rather, it gives an alternative point of view—one that lays bare
the structural reason behind the numerical coincidences. The proof, given in Sect. 2,
uses a modicum of Differential Graded (henceforth abbreviated to DG) homological
algebra, recapped in Sect. 1.
As one application of the theorem in Sect. 5, we describe how to deduce it from
certain results of Carlson [8], Friedlander and Pevtsova [11], and Suslin [15] that
underlie the theory of rank varieties for finite groups and group schemes.
The theorem above also leads to the notion of a support set of M with respect to
the homomorphism π : P → R, denoted Vπ (M ). It is a subset of k c , where c is the
minimal number of elements required to generate the ideal I . We prove:
When I contains a regular element and M is finitely generated, the
subset Vπ (M ) ⊆ k c is closed in the Zariski topology.
This result is contained in Theorem 3.4 which identifies the support set of M as
the algebraic set defined by the annihilator of a certain module, over a polynomial
ring k[s1 , . . . , sc ], naturally associated to M . Up to radical, this annihilator ideal can
be described explicitly, as we illustrate in Sect. 4.

1 Differential Graded Algebra

In this section, we recall what little is needed by way of constructs and results
concerning DG algebras and DG modules. Our reference for this material is [3].
Let P be a commutative ring and A a DG P-algebra; it will be implicit that A is
graded commutative and satisfies Ai = 0 for i < 0.

1.1 Tensor Products

When A and B are DG P-algebras, so is A ⊗P B, with standard differential and

products
Restricting Homology to Hypersurfaces 3


(a ⊗ b)(a ⊗ b ) = (−1)|a ||b| aa ⊗ bb .

We identify A and B with their images in A ⊗P B and write ab instead of a ⊗ b.

1.2 Divided Powers

For w ∈ A2d with d ≥ 1, we say that (w (i) ∈ A2di )i0 is a sequence of divided powers
of w if w (0) = 1, w (1) = w, and there are equalities

(i + j)! (i+j)
w (i) w (j) = w and ∂(w (i) ) = ∂(w)w (i−1) for all i, j ≥ 0 .
i!j!

Induction on i yields w i = i!w (i) for i ≥ 0. Thus, w has a unique sequence of

divided powers if char(k) = 0, or if char(k) = p > 0 and Aj = 0 for j ≥ 2dp − 1.
However, not every element of even positive degree has divided powers in general.
If v and w have divided powers, then so do aw, for a ∈ P, and v + w with

(aw)i = ai w (i) and (v + w)(h) = v (i) w (j) .
i+j=h

1.3 The Koszul Complex

Let t1 , . . . , tn be elements in P and K the Koszul complex ont. Thus K is a DG

P-algebra with underlying graded algebra the exterior algebra P K1 , where K1 is
a free P-module with basis x1 , . . . , xn , and differential ∂ defined by the condition
∂(xi ) = ti . Then H0 (K) = P/(t), so K comes equipped with a canonical morphism
of DG P-algebras ε : K → P/(t).
Every element of K2d with d ≥ 1 has a sequence of divided powers. Only those
for d = 1 are needed here, and we proceed to define them ad hoc.
Write each w ∈ K2 as w = 1≤a<b≤n ra,b xa xb with ra,b ∈ P and set

w (i) := (ra1 ,b1 · · · rai ,bi ) xa1 xb1 · · · xai xbi for i ≥ 1.
1≤aj <bj ≤n
1≤j≤i

The right-hand side is well defined because the expression of w is unique. A direct
computations shows that ∂(w(i) ) = ∂(w)w (i−1) holds for i ≥ 0.
4 L. L. Avramov and S. B. Iyengar

1.4 The Tate Construction

Let P y be a graded-free P-module with basis {y(i) }i0 , where |y(i) | = 2i, and define
a P-linear multiplication table by

(i + j)! (i+j)
y(0) = 1 and y(i) y(j) = y for all i, j ≥ 0 .
i!j!

By construction, the element y has a sequence of divided powers, namely {y(i) }.

Let A be a DG P-algebra and z ∈ A1 a cycle. We write A y | ∂(y) = z for the DG
P-algebra with underlying graded algebra A ⊗P P y , with differential extending the
one on A and satisfying

∂(y(i) ) = zy(i−1) for all i ≥ 1

We abbreviate A y | ∂(y) = z to A y , if the differential on y is clear, or irrelevant.

If an element a ∈ A2 has a sequence of divided powers, then so do elements of
the P-submodule of A y generated by a and y; see Sect. 1.2.
The map A → A y assigning a ∈ A to ay(0) is one-to-one and a morphism of DG
P-algebras. For i ≥ 0, it induces a homomorphism

Hi (A) −→ Hi (A y )

that is bijective for i = 0 and surjective for i = 1, with kernel the ideal generated by
the class of z. The results below and its proof are standard; see [3, Sect. 6]. Details
of the proof are given for ease of reference.
Lemma 1.1 Let z be a cycle in A1 .
(1) If an element w ∈ A2 has a sequence of divided powers, then the DG P-algebras
A y | ∂(y) = z and A y | ∂(y) = z + ∂(w) are isomorphic.
(2) If the class of z is a basis for the H0 (A)-module H1 (A) and Hi (A) = 0 for i ≥ 2,
then the canonical map

A y | ∂(y) = z −→ H0 (A)

is a quasi-isomorphism.
Proof It is readily verified that the A-linear map

A y | ∂(y) = z −→ A y | ∂(y) = z + ∂(w)

that assigns y(i) to (y + w)(i) is a morphism of DG P-algebras; it is bijective, with

inverse the A-linear map that assigns y(i) to (y − w)(i) . This justifies (1).
The A-linear map  : A y →  2 A y assigning y(i) to y(i−1) is a morphism of
DG A-modules, with kernel A, identified as the DG A-submodule Ay(0) of A y . Thus,
Restricting Homology to Hypersurfaces 5

there is an exact sequence of DG A-modules


0 −→ A −→ A y −→  2 A y −→ 0 .

In homology, this yields that H0 (A y ) ∼

= H0 (A), an exact sequence

ð
0 −→ H2 (A y ) −→ H0 (A y ) −→ H1 (A) −→ H1 (A y ) −→ 0

of H0 (A)-modules where ð maps the class of 1 to the class of z, and isomorphisms

Hi (A y ) ∼
= Hi−2 (A y ) for i ≥ 3.

The hypothesis yields that ð is an isomorphism, so H2 (A y ) = 0 = H1 (A y ) and

then the isomorphisms above entail Hi (A y ) = 0 for all i ≥ 1. 

