Appendix

Hyperhomology
This appendix offers a crash course in hyperhomological algebra. The aim is, first
of all, to provide an easy reference for readers who are used to plain modules and
do not feel entirely at home among complexes. Of course, this aim could also be
accomplished by referring consistently to one or two textbooks on the subject - -
if only such existed. While there are numerous introductions to the technical side
of hyperhomology, including the construction of derived categories, a thorough
and coherent introduction to homologicai dimensions (not to mention depth,
Krull dimension, and other invariants from commutative algebra) for complexes
has yet to be published. The appearance of an introductory text like that may
not be imminent, but this appendix is based on Foxby's notes [33], and their
eventual publication will render it obsolete.

In general we omit references and proofs for standard definitions and results:
they can be found almost everywhere in the literature. Specific references
are, however, given for the more special results on homological dimensions
and other invariants for complexes; the primary sources are [35, 36] (Foxby)
and [7] (Avramov and Foxby).
The author's favorite sources for background stuff are
* [49] (Matsumura) and [12] (Bruns and Herzog) for commutative algebra,
• [13] (Cartan & Eilenberg) and [60] (Weibel) for homological algebra, and
• [47] (MacLane) for categories.
But also [20] (Eisenbud) has definitions, usually accompanied by illustrative
examples and exercises, of many basic and advanced notions in commutative
algebra and homological algebra.

T h e blanket assumptions (see page 9) are still in force; in particular, R is

always assumed to be a non-trivial, c o m m u t a t i v e , a n d N o e t h e r i a n ring.
Needless to say, most results in this appendix hold in a far more general setting,
but to keep it simple we only state what we need.
160 APPENDIX. HYPERHOMOLOGY

A.1 Basic Definitions and Notation

The definitions given in this section are all standard, and the same can be said
about most of the notation.
Further details (in particular a b o u t the central notion of equivalence of com-
plexes) can be extracted from the first chapters in [41] and [43].

(A.I.1) Complexes. An R-complex X is a sequence of R - m o d u l e s X~ and

R-linear m a p s 0 x , ~ E Z,

Off+l) 0x
X = ... ) Xt+I X~ ~) X~-I )...

The module X~ is called the module in degree g, and the m a p O~¢ : Xl ~ X l - 1

is the ~-th differential; composition of two consecutive differentials always yields
the zero-map, i.e., cOxcq~l = 0 for all g E Z. The degree of an element x is
denoted by Ix[, i.e.,

Ixl= xeXt.
A complex X is said to be concentrated in degrees u , . . . , v if X ! = 0 for
> u and ~ < v. A complex X concentrated in degree zero is identified with the
module X0, and a module M is thought of as the complex

M = 0--+ M - - ~ 0 ,

with M in degree zero. Of course, the complex M has 0 in all degrees except
(possibly) degree zero, i.e., M . . . . --+ 0 --+ 0 --+ M -~ 0 --+ 0 ~ - - - , but we
never write superfluous zeros. In line with this the zero-complex is denoted by
0.

(A.1.2) Homology. For an R - c o m p l e x X and g E Z the following notation is

used:

zX = KerOX;
Bx = Im0~l; and
C x = Coker 0 ~ 1 .

Both B x and Z x are submodules of Xe, and B x C_ Z x because 0;x 0 x+1 = o.

The residue class module

H i ( X ) = Zx / B x

is called the homology module in degree ~, and the homology complex H ( X ) is

defined by setting

H(X)e = He(X) and O~ (x) = 0

for ~ E Z.
A.1. BASIC DEFINITIONS AND NOTATION 161

Note t h a t H ( H ( X ) ) = H ( X ) for any complex X , and H ( M ) = M when M is

a module.
A complex is said to be homologically trivial when H ( X ) = 0. Thus, a ho-
mologically trivial R - c o m p l e x is what is classically called an exact sequence of
R-modules.

(A.1.3) Shift. If m is an integer and X is a complex, then ~'~mx denotes the

complex X shifted m degrees (to the left); it is given by

(~mx)e = Xt_m and a~ mx = (-1)mOX_m

for t 6 Z. Note that

(A.1.3.1) H t ( ~ m X ) = He-re(X).

For example, if M is a module, then the complex ~ ' n M has M in degree m

and 0 elsewhere.

(A.1.4) M o r p h i s m 8 . A morphism a: X ~ Y of R-complexes is a family a =

(ae)tez of R-linear maps a~ : X t -+ lit making the diagram

) A~t+l ) Xt ) Xt-1 ) "'"

) Yt+l ) :1~ > Ye-1 >""

o~+1 o~

commutative. T h a t is, OYat = a t _ l O X for all g 6 Z.

For an element r E R and an R - c o m p l e x X the morphism r x : X ~ X is the
homothety given by multiplication by r. In line with this we denote the identity
morphism on X by l x .
Any morphism a : X -~ Y induces a morphism H ( a ) : H ( X ) -+ H(Y) in ho-
mology. The homology H is a functor in the category of all R-complexes and all
morphisms of R-complexes.

(A.1.5) I s o m o r p h i s m s . A morphism a : X -~ Y of R-complexes is said to be

an isomorphism when there exists a morphism a -1 : Y --~ X such t h a t a a -1 =
1y and a - l a = l x . Isomorphisms are indicated by the symbol ~ next to their
arrows, and two complexes X and Y are isomorphic, X ~- Y in symbols, if and
only if there exists an isomorphism X - > Y.
Note that a : X -~ Y is an isomorphism of R-complexes if and only if all the
maps ae : Xt ~ Yt are isomorphisms of R-modules. In particular, two modules
are isomorphic as complexes if and only if they are so as modules.
If a : X --+ Y is an isomorphism, then so is the induced morphism in homo-
logy, and H ( a ) -1 = H ( a - 1 ) .
162 APPENDIX. HYPERHOMOLOGY

(A.1.6) Quasi-isomorphisms. A morphism a : X ~ Y is said to be a quasi-

isomorphism if the induced morphism H(a) : H(X) ~ H(Y) is an isomorphism.
Quasi-isomorphisms are also called homology isomorphisms, and they are indi-
cated by the symbol _~ next to their arrows.
Note that all isomorphisms are quasi-isomorphisms. A morphism of com-
plexes concentrated in degree zero (that is, a homomorphism of modules) is a
quasi-isomorphism if and only if it is an isomorphism.

(A.1.7) Homologieal Position and Size. The numbers supremum, infimum,

and amplitude:

s u p X = sup{g E Z I He(X) • 0},

i n f X = inf{g E Z[ He(X) ~ 0}, and
amp X = sup X - inf X

capture the homological position and size of the complex X. By the conventions
for supremum and infimum of the empty set it follows that s u p X = - c o and
inf X = co if X is homologically trivial; otherwise we have

co > s u p X _> i n f X _> - c o .

Let X be any complex; it is immediate that

(A.1.7.1) Z~~ = B x for g < i n f X ,

and

(A.1.7.2) B x ~- X t + a / Z f + 1 = C~_ 1 for g _~ sup X.

In particular, if X is homologically trivial, then

(A.1.7.3) Z x = B f ~ Cf+ 1

for all g E Z.

(A.1.8) The Category of R-complexes. We use the notation C(R) for the
category of all R-complexes and all morphism of R-complexes. Recall that a
full subcategory S of C(R) is defined by specifying its objects, the arrows in
S are simply all morphisms between the specified objects. We shall consider a
number of full subcategories S of C(R) (the first ones are introduced below); of
course the notation X E S means that X is an object in S, and for two full
subcategories the notation $1 C $2 means that every object in S1 is also an
object in $2.

(A.1.9) Categories of ( Homologieally) Bounded Complexes. An R-com-

plex X is said to be bounded to the left if there is an integer u such that Xt = 0
for all g > u; similarly X is bounded to the right if there is an integer v such that
A.1. BASIC DEFINITIONS AND NOTATION 163

Xe = 0 for all e < v. A complex which is bounded to the right as well as to the
left is said to be bounded.
We define the full subcategories CE (R), C~ (R), Co (R), and C0(R) of C(R) by
specifying their objects as follows:
CE (R): complexes bounded to the left;
C~(R): complexes bounded to the right;
Co(R): bounded complexes; and
Co(R): modules (considered as complexes concentrated in degree 0).
An R-complex X is said to be homologically bounded (to the left/right) when
the homology complex H(X) is bounded (to the left/right). We also consider
the following full subcategories of C(R):
C(E) (R): complexes homologically bounded to the left;
C(~) (R): complexes homologically bounded to the right;
C(o) (R): homologically bounded complexes; and
C(0)(R): complexes with homology concentrated in degree zero.
Note that these last four subcategories can be characterized as follows:

X E C(F)(R) ~ supX < co;

X E C(-7)(R) ~ infX > -oc;
X E C(D)(R ) ~ a m p X < co; and
XEC(0)(R) ~ supX_<infX.

(A.1.10) A Menagerie of Categories. We also consider the following full

subcategories of C(R):
Cf(R): complexes of finite modules;
c(f) (R): complexes with finite homology modules;
CI (R): complexes of injective modules;
CF(R): complexes of flat modules;
CP (R): complexes of projective modules;
cfP(R): complexes of finite projective modules; and
CL(R): complexes of finite free modules.
Superscripts and subscripts are freely mixed to produce new notation. E.g.,
c((f~ (R): homologically bounded complexes with finite homology modules
(for short, complexes with finite homology);
c0f(R): finite modules;
CL(R): bounded to the right complexes of finite free modules; and
CoI(R): injective modules.
The principle behind this notation is that subscripts indicate boundedness con-
ditions on the complexes, while superscripts indicate conditions on the modules
164 APPENDIX. HYPERHOMOLOGY

of the complexes. We use (or rather allow) any combination

C (n) = Ca(R) n C (R)

of subscripts (c-, Z, D, or 0) and superscripts (f, I, F, P, fP, or L).
Sub- and superscripts in parentheses indicate the corresponding conditions
on the homology complexes. That is,

X ~ C(~)
(~) (R) H(X) E C~(R).

(A.I.ll) Equivalence. Equivalence of R-complexes can be defined as follows:

two complexes X and Y are equivalent, we write X ~- Y, if and only if there
exists a third complex Z and two quasi-isomorphisms: X ~-) Z ~ ~- Y. In
particular, the existence of a quasi-isomorphism X - ) Y implies that X and
Y are equivalent. Equivalence _~ is an equivalence relation in C(R).
The following implications always hold:

X ~
= Y ~ X ~ Y ~ H(X) ~= H(Y) ,

but the reverse implications do not hold in general. For R-modules M and N,
however, we have

M~-N ~ M'.,N;
that is, equivalent objects in Co(R) are "just" isomorphic modules.
Also note that a complex is homologically trivial if and only if it is equivalent
to the zero-complex. That is,

X'-'0 ~ H(X)=0.

The symbol .-. denotes equivalence up to a shift. That is,

X~Y .: ?, X_~NmY for s o m e m E Z .

(A.l.12) Notation for Equivalence Classes. Let X be an equivalence class

of R-complexes; all representatives of X have isomorphic homology modules, so
it makes sense to say, e.g., that X has finite homology: meaning, of course, that
every representative of X belongs to C((r))(R). Generally speaking, notation of
(&)
the form X E C(~) (R), cf. (A.I.10), will make sense but X E C~(R) will not!
Sensible notation should be interpreted like this:
(&)
X E C((~))(R) ~ Z EC(~)(R) for some representative X of X
(&)
X E C(~) (R) for every representative X of X.

