You are on page 1of 76

Aoo RING S P A C E S AND A L G E B R A I C K - T H E O R Y

J.P. M a y

In [ZZ], W a l d h a u s e n introduced a certain functor

A(X), w h i c h he thought of as

the algebraic K-theory of spaces X, with a view towards applications to the study of the concordance groups of P L manifolds a m o n g other things. Actually, the m o s t

conceptual definition of A(X), f r o m w h i c h the proofs w o u l d p r e s u m a b l y flow m o s t smoothly, w a s not m a d e rigorous in [2Z] on the grounds that the prerequisite theory

of rings up to all higher coherence homotopies w a s not yet available. A s W a l d h a u s e n pointed out, m y theory of E co ring spaces [i2] gave a success-

ful codification of the stronger notion of c o m m u t a t i v e ring up to all higher c o h e r e n c e homotopies. W e begin this paper by pointing out that all of the details

n e c e s s a r y for a c o m p r e h e n s i v e treatment of the w e a k e r theory appropriate in the absence of commutativity are already implicit in [IZ]. Thus w e define Aco ring spaces in section i, define Aco ring spectra in section 2, and s h o w h o w to pass b a c k and forth between these structures in section 3. for an intuitive s u m m a r y will imitate. T h e general theory does not i m m e d i a t e l y imply that Waldhausen's proposed definition of A(X) can n o w be m a d e M rigorous. n X O n e m u s t first analyze the structure of the E 0o The reader is referred to[13]

ring theory that the present Aoo ring theory

present on the topological space an A co ring space X.

of ( n × n)-matrices with coefficients in a suitable additive

T h e r e is no difficulty in giving M n X

structure, but it is the multiplicative structure that is of interest and its analysis requires considerable work. We co prove in section 4 that M n X is a multiplicative Technically, co

Aco space and c o m p a r e these A the f r e e d o m to use different A

structures as n varies in section 6.

co

operads is crucial to the definition of these A

structures, and a curious change of operad pairs trick is n e e d e d for their

241

c o m p a r i s o n as

n varies.

We M

study the relationship b e t w e e n the additive and n X in section 5. It turns out that M n X is definitely

multiplicative structures on

not an Aoo ring space, although it m a y

satisfy the requirements of an appropriate

strong h o m o t o p y generalization of this notion. With this theoretical b a c k g r o u n d in place, w e find ourselves in a position to develop a far m o r e general theory than w o u l d be needed solely to obtain the T h u s w e construct the algebraic K-theory of A co W e take the h o m o -

algebraic K-theory of spaces. ring spaces X in section 7.

T h e basic idea is simple enough. KX

topy groups of the plus construction

on the telescope of the classifying spaces T h e technical w o r k here involves the

of the Aoo spaces of unit c o m p o n e n t s F M n X .

construction of the relevant compatible classifying spaces and of a modified telescope n e c e s s a r y for functoriality. In section 8, w e analyze the effect of restricting of m o n o m i a l matrices. The

this chain of functors to the sub Aoo spaces F n X

resulting plus construction turns out to be equivalent to the zero c o m p o n e n t of Q(BFXJI{0}), denotes colim where n n f~ ~ Y. FX is the Aoo space of unit c o m p o n e n t s of X and QY

T h e proof involves the generalization of a standard con-

sequence of the Barratt-Quillen t h e o r e m for w r e a t h products of m o n o i d s to w r e a t h products of Aao spaces together with a c o m p a r i s o n b e t w e e n the Aoo spaces ~nIFX. This completes the proofs in sections necessarily consumer the factory statement addressed who wishes the development of the technical relegated machinery. Of c o u r s e , are F n X and

3 - 8 ( a l l of w h i c h a r e to t h o s e i n t e r e s t e d to i n s p e c t

to t h e e n d s of t h e s e c t i o n s ) The

in a c l o s e l o o k a t t h e m a c h i n e r y . product

the finished section

without taking the tour through sections Z a n d 3 ( u p to t h e as far as Remarks

i s i n v i t e d to f i r s t of Proposition

read

1 and skim

3 . 7 ) , t h e n s k i p to s e c t i o n 7 a n d r e a d

7.4, next read section 8 as far as R e m a r k s

8.5, and finally turn to sections 9-12.