1.5 Acyclic Closures

Let t be a finite set of elements generating an ideal n of P and K the Koszul complex
on t. Let z1 , . . . , zc be cycles in K1 that generate the H0 (K)-module H1 (K). Iterating
the construction from Sect. 1.4 yields a DG P-algebra

T := K y1 , . . . , yc | ∂(yi ) = zi .

By construction, H0 (T ) = P/n and H1 (T ) = 0. A repetition of this procedure,

involving also the adjunction of exterior variables leads to a DG P-algebra A con-
taining T as a DG subalgebra and satisfying Hi (A) = P/n and Hi (A) = 0 for i ≥ 1.
This is an acyclic closure of P/n; see [3, Chap. 6].
As noted in Sect. 1.3, in the DG P-algebra K, each element of even positive
degree has a system of divided powers; the ones on K2 were described explicitly.
These induce a sequence of divided powers on the elements of degree two in T , see
Sect. 1.4, and hence also in A.

2 Hypersurfaces

Let P be a commutative noetherian ring, n a maximal ideal of P; set k = P/n. We

fix an ideal I of P contained in n and set R = P/I . For every f ∈ I , let

πf : P/(f ) −→ R

be the canonical homomorphism of rings.

6 L. L. Avramov and S. B. Iyengar

Theorem 2.1 Let I ⊂ P be an ideal containing a regular element, set R := P/I and
let M be an R-complex.
(1) For each f ∈ I , there exists an h ∈ nI such that f + h is P-regular.
(2) Let f and g be regular elements lying in I . If f − g is in nI , then there is an
isomorphism of graded k-vector spaces

Tor P/(f ) (k, M ) ∼

= Tor P/(g) (k, M ) .

(3) If f is a regular element lying in nI , then there is an isomorphism of graded

k-vector spaces

Tor P/(f ) (k, M ) ∼

= Tor P (k, M ) ⊗k k y , with |y| = 2.

The proof is given at the end of this section. Part (1) is well known and part (3) is
essentially contained in the work of Shamash [14, Theorem 1].
Throughout the rest of the section, t := t1 , . . . , tn denotes a finite generating set
for the ideal n and A denotes an acyclic closure of k over P; see Sect. 1.5.

2.1 Degree One Cycles

Set C := A ⊗P R; this is a DG R-algebra. As A is a free resolution of k, there is an

isomorphism of graded k-spaces

H(C) ∼
= Tor P (k, R) .

Expressing elements of I as linear combinations of t1 , . . . , tn gives rise to degree one

cycles of C. We give a compact description of that process by using the set

Z := {(f , x) ∈ I ⊕ A1 | f = ∂(x)} .

This is a P-submodule of I ⊕ A1 .
Lemma 2.2 The assignments f ← (f , x) → x ⊗ 1 define surjective P-linear maps

ε ζ
I←− Z −→ Z1 (C) .

An element z ∈ Z satisfies ε(z) ∈ nI if and only if ζ(z) is in ∂2 (C).

Proof Let x be an arbitrary element of A1 .
The element x ⊗ 1 is a cycle in C1 if and only if ∂(x) lies in I . It follows that ζ is
well defined and that both ζ and ε are surjective.
Fix z := (f , x) in Z, so ζ(z) = x ⊗ 1. Then, ζ(z) is in ∂(C) if and only if x lies in
∂(A) + IA1 . Thus, ζ(z) in ∂(C) yields f = ∂(x) ∈ ∂(IA1 ) = nI .
Restricting Homology to Hypersurfaces 7


Conversely,
 f in nI gives f = ∂( ai xi ) with {xi }ni=1 in A
1 and {ai }i=1 in I , so that
n

x − ai xi is a cycle in A1 . As H1 (A) is zero, we get x − ai xi = ∂(w) for some

w ∈ A, whence x ⊗ 1 = ∂(w ⊗ 1). 

2.2 Hypersurfaces

Fix an element f in I , set P := P/(f )P and consider the DG P-algebra A := A ⊗P

P = A/fA. Choose an element zf ∈ A1 such that ∂(zf ) = f . The residue class of zf
in A is a cycle, denoted z f . Set

Bf := A y | ∂(y) = z f .

This has a canonical augmentation Bf → k, which is a morphism of DG A-algebras.

Lemma 2.3 If f is a regular element, then Bf → k is a quasi-isomorphism.

Proof Applying Lemma 2.2 with (f ) in place of I , we see that H1 (A) is generated
by the class of zf and is zero if and only if f lies in nf ; the latter possibility is ruled
out by the hypothesis that f is regular. By using the resolution

f
0→P−
→ P → 0,

of P over P we get 
k for i = 0, 1
Tor Pi (k, P) ∼
=
0 otherwise

Since H(A) ∼
= Tor P (k, P), it remains to apply Lemma 1.1. 

Proof of Theorem 2.1 The zero divisors in P are the elements of its associated primes;
call them p1 , . . . , pn . Since I has a regular element so does nI , and hence the latter
is not contained in ∪ni=1 pi . Thus, (1) follows from [13, Theorem 124].
In the next steps, the notation and constructions from Sects. 2.1 and 2.2 will be
used. In particular, A is a DG P-algebra resolution of k, and C := A ⊗P R. Moreover
Bf is a DG P-algebra resolution of k.

Claim 1 Tor P/(f ) (k, M ) is the homology of the DG R-module

C y | ∂(y) = zf ⊗ 1 ⊗R M

Indeed, Tor P/(f ) (k, M ) is the homology of the DG R-module Bf ⊗P/(f ) M . Asso-
ciativity of tensor products yields isomorphisms of R-complexes
8 L. L. Avramov and S. B. Iyengar

Bf ⊗P/(f ) M ∼
= ((A/fA) y | ∂(y) = z f ⊗P/(f ) R) ⊗R M
∼
= C y | ∂(y) = zf ⊗ 1 ⊗R M

This justifies the claim.

(3) When f ∈ nI holds, the element zf is in IA and hence its residue class in C is
zero. The preceding claim then yields

Tor P/(f ) (k, M ) ∼

= H(C y | ∂(y) = 0 ⊗R M )
= H(C ⊗R M ) ⊗R R y
= H(A ⊗P M ) ⊗R R y
∼
= Tor P (k, M ) ⊗R R y
∼
= Tor P (k, M ) ⊗k k y

(2) Recall that the element g is also a regular element in I , with f − g in nI .

Claim 2 There is an isomorphism of DG R-algebras

C y | ∂(y) = zf ⊗ 1 ∼
= C y | ∂(y) = zg ⊗ 1 .