We sometimes identify a complex with its equivalence class, and - - abusing

the notation slightly - - we may write, e.g., X ~ = X intending, of course, that
the complex X ~ represents the equivalence class X.
A.1. BASIC DEFINITIONS AND NOTATION 165

(A.l.13) R e m a r k s . The derived category of the category of R - m o d u l e s - - usu-

ally denoted by D(R) - - is the localization of C(R) at the class of all quasi-
isomorphisms. T h a t is, the quasi-isomorphisms are formally inverted and be-
come the isomorphisms in D(R); the notation X _~ Y then means t h a t X and
Y are isomorphic objects in D(R).
The classical references for derived categories are [59] (Verdier) and [41]
(Hartshorne/Grothendieck), but most readers will probably find chapter 10
in [60] (Weibel) more accessible.
Now, we don't use derived categories, but [41] is still well worth perusing in
order to develop a gut feeling for the preliminary notion of equivalent complex-
es - - and so is [43] (Iversen).

(A.1.14) ~runcations. Let X be an R - c o m p l e x and let u, v be integers. The

hard left-truncation, E u X , of X at u and the hard right-truncation, X v ~ , of X
at v are given by:

o~x oL1 o~_2

Eu x = 0 ) Xu ) Xu-1 > X~,-2 > ... and

Xv-n . . . . °~x+8>Xv+~ °~x+=>Xv+I °~x+5 X~ ---+ O.

The soft left-truncation, C u X , of X at u and the soft right-truncation, XvD,

of X at v are given by:

CuX = 0 ) Cf Xu-1 >Xu-2 > ... and

x ox ox
X v D = "'" 0~+8> Xv+2 ~+2>Xv+l ~+~> Z x > 0.

The differential 0 x is the induced m a p on residue classes.

It is easy to see t h a t

0 f o r g > u,
He(EuX)= Zx forg=u, and
Hi(X) for g < u.

For u _< inf X the natural inclusion of ~ z u X into E u X ,

0 > Zx > 0

1 1 1 > ...
0 > Xu > Xu-1

is, therefore, a quasi-isomorphism. In particular,

(A.1.14.1) E u X - - ~ , u Z uX for u _ ~ i n f X .
166 APPENDIX. HYPERHOMOLOGY

It is also evident that

fore>u, and
He(CuX) = (
t He(X) f o r e _< u;

so the canonical morphism X --~ c u X ,

) Xu+l ) Xu ) Xu_l ) ...

1 1 1
0 ) CXu ) Xu_ 1 ) "'"

is a quasi-isomorphism if and only if u >_ sup X. In particular,

(A.1.14.2) C u X _~ X for u > sup X.

It is equally easy to see that

(A.1.14.3) Xv-q ~- ~v cvX for v _> s u p X ; and

(A.1.14.4) X v D ~-- X for v < inf X.

(A.l.15) R e m a r k . It follows by (A.1.14.2) and (A.1.14.4) that any R-complex

X is equivalent to a complex X ~ with X~ = 0 for e > sup x and e < inf X. In
particular, it follows that

(h.l.15.1) X E C(0)(R) ¢==~ X _~ H0(X).

(A.l.16) S h o r t Exact Sequences. Consider three R-complexes X, Y, and Z,

and morphisms a : X --->Y and/3: Y ~ Z. We say that

(A.l.16.1) 0 -~ X --% Y ~--~Z -~ 0

is a short exact sequence of R-complexes if it is exact in each degree. T h a t is,

o x t 2-# Ye zt o

is a short exact sequence of R-modules for each e G z .

A short exact sequence like (A.1.16.1) induces a long exact sequence of ho-
mology modules

... He+,(/3)> He+I(Z) zx~+l> He(X) Ht(c~))He(Y) Ht(/3))He(Z) '~t> ...

Note that homological triviality of two complexes in a short exact sequence

implies homological triviality of the third.
A.1. BASIC DEFINITIONS AND NOTATION 167

(A.l.17) R e m a r k . For any X E C(R) and n E Z the diagram

0 ) Xn-1 ) Xn-2 ) "'"

l
"'" ) Xn+l ) Xn ) Xn-1 ) Xn-2 ) "'"

1 l 1
• .. ) Xn+l ~ X~ ~ 0

is commutative, so we have a short exact sequence of complexes:

0 ---+ E n - I X ~X ~XnZ ~ 0,

which is split (even trivial) in each degree.

(A.l.18) M a p p i n g C o n e s . To a morphism a : X --+ Y we associate a complex,

A4(a), called the mapping cone of a. It is given by

M(a)e = Yl @ X t - x and
+ x
= , -Ot_:(xt-:)).

(A.l.19) L e m m a . A morphism a: X ~ Y of R - c o m p l e x e s is a quasi-isomor-

phism if and only if the mapping cone M ( a ) is homologically trivia/.

Proof. It is easy to check that the inclusion Y ~ M ( a ) and the (degree-wise

projection) M ( a ) -~ Z I X are morphisms and make up a short exact sequence
of complexes

0 -~ Y -~ M ( a ) -* E I X ~ O.

In the induced long exact sequence of homology modules

• .. -+ H t + l ( M ( a ) ) -+ Ht+I(E1X) /xe+l) Ht(Y) --} H d M ( a ) ) - + . . .

we have Ht+l ( ~ I X ) = Ht(X), cf. (A.1.3.1), and the connecting map At+x is just
the induced map Ht(a): Ht(X) -+ He(Y). The assertion is now immediate. []

(A.1.20) I n d u c e d F u n c t o r s . Any additive module functor T: C0(R) --} C0(R')

induces a functor T: C(R) -~ C(R') on complexes.
Let X E C(R). If the functor T is covariant, then T(X) E C(R') is given by

(A.1.20.1) T(X)t = T(Xt) and O~ (x) = T(0~¢);

and if T is contravariant, then T(X) is given by

(A.1.20.2) T(X)t = T ( X - t ) and 0 T(x) = T(0_,+I)

. x
168 APPENDIX. HYPERHOMOLOGY

If T is exact, then H(T(X)) = T(H(X)) for every X E C(R), and T pre-

serves quasi-isomorphisms and equivalences. For example, the functor - ®R Rp
(localization at p) is exact for every prime ideal p E Spec R.
If the module functor T is faithful (i.e., T is "injective on homomorphisms'),
then, in particular, M = 0 if T(M) = 0, so T(M) i~ 0 ¢¢ M ~ 0 as T is
additive. Thus, if T is faithful, exact, and covariant, then the induced functor
on complexes preserves suprema and infima:

(A.1.20.3) sup (T(X)) = supX and inf (T(X)) = infX.

If T is faithful, exact, and contravariant, then the induced functor "swaps"

suprema and infima:

(A.1.20.4) sup (T(X)) -- - i n f X and inf (T(X)) = - supX.

A.2 Standard Functors and M o r p h i s m s

Ever since the highly influential book [13] by Cartan and Eilenberg appeared,
the functors Hom and tensor product - - and the associated standard homomor-
phisms - - have formed the core of almost any course in homological algebra.
In this section we review Hom, tensor product, and standard morphisms for
complexes.

(A.2.1) Homomorphisms. For R-complexes X and Y we define the homomor-

phism complex Homn(X, Y) e C(R) as follows:

n o m n ( X , Y ) t = H Homn(Xp, Yp+e)
pEZ

and when ¢ = (¢p)pez belongs to HomR(X,Y)I the family oS°mn(X'Y)(¢) in

Homn(X, Y)t-1 has p-th component

coH°mR(X'Y) (~/))p _~ (OYp+£~)p -- (--1)tCp_lO X.

If V and W are fixed R-complexes, then Homn(V,-) and H o m n ( - , W )

are functors in C(R). No ambiguity arises when one or both involved com-
plexes are modules. If M E Co(R) and X E C(R), then the homomorphism
complexes HomR(M,X) and Homn(X, M) agree with the complexes yielded
by applying, respectively, H o m n ( M , - ) and H o m u ( - , M) to X. In particular,
for M, N e C0(R) the homomorphism complex HomR(M, N) is concentrated in
degree zero, where it is the module Homn(M, N).
The covariant functor Homn(V, - ) commutes with shift and mapping cones:

(A.2.1.1) Homn(V, ~my) = EmHomn(V,y); and

(A.2.1.2) .£d(Homn(V, a)) = Homn(V,.h4(a)).
A.2. STANDARD FUNCTORS AND MORPHISMS 169

For the contravariant functor Homit(-, W) we have the following:

(A.2.1.3) Homit(Emx, w ) ~ E-mHomn(X, W); and

(A.2.1.4) .A4(Uomit(a, W)) ~ EiHomit(.h4(a), W).

When X E C-7(R) and Y E Cr-(R) all the products 1-IpezHomit(Xp, Yp+t) are
finite; the next two lemmas are, therefore, direct consequences of the similar
results for modules.

(A.2.2) L e m m a . If X E c f (n) and Y E c f (n), then H o m n ( X , Y ) E Cf (R).

(A.2.3) L e m m a . If X E Cf (R), Y E Co(R), and p is a prime ideal in R, then

there is an isomorphism of Rp-complexes:

Homn(X, Y), -~ Homnp (X,, Yp).

(A.2.4) Tensor Products. For R-complexes X and Y we define the tensor

product complex X ®R Y E C(R) as follows:

(X ®n Y)e = H Xp ®it Ye-p

pEZ

and the g-th differential O~

x®RY is given on a generator Xp ® yt_p in (X ®it Y)t
by

~.~p ® Ye-p) = ® Ye-p

which is an element in (X ®it Y)t-i.
The tensor product is commutative: for complexes X and Y the commuta-
tivity isomorphism

(A.2.4.1) ~-xy : X ®it Y ~- > Y ®it X,

with the map in degree e given on generators by

(A.2.4.2) TXy e(Xp ® Ye-p) = (--1)P(t-P)Yt-p ® Xp,

is natural in X and Y.
If V is a fixed R-complex, then V ®it - (and thereby - ®it V) is a functor
in C(R). No ambiguity arises when one or both involved complexes are mo-
dules. If M E Co(R) and X E C(R), then the tensor product complex M ®it X
agrees with the complex yielded by applying M ®it - to X. In particular, for
M, N E C0(R) the tensor product complex M ®it N is concentrated in degree
zero, where it is the module M ®R N.
The tensor product functor V ®it - commutes with shift and mapping cones:
(A.2.4.3) V ®it (EmY) -: E m ( v ®It Y); and
(A.2.4.4) M ( V ®it a) ~- Y ®it M(a).
170 APPENDIX. HYPERHOMOLOGY

The first lemma below is a direct consequence of the corresponding result for
modules; and so is the second, because all the sums ]-Ipez Xp ®R Yl-p are finite
when X and Y are bounded to the right.

(A.2.5) L e m m a . If X and Y are R-complexes, and p is a prime ideal in R,

then there is an isomorphism of Rp-complexes:
(X ®R Y)p ~- Xp ®R~ Yp.

(A.2.6) L e m m a . If X and Y belong to c f (n), then also X ®n Y • Of(R).

The five standard h o m o m o r p h i s m s of m o d u l e s (see page 11) induce five

n a t u r a l m o r p h i s m s of complexes; these axe described below in (A.2.7)-
(A.2.11).
In the rest of this section S denotes an R-algebra, and in most applications
S will be R itself.

(A.2.7) Associativity. Let Z, Y • C(S) and X • C(R). Then Z ®s Y • C(R)

and Y ®R X • C(S), and

gzvx: (Z®sY)®RX ~-> Z ® s ( Y ® n X )

is a natural isomorphism of S-complexes.

(A.2.8) Adjointness. Let Z , Y • C(S) and X • C(R). Then Z ®s Y • C(R)

and HomR(Y, X ) • C(S), and

P z g x : Homn(Z @s Y,X) -~ } Horns(Z, HomR(Y,X))

is a natural isomorphism of S-complexes.