W e begin the homotopical and homological analysis of our functors in section 9, giving general h o m o t o p y invariance properties and pointing out a

242

spectral sequence converging to H , K X computation of its Ez-term.

and s o m e general formulas relevant to the

W e discuss e x a m p l e s and display various natural m a p s and d i a g r a m s section I0. On discrete rings, our theory reduces to Quillen's [i8, 5 ]. F o r X -~ ~0 X

in

general Aco ring spaces X, the discretization m a p spaces.

is a m a p of Aoo ring to

This establishes a natural m a p f r o m the n e w algebraic K - g r o u p s of X

Quillen's algebraic K - g r o u p s of ToM.

To illustrate the force of this assertion, w e 10,7.

record the following trifling consequence of the d i a g r a m displayed in T h e o r e m Corollary. T h e usual m a p

f r o m the qt_hhstable h o m o t o p y group of spheres

to the qth algebraic K - g r o u p of Z factors through the qth algebraic K - g r o u p of X for any Aoo ring space X such that ~0 X = Z.

O n topological rings, our theory reduces to that of W a l d h a u s e n [ZZ,§ I], or rather, to the topological analog of his simplicial theory. However, it should be completely

pointed out that virtually all of the proofs in the earlier sections b e c o m e trivial in this special case. intuition, that m u c h

T h e force of the theory is the translation of the obvious

that is true for rings remains true for Aco ring spaces, into The crucial question is, h o w m u c h ? W e have already

rigorous mathematics.

observed that the matrix "ring" functor M n does not preserve Aoo ring spaces, and other e x a m p l e s Remarks of p h e n o m e n a which do not directly generalize are discussed in does in fact ring spaces

i0.3 and IZ.4.

W h a t I,find truly remarkable is h o w very m u c h of interesting A co

generalize.

The point is that there are vast n u m b e r s

which are far r e m o v e d

f r o m our intuition of ~vhat a ring looks like. f r o m the fact that Eco ring F o r instance, the category of ring R

P e r h a p s the m o s t fascinating e x a m p l e s c o m e spaces are AGO ring spaces by neglect of structure.

finitely generated projective m o d u l e s (or free modules) over a c o m m u t a t i v e gives rise to an Eoo ring space in which the addition c o m e s m o d u l e s and the multiplication c o m e s

f r o m the direct s u m of The h o m o t o p y

f r o m the tensor product.

groups of this space are Quillen's algebraic K - g r o u p s of R, and the present theory

243

gives rise to a second order algebraic K-theory based on its I shall say a little bit m o r e about such e x a m p l e s

Aoo ring

structure.

in section i0, but I should add at

once that I have not yet had time even to contemplate the p r o b l e m of m a k i n g actual c alc ulations. We specialize to obtain the algebraic K - t h e o r y of topological spaces A(X) to be KQ(I]xJ~{0}), as proposed by Waldhausen. type of X The in

section ii, defining

general theory gives an i m m e d i a t e

calculation of the rational h o m o t o p y

A(X), and various other properties c l a i m e d by W a l d h a u s e n specialization.

also drop out by

In particular, w e discuss the algebraic K - t h e o r y of X with coeffiring and give a complete account of the

cients in a (discrete) c o m m u t a t i v e

stabilizationof the various algebraic K-theories of X to generalized h o m o l o g y theories, this being based on a general stabilization t h e o r e m given in the Appendix. ideas, it The

While the theory discussed above w a s inspired by Waldhausen's is logically independent of his w o r k and should be of independent interest.

connection with his theory is w o r k in progress and is discussed v e r y briefly in section 12. T h e basic point to be m a d e is that, at this writing, there exist two

algebraic K-theories of spaces, the one developed here and the one rigorously defined by Waldhausen, and a key remaining p r o b l e m is to prove their equivalence .

w o u l d like to thank M e l R o t h e n b e r g for insisting that I try to m a k e Waldhausen's Bob Thomason ideas rigorous and for m a n y have also been v e r y useful. helpful discussions. I am Conversations with

profoundly indebted to W a l d h a u s e n

for envisioning the possibility of such a theory as that presented here.