Indeed, Lemma 2.2 provides elements (f , x) and (g, y) in the module Z defined
in Sect. 2.1, and w in C2 satisfying x ⊗ 1 − y ⊗ 1 = ∂(w). Since w has a sequence
of divided powers (see Sect. 1.5), Lemma 1.1 yields the desired isomorphism.
The isomorphism in the Claim 2 induces an isomorphism of R-complexes

C y | ∂(y) = zf ⊗ 1 ⊗R M ∼
= C y | ∂(y) = zg ⊗ 1 ⊗R M .

Taking homology, and recalling Claim 1, yields the desired isomorphism. 

3 Support Sets

Let (P, n, k) be a (commutative noetherian) local ring, with maximal ideal n and
residue field k. Let π : P → R be a surjective homomorphism of rings such that the
ideal I := Ker(π) contains a regular element. Throughout, M will be an R-module;
the results presented below all carry over to complexes; see Remark 3.5.

3.1 Support Sets

We view the k-vector space I /nI as endowed with the Zariski topology. Given f ∈ I
we write [f ] for its residue class in I /nI . The homological support set of M with
Restricting Homology to Hypersurfaces 9

respect to π is the subset of I /nI described by

⎧ ⎫
⎪
⎨ there exists a regular element g ∈ I with ⎪
⎬
Vπ (M ) := [f ] ∈ I /nI f − g ∈ nI such that Tor iP/(g) (k, M ) = 0 ∪ {0} .
⎪
⎩ ⎪
⎭
for infinitely many integers i

Thanks to Theorem 2.1, one can test whether [f ] ∈ Vπ (M ) holds by considering

Tor P/(g) (k, M ) for any regular element g in I with [f ] = [g]. Evidently, if [f ] is
in Vπ (M ), then so is [λf ] for any non-zero element λ in k. Hence, Vπ (M ) is an
affine cone with vertex at {0}. Therefore, one could also consider the homological
support set as a subset of the projective space associated to I /nI . See Remark 3.6 for
antecedents of this construction.
The condition on the vanishing of Tor appearing in the definition of the homo-
logical support set of M has a more familiar interpretation, at least under additional
conditions on the module M . This is explained in the next paragraph.

3.2 Projective Dimension

We write proj dimR M for the projective dimension of the R-module M . Assume M
is finitely generated or m-torsion; the latter condition means that for each u ∈ M
there is an integer l such that ml u = 0. Then

proj dimR M < ∞ ⇐⇒ Tor Ri (k, M ) = 0 for i  0 ; (1)

see, for example, [6, Propositions 5.3F and 5.5].

3.3 Alternative Description of Support

Fix a minimal generating set f1 , . . . , fc for the ideal I . The residue classes [f1 ], . . . , [fc ]
form a basis for I /nI as a k-vector space. We will use this to identify I /nI with k c
whenever necessary, or convenient.
Given a point a := (a1 , . . . , ac ) in k c , it follows from Theorem 2.1(1) that there
is an element a := (a1 , . . . , ac ) in P c such that

i = ai mod n for i = 1, . . . , c;
(1) a
(2) ai fi is regular in P.
We will call such an element a a lifting of a to P. One can then also describe the
subset Vπ (M ) from Sect. 3.1 as
10 L. L. Avramov and S. B. Iyengar
 
there exists a lifting a of a with
Vπ (M ) = a ∈ k c 
proj dimP/(f ) (M ) = ∞ for f = ai fi

In particular, the right-hand side does not depend on the choice of a minimal gener-
ating set for I . The subset Vπ (M ) ⊆ k c is closed when M is finitely generated; this is
contained in Theorem 3.4. For the proof of this result, we require a statement about
complexes over regular rings, presented below.

3.4 Complexes over Regular Rings

By a regular ring we mean a commutative noetherian ring S such that the local rings
Sp are regular for all p in Spec S. For any p ∈ Spec S, We write k(p) for the residue
field, (S/p)p , at p.

Lemma 3.1 Let S be a regular ring and let X be a complex of S-modules such that
Hi (X ) is finitely generated for each i and equals 0 for i  0. Fix a prime ideal p in
S. The following conditions are equivalent.
(1) Tor Si (k(p), X ) = 0 for i  0;
(2) k(p) ⊗S Hi (X ) = 0 for i  0;
(3) Hi (X )p = 0 for i  0.

Proof Since the action S on k(p) factors through the localization homomorphism
S → Sp and Sp is flat as an S-module, one has isomorphisms

Tor S (k(p), X ) ∼
= Tor Sp (k(p), Xp )
k(p) ⊗S Hi (X ) ∼ = k(p) ⊗S Hi (X )p
p

Moreover, Xp is a complex of Sp -modules with (Xp )i = 0 for i  0. Thus, replacing

S and X by their localizations at p, we assume S is local and p is its maximal ideal.
(1) ⇐⇒ (3) Let E be the Koszul complex on a minimal set of generators of the
ideal p. Since E ∼
= k in the derived category of S, the hypothesis in (1) equivalent
to Hi (E ⊗S X ) = 0 for i  0. Using the fact that the S-modules Hi (X ) are finitely
generated, a standard argument now implies Hi (X ) = 0 for i  0; see, for example,
[10, Sect. 1.3].
(2) ⇐⇒ (3) This is by Nakayama’s Lemma. 

3.5 On Being Closed

Let P[s] be the polynomial ring over P in indeterminates s := s1 , . . . , sc and

Restricting Homology to Hypersurfaces 11


c
 := P[s]/(
Q f) where 
f := s i fi .
i=1

The ideal I contains a P-regular element so there is no non-zero element in P that

annihilates each of f1 , . . . , fc . Therefore, 
f is a regular element in P[s]; see, for
example, [1, Chap. 1, Exercise 2]. Thus, the complex of P[s]-modules


f
0 −→ P[s] −→ P[s] −→ 0 (2)


is a free resolution of Q.
For each a := (a1 , . . . , ac ) in P c form, the homomorphism of P-algebras

εa : P[s] −→ P given by si → ai .

In what follows, we write Pa for P viewed as a P[s]-module via εa . Base change
 gives a commutative square
along the canonical surjection P[s] → Q

εa
P[s] P


Q Q P/(f )

where f = ai fi .

Lemma 3.2 When the element f := ai fi is P-regular, one has

 Q for i = 0
Tor P[s]
i (Pa , Q) =
0 otherwise

 over P[s], one gets that Tor P[s] (Pa , Q)

Proof Using the free resolution (2) of Q  is
the homology of the complex

f
0 −→ P −→ P −→ 0 

Set M [s] := P[s] ⊗P M . Evidently f · M [s] = 0 so M [s] is an Q-module.