(A.2.9) Swap. Let Z, Y • C(S) and X • C(R). Then HomR(X, Y) • C(S) and
Uoms(Z, Y) • C(R), and

g z x g : noms(Z, Hom~(X,r)) -~ ~ HomR(X, Homs(Z, Y))

is a natural isomorphism of S-complexes.

(A.2.10) Tensor Evaluation. Let Z , Y E C(S) and X belong to C(R). Then

Horns(Z, Y) E C(R) and Y ®n X E C(S), and
w z y x : Homs(Z, Y) ®R X --~ Horns(Z, Y ®n X)
is a natural morphism of S-complexes. The morphism is invertible under each
of the next two extra conditions:
• Z E C~(S), Y E C~(S), and X E C~(R); or
• Z e CWn(S), Y • CE(S ), and X • CD(R).
A.3. RESOLUTIONS 171

(A.2.11) H o m E v a l u a t i o n . Let Z, Y E C(S) and X belong to C(R). Then

H o m s ( Z , Y ) E C(R) and HomR(Y,X) E C(S), and

8zyx: Z®sHomR(Y,X) ~ HomR(Homs(Z,Y),X)

is a natural morphism of S-complexes. The morphism is invertible under each

of the next two extra conditions:
• Z E Crop(S), Y E C~(S), and X E Cr(R); or
• Z E C~m(S), Y E Cz(S ), and X E Co(R).

(A.2.12) Proof of (A.2.7)-(A.2.11). Basically, the morphisms are defined by ap-

plying the corresponding homomorphisms of modules in each degree. However,
by the "universal sign rule" a sign (-1) pq is introduced whenever two elements
of degrees, respectively, p and q are interchanged. In short (but suggestive)
notation we can write the definitions as follows:

(azyx) (z®y)®x , ~ z®(y®x);

(pzvx) ¢ , [z [y ¢(z ® y)]];
(qzxr) ¢ , ~ [x ~ [z ~ (-1)l~tlzl¢(z)(x)]];
(wzyx) ¢®x , ~ [z ~-~ (-1)lxllzl¢(z) ®x]; and
(tgzyx) z ®¢ , ~ [• ~ (-1)lzl(l¢l+l~l)¢zg(z)].

Of course, it must be verified that the modules have the right form, so that these
definitions make sense; they do. It must also be checked that the degree-wise
maps commute with the differentials; they do. Finally, it must be verified that
the extra conditions listed in (A.2.10) and (A.2.11) ensure that the degree-wise
maps axe invertible. This boils down to the direct sums and products (making
up the modules in the Hom and tensor product complexes) being finite in each
degree; and they are. Details are given in [33, Chapter 5].

A.3 Resolutions
To do hyperhomological algebra we must first establish the existence of resolu-
tions. It should be emphasized right away that we are aiming for a relative
homologieal algebra: we will not resolve objects in C(R) by projective, in-
jective, or flat objects in that category, but rather by complexes of projective,
injective, or fiat modules.

(A.3.1) Definitions. We define resolutions for (appropriately bounded) com-

plexes as follows:
(I) An injective resolution of a complex Y E CCr-)(R) is a quasi-isomorphism
t: Y ~-~ I where I E C~(R).
172 APPENDIX. HYPERHOMOLOGY

(F) A fiat resolution of a complex X E C(z)(R) is a quasi-isomorphism

~: F -~) X where F 6 c F ( R ) .
(P) A projective resolution of a complex X 6 C(z) (R) is a quasi-isomorphism
7r:P -~XwhereP6CP(R).
(L) A resolution by finite free modules of a complex Z 6 C((~f~(R) is a quasi-
isomorphism A: L -~) Z where L 6 CL(R).

In (A.3.11) we shall see how these definitions relate to the usual concepts of
injective, flat, and projective resolutions of modules.

(A.3.2) T h e o r e m ( E x i s t e n c e o f R e s o l u t i o n s ) . The following hold:

(I) Every complex Y E C(z)(R ) has an injective resolution Y ~-) I with
It = 0 for g > sup Y.
(P&F) Every complex X 6 C(~) (R) has a projective, and thereby a fiat, reso-
lution X ( ~- P with Pt = 0 for g < inf X.
(L) Every complex Z E c((f](R) has a resolution by finite free modules
Z ( ~- L with Lt = O for g < inf X .

Proof. Various versions can be found in [13,41,43,56]; see [7, 1.7] for further
guidance.

(A.3.3) L e m m a . If 0 -~ Y ' -~ Y -+ Y " -+ 0 is a short exact sequen-

ce in C(z)(R), then there exists a short exact sequence 0 -+ I' -~ I -~ I " ~ 0
in C~(R), where I', I, and I " are injective resolutions of, respectively, Y ' , Y ,
and Y " .

Proof. [43, Proposition 6.10].

(A.3.4) Lemma. If 0 --+ X ' -+ X -+ X" --+ 0 is a short exact sequence in

CL-~)(R), then there exists a short exact sequence 0 -4 P' -+ P -~ P" -+ 0 in
C~r(R), where P', P, and P" are projective resolutions of, respectively, X', X,
and X".

Proof. See Proposition 6.10 ° (,o, is for 'opposite') on page 67 in [43].

Complexes of injective and projective modules have convenient lifting properties

described by the next two lemmas.

(A.3.5) L e m m a . If Y and I are equivalent complexes, and I belongs to C~(R),

then there exists a quasi-isomorphism Y ~- ~ I; that is, I is an injective resolu-
tion of Y .

Proof. See [7, 1.1.I and 1.4I].

A.3. RESOLUTIONS 173

(A.3.6) L e m m a . /f X and P are equivalent complexes, and P belongs to CP (R),

then there exists a quasi-isomorphism P ~- ~ X; that is, P is a projective reso-
lution of X .
Proof. See [7, 1.1.P and 1.4.P].

In (A.3.9)-(A.3.10) we define the s t a n d a r d homological d i m e n s i o n s for com-

plexes, and it is n o t done b y w a y of r e s o l u t i o n s but rather by way of equiv-
alence. Now, by the last two lemmas this is of no importance for the projective
and injective dimensions, but it still takes a short argument to see that the
new definitions agree with the usual ones for modules; this argument is given in
(A.3.11).

(A.3.7) Definitions. The full subcategories 7)(R), 2:(R), and Y(R) of C(R) are
defined as follows:
YEZ(R) ~ 3IECI(R):Y~-I;
X E 7~(R) ~ 3PECP(R):X~_P; and
X E ~'(n) ~ BFECF(R):X~-F.
We also use the notation P(R), Z(R), and ~-(R) with superscripts and subscripts
following the general rules from (A.1.10). See also (A.3.12).

(A.3.8) Definition. For Y E C(E)(R ) the injective dimension, idn Y, is defined

as

i d n Y = inf {sup {t E Z[I_~ ¢ O}[Y ~ I E C~(R)}.

Note that

idnY e {-co} U ZU {co};

idnY >_ - i n f Y ;
idnY=-co ,', ;, Y ~_ O; and
idn Y < co -', ',, Y E 2:(R).

(A.3.9) Definition. For X E C(-7)(R) the projective dimension, p d n X , is de-

fined as
p d n X = inf {sup {~ E Z [ P / ~ O} I X _~ P E CP(R)}.
Note that

p d n X e {-co} U Z U {co};
pd n X > sup X;
pd R X = - c o ,,, ,,, X ~_ O; and
pd R X < c o ~ XET~(R).
174 APPENDIX. HYPERHOMOLOGY

(A.3.10) Definition. For X E C(~)(R) the fiat dimension, fdn X, is defined as

fdR X = inf {sup {g E Z [ F~ ~ 0} [ X _~ F E CF(R)}.

Note that

x e u s u
p d R X >_fdRX _> supX;
fdRX=-c~ ¢=~ X_0; and
fdnX<c~ ~ XEgV(R).

The last result of this section shows that the definitions above extend the usual
notions of projective, injective, and flat dimension of modules.

(A.3.11) P r o p o s i t i o n . The following hold:

(I) I f the R - m o d u l e N is equivalent to I E C~(R), then the truncated com-
plex

C o / = 0 ~ C~ - + 1 _ 1 ~ I-2 ~ ... -+ I~ ~ ...

is a (usua/) injective resolution of N .

(P) If the R - m o d u l e M is equivalent to P E CP(R), then the truncated com-
plex

PoD .... +Pt+ ""-+P2 - ~ P 1 -~ z P -+ 0

is a (usual) projective resolution of M .

(F) If the R - m o d u l e M is equivalent to F E CF(R), then the truncated com-
plex

FoD . . . . ~ Fe ~ " " - ~ F2 -+ F1 -"> Z F -"~ 0

is a (usual) fiat resolution of M .

Proof. If N is equivalent to I E C~(R), then s u p / = s u p N _< 0 so, by

(A.1.14.2), N is also equivalent to Co/. That is, N is isomorphic to the ker-
nel Ker(CoI -+ I-1), and we just have to prove that CoI is injective. But this is
immediate: set u = sup {g E Z [ It ¢ 0), and consider the exact sequence

O---~I~-+Iu-1 --+ • • ' --~ Io --+ C0I ---~0.

The modules Iu, I ~ - 1 , . . . , I0 are injective, and hence so is the cokernel C/. This
proves (I), and the proofs of (P) and (F) are similar. []

(A.3.12) R e m a r k . It follows by the Proposition that, e.g., ~-0(R) is (naturally

identified with) the full subcategory of modules of finite flat dimension, and
7~f0(R) is (naturally identified with) the full subcategory of finite modules of
finite projective dimension.
A . 4 . (ALMOST) DERIVED FUNCTORS 175

A.4 (Almost) Derived Functors

The notations R H o m n ( - , - ) and - ® ~ - are usually used for, respectively,
the right and left d e r i v e d f u n c t o r s of the homomorphism and tensor product
functors for R-complexes. For objects (complexes) X and Y in the derived cat-
egory RHomR(X, Y) and X ®L y are uniquely determined up to isomorphism
in 7:)(R). In this book we use this very same notation for certain equivalence
classes of R-complexes; they are defined in (A.4.2) and (A.4.11) below. Not
only is this permissible, in as much as we never use the true derived functors,
but it is also very convenient, because experienced users of the derived cate-
gory will certainly want to think of these gadgets as "real complexes". Likewise
these readers will interpret equalities of equivalence classes of R-complexes as
isomorphisms in the derived category D(R).
The first paragraph collects a series of results from [43, Part I] and [7, Sec-
tion 1].

(A.4.1) P r e s e r v a t i o n o f Q u a s i - I s o m o r p h i s m s a n d Equivalences. If P E
CP(R) and I E C~(R), then the functors H o m n ( P , - ) and H o m n ( - , I) preserve
quasi-isomorphisms and, thereby, equivalences:

X -% Y Homn(P, X) ~- ) Homn(P, Y);

c~ Homn(P,a)
X ~_ y ==, Homn(P, X) _~ Homn(P, Y);
x r Homn(]~;I) -~ ) Homn(X,I); and
(2 Homn(a,l)
X~_Y Homn(X, I) _~ Homn(Y, I).

For fixed complexes V and W the restrictions of the functors H o m n ( V , - )

and H o m n ( - , W ) to, respectively, C~(R) and C~(R) also preserve quasi-
isomorphisms and equivalences. That is, for I, I' E C~(R) there are implications:

I -~4 I' ~ Homn(V, I) _-z ) Homn(V, I'); and

t Homn(V,Q
I ~_ I' ==* Homn(V,I) ~- n o m n ( V , I ' ) .