244 Contents §1. A ring spaces

oo

§z
g3 §4 g5 56

Aco r i n g s p e c t r a The recognition principle M a t r i c e s w i t h e n t r i e s in A m r i n g s p a c e s

Strong h o m o t o p y ( ~ , ~ ) - s p a c e s a n d m a t r i x r i n g s
The comparison b e t w e e n MnX and Mn+l X

§7
§8 g9

T h e a l g e b r a i c K - t h e o r y of A m r i n g s p a c e s Monomial matrices and Q0(BFX ii {0})

S o m e h o m o t o p i c a l a n d h o m o l o g i c a l p r o p e r t i e s of KX Examples, natural maps, and formal properties of KX

§10
§ll §lZ

T h e a l g e b r a i c K - t h e o r y of s p a c e s N o t e s on W a l d h a u s e n ' s w o r k S t a b i l i z a t i o n s of f u n c t o r s to h o m o l o g y t h e o r i e s

Appendix.

245

§ I.

A

oo

ring spaces

W e begin by recalling the definitional f r a m e w o r k [I2,Vl § I and § Z] and a m o r e [i0,§i-3].

of [iZ]~ details are in

leisurely discussion of s o m e of the m a i n ideas is in

The notion of an Eoo ring space is based on the notion of an operad pair (~,~), which consists of an "additive" operad and an action of ~ in J on C • A n operad ~, a "multiplicative" operad ~ ,

C has associated to it a m o n a d

(C,~, ~])

, the category of (nice) based spaces.

There is a notion of an action of

on a space C on X.

X, and this is equivalent to the standard notion of a~* action of the m o n a d These notions apply equally well to ~ . Actions by I, and a ~0-space is a When ~ are thought of as

multiplicative, with basepoint

/~-space with a second ~ acts on ~0-spaces. and ~]:X - ~ C X ~, the That is, are

basepoint 0 which behaves as zero under the action. monad CX C restricts to a m o n a d ~0-space of if X is a in

.~0[ J ], the category of and ~:CCX -~CX

is a

~0-space

then m a p s monad C

~o-Spaces.

A n action of ( C , ~ )

on a space X is an action of the ~-space and a The

in ~ 0 [ ~

] on X.

That is, a (~, ~ ) - s p a c e is both a is a m a p of

~o-space

such that the additive action C X -~ X

~o-Spaces.

last condition encodes distributivity homotopies in a simple conceptual way, and the multiplicative theory is to be thought of as obtained f r o m the additive theory by a change of ground categories f r o m spaces to ~ o - s p a c e s .

246

is an E

oo

operad if its jth space group ~.. j An E

C(j) is contractible and is acted on space is a space together with an action oo CX for any space X. T h e s e are co operad

freely by the s y m m e t r i c by any Eco operad commutative

~ . Examples

are the spaces

m o n o i d s up to all coherence homotopies. oo operads. An E Co

(~ , /2 ) is an E

pair if ~= and ~ are both E anactionby ~0-space X. any E co

ring space is a space together with are the spaces CX for any

operad pair

(~, ~).

Examples

T h e s e are c o m m u t a t i v e

semi-rings up to all coherence homotopies

(semi-ring because additive inverses are not built in). A theory of A A co oo spaces is developed in [i0, §3 and 13]. A n operad ~ (j) is contractible. ~ is an Co

operad if w 0 ~(j) is ~. and each c o m p o n e n t of j

An A

space is a space together with an action by any Aoo operad up to all coherence homotopies. lent to monoids,

~ . T h e s e are m o n o i d s co spaces are equiva-

As explained in [i0,p. 134], A

hence admit classifying spaces; there is also a direct delooping

construction independent of the use of monoids. Definition l.l. E
(3o operad and ~

A n operad pair
(DO operad.