Proposition 3.3 With notation as above, the following statements hold.


(1) When p := proj dimP M is finite, for any Q-module N one has
 
Tor i+2 (N , M [s]) ∼
Q Q
= Tor i (N , M [s]) for i ≥ p.
12 L. L. Avramov and S. B. Iyengar


(2) When i ai fi is regular in P, one has


Q M for i = 0
Tor i (Q, M [s]) =
0 otherwise

Proof The argument uses the standard first quadrant change of rings spectral
 and a P[s]-module L.
sequence associated to the homomorphism P[s] → Q,

2
Ei,j
Q
:= Tor i (Tor P[s]  P[s]
j (L, Q), M [s]) =⇒ Tor i+j (L, M [s])

(1) Set L := N in the spectral sequence. From (2) and 

f N = 0, one gets that

 ∼ N for j = 0, 1
Tor jP[s] (N , Q) =
0 otherwise

2
Thus, the Ei,j = 0 for j = 0, 1, so the spectral sequence unwinds to an exact sequence


Q 
Q
· · · → Tor P[s]
i+2 (N , M [s]) → Tor i+2 (N , M [s]) → Tor i (N , M [s])

→ Tor P[s]
i+1 (N , M [s]) → · · ·

Since proj dimP M = p, one has proj dimP[s] M [s] = p, so the first and the last
modules in the sequence above are zero for i ≥ p. This yields (1).
(2) Consider the spectral sequence above with L := Pa . In view of Lemma 3.2,
this spectral sequence collapses on the second page and yields isomorphisms

Tor i (Q, M [s]) ∼
Q P[s]
= Tor i (Pa , M [s]) for each i.

Let F be a free resolution of M as a P-module; then F[s] := P[s] ⊗P F is a free

resolution of M [s] over P[s]. Associativity of tensor products yields

→ P ⊗P M ∼
Pa ⊗P[s] F[s] − =M

of P-complexes. This yields


M for i = 0
Tor P[s]
i (Pa , M [s]) =
0 otherwise

The preceding computations justifies the assertion in (2). 

The statement and proof of the next result involves constructions introduced in
Sect. 3.5. We identify k[s] and the ring of k-valued algebraic functions on I /nI by
mapping si to the ith coordinate function of the k-basis {[f1 ], · · · , [fc ]} of of I /nI .
Restricting Homology to Hypersurfaces 13

Theorem 3.4 Assume that k is algebraically closed and that I contains a regular
element. Fix an integer d ≥ depth P. Let M be a finitely generated R-module and J
the annihilator of the k[s]-module

Q 
Q
Tor d (k[s], M [s]) ⊕ Tor d +1 (k[s], M [s]) .

If proj dimP M = ∞, then Vπ (M ) = k c . Otherwise, it is given by

Vπ (M ) = {(a1 , . . . , ac ) ∈ k c | h(a) = 0 for all polynomials h(s) ∈ J }.

In particular, Vπ (M ) is a Zariski-closed subset of k c .

Proof Set p := proj dimP M . If proj dimP/(f ) M is finite for some regular element
f ∈ I , then p is finite as well. Therefore, Vπ (M ) = k c when p = ∞; see Sect. 3.3.
For the rest of the proof, we assume that p is finite.

a ∈ Ak . Let a ∈ P be a lifting of a, as in Sect. 3.3, and set Q := P/(f ),
c c
Fix a point
where f = i ai fi . Let G be a free resolution of M [s] over Q  and set

 G.
X := k[s] ⊗Q

By construction, this is a complex of free k[s]-modules and satisfies


H(X ) ∼
= Tor Q (k[s], M [s]) .

In particular, the k[s]-module Hi (X ) is finitely generated for each i, and 0 for i < 0.

Claim 1 Let εa : k[s] → k be the homomorphism of k-algebras with εa (xi ) = ai for

each i. There is an isomorphism of graded k-vector spaces

Tor Q (k, M ) ∼
= Tor k[s] (ka , X ) .

Indeed, by Proposition 3.3(2), the complex Q ⊗Q  G of free Q-modules is a resolution

of M , so Tor Q (k, M ) is the homology of the complex

k ⊗Q (Q ⊗Q ∼
 G) = k ⊗Q
∼
 G = ka ⊗k[s] (k[s] ⊗Q
 G) = ka ⊗k[s] X .

The second isomorphism holds because the composed map Q  → Q → k factors

εa

as Q → k[s] − → k. Taking homology and keeping in mind that the k[s]-complex X
consists of free modules and is concentrated in non-negative degrees, one gets the
stated isomorphism.

Claim 2 Set ma := Ker(εa ). One has

a ∈ Vπ (M ) ⇐⇒ ma ⊇ J .
14 L. L. Avramov and S. B. Iyengar

Indeed, consider the following chain of equivalences, where the first one holds by
definition, and the second is by Claim 1.
Q
/ Vπ (M ) ⇐⇒ Tor i (k, M ) = 0 for i  0
a∈
⇐⇒ Tor k[s]
i (ka , X ) = 0 for i  0
⇐⇒ Hi (X )ma = 0 for i  0.

The last one is given by Lemma 3.1, applied with S := k[s] and p := ma . Since
proj dimR M = p is finite, the Auslander–Buchsbaum Equality implies d ≥ p. Propo-
sition 3.3(1), then gives the first equivalence below.

Hi (X )ma = 0 for i  0 ⇐⇒ Hd (X )ma ⊕ Hd +1 (X )ma = 0

⇐⇒ ma  J .

The second one is standard. Putting together the string of equivalences above yields
the contrapositive of the desired claim.

Since k is algebraically closed, Hilbert’s Nullstellensatz implies that every max-

imal ideal of k[s] is of the form ma , for some a ∈ k c , and that

ma ⊇ J ⇐⇒ h(a) = 0 for all polynomials h(s) ∈ J .

Combining this with the conclusion in Claim 2 yields the desired result. 

Remark 3.5 Theorem 3.4 carries over, with essentially the same proof, to the case
when M is an R-complex with H(M ) finitely generated. The crucial point is that, in
the notation above, the natural morphism of Q-complexes

 M [s] −→ M
Q ⊗LQ

is a quasi-isomorphism. This can be verified using a spectral sequence as in the proof

of Proposition 3.3(2).