Similarly for P, P' E CP(R):

P - - ~ P' ~ nomn(P',W) -~ ~ Homn(P,W); and

lr Homn(lr,W)
P - ~ P' ~ n o m n ( P , W ) _~ HomR(P',W).

If F E C~(R), then the functor F ®n - (and the isomorphic - ®n F) pre-

serves quasi-isomorphisms and, therefore, equivalences:

X--% Y ~ F®nX ~- > F ® n Y ; and

a F@na

X~-Y ~ F®nX~-F®n~:
176 APPENDIX. HYPERHOMOLOGY

For a fixed complex V the restriction of the functor V ®R - (and the isomorphic
- ®R V) to C~ F (R) also preserves quasi-isomorphisms and equivalences. That is,
if F, F I E CF(R), then

F - ~ F' ~ V ®R F ~- > V ®R FI; and

F~F ~ ~ V®RF~-V®RF ~-

(A.4.2) Definition. Let X and Y be R-complexes and assume that X E

C(n)(R) and/or Y e C(E)(R ). By RHomR(X,Y) we denote the equivalence
class of R-complexes represented by HomR(P, Y) and/or HomR(X, I), where
X _~ P E CP(R) and/or Y __ I e CI(R). The equivalence preserving properties
described above ensure that RHomR(X, Y) is well-defined and only depends on
the equivalence classes of X and Y.

(A.4.3) E x t Modules. For R-modules M, N and m E Z the isomorphism class

H_m(RHomR(M, N)) is known as the m-th Ext module, i.e.,
E x t , ( M , N) = H - m ( R H o m R ( M , N)).

(A.4.4) L e m m a . If X E c((f~(R) and Y E c((f~(R), then also RHomR(X,Y)

belongs to C~ (R ).

Proof. An easy consequences of the definition and (A.2.2), see also [8, (1.2.2)].

(A.4.5) L e m m a . If X E c((f~(R), Y E C(c)(R), and p is a prime ideal in R,

then there is an equality of equivalence classes of Rp-complexes:
RHomR(X, Y)p = RHomRp (Xp, Yp).
Proof. The equality is an easy consequences of the definition and (A.2.3), see
also [7, Lemma 5.2(b)].

It makes sense to talk about the supremum of an equivalence class, cf. (A.1.12),
and the next result is very useful.

(A.4.6) P r o p o s i t i o n . If X E C(-7)(R) and Y E C(E)(R ), then R H o m R ( X , Y )

belongs to C(E)(R ) and there is an inequality:
(A.4.6.1) sup (RHomR(X, Y)) _< sup Y - inf X.
Earthermore, assume that both X and Y are homologically non-trivial, and set
s = supY and i = infX. Then H s _ i ( R H o m n ( X , Y ) ) is represented by the
module Homn (Hi (X), H8 (Y)), so

(A.4.6.2) sup (RHomn(X, Y)) -- sup Y - inf X

HomR(Hi(X), Hs(Y)) ~ 0.
A.4. (ALMOST) DERIVED FUNCTORS 177

Proof. [35, Lemma 2.1(1)].

(A.4.7) L e m m a . Let 0 -4 K -4 H -4 C -4 0 be a short exact sequence of R - m o -

dules. I f X E C(~) (R), then there is a long exact sequence of homology modules:

• .. -4 Ht(RHomR(X, K)) -4 Ht(RHomR(X, H)) -4

Ht(RHomR(X, C)) -4 H t _ I ( R H o m R ( X , K ) ) -4 ...

Proof. Choose a projective resolution X ~ ~- P E CP(R), then

0 -4 HomR(P, K) -4 HomR(P, H) -4 HomR(P, C) -4 0

is a short exact sequence of complexes, and the associated long exact sequence

• .. -4 Ht(HomR(P, K)) -4 Ht(HomR(P, H)) -4

Ht(HomR(P, C)) -4 Ht-1 (HomR(P, K)) - 4 . . .

is the desired one. []

(A.4.8) L e m m a . Let 0 -4 K -4 H -4 C -4 0 be a short exact sequence of R - m o -

dules. If Y E C(E ) (R), then there is a long exact sequence of homology modules:

• . . -4 Ht(RHomR(C, Y)) -4 Ht(RHomR(H, Y)) -4

Ht(RHomR(K, Y)) -4 Ht-1 (RHomR(C, Y)) -4 . . .

Proof. Similar to the proof of (A.4.7), only this time choose an injective resolu-
tion of Y. []

(A.4.9) R e m a r k s . If X = M and Y = N are modules, then the long exact

sequences in (A.4.7) and (A.4.8) are just the usual long exact sequences of
Ext modules:

• .. -4 E x t , ( M , K) -4 E x t , ( M , H) -4 E x t , ( M , C) -4 Ext~+l (M, K) - 4 - . .

and

• -- -4 E x t , ( C , N) -4 E x t , ( H , N) -4 E x t , ( K , N) -4 Ext~+1 (C, N) - 4 . . . .

(A.4.10) F a i t h f u l l y I n j e c t i v e M o d u l e s . If E is an injective R-module, then,

for any R-complex X, the equivalence class RHomR(X, E) is represented by
HomR(X, E). If E is faithfully injective (i.e., the functor HomR(-, E) is faithful
and exact), then we have the following special case of (A.1.20.4):

sup (RHomR(X, E)) = sup (HomR(X, E)) = - inf X

and
inf (RHomR(X, E)) -- inf (HomR(X, E)) = - sup X.
178 APPENDIX. HYPERHOMOLOGY

If (R, m, k) is local, then ER(k), the injective hull of the residue field, is a
faithfully injective R-module, and the (module) functor HomR(-,ER(k)) (as
well as the induced functor on complexes) is called the Matlis duality functor.
Every ring R admits a faithfully injective module E, e.g.,

E= H ER(R/m).
mEMax R

(A.4.11) Definition. Let X and Y be R-complexes and assume that X E

C(~)(R) and/or Y e C(~)(R). By X ®L y we denote the equivalence class of
R-complexes represented by F ®R Y and/or X ®R F', where X _~ F E CF(R)
and/or Y ~_ F' E CF(R). The equivalence preserving properties described in
(A.4.1) ensure that X ®L y is well-defined and only depends on the equivalence
classes of X and Y.

(A.4.12) Tor Modules. For R-modules M, N and m E Z the isomorphism class

Hm(M ®L N) is known as the m-th Tot module, i.e.,
TorRm(M, N) = Hm(M ®L N).

(A.4.13) L e m m a . If X , Y E Ct(f~(R), then also X ®L y E C~f~(n).

Proof. An easy consequences of the definition and (A.2.6), see also [8, (1.2.1)].

(A.4.14) L e m m a . Let X and Y be R-complexes and p be a prime ideal in R.

If one of the complexes is homologically bounded to the right, then there is an
equality of equivalence classes of Rv--complexes:
(x Y)p = xp Yp.
Proof. The equality is an easy consequences of the definition and (A.2.5), see
also [7, Lemma 5.2(a)].

(A.4.15) Proposition. If X , Y E C(~)(R), then also X ®~ Y E C(~)(R) and

there is an inequality:
(A.4.15.1) inf (X ®~ Y) > i n f X + infY.
Furthermore, assume that X and Y are both homologically non-trivial, and set
i = inf X and j = inf Y. Then Hi+j (X ®~ Y) is represented by the module
Hi(X) ®n Hi(Y), so
(A.4.15.2) inf(X®~Y)=infX+infY -: ~ Hi(X)®nHj(Y)~O.
Proof. [35, Lemma 2.1(2)].

The next Corollary is an immediate consequence of Nakayama's lemma and

(A.4.15.2); it is sometimes called 'Nakayama's lemma for complexes'.
A.4. (ALMOST) DERIVED FUNCTORS 179

(A.4.16) Corollary. Let (R, m, k) be local. If X and Y belong to c~f~(R), then

inf (X ®~ Y) = inf X + inf Y.

(A.4.17) L e m m a . Let 0 ~ K ~ H ~ C -~ 0 be a short exact sequence of

R-modules. If X • C(-n)(R), then there is a long exact sequence of homolo-
gy modules:

• "" --~ Ht+l(C ®L X) -+ Ht(K ®~ X) --+ Ht(H ®L X) --+ Ht(C ®L X) --+...

Proof. Similar to the proof of (A.4.7). []

(A.4.18) R e m a r k . If X = M is a module, then the long exact sequence in

(A.4.17) is just the usual long exact sequence of Tot modules:

• ..-~ Tor~+l(C, M ) + Tor~(K, M) + Tor~(H,M) + Tor~(C, M) + . . .

The s t a n d a r d i s o m o r p h i s m s of c o m p l e x e s (A.2.4.1) and (A.2.7)-(A.2.11)

i n d u c e six identities of equivalence classes; these are described below in
(A.4.19)-(A.4.24). As usual S is an R-algebra; and for an equivalence class X
of R-complexes we write X • C(S) if X has a r e p r e s e n t a t i v e in C(S).

(A.4.19) Commutativity. Assume that X • C(-~)(R) and Y • C(R). Then

there is an identity of equivalence classes of R-complexes:

X®~Y=Y®~X.

(A.4.20) Associativity. Assume that Z • C(2)(S), Y • C(S), and X belongs

to C(-~)(R). Then Z @L y • C(R) and Y ®L X • C(S), and there is an identity
of equivalence classes of S-complexes:

(Z ®~ Y) ®L X = z ®L ( y @~ X).

(A.4.21) A d j o i n t n e s s . Assume that Z • C(2)(S), Y • C(S), and X • C(E)(R ).

Then Z @~ Y • C(R) and RHomR(Y, X ) • C(S), and there is an identity of
equivalence classes of S-complexes:

RHomR(Z ®L y , X) = RHoms(Z, RHomR(Y, X)).

(A.4.22) Swap. Assume that Z E C(~)(S), Y e C(S), a n d X belongs toC(2)(R ).

Then RHomR(X, Y) e C(S) and RHoms(Z, Y) • C(R), and there is an identity
of equivalence classes of S-complexes:

RHoms(Z, RHomR(X, Y)) = RHomR(X, RHoms(Z, Y)).

180 APPENDIX. HYPERHOMOLOGY

(A.4.23) Tensor Evaluation. Assume that Z E c((f~(s), Y E C(D)(S), and

X e C(-~)(R). Then R H o m s ( Z , Y ) E C(R) and Y ®~ X E C(S), and there is an
identity of equivalence classes of S-complexes:

RHoms(Z, Y) ®~ X = RHoms(Z, r ®~ x ) ,

provided that Z E P(f)(s) or X E Jr(R).

(A.4.24) H o m Evaluation. Assume that Z E C(f~(S), Y E d(D)(S), and X

belongs to C(F)(R). Then R H o m s ( Z , Y ) E C(R) and RHomn(Y,X) E d(S),
and there is an identity of equivalence classes of S-complexes:

Z ®~ RHomR(Y, X) = RHomR(RHoms(Z, Y), X),

provided that Z E P(f)(s) or X E Z(R).

Proof of (A.4.19)-(A.4.24). The equalities are straightforward consequences of

the standard morphisms, but the reader may want to check with [7, Lemma 4.4].

A.5 Homological Dimensions

The standard homological dimensions were defined in section A.3. In this section
we collect a number of results that allow us to compute the dimensions in terms
of (almost) derived functors.