Remark 3.6 When P is a domain the homological support set of M , as in Sect. 3.1, is
the one introduced by Jorgensen [12], which extends the notion of varieties for mod-
ules over complete intersections due to Avramov [2]. The last part of Theorem 3.4—
that Vπ (M ) is closed—was proved by Jorgensen [12, Theorem 2.2] under the addi-
tional assumption that P is a domain. The argument in op. cit. is quite different from
the one presented above, which builds on a idea in the proof of [4, Theorem 3.1].
Restricting Homology to Hypersurfaces 15

4 Defining Equations

The main result of this section, Theorem 4.3, gives equations that define the homo-
logical support set of modules over complete intersections. The statement, and its
proof, involves some linear algebra over commutative rings, recalled below.

Remark 4.1 Let S be a local ring and δ : U → V a homomorphism of finite free S-

modules. For an integer r ≥ 0, let Ir (δ) denote the ideal generated by the r × r minors
of a matrix representing δ in some bases for U and V ; the ideal Ir (δ) is independent
of these choices. The following conditions are equivalent for any integer r ≥ 1.
(1) Ir (δ) = S;
(2) Im(δ) shares with V a free direct summand of rank r;
(3) Coker(δ) can be generated, as an S-module, by rank S V − r elements.
Assume in addition that S is a domain. With S0 the field of fractions of S, the rank of an
S-module M is the rank of the S0 -vector space S0 ⊗S M . Then, when rank Im(δ) ≤ r,
the conditions above are equivalent to
(4) Coker(δ) is free of rank equal to rank S V − r.
These assertions are easy to verify, given that Ir (δ) = S if and only if there exists a
choice of bases for U and V such that δ is represented by a matrix of the form
 
Ir 0
0 B

where Ir is the identity matrix of size r; see also [7, Lemmas 1.4.8, 1.4.9].
√
The results below, where J denotes the radical of an ideal J , concerns differential
modules; cf. [5, Remark 1.6], and also [5, Example 1.7] that shows that the hypothesis
that S is regular is needed.

Proposition 4.2 Let S be a regular ring and δ : S 2r → S 2r an S-linear map with

δ 2 = 0. The annihilator J of the S-module Ker(δ)/ Im(δ) satisfies
√ 
J = Ir (δ) .

Proof Set H := Ker(δ)/ Im(δ). The radical of an ideal is the intersection of the
prime ideals containing it, and a prime ideal p of S contains J precisely when Hp is
non-zero. Thus, the desired statement is equivalent to:

Hp = 0 ⇐⇒ Ir (δ) ⊂ p .

Replacing S by Sp and δ by δp , it thus suffices to prove that when S is a regular local

ring, one has H = 0 if and only if Ir (δ) = S.
There are exact sequences of S-modules
16 L. L. Avramov and S. B. Iyengar

0 −→ Ker(δ) −→ S 2r −→ Im(δ) −→ 0 , (3)

0 −→ Im(δ) −→ Ker(δ) −→ H −→ 0 , (4)
0 −→ H −→ Coker(δ) −→ Im(δ) −→ 0 . (5)

In particular, there are (in)equalities

rank S Ker(δ) + rank S Im(δ) = 2r and rank S Ker(δ) ≥ rank S Im(δ) .

Therefore, one has

rank S Ker(δ) ≥ r ≥ rank S Im(δ) ,

and both inequalities become equalities when H = 0.

Assume H = 0 so that rank S Im(δ) = r. Since the S-modules Coker(δ) and Im(δ)
are isomorphic, by (5), it remains to prove that Im(δ) is free; then Remark 4.1 would
yield Ir (δ) = S.
For each integer n ≥ 1, there is an exact sequence of S-modules

0 −→ Im(δ) −→ S 2r −→ · · · −→ S 2r −→ Im(δ) −→ 0 ,

of length n. In particular for n = dim S + 2, one sees that Im(δ) is a (dim S)-th
syzygy module and hence free, as S is a regular local ring. This is as desired.
Assume Ir (δ) = S. Since S is a domain and rank S Im(δ) ≤ r, it follows from
Remark 4.1 that Coker(δ) is free of rank r. This implies that, with l the residue field
of S, the natural maps
∼
= ∼
=
Coker(δ) ⊗S l −→ Coker(δ ⊗S l) and Im(δ) ⊗S l −→ Im(δ ⊗S l)

are isomorphisms. Thus, applying (−) ⊗S l to (5) and keeping in mind that Im(δ) is
free, one then gets an exact sequence

0 → H ⊗S l → Coker(δ ⊗S l) → Im(δ ⊗S l) → 0 ,

The isomorphisms above imply that both l-vector spaces on the right have rank r. It
follows that H ⊗S l = 0; thus H = 0, by Nakayama’s Lemma. 

4.1 Complete Intersections

Let (P, n, k) be a regular local ring of Krull dimension d and containing its residue
field, k, as a subring. Let I be the ideal generated by a P-regular sequence f1 , . . . , fc
in n2 and set R := P/I .
As in Sect. 3.5, let  P[s] be the polynomial ring over P in indeterminates s :=
s1 , . . . , sc , set   := P[s]/(
f := i si fi and Q  is a hypersurface, the Q-
f ). Since Q 
Restricting Homology to Hypersurfaces 17

 Q
module Q/n ∼ = k[s] has a free resolution G with the property that G i ∼
=Q2d −1 and
∂i = ∂i+2 for all i ≥ c so that the complex G ≥c has the form

B A B A
· · · −→ G c+3 −→ G c+2 −→ G c+1 −→ G c

The matrices A and B come from a matrix factorization of 

f ; see [9].

Given an R-module M , the matrices A and B define Q-linear endomorphisms of
d −1
M [s]2 and hence an endomorphism
 
0 A d d
: M [s]2 −→ M [s]2 .
B 0

When M has finite rank as a k-vector space, one can view this as an endomorphism
of free k[s]-modules of rank (rank k M )2d ; we write C(M ) for this map.

Theorem 4.3 Let R be as in Sect. 4.1. If M is an R-module with rank k M finite,

VR (M ) is defined by the vanishing of the ideal Ir (C(M )), where r := (rank k M )2d −1 .

Proof Note that, Tor Q (k[s], M [s]) is the homology of the complex G ⊗Q
 M [s],
which for i ≥ c reads
d −1 B d −1 A d −1
· · · −→ M [s]2 −→ M [s]2 −→ M [s]2 .

Q 
Q
Since depth P = d ≥ c, it follows that Tor d +1 (k[s], M [s]) ⊕ Tor d (k[s], M [s]) is the
middle homology of the sequence

d C(M ) d C(M ) d
M [s]2 −−−−→ M [s]2 −−−−→ M [s]2

of k[s]-modules. The desired statement is thus a consequence of Theorem 3.4, and

Proposition 4.2 applied with S := k[s] and δ := C(M ). 