(A.5.1) ID T h e o r e m . Let Y E C(E)(R ) and n E Z. The following are equiva-

lent:
(i) Y is equivalent to a complex I E C~(R) concentrated in degrees at
/east - n ; and I can be chosen with It = 0 for f > sup Y.
(ii) idR Y < n.
(iii) n >_ - s u p U - inf (RHomR(U, Y)) for all U ~ 0 in C(D)(R ).
(iv) n >_ - i n f Y and H_(n+I)(RHomR(T, Y) ) = 0 for all cyclic modules T.
(v) n _> - i n f Y and the module Z I_n is injective whenever Y _" I E d I (R).

Proof. See Theorem 2.4.I and Corollary 2.7.I in [7].

A.5. HOMOLOGICALDIMENSIONS 181

(A.5.2) ID Corollary. For Y • C(r-)(R) there are equalities:

idn Y = sup { - s u p U - inf (RHomn(U, Y ) ) [ U • C(n)(R) A U ~ 0}

(A.5.2.1)
= sup { - i n f (RHomR(T, Y ) ) [ T e Co(R) cyclic};

and the following are equivalent:

(i) Y e 2:(n).
(ii) RHomn(U,Y) • C(D)(R) for all V • C(o)(R ).
(iii) RHomn(T,Y) • C(n)(R) for all T • Co(R).

Proof. See section 2.I in [7].

(A.5.3) P D T h e o r e m . Let X E C(-~)(R) and n E Z. The following are equiva-

lent:
(i) X is equivalent to a complex P E CP(R) concentrated in degrees at
most n; and P can be chosen with Pt = 0 for ~ < inf X.
(ii) pd R X < _ n .
(iii) n >_infU - inf (RHomR(X, U)) for all U 7~ 0 in C(D)(R ).
(iv) n >_supX and H_(n+I)(RHomn(X,T)) = 0 for all R-modules T.
(v) n > supX and the module C P is projective whenever X ~_ P e CP(R).

Proof. See Theorem 2.4.P and Corollary 2.7.P in [7].

(A.5.4) P D Corollary. For X E C(n)(R) there are equalities:

p d n X = sup {inf U - inf (RHoma(X,U)) [U E C(o)(R) A U ~ 0}

(A.5.4.1)
= sup { - inf (RHomn(X, T)) [ T e C0(R)};

and the following are equivalent:

(i) X e P(R).
(ii) RHomR(X, U) e C(n)(R) for all U • C(o)(R ).
(iii) Rnomn(X,T) e C(o)(R) for all T e Co(R).
Furthermore, the following hold if X E c((f~ (R):

(A.5.4.2) X E ~(f)(R) ~ 3 P e 6~(R) : X ~ P;

and

(A.5.4.3) pd n X = sup { - inf (RHomR(X, T)) I T e c0f(R)}.

Proof. See section 2.P and Proposition 5.3.P in [7].

182 APPENDIX. HYPERHOMOLOGY

(A.5.5) F D T h e o r e m . Let X • C(~)(R) and n • Z. The following are equiva-

lent
(i) X is equivalent to a complex F • CF(R) concentrated in degrees at
most n; and F can be chosen with Ft = 0 for g < inf X.
(ii) fdR X _< n.
(iii) sup(U ®~ X) - supU <_n for all U ~ 0 in C(n)(R).
(iv) n >_sup X and Hn+I (T ®~ X) = 0 for all cyclic modules T.
(v) n _> supX and the module C F is fiat whenever X ~_ F • CF (R).
Proof. See Theorem 2.4.F and Corollary 2.7.F in [7].

(A.5.6) F D Corollary. For X E C(~) (R) there axe equalities:

fdR X = sup {sup (U @~ X) - sup U [ U e C(m)(n) A U ~ 0}

(A.5.6.1)
= sup {sup (T ®~ X) I T • Co(R) cyclic};
and the following axe equivalent:
(0 X • 7(R).
(ii) U ®~ X • C(D)(R) for all U • C(D)(R ).
(iii) T ®~ X • C(n)(R) for all T • Co(R).
Proof. See section 2.F in [7].

(A.5.7) Theorem (Homological Dimensions over Local Rings).

Let (R, m, k) be local. There is an equality of full subcategories:

(A.5.7.1) ~(f)(R) -- .T'(f)(R);

the following hold for X e and r e

(A.5.7.2) pd n X = fdn X = sup (X ®~ k);

(A.5.7.3) pd R X = - inf (RHomR(X, k)); and
(A.5.7.4) idR Y = - inf (RHomR(k, Y)).

Proof. See Corollary 2.10.F and Proposition 5.5 in [7].

(A.5.8) T h e o r e m (Stability). If X, X ' E 3:(R) and Y, Y' e Z(R), then

(A.5.S.1) fdR(X ®~ X') _<fdR X + fdR X';

(A.5.8.2) idR(RHomR(X, Y)) <_ fdR X + idR Y;
(A.5.8.3) idn(Y ®~ X) < idR Y - infX; and
(A.5.8.4) fdR(RHomR(Y, Y')) _< idR Y + sup Y'.

Proof. See Theorems 4.1 and 4.5 in [7].

A . 6 . DEPTH AND WIDTH 183

(A.5.9) L e m m a . If X • C(D)(R ) is equivalent to P • CP(R), n >_ supX, and

N • Co(R), then
1 P
ExtR(Cn, N) = H-(n+I) (RHomR(X, N)).

Proof. Since n _> s u p X = s u p P we have Pn-7 -~ E n Cn,

P cf. (A.1.14.3), and
RHomR(C~,N) is, therefore, represented by H o m R ( E - n ( P n ~ ) , N ) . The iso-
morphism class ExtR(Cn,
1 P N) is then represented by

H-1 (HomR(~ -n (Pn'7), N)) = H-1 (F~nHomR(Pn~, N))

= H-(n+i) (HomR(Pn~, N))
= H-(n+l)( c - n H o m R ( P , N))
= H-(n+l) (HomR(P, N));

cf. (A.2.1.3), (A.1.3.1), and (A.1.20.2). Since the complex HomR(P,N) rep-
resents RHomR(X,N), we have ExtR(Cn, 1 P N) = H-(n+I)(RHomR(X,N)) as
wanted. []

A.6 D e p t h and W i d t h
The invariants depth and width for modules (the latter is sometimes called
codepth or Tot-depth) have been extended to complexes by Foxby [36] and
Yassemi [63]. Depth for complexes has also been studied by Iyengar in [44].

(A.6.1) Depth. If (R,m,k) is local and Y E C(r-)(R), then the depth of Y is

defined as:

depth R Y = - sup (RHomR(k, Y)).

For finite modules this definition agrees with the classical one (the maximal
length of a regular sequence).
For every prime ideal p in R (not necessarily a local ring) and every complex
Y E C(r)(R) there are inequalities:

(A.6.1.1) depthRp Yp > - sup Yp > - sup Y.

Furthermore, if Y is homologically non-trivial and s = sup Y, then

(A.6.1.2) p e AssR(H,(Y)) ~ depthRp Yv = - s u p Y.

(A.6.2) L e m m a . If R is local and Y E C(ff~(R), then

depth R Y <_ depthRp Yp + dim R/p

for every p E Spec R.

184 APPENDIX. HYPERHOMOLOGY

Proof. For finite modules the inequality is a consequence of [11, Lemma (3.1)];
a proof for complexes is given in [33, Chapter 13].

(A.6.3) Width. If (R,m,k) is local and X E C(~)(R), then the width of X is

defined as:

widthR X = inf (X ®~ k).

For modules this is in agreement with the usual definition.

For every prime ideal p in R (not necessarily a local ring) and every complex
X E C(~)(R) there are inequalities:

(A.6.3.1) widthRp Xp _> infXp _> infX.

If R is local and X E c((f~(R), then

(A.6.3.2) widthR X = inf X.

(A.6.4) L e m m a . Let R be local. If X E C(~)(R) and Y E C(E)(R), then

depthR(RHomR(X , Y)) = widthR X + depth R Y.

Proof. [63, Theorem 2.4(a)].

(A.6.5) L e m m a . Let R be local. If X , Y E C(~)(R), then

widthR(X ®~ Y) = widthR X + widthR Y.

Proof. [63, Theorem 2.4(b)].

(A.6.6) T h e o r e m . If R is local and X E C(D)(R), then

widthR X < co ¢ = : > depth R X < co.

Proof. Follows by [36, Proposition 2.8].

(A.6.7) T h e o r e m . If R is local, U E ~(R), and X E C(o)(R), then

depthR(U ®~ X) = depth R U + depth R X - depth R.

Proof. See [37, Lemma 2.1] or [44, Theorem 4.1].

A . 7 . NUMERICAL AND FORMAL INVARIANTS 185

A.7 N u m e r i c a l and Formal Invariants

In this section we review a practical technique for manipulating certain invariants
for complexes over local rings.

(A.7.1) B e t t i N u m b e r s . Let X be any R-complex. If (R, m, k) is local, then

/~mR(X) = rankk(Hm(X ®~ k))

is the m-th Betti number of X . Note that i f X • C ~ ( R ) , then f ~ ( X ) • No for

all m • Z, cf. (A.4.13).
For M • C0(R) the definition reads: f~mR(M) = rankk(Tor~(M, k)); and if
M • Cfo(R), then/~0R(M) = rankk M / m M is the minimal number of generators
for M.
In general, for p • SpecR the m-th Betti number of X at p i s / ~ ( p , X ) =
/~m
Rp (Zp). Note that if (R, m, k) is local, then/~mR(m, X) =/~mR(X).

(A.7.2) P r o p o s i t i o n . For X • C(~) (R) there is an equality:

fdRX ----sup{m • Z I3 p • SpecR :/~mR(p,X) # 0}.

Proof. [7, Proposition 5.3.F].

(A.7.3) B a s s Numbers. Let Y be any R-complex. If (R, m, k) is local, then

#~ (Y) = rankk (H-m (RHomR (k, Y)))

is the m-th Bass number of Y. Note that if Y e c((f~(R), then #~(Y) E No for
all m • Z, cf. (A.4.4).
For N • C0(R) the definition reads #~(N) = rankk(Ext~(k, N)). For brevity
we set # ~ = ~u~(R).
In general, for p • SpecR the m-th Bass number of Y at p is #~(p,Y) =
#Rm (Yp). Note that if (R, m, k) is local, then #~(m, Y) = #~(Y).

(A.7.4) P o i n c a r d series. Let (R, m, k) be local. The Poincard series, P)~(t),

of a complex X e C~f~(R) is defined as:

P~(t) = ~ f~(X)tm;
mEZ

it is a formal Laurant series with non-negative integer coefficients.

It follows by (A.5.7.2) and (A.4.16) that

(A.7.4.1) degPRx(t) = p d n X and ordPRx(t) = i n f X .

186 APPENDIX. HYPERHOMOLOGY

(A.7.5) B a s s aeries. Let (R,m,k) be local. The Bass series, IX(t), of a com-
plex Y • C((Ef~(R) is defined as:

IX(t) = E "R"my'~tm~' J ,
m6Z
it is a formal Laurant series with non-negative integer coefficients.
It follows by (A.5.7.4) and the definition of depth that
(A.7.5.1) degIV(t) = idRY and ordIRV(t) = depthRY.

(A.7.6) T h e o r e m . If R is local and X , Y • c((f~iR), then there is an equality

of formal Laurant series:
pR®~y (t) = P~ (t) PRv(t).

Proof. See [35, Theorem 4.2(a)] or [8, Lemma (1.5.3)(a)].

(A.7.7) T h e o r e m . If R is local, Z 6 C~f~(R), and Y • C((f~(R), then there is

an equality of formal Laurant series:
I~a~°mR(x'Y) (t) = P~ (t) I X it).
Proof. See [35, Theorem 4.1(a)] or [8, Lemma (1.5.3)(b)].

(A.7.8) T h e o r e m . If R is local, X • P(f)(R), and Y • c~f~(R), then there is

an equality of formal Laurant series:
R
PRHoma(X,Y) (t) =
p)R(t-1) p~(t).
In particular,
inf (RHomR(X, Y)) = inf Y - pd R X.
Proof. [15, Corollary (2.14)]. (The result is due to Foxby.)

The proofs of (A.7.6)-iA.7.8) all use what Foxby calls 'accounting principles'.

(A.7.9) L e m m a ( A c c o u n t i n g Principles). Let R be a local ring with residue

field k. If X • C(D)(R ) and V • C(k), then
(A.7.9.1) sup (V ®~ X) = sup Y + sup (X ®~ k); and
(A.7.9.2) inf iV ®~ X) = inf V + inf (X ®~ k).
/f Y • C(E)(R) and V • C(k), then
(A.7.9.3) sup (RHOmR(V, Y)) = sup (RHoma(k, Y)) - inf V; and
(A.7.9.4) inf (RHomR(V, Y)) = inf (RHomR(k, Y)) - sup V.
Proof. See, e.g., the proofs of Theorems 4.1 and 4.2 in [35].
A.8. DUALIZING COMPLEXES 187

A.8 Dualizing Complexes

The definition of dualizing complexes goes back to [41]. In the literature these
complexes are usually taken to be bounded complexes of injective modules,
cf. part (1) in the definition below; for the proofs in this book, however, we
usually need the (infinite) projective resolution instead, and this accounts for
the formulation of part (2).

(A.8.1) Definition. Let R be a local ring. A complex D E C[~(R) is dualizing

for R if and only if
(1) it has finite injective dimension, i.e., D E z(f)(R); and
(2) if P E CP(R) is a projective resolution of D, then the homothety
morphism X~ : R -+ HomR(P, P) is a quasi-isomorphism.

(A.8.2) R e m a r k . To see that this definition of dualizing complexes makes sense,

take two projective resolutions P, P' E CP(R) of R. By (A.3.6) there is then
a quasi-isomorphism ~r: P -~ P', and using the quasi-isomorphism preserving
properties described in (A.4.1) we establish a commutative diagram

R xg, ) Homn(P', P')

HomR(P,P) H°mR(Pdr)) HomR(P,P')

which shows that Xp
R is a quasi-isomorphism if and only if XRp, is so.

(A.8.3) T h e o r e m (Existence a n d U n i q u e n e s s ) . Let R be a local ring. The

following hold:
(A.8.3.1) R is Gorenstein if and only if the R-module R is dualizing for R.
(A.8.3.2) If R is a homomorphic image of a Gorenstein ring, then R has a dua-
lizing complex.
(A.8.3.3) If both D and D' are dualizing complexes for R, then D ,,~ D'.
(A.8.3.4) If D is a dualizing complex for R and p E Spec R, then Dp is dualizing
for Rp.

Proof. See, respectively, Proposition 3.4, §10, Theorem 3.1, and §8 in [41, Chap-
ter V] for the original results; or refer to sections 2 and 6 in [15].

(A.8.4) S u p p o r t a n d D i m e n s i o n f o r Complexes. Let X be any R-complex;

the support of X is the set

SuppR X = {p E SpecR ] Xp ~ 0} = U SuppR(Ht(X))"

tEZ
188 APPENDIX. HYPERHOMOLOGY

For modules this agrees with the usual definition.

Note that

(A.8.4.1) X ~ 0 ¢==~ Supp R X ~ 0.

The (Krull) dimension of an R-complex X is defined as

dimR X = sup {dim R/p - inf Xp I P • Spec R}

-- sup { d i m R / p - infXp I P • SuppR X}-
Also this definition [36, Section 3] extends the usual concept for modules.

(A.8.5) T h e o r e m ( B i d u a l i t y ) . Let R be a local ring and assume that D is a

dualizing complex for R. For every complex Z • C~f)) (R ) the biduality morphism

~Dz : Z ~ RHomR(RHomR(Z,D),D)
is then a quasi-isomorphism, and the following equalities hold:
(A.8.5.1) inf (RHomR(Z, D)) -- depth R Z - depth R D; and
(A.8.5.2) sup (RHomR(Z, D)) -- dimR Z - depth R D-
In particular, we have
(A.8.5.3) amp D -- dim R - depth R = cmd R.

Proof. See [41, Proposition V.2.1] and [36, Proposition 3.14], or refer to section 3
in [15].

(A.8.6) R e m a r k . If R is a local ring, and D is a dualizing complex for R, then

it follows by (A.8.3.4) that

(A.8.6.1) SuppR D = Spec R.

In our applications of the next two lemmas it is always a dualizing complex that
plays the role of Z.

(A.8.7) L e m m a . Assume that Z • c((f~(R) and Y • C(E)(R ) axe both homo-

logically non-trivial. IfAssR(Hsupv(Y)) C_ Supp R Z (e.g., suppR Z = SpecR),
then
sup Y - sup Z < sup (RHomR(Z, Y)).

Proof. See [35, Proposition 2.2].

(A.8.8) L e m m a . Assume that Z • c((f~ (R) and X • C(3 ) (R) are both homolo-
gically non-trivial. If Supp R Z = Spec R, then
sup Z + inf X _> inf (Z ®L X).
A.8. DUALIZING COMPLEXES 189

Proof. Follows from (A.8.7); see [15, Lemma (4.11)].

In the rest of this section we work to establish three lemmas - - (A.8.11), (A.8.12),
and (A.8.13) - - which allow us to conclude that a morphism a, between appropri-
ately bounded complexes, is invertible in the derived category if RHomn(a, R),
D ®~ a, or RHomn(D, a) is so. Those who are familiar with the derived cate-
gory can skip the rest of this section and refer to [8, Lemma (1.2.3)(b)] instead.
For the proofs of (A.8.11)-(A.8.13) we need some extra properties of depth
and width.

(A.8.9) L e m m a . /f Y E Cff~ (R) and p E Spec R, then

pESupp nY ~ .~ depthnpYp<co.

Proof. Follows, e.g., by [16, Corollary (5.2)].

(A.8.10) L e m m a . If X 6 C(3)(R), then

X;~0 ~ ( p E S p e c R I widthnpXp < c o } ~ 0 .

Proof. Follows, e.g., by [36, Lemma 2.6].

(A.8.11) L e m m a . Assume that Y E C ~ ( R ) with SuppRY = SpecR, and let

I E C~ (R) be an injective resolution of Y. If a : V -~ V' is a morphism in
Cff~ (R), and HomR(a, I) is a quasi-isomorphism, then a is a quasi-isomorphism.

Proof. If HOmR((~,I) is a quasi-isomorphism, then the mapping cone

2~4(Homn(a,I)) is homologically trivial, cf. (A.l.19), and by (A.2.1.4) also
the complex Homn(Ad(a),I) is homologically trivial. We want to prove
that .h//(~) _~ 0, so we assume that SuppR2~4(~ ) ¢ 0 and work to estab-
lish a contradiction, cf. (A.8.4.1). The complex HomR(J~//(a),I) represents
RHomn(.~'/(a), Y), and ~4(~) belongs to C~(~)~(R) because both V and V' do so.
For p E SuppRM(a ) it now follows by (A.4.5), (A.6.4), (A.6.3.2), and (A.8.9)
that

depthRp (RHomR(M(a), Y))p = depthRp (RHomRp (M(a)p, Yp))

= widthRp M(a)p + depthRp Yp
= inf fl4(a)p + depthR~ Yp
((x).

But RHomR(A4(a), Y)p is homologically trivial, and this means, in particular,

that depthR, (RHomR(2~4(a),Y))p = co, whence the desired contradiction has
been obtained. []
190 APPENDIX. HYPERHOMOLOGY

(A.8.12) L e m m a . Assume that Z E c((f~(R) with SuppRZ = SpecR, and let

P E CP(R) be a projective resolution of Z. I r a : V ~ V' is a morphism in
C(-~)(R), and P ®R a is a quasi-isomorphism, then a is a quasi-isomorphism.
Proof. If P ®R a is a quasi-isomorphism, then the mapping cone M ( P ®n a) is
homologically trivial, cf. (A.1.19). By (A.2.4.4) we have .M(P ®R a) "~ P ®R
M ( a ) , and the latter complex represents Z ®~ M(a). The mapping cone of
a belongs to C(n)(R) because both V and V' do so. Assume that M ( a ) ~ 0
and choose by (A.8.10) a prime ideal p such that widthRp M(a)p < oo. It now
follows by (A.4.14), (A.6.5), and (A.6.3.2) that

n .M(a))p = widthRp (Zp ®~p A4(a)p)

widthRp (Z ®L
= widthnp Zp + widthnp .M(a)p
= inf Zp + widthnp .M(a)p
<oo.

But Z ®L M ( a ) is homologically trivial, so widthRp (Z ®~ M(a))p = oo. Thus,

we have reached a contradiction, and we conclude that ~4(a) is homologically
trivial as wanted. []

(A.8.13) L e m m a . Assume that Z E c((f~(R) with SuppnZ = SpecR, and let

P E CP(R) be a projective resolution of Z. I r a : W --+ W' is a morphism in
C(E)(R), and HomR (P, a) is a quasi-isomorphism, then a is a quasi-isomorphism.
Proof. If HomR(P,a) is a quasi-isomorphism, then the mapping cone
M(HomR(P,a)) is homologically trivial, cf. (A.l.19), and by (A.2.1.2) we
have .A4(HomR(P,a)) = HomR(P,.A4(a)). The latter complex represents
RHomR(Z,J~4(a)), and M ( a ) belongs to C(r)(R) as both W and W' do so.
If M ( a ) is homologically non-trivial, we can choose a prime ideal p associated
to the top homology module; by (A.4.5), (A.6.4), (A.6.3.2), and (A.6.1.2) we
then have

depthnp (RHomR(Z, .A4(a)))p = depthnp (RHomnp (Zp, M(a)p))

= widthnp Zp + depthnp M(a)p
= inf Zp - sup M (a)
< (:x~.

But RHomR(Z,,~4(a))p is homologically trivial and has, therefore, infinite

depth. Thus, we have reached a contradiction, and we conclude that A4(a)
is homologically trivial, i.e., a is a quasi-isomorphism. []
Bibliography
[1] Maurice Auslander, Anneaux de Gorenstein, et torsion en alg~bre commuta-
tive, Secretariat mathdmatique, Paris, 1967, S~minaire d'Alg~bre Commu-
tative dirig~ par Pierre Samuel, 1966/67. Texte r~dig~, d'apr~s des exposes
de Maurice Auslander, par Marquerite Mangeney, Christian Peskine et Lu-
cien Szpiro. l~cole Normale Sup~rieure de Jeunes Filles.
[2] Maurice Auslander and Mark Bridger, Stable module theory, American
Mathematical Society, Providence, R.I., 1969, Memoirs of the American
Mathematical Society, no. 94.
[3] Maurice Auslander and David A. Buchsbaum, Homological dimension in
Noetherian rings, Proc. Nat. Acad. Sci. U.S.A. 42 (1956), 36-38.
[4] , Homological dimension in local rings, Trans. Amer. Math. Soc. 85
(1957), 390-405.
[5] , Homological dimension in Noetherian rings. II, Trans. Amer. Math.
Soc. 88 (1958), 194-206.
[6] Luchezar L. Avramov, Ragnar-Olaf Buchweitz, Alex Martsinkovsky, and
Idun Reiten, Stable cohomological algebra, in preparation.
[7] Luchezar L. Avramov and Hans-Bjorn Foxby, Homological dimensions of
unbounded complexes, J. Pure Appl. Algebra 71 (1991), no. 2-3, 129-155.
[8] , Ring homomorphisms and finite Gorenstein dimension, Proc. Lon-
don Math. Soc. (3) 75 (1997), no. 2, 241-270.
[9] , Cohen-Macaulay properties of ring homomorphisms, Adv. Math.
133 (1998), no. 1, 54-95.
[10] Hyman Bass, Injective dimension in Noetherian rings, Trans. Amer. Math.
Soc. 102 (1962), 18-29.
[11] , On the ubiquity of Gorenstein rings, Math. Z. 82 (1963), 8-28.
[12] Winfried Bruns and Jiirgen Herzog, Cohen-Macaulay rings, revised ed.,
Cambridge University Press, Cambridge, 1998, Cambridge studies in ad-
vanced mathematics, vol. 39.
192 BIBLIOGRAPHY

[13] Henri Cartan and Samuel Eilenberg, Homological algebra, Princeton Uni-
versity Press, Princeton, N. J., 1956.
[14] Leo G. Chouinard, II, On finite weak and injective dimension, Proc. Amer.
Math. Soc. 60 (1976), 57-60 (1977).
[15] Lars W. Christensen, Semi-dualizing complexes and their Auslander cate-
gories, to appear in Trans. Amer. Math. Soc.