Remark 4.4 Theorem 4.3 should be compared to [4, Theorem 3.2] that describes
equations defining VR (M ) in terms of data derived from a free resolution, say F, of
M over P, instead of the resolution of k[s] over Q.  In detail, one can construct a
system of higher homotopies, a family of endomorphism (not chain maps) of F that
encode the data that, because M is an R-module, multiplication by f1 , . . . , fc on F is a
commuting family of morphisms that are homotopic to zero. Then [4, Theorem 3.2],
expresses Vπ (M ) as the zero set of appropriate minors of the matrix involving these
higher homotopies. In contrast, the C(M ) appearing in Theorem 4.3 is universal, in
that it involves only a matrix factorization of 
f , and matrices describing the action
of the elements of P on M , viewed as k-vector space.

We finish by describing the free resolution of k[s] over Q  and, in particular, the
matrices A and B, when R is a truncated power series ring; group algebras of abelian
p-groups over a field of positive characteristic p have this form.
18 L. L. Avramov and S. B. Iyengar

Example 4.5 Let k be a field, P := k[|t1 , . . . , td |] the power series ring over k in
indeterminates t := t1 , . . . , td , and set

R := k[|t1 , . . . , td |]/(t1u1 , . . . , tcuc ) ,

where ui ≥ 2 for each i.

 on t; thus E = Q  x | ∂(xi ) = ti . The element
 LetuE be the Koszul complex over Q
i s i t i
i −1
xi in E1 is evidently a cycle. Since t is a regular sequence in P[s] and  f is
a regular element contained in (t), the complex
  u −1 
E y | ∂(y) = si ti i xi
i


is a Q-free 
resolution of Q/(t), that is to say, of k[s]; see Lemma 2.2. For any integer
n ≥ c, one has 
E2i when n is even
E y n := i0
i0 E2i+1 when n is odd


In either parity, it is a finite free Q-module of rank 2c−1 . Moreover, after a suitable
choice of bases, the differential on E y is given by

∂even := A and ∂odd := B .

 described below.
with A, B are square matrices of size 2c−1 , and coefficients in Q,
The rows and columns of A and B are indexed by subsets of the sequences

U := {h = (h1 , . . . , hc ) | h ∈ {0, 1}c }

ordered weighted lexicographically. 

The matrix A has rows indexed  by sequences h in U with hi odd and columns
indexed by sequences j with
 ji even. The entry in A in position (h, j) is 0 when
i |ji − hi | 
= 1; otherwise, there is a unique integer p such that |jp − hp | = 1, and
setting n = i<p ji , one has
 u −1
(−1)n sp tp p if hp = 1 and jp = 0
Ah,j :=
(−1)n tp if hp = 0 and jp = 1

The rows of B are indexed by
 sequences h with hi even and the columns are
indexed by sequences j with ji odd. Its entries are defined exactly as for A.
Restricting Homology to Hypersurfaces 19

5 Group Algebras of Elementary Abelian Groups

Let k be a field of positive characteristic p, and P := k[s] the polynomial ring in

p p
indeterminates s := s1 , . . . , sc . Set I := (s1 , . . . , sc ) and R := P/I . The k-algebra
R is the group algebra of an elementary abelian p-group and, more generally, of a
quasi-elementary finite group scheme [11]. The ring R admits a coproduct making it
into a Hopf algebra, but this structure plays no role in what follows.
Set n := (s); this is a maximal ideal of P. Each u ∈ n yields a homomorphism

ηu : k[t]/(t p ) −→ R such that ηu (t) = u mod I .

For such a u one has up ∈ I and hence also a surjective homomorphism

πup : P/(up ) −→ R

of k-algebras; see Sect. 2. In what follows it is expedient to write M ↓u for M viewed

as a k[t]/(t p )-module via ηu . The proof of the next result extracts an argument from
that for [4, Theorem 3.2], and is given after Lemma 5.4.
Theorem 5.1 For any R-module M one has

M ↓u is free ⇐⇒ flat dimP/(up ) (M ) < ∞ .

In the equivalence above, the invariant on the right is the flat dimension.

5.1 Flat Dimension

Let Q be a commutative noetherian ring and N a Q-module. Recall that flat dimQ (N ),
the flat dimension of N , is the length of the shortest flat resolution of N over Q. Evi-
dently flat dimQ (N ) ≤ proj dimQ (N ); equality holds when N is finitely generated.
The Q-module N is J -torsion, for some ideal J in Q, if for each u ∈ N there is an
integer l such that J l u = 0. If N is n-torsion for a maximal ideal n of Q, then
Q
flat dimQ N < ∞ ⇐⇒ Tor i (Q/n, N ) = 0 fori  0 . (6)

This follows form [6, Propositions 5.3.F], keeping in mind that for any prime ideal
q of Q, when q = n, one has Tor Q ((Q/q)q , Nq ) = 0.
We record a few consequences of Theorem 5.1.
Corollary 5.2 If an element w ∈ n satisfies u − w ∈ n2 , then M ↓w is free if and
only if is M ↓u .
Proof Evidently nl annihilates R, and hence M , for l  0; thus M is n-torsion. Given
(6), the desired result follows from Theorems 2.1(2) and 5.1. 
20 L. L. Avramov and S. B. Iyengar

5.2 Rank Varieties

In this paragraph we assume that k is algebraically closed. Let Ack be the affine space
over k, of dimension c. We write a for a point (a1 , . . . , ac ) in this space. For each
R-module M set
 

VRr (M ) := a ∈ Ack M ↓u is not free for u = ai si ∪ {0} .
i

This is called the rank variety of M . It was introduced by Carlson who also proved
that it is a closed subset of Ack ; see [8, §4]. The result below, which relates the rank
variety of M to its homological support set in the sense of Sect. 3.1, is contained
in [2, Theorem 7.5]. The argument in op. cit. gets to the stated bijection via the
cohomological variety of M . Our proof, using Theorems 2.1 and 5.1, is more direct.
p p
Corollary 5.3 The map Ack → Ack sending (a1 , . . . , ac ) to (a1 , . . . , ac ) is bijective
and for each R-module M it restricts to a bijection
∼
=
VRr (M ) −→ VR (M ) .

Proof The given map is a bijection on Ack because k is algebraically closed. The
assertion about the varieties associated to M is a translation of Theorem 5.1. 

A classical observation—see [13, Part III]—will be used in the proof of Theo-

rem 5.1. It is a special case of a statement that holds for smooth morphisms.

5.3 Polynomial Extensions

Let S be a commutative ring, S[x] a polynomial ring in d indeterminates x, and M a

S[x]-module.

Lemma 5.4 The following inequalities hold:

flat dimS M ≤ flat dimS[x] M ≤ flat dimS M + d .

Proof The inequality on the left holds because any flat resolution over S[x] is also a
flat resolution over S, due to the freeness of S[x] as a S-module.
For the inequality on the right, we may assume that flat dimS M is finite. Induction
on d then shows that it suffices to deal with the case when S[x] is a polynomial ring in
a single indeterminate. In the ring S[x] ⊗S S[x], the element x ⊗ 1 − 1 ⊗ x is regular,
so we have an exact sequence of S[x] ⊗S S[x]-modules

x⊗1−1⊗x
0 −→ (S[x] ⊗S S[x]) −−−−−−→ (S[x] ⊗S S[x]) −→ S[x] → 0
Another random document with
no related content on Scribd:
 T
Taffy’s imagination (fear of the dark)
MP25627.
Taj Mahal.
MP26015.
The Take.
LP43762.
Take my life.
R593268.
The Taking of Pelham one two three.
LP43957.
Tale of King Midas.
MP25885.
Tale of Rumpelstiltskin.
MP25905.
Tales of Manhattan.
LU3678.
The Talking magpies in Flying south.
R591491.
Tanbark champions.
R583809.
Tandem Productions, Inc.
LP43835 - LP43853.
LP43965 - LP43981.
Tantalus, Inc.
MP25507.
Tawny Pipit.
R593272.
Tax audit.
LP43849.
Taylor, Stanley B.
MP25706.
The Teachers’ crisis.
MP26037.
Teaching Films, Inc., division of A-V Corporation. SEE A-V
Corporation. Teaching Films, Inc.
Technique for lifting.
MP25852.
Ted Ziegler and his folding dome meet Bucky Fuller.
MU9077.
Teenage relationships: Venessa and her friends.
MP25908.
Telephone techniques to get appointments.
MP25990.
Tell ’em where to get it.
MP26055.
Tenafly.
LP43708.
Tender years.
R589461.
Tennis in rhythm.
R584970.
Tennis town.
R590801.
Tenth Avenue angel.
R594212.
Terminal man.
LP43787.
Terr-Filmkunst.
R584014.
Territoriality in reef animals.
MP25866.
Terry-Toons, Inc.
R591488 - R591494.
The Texas chain saw massacre.
LU3676.
Texas. Criminal Justice Division.
MP25813 - MP25814.
MP25993 - MP25996.
Text of light.
MP25816.
Texture Films, Inc.
LP43674.
LP43675.
The T-group experience for individual and organizational
effectiveness, film 1.
MP25609.
The T-group experience for individual and organizational
effectiveness, film 2.
MP25610.
The T-group experience for individual and organizational
effectiveness, film 3.
MP25611.
Thailand: past and present.
MP25705.
That Hagen girl.
R592812.
That’s entertainment.
LP43640.
That’s my gal.
R582730.
That’s my man.
R582729.
They’re off.
R580998.
They steal cars, don’t they?
LP43697.
They won’t believe me.
R584159.
Thicker.
MU9025.
Thin air.
LP43699.
Third world.
MP25536.
This happy breed.
R593271.
This sporting life.
LF162.
This time for keeps.
R588088.
Thomas (Tom) Organization.
MP25720.
Thorne Films, Inc.
MP25856 - MP25874.
Three-cornered cage.
LP43800.
Three Musketeers.
LF190.
Three sisters.
LF184.
Threshold: the Blue Angels experience.
MP25920.
Thunder in the valley.
R583806.
Thunder Mountain.
R584161.
Tiara Tahiti.
LF168.
Tikki tikki tembo.
LP43684.
A Time for learning.
MP26010.
Time, Inc.
MP25747.
MP26035 - MP26047.
R584088 - R584098.
R587470.
Time of darkness.
LP43642.
Time to discover.
MP25750.
A Time to heal.
MP25919.
Tizzard, Edward G.
MP25750.
T-men in action.
MP26044.
Tobis Klangfilm.
R583554.
R584010.
Tobis Zeigt.
R584006.
R584008.
Tomorrow Entertainment, Inc.
LP43755.
Tomorrow’s Mexico.
R584090.
Tonight at 8:30.
LF199.
Tonometry.
MP25959.
Too long to wait.
LP43950.
Tools for cutting: stone axes to lasers.
MP25730.
Tooth or consequences.
R582964.
To Paris with love.
LF166
Top Choice.
MP25543.
Topol, John.
MU9037.
To ride a yeller horse.
LP43876.
To stalk the night.
LP43701.
Touchstone.
LP43928.
Touch your face.
MU9164.
The Town tamers.
LP43871.
Trailing danger.
R588501.
A Trail of bloodshed.
LP43869.
Trail of the lonesome Pink.
LP43766.
Trail to San Antone.
R588538.
Train, Middleton G. C.
MU9024.
Training the construction laborer.
MP25714.
Transition.
MP25565.
Transition-lift.
MP25564.
Transparent animals.
MP25660 - MP25663.
MP26057 - MP26058.
Trap that failed.
R587222.
The Treasure hunt (teamwork)
MP25624.
The Trespasser.
R582739.
Trevor, Elleston.
LF202.
Trial by knife.
LP43654.
Trial of Donald Duck.
R589193.
Triangles.
LP43894.
Tricks are not treats.
LP43819.
Triode vacuum tube voltmeters.
MP25525.
Tropical biology.
 MP25811.
Tropical harmony.
R584821.
Trouble with women.
R587023.
Trustees of Trust under the Will of Mark J. Hellinger.
R585843.
Truth.
MP25749.
Try to die on time.
LP43823.
Tuesday movie of the week.
LP43716.
Tundra is a wide road.
MU9030.
Turkey’s 100 million.
MP26042.
Turn the key softly.
LF192.
Twentieth Century-Fox Film Corporation.
LF190.
LP43734.
LP43735.
LP43736.
LP43738.
LF43740.
LP43763.
LP43773.
LP43774.
LP43879.
LU3678.
R583794 - R583816.
R584187 - R584188.
R593725 - R593750.
Twenty-one cent bet and a positive slump eliminator.
MP25570.
Twenty thousand streets under the sky.
LF163.
Twilight on the Rio Grande.
R588539.
Two.
LP43909.
Two blondes and a redhead.
R591426.
Two Cities Films, Ltd.
LF158.
LF166.
LF178.
R590127.
R590128.
R593275.
 U
UFA Filmkunst.
R583549.
R583552.
R585555.
R583556.
R584007.
R584559.
UFA Filmkunst Zeigt.
R584009.
U.F.O.’s.
LP43905.
Uncle Silas.
R593275.
Uncle Tom’s cabana.
R581040.
Unconquered.
R587026.
Under Colorado skies.
R590897.
Undercover agent.
MU9161.
Under the Tonto Rim.
R586033.