[16] , Sequences for complexes, to appear in Math. Scand.

[17] , Gorenstein dimensions, Master's thesis, Matematisk Institut,
Kobenhavns Universitet, 1996.
[18] , Functorial dimensions, Ph.D. thesis, University of Copenhagen,
1999.
[19] Lars W. Christensen and Anders Frankild, Functorial dimensions and
Cohen-Macaulayness of rings, in preparation.
[20] David Eisenbud, Commutative algebra with a view toward algebraic geome-
try, Springer-Verlag, New York, 1995, Graduate Texts in Mathematics, vol.
150.
[21] Edgar E. Enochs, Injective and fiat covers, envelopes and resolvents, Israel
J. Math. 39 (1981), no. 3, 189-209.
[22] Edgar E. Enochs and Overtoun M. G. Jenda, On Gorenstein injective mod-
ules, Comm. Algebra 21 (1993), no. 10, 3489-3501.

[23] , Mock finitely generated Gorenstein injective modules and isolated

singularities, J. Pure Appl. Algebra 96 (1994), no. 3, 259-269.
[24] , Gorenstein flat preenvelopes and resolvents, Nanjing Daxue Xuebao
Shuxue Bannian Kan 12 (1995), no. 1, 1-9.
[25] , Gorenstein injective and projective modules, Math. Z. 220 (1995),
no. 4, 611-633.
[26] , Resolutions by Gorenstein injective and projective modules and
modules of finite injective dimension over Gorenstein rings, Comm. Algebra
23 (1995), no. 3, 869-877.

[27] __, Gorenstein injective and flat dimensions, Math. Japon. 44 (1996),
no. 2, 261-268.
[28] m , Coliftings and Gorenstein injective modules, J. Math. Kyoto Univ.
38 (1998), no. 2, 241-254.

[29] , Gorenstein injective modules over Gorenstein rings, Comm. Alge-

bra 26 (1998), no. 11, 3489-3496.
BIBLIOGRAPHY 193

[30] , Gorenstein injective dimension and Tot-depth of modules, Arch.

Math. (Basel) 72 (1999), no. 2, 107-117.
[31] Edgar E. Enochs, Overtoun M. G. Jenda, and Bias Torrecillas, Gorenstein
flat modules, Nanjing Daxue Xuebao Shuxue Bannian Kan 10 (1993), no. 1,
1-9.
[32] Edgar E. Enochs, Overtoun M. G. Jenda, and Jinzhong Xu, Foxby duality
and Gorenstein injective and projective modules, Trans. Amer. Math. Soc.
348 (1996), no. 8, 3223-3234.
[33] Hans-Bjorn Foxby, Hyperhomological algebra ~ commutative rings, notes in
preparation.
[34] , Gorenstein modules and related modules, Math. Scand. 31 (1972),
267-284 (1973).
[35] , Isomorphisms between complexes with applications to the homolo-
gical theory of modules, Math. Scand. 40 (1977), no. 1, 5-19.
[36] , Bounded complexes of flat modules, J. Pure Appl. Algebra 15
(1979), no. 2, 149-172.
[37] , Homological dimensions of complexes of modules, S6minalre
d'Alg~bre Paul Dubreil et Marie-Paule Malliavin, 32~me ann6e (Paris,
1979), Springer, Berlin, 1980, Lecture Notes in Mathematics, no. 795,
pp. 360-368.
[38] , A homological theory for complexes of modules, Preprint Series 1981
nos. 19a&b, Matematisk Institut, K~benhavns Universitet, 1981.
[39] , Gorenstein dimension over Cohen-Macaulay rings, Proceedings of
international conference on commutative algebra (W. Bruns, ed.), Univer-
sit/it Onsabriick, 1994.
[40] Evgeniy S. Golod, G-dimension and generalized perfect ideals, Trudy Mat.
Inst. Steklov. 165 (1984), 62-66, Algebraic geometry and its applications.
[41] Robin Hartshorne, Residues and duality, Springer-Verlag, Berlin, 1966, Lec-
ture notes of a seminar on the work of A. Grothendieck, given at Harvard
1963/64. With an appendix by P. Deligne. Lecture Notes in Mathematics,
no. 20.
[42] Takeshi Ishikawa, On injective modules and flat modules, J. Math. Soc.
Japan 17 (1965), 291-296.
[43] Birger Iversen, Cohomology of sheaves, Springer-Verlag, Berlin, 1986, Uni-
versitext.
[44] Srikanth Iyengar, Depth for complexes, and intersection theorems, Math. Z.
230 (1999), no. 3, 545-567.
194 BIBLIOGRAPHY

[45] Christian U. Jensen, On the vanishing oflim (i), J. Algebra 15 (1970), 151-
166.

[46] Daniel Lazard, Autour de la platitude, Bull. Soc. Math. France 97 (1969),
81-128.

[47] Saunders MacLane, Categories for the working mathematician, Springer-

Verlag, New York, 1971, Graduate Texts in Mathematics, vol. 5.

[48] Eben Matlis, Applications of duality, Proe. Amer. Math. Soc. 10 (1959),
659-662.

[49] Hideyuki Matsumura, Commutative ring theory, second ed., Cambridge

University Press, Cambridge, 1989, Cambridge studies in advanced mathe-
matics, vol. 8.

[50] H5 Dinh Du~'n, A note on Gorenstein dimension and the Auslander-

Buchsbaum formula, Kodai Math. J. 17 (1994), no. 3, 390-394, Workshop
on Geometry and Topology (Hanoi, 1993).
[51] Christian Peskine and Lucien Szpiro, Dimension projective finie et coho-
mologie locale. Applications d la ddmonstration de conjectures de M. Aus-
lander, H. Bass et A. Grothendieck, Inst. Hautes Etudes Sci. Publ. Math.
(1973), no. 42, 47-119.

[52] Michel Raynaud and Laurent Gruson, Crit~res de platitude et de projec-

tivitd. Techniques de "platification" d'un module, Invent. Math. 13 (1971),
1-89.
[53] Paul Roberts, Le thdor~me d'intersection, C. R. Acad. Sci. Paris S~r. I
Math. 304 (1987), no. 7, 177-180.
[54] Jean-Pierre Serre, Sur la dimension homologique des anneaux et des mod-
ules noethdriens, Proceedings of the international symposium on algebraic
number theory, Tokyo & Nikko, 1955 (Tokyo), Science Council of Japan,
1956, pp. 175-189.

[55] Rodney Y. Sharp, Finitely generated modules of finite injective dimension

over certain Cohen-Macaulay rings, Proc. London Math. Soc. (3) 25 (1972),
303-328.
[56] Nicolas Spaltenstein, Resolutions of unbounded complexes, Compositio
Math. 65 (1988), no. 2, 121-154.

[57] Yasuji Takeuchi, On finite Gorenstein dimension, Kobe J. Math. 4 (1987),

no. 1, 103-106.

[58] , On weakly Gorenstein dimension, Kobe J. Math. 5 (1988), no. 2,

271-277.
BIBLIOGRAPHY 195

[59] Jean-Louis Verdier, Cat@odes ddrivdes, quelques rdsultats (dtat 0),

SGA 4½, Springer, Berlin Heidelberg New York, 1977, pp. 262-311.
[60] Charles A. Weibel, An introduction to homological algebra, Cambridge Uni-
versity Press, Cambridge, 1994, Cambridge studies in advanced mathemat-
ics, vol. 38.
[61] Jinzhong Xu, Flat covers of modules, Springer-Verlag, Berlin, 1996, Lecture
Notes in Mathematics, no. 1634.
[62] Siamak Yassemi, G-dimension, Math. Scand. 77 (1995), no. 2, 161-174.
[63] , Width of complexes of modules, Acta Math. Vietnam. 23 (1998),
no. 1, 161-169.
List of Symbols

Z, 10 Gpd R X, 106
N, 10 c°F(R), 120
No, 10 Gfdn X, 120
( x l , . . . , xt), 10 Tdn X, 127
(R, m, k), 10 CGI(R), 139
k, 10 Gidn Y, 141
k(p), 10 0tx, 160
ER(M), 10 Ixl, 160
zn M, 10 ZX, 160
z R, 10 B~, 160
Annn M, 10 C t , 160
~PNM~ 11 Ht(X), 160
PPNM, 12 H(X), 160
qPMN, 12 E m x , 161
o3PNM, 12 rx, 161
OPNM, 12 l x , 161
5M, 17 ~-, 161
G(R), 18 - , 162
M*, 18 supX, 162
M**, 18 infX, 162
-*, 18 ampX, 162
MT, 18 C(R), 162
G-dimn M, 22 ~, 164
depth R M, 32 ,-~, 164
6Y , 42 D(R), a65
XRy, 42 r u X , 165
Tt(R), 44 X v ~ , 165
X*, 44 CuX, 165
Ca(R), 52 XvD, 165
G-dimn X, 52 Homn(X, Y), 168
7~, 66 X ®n Y, 169
A(R), 67 azgx, 170
~z, 72 PzYx, 170
B(R), 73 ~zxy, 170
cOP(R), 104 wzgx, 170
198 LIST OF SYMBOLS

OZyX, 171
P(R), 173
Z(R), 173
Y(R), 173
idR Y, 173
pdRX, 173
fdRX, 174
RHomR(X. Y). 176
Ext,(M. N). 176
X ®~ Y. 178
Tor.~(M. N). 178
depth R Y. 183
widthR X. 184
~ ( X ) . 185
m y 185
#R,
m
185
SuppR X, 187
Index

AB formula, 5 of modules, 110

for flat dimension, 132 and localization, 68
for G-dimension, 56 and short exact sequences, 70,
of finite modules, 35 85
for Gorenstein flat dimension, for Gorenstein rings, 70, 80, 88
134 modules in, 85
for projective dimension of and short exact sequences,
finite modules, 13 85
for restricted Tor-dimension, bounded complex of, 71
129 Auslander's zero-divisor conjecture,
of complexes with finite 39
homology, 130 Auslander-Bridger formula, 35, see
accounting principles, 186 also AB formula
adjointness Auslander-Buchsbaum formula, 13,
isomorphism of complexes, see also AB formula
170 Avramov, Luchezar L., 52, 58, 99,
identity induced by, 179 159
isomorphism of modules, 12
amplitude inequalities, 81 Bass, Hyman, 5, 7
amplitude of complex, 162, 163 Bass class, 73, 158
annihilator, 10 and finite Gorenstein injective
associativity dimension, 143
isomorphism of complexes, of modules, 145
170 and finite injective dimension,
identity induced by, 179 73
isomorphism of modules, 11 and localization, 73
Auslander, Maurice, 1-3, 17, 63, 97 and short exact sequences, 76,
Auslander class, 67, 158 87
and finite flat dimension, 68 for Gorenstein rings, 75, 80, 88
and finite G-dimension, 70 modules in, 86
and finite Gorenstein flat and short exact sequences,
dimension, 122 87
of modules, 125 bounded complex of, 76
and finite Gorenstein Bass conjecture, 7, 147
projective dimension, 107 Bass formula, 13
200 INDEX

for Gorenstein injective complete resolution by finite free

dimension, 147 modules, 93, 93, 94, 118
Bass numbers, 185 complex(es), 160
Bass series, 186 amplitude of, 162
Betti numbers, 185 bounded to the left/right, 162
bidual module, 18 concentrated in certain
biduality, 18, 19, 22, 42, 43, 50, 188 degrees, 160
homomorphism of modules, 17 equivalent, 164
morphism of complexes, 42 up to a shift, 164
Bridger, Mark, 2, 17, 63 homologically bounded (to the
Buchsbanm, David A., 3 left/right), 163
Buchweitz, Ragner-Olaf, 2 homologically trivial, 161,164
infimum of, 162
canonical module, s e e dualizing isomorphic, 161
module modules considered as, 160
canonical representation shifted, 161
short exact sequence of, 166
of RHomR(RHomR(X, R), R),
43, 43, 68 supremum of, 162
truncated, 165
of RHomR(D,D ®~ X), 66,
with finite homology, 163
67, 68
of D ®L RHomR(D,Y), 72,
degree of element in complex, 160
72
depth, 10, 183
category of R-complexes, 162 and Bass series, 186
full subcategories of, 162-164 of modules, 32
Change of Rings Theorem for derived category, 82, 165, 175
G-dimension, 56 differential, 160
of finite modules, 39, 51 dimension, s e e Krull dimension
Chouinard, Leo G., II, 5, 134, 147 dual module, 18, 19
CM rings, s e e Cohen-Macaulay duality functor (algebraic), 18, 44
rings dualizing, 18
codepth, s e e width dualizing complex, 65, 81,187
Cohen-Macaulay defect, 14 dualizing module, 76, 83, s e e a l s o
at most 1, 131 dualizing complex
Cohen-Macanlay rings, 14, 40, 63,
81, 112, 130, 131 Enochs, Edgar E., 1, 2, 91, 99, 100,
commutativity 105, 113, 115, 120, 126,
isomorphism of complexes, 135, 136, 147, 148, 152
169 equivalence of complexes, 164
identity induced by, 179 preservation of, 168, 175
complete flat resolution, 113, 114, up to a shift, 164
115, 118, 153, 156 exact sequence, 161
complete injective resolution, 135, Ext modules, 176
136, 153, 156 long exact sequence of, 177
complete projective resolution, 97,
98, 114, 118, 156 faithful functor, 168
INDEX 201

faithfully injective module, 177 and reflexive complexes, 54

finer invariant, 3 of finite modules, 22, 48, 57,
finite free modules 59-61, 63
belong to the G-class, 18 and change of rings, 38, 51
finite module, 10 and localization, 30
fiat dimension, 68, 89, 123, 131, in short exact sequence, 27,
174, 182 50
and Betti numbers, 185 modulo regular sequence, 34
of complexes with finite over Cohen-Macaulay rings,
homology, 182 63
of modules, 13 over residue class ring, 32
fiat modules, 14 refinement of projective
are Gorenstein fiat, 113 dimension, 28
fiat preenvelope, 100 over Gorenstein rings, 36, 56
fiat resolution, 172 refinement projective
existence of, 172 dimension, 55
of module, 174 G-resolution, 22, 57
Foxby, Hans-Bjorn, 1, 2, 5, 52, 58, existence of, 22
76, 82, 90, 112, 113, 126, length of, 22
131, 134, 158, 159, 183, GD Theorem, 54
186 for finite modules, 24
Foxby equivalence, 1, 76 T~ version, 57
over Cohen-Macaulay rings, 87 GD-GPD equality, 108
over Gorenstein rings, 79
GD-PD inequality, 55
G-class, 18 for finite modules, 29
and localization, 30 GFD Corollary, 122
dual modules in, 18 for modules, 125
finite free modules in, 18 GFD Theorem, 121
finite projective modules in, 21 for modules, 124
modules in, 94 GFD-FD inequality, 123
and short exact sequences, GFD-GPD inequality, 123
2O GID Corollary, 143
are Gorenstein flat, 119 for modules, 145
are Gorenstein projective, 98 GID Theorem, 142
are reflexive as complexes, for modules, 145
48 GID-ID inequality, 143
are torsion-free, 18 Golod, Evgeniy, S., 2, 40, 52
modulo regular sequence, 33 Gorenstein fiat dimension, 120,
non-projective modules in, 21, 120-125, 133-134,
95 148-151,155-157
of residue class ring, 31 and localization, 122
G-dimension, 52, 52-63, 70, 108, and the Auslander class, 122,
132 125
and change of rings, 56 finer invariant than Gorenstein
and localization, 55 projective dimension, 122
202 INDEX

of complexes with finite Gorenstein projective modules, 97,

homology, 123 109, 156
of modules, 124, 125, 134 and localization, 97, 110
over Gorenstein rings, 124 and short exact sequences, 103
refinement of flat dimension, and the Auslander class, 100,
122 105
Gorenstein flat modules, 113, 124, are Gorenstein flat, 114
149, 150, 153, 156 finite, 98
and localization, 114 Gorenstein projective resolution,
and short exact sequences, 117 109, 100
and the Auslander class, 115, Gorenstein rings, 6, 14, 70, 75, 80,
120 88, 187
and finite flat dimension, 6, 79,
finite, 119
148
Gorenstein flat resolution, 124, 124
and finite G-dimension, 36, 56
Gorenstein injective dimension, and finite Gorenstein flat
141,141-152, 155-157 dimension, 124, 148
and localization, 146 and finite Gorenstein injective
and the Bass class, 143, 145 dimension, 144, 148
of complexes with finite and finite Gorenstein
homology, 147 projective dimension, 108
of finite modules, 147 and finite injective dimension,
of modules, 145, 147, 152 6, 14, 79, 148
over Gorenstein rings, 144 and Foxby equivalence, 79
refinement of injective Gorenstein Theorem
dimension, 143 A version, 70
Gorenstein injective modules, 135, B version, 75
145, 149, 150, 153, 156 Foxby equivalence version, 79
and localization, 135, 146 GD version, 36
and short exact sequences, 139 GFD version, 124
and the Bass class, 137, 139 GFD/GID version, 148
width of, 150 GID version, 144
Gorenstein injective resolution, GPD version, 108
144, 144, 145 PD/ID version, 6
T~ version, 56
Gorenstein projective dimension,
special complexes version, 80
106, 106-112, 156
special modules version, 88
and localization, 110
GPD Corollary, 107
and the Auslander class, 107, for modules, 110
110 GPD Theorem, 106
of complexes with finite for modules, 109
homology, 108, 123 GPD-PD inequality, 108
of modules, 109, 110, 112
over Gorenstein rings, 108 Hom complex, s e e homomorphism
refinement of projective complex
dimension, 108 Horn evaluation, 17, 21, 42, 43, 72
INDEX 203

homomorphism of modules, 12 Krull dimension, 10

morphism of complexes, 171 of complexes, 188
identity induced by, 180 of rings, 13, 63
Hom vanishing corollary, 11
Hom vanishing lemma, 11 ladder, 10
homology complex, 160, 164 local ring, 10
homology isomorphism, s e e localization at prime ideal, 168
quasi-isomorphism
homology modules, 160 M-regular element, 30, 33
long exact sequence of, 166, M-sequence, 30, 33
177, 179 mapping cone, 26, 29, 167, 167
homomorphism complex, 168 Matlis duality, 178
and mapping cones, 168, 169 modules considered as complexes,
and shift, 168, 169 160, 163
homothety, 161 morphism of complexes, 161
homothety morphism, 42, 187 identity, 161
induced in homology, 161
identity morphism, 161
induced functor on complexes, 167 Nakaya.ma's lemma, 11
infimum for complexes, 178
of complex, 162, 163 natural map, 10
of empty set, 10
injective dimension, 73, 143, 173, Peskine, Christian, 7, 40, 52
180-182 Poincar~ series, 185
of complexes with finite preservation of quasi-isomorphisms
homology, 182 and equivalences, 168, 175
and Bass series, 186 projective dimension, 44, 45, 55,
of modules, 13 89, 108, 173, 181
injective hull, 10 of complexes with finite
injective modules, 14 homology, 182
are Gorenstein injective, 135 and Poincar~ series, 185
faithfully so, 177 of finite modules, 13, 29
injective precover, 136 and change of rings, 38
injective resolution, 171, 172 of modules, 13
existence of, 172 over regular rings, 3
of module, 174 projective modules, 14
intersection theorems, 39 are Gorenstein projective, 97
Ishikawa, Takeshi, 7, 149 finite, 14
isomorphism belong to the G-class, 21
of complexes, 161 projective resolution, 172, 173
of modules, 161 existence of, 172
Iyengar, Srikanth, 29, 183 of module, 174

Jenda, Overtoun M. G., 2, 91, 99, quasi-isomorphism(s), 162

113, 135, 147, 148, 152 and mapping cones, 167
Jensen, Christian U., 90 preservation of, 168, 175
204 INDEX

R-complex, s e e complex standard morphisms, 170-171

R-sequence, 33 support, 187
refinement, 4 supremum
reflexive complex(es), 44 of complex, 162, 163
and finite G-dimension, 48, of empty set, 10
54, 61, 63 swap
and finite projective isomorphism of complexes,
dimension, 44 170
and localization, 45 identity induced by, 179
and short exact sequences, 46 isomorphism of modules, 12
belong to the Auslander class, Szpiro, Lucien, 7, 40, 52
69
over Gorenstein rings, 56 Takeuchi, Yasuji, 63
reflexive module, 44, 50 TD-FD inequality, 131
regular element, 30 TD-GD inequality, 132
regular rings, 14, 37 TD-GFD inequality, 134
and finite projective tensor evaluation, 67
dimension, 3, 14 homomorphism of modules, 12
Regularity Theorem, 3 morphism of complexes, 170
Reiten, Idun, 2 identity induced by, 180
residue field, 10 tensor product complex, 169
resolution by finite free modules, and mapping cones, 169
172 and shift, 169
existence of, 172 Tor modules, 178
restricted Tot-dimension, 127, long exact sequence of, 179
127-131 Tor-depth, s e e width
and localization, 128 Torrecillas, Blas, 2, 113
of complexes with finite torsion, 18
homology, 130, 132 -free, 18, 19, 33
over Cohen-Macanlay rings, -less, 19
130 truncations, 165
over Gorenstein rings, 128
refinement of flat dimension, width, 184
131
Xu, Jinzhong, 2, 99, 105, 120, 126,
refinement of G-dimension,
135
132
refinement of Gorenstein flat Yassemi, Sia.mak, 2, 41, 58, 147,
dimension, 134 183
Roberts, Paul C., 7
zero-complex, 160, 164
Serre, Jean-Pierre, 3 zero-divisors, 10, 39
Sharp, Rodney Y., 76, 90
snake lemma, 11
stability, 7, 14, 45, 58, 182
standard homomorphisms, 11-12
standard identities, 179-180