Unekis, Richard.
LP43734.
The Unfaithful.
R583122.
The Unfinished dance.
R583044.
Unicorn Productions, Inc.
MP25705.
Union Pacific Railroad.
MP25709.
Union Pacific Railroad. Department of Public Relations. Audio-
Visual Services.
MP25709.
Union Pacific Railroad. Safety Department.
MP25709.
United Artists.
LP43802.
United Artists Corporation.
LP43679.
LP43726.
LP43727.
LP43728.
LP43765.
LP43766.
LP43767.
LP43775.
LP43783.
LP43784.
LP43785.
LF43888.
LP43919.
LP43920.
LP43921.
LP43957.
United Artists Television, Inc.
R580300 - R580305.
R583122 - R583125.
R585005 - R585013.
R587017 - R587022.
R590795 - R590803.
R592812 - R592817.
United Brands Company.
 R591937.
R592080 - R592088.
United States. Department of Justice. Law Enforcement Assistance
Administration.
MP25993 - MP25996.
United States. Department of the Interior. National Park Service.
MP25884.
United States Golf Association.
LP43794.
Units of conviction.
MP25981.
Universal.
LP43806 - LP43809.
Universal City Studios, Inc.
LP43697 - LP43705.
LP43707 - LP43708.
LP43710 - LP43725.
LP43756.
LP43804 - LP43805.
LP43932 - LP43938.
Universal City Studios, Inc. Emka Division.
R587023 - R587027.
R593022 - R593023.
Universal-International.
LF181.
Universal International Pictures, Inc.
R584818.
Universal Pictures.
LP43743.
LP43768.
LP43769.
LP43776.
LP43777.
R583192 - R583202.
 R584814 - R584840.
R587707 - R587716.
R590970 - R590980.
R591746 - R591754.
Universal Pictures Company, Inc.
R583192.
R583194 - R583202.
R584814 - R584817.
R584819 - R584840.
R585843.
R587707 - R587716.
R590970 - R590980.
R591746 - R591754.
R593269 - R593273.
Universal Pictures, Ltd.
LP43770.
Universal Productions France, S.A.
LP43786.
Universal Studios.
LP43786.
University of Louisville (KY) National Crime Prevention Institute.
MP25812.
University of Michigan, Ann Arbor.
MP25965.
University of Michigan, Ann Arbor. Audio-Visual Education Center.
MP25965.
University of Michigan, Ann Arbor. Department of Electrical and
Computer Engineering.
MP25965.
University of Oregon, Portland. Medical School. Division of
Instructional Aids.
MP25775.
University of South Carolina, Columbia. Division of Educational
Services.
 MP25562.
UniWorld Group, Inc.
MP25746.
Unsuspected.
R590796.
Unusual behavior patterns in fishes.
MP25874.
Up here looking down.
MP25614.
Up n’ atom.
R582972.
Uptown Saturday night.
LP43810.
Urban design issues in Texas.
MP25808.
Urchins of the coral reef.
MP25863.
Use of art therapy in a vocational milieu.
MP25825.
 V
Vacant land.
MU9028.
Vacation days.
R584642.
Vacation magic.
R593732.
Vacations (two weeks a year)
R593749.
Valley of fear.
R585845.
Valse des toreadors.
LF160.
Vanished traces.
R584008.
Vantage Communications, Inc.
MP25675.
Variety girl.
R587027.
Velveteen rabbit.
LP43908.
Vera paints Ibiza in the sun.
LP26034
Verdon Productions, Ltd.
LP43810.
Verwehte Spuren.
R584008.
A Very missing person.
LP43724.
Very natural thing.
MP25546.
Viacom international, Inc.
R591488 - R591494.
Videocom, Inc.
MP25976 - MP25992.
View from the crib.
MP25819.
Vigilante.
R585734 - R585737.
R587216 - R587224.
Vigilantes return.
R584815.
Violence.
R591328.
Virginia Polytechnic Institute and State University, Blacksburg.
MP25820.
Virginia Polytechnic Institute and State University, Blacksburg.
College of Agriculture and Life Sciences.
MP25820.
Virginia Polytechnic Institute and State University, Blacksburg. Film
Unit.
MP25820.
Virtual storage concepts.
MP25582 - MP25585.
MP25737.
MP25952.
Vision Associates, Inc.
MP25558.
Visiting Virginia.
R589600.
The Visitor.
LP43979.
Visual Education Centre.
MP25893.
MP25894.
Visual Education Centre, a division of Public Media, Inc. SEE Public
Media, Inc. Visual Education Centre.
Visual Instruction Productions, a department of Victor Kayfetz
Productions, Inc. SEE Kayfetz (Victor) Productions, Inc. Visual
Instruction Productions.
Vitaphone Corporation.
R580303.
R580304.
R580305.
R583124.
R583125.
R585007 - R585013.
R587018.
R587021.
R587022.
R590798 - R590803.
R592815 - R592817.
Vivian’s problem.
LP43841.
Voice of the turtle.
R590797.
Volley oop.
R582974.
Voltmeter multiplier actions.
MP25518.
Volunteer.
MP25735.
Vortex, Inc.
LU3676.
VSAM concepts and access method services usage (DOS/VS)
MP25779.
VSAM macro coding and debugging.
MP25778.
 W
Waber, Bernard.
LP43690.
Wade (Roger) Productions, Inc.
LP43890 - LP43903.
Walking home.
MU9153.
Walking safely to school.
MP26011.
Wall, Robert E.
MU9042.
Wallis (Hal) Productions, Inc.
R585723.
Walter Kerr on theater.
LP43883.
Walter’s holiday.
LP43835.
Waltz of the toreadors.
LF160.
W and W Films.
LP43794.
The Wankel engine—an analysis.
MP25688.
Wanted—more homes.
R584089.
Warner, Douglas.
LF207.
Warner Brothers.
LP43678.
 LP43811.
LP43889.
Warner Brothers, Inc.
LP43741.
LP43744.
LP43745.
LP43757.
LP43764.
LP43778.
LP43787 - LP43791.
LP43810.
LP43812.
LP43813.
LP43913.
LP43939.
MP25503.
R583818.
R591573.
Warner Brothers Pictures Corporation.
R580302.
Warner Brothers Pictures, Inc.
R580301.
R583122.
R583123.
R585005.
R585006.
R587019.
R590797.
R592812.
R592813.
Warner Brothers Productions, Ltd.
LP43913.
Warner Communications.
LP43764.
Warner Communications Company.
LP43745.