(¢, ~ ) is an A
CO

co

operad pair if ~ is an
with an

is an A

An A

ring space is a space together

action by any A ~ 0-space X.

oo

operad pair ( ~, ~ ). E x a m p l e s

are the spaces

CX

for any

This is the desired notion of a ring (or rather, semi-ring) up to all coherence homotopies. but it s e e m s There is also a notion with both ~ and ~ A Co operads,

unprofitable to study rings up to h o m o t o p y for which not even addition

is h o m o t o p y commutative. Note that the product of an E T h u s if (~, ~ ) is an A co is an E co (3o operad and an A 5' is an A co operad is an A oo operad.

operad pair and ~'X

co

operad, then (~, ~' X ~ )

operad pair, the action of on C

/~ on ~ being obtained by pullback f r o m again by

the action of ~ pullback, E co

• Since a (C, ~ ) - s p a c e is a ( C, ~' X ~)-space, co ring spaces.

ring spaces are A

247

S o m e discussion of discrete operads m a y clarify ideas. operads 7~Z and ~ monoid and an An such that ~ (j) = ~. and J

There are ~7~ -space is a act on ~ .

~ (j) is a point. A n Both ~ and ~

~ -space is a commutative monoido

( ~ , 7)q)-space is a semi-ring and an (~, ~)-space is a commutative semi-

ring. Say that a m a p of operads is an equivalence if the underlying m a p of jth spaces is a homotopy equivalence for each j. A n Eoo operad equivalence equivalence ~ admits an evident

~ -~ ~ . A n operad ~ is an Aoo operad if and only if it admits an ~ -~ ~7~ . Thus Aoo operad pairs and Eoo operad pairs m a p by and ( ~ , ~ ) °

equivalences onto the respective operad pairs (~, 7~) R e m a r k 1.2.

I would like to correct an annoying misprint in the crucial definition,

[IZ,VI. 1.6], of an action of ~ on ~ . In (a') of the cited definition, the displayed formula is missing s o m e symbols, ~(k(g;c I..... Ck); dI being written for k(g, di), and should read

× k(g;di))v I, S(j i .....Jk)

= k(g;e I ...... ek)

§ Z Aoo ring spectra A n Eoo space determines a spectrum and thus a cohomology theory. The

notion of Eoo ring spectrum encodes the additional multiplicatlve structure on the spectra derived from the underlying additive Eoo spaces of Eoo ring spaces. Aoo ring spaces also have underlying additive Zoo spaces, and w e have an analogous notion of Aoo ring spectrum. definitions. Let ~ Only the multiplicative operad ~ appears in these

denote the linear isometries Eco operad of [IZ,I.I.Z ]. F o r good

and sufficient reasons explained in [IZ,IV§I ], w e a s s u m e given a m a p of operads -~ ~ . In the cited section, ~ was a s s u m e d to be an Eoo operad. For example, We may

instead a s s u m e that ~

is an Aoo operad.

~ might be the product

of an Aoo operad and an Eco operad which m a p s to ~ .

N o w the assumption that

was an Eoo operad played no mathematical role whatever in the definition,

248

[!2,1V.l.l], of a map /J -~ ~

/~ -spectrum.

This notion of an action by

~

only required the

and is thus already on hand in our ACO context. A n Aco ring s p e c t r u m is a of operads /~ -~ /J-spectrum over any Aoo

Definition Z.I0 operad

/~ with a given m o r p h i s m

A s explained in [iZ,p.68-70], an Aco ring s p e c t r u m is a (not necessarily commutative) space is a ring s p e c t r u m with additional structure. n0-space. The formal l e m m a s In particular, its zero th

[IZ,IV 1.4-1.9] apply verbatim to . We summarize the conclusions

-spectra for any operad /~ which m a p s to ~ they yield. Recall that a (coordinate-free) s p e c t r u m each finite-dimensional sub inner product space associative and nnital s y s t e m of h o m e o r n o r p h i s m s gonal to W ; here E 0 = E{0}.

E V

consists of a space of Rco

EV

for

together with an for V ortho-

E V -- ~2WE(v + W ) Q Co

T h e stabilization functor

f r o m spaces to

spectra is defined by QCOX : {Q~vX IV C R°°}, w h e r e QX : colim ~v ~V~Vx;

here the loop and suspension functors sphere

f~v and

are defined in t e r m s of the The inclusion ~]:X -~ Q X and

tV, the one-point compactification of V. ~: Q Q X ~ QX

colimit of loops on evaluation m a p s and the analogous colimit m a p sphere operad spectrum ~

give a m o n a d

((],M,I]) in ~ , on E O. The

~: Q E 0 -~ E 0

gives an action of Q S O is a

S is defined to be

Qoo SO, and

/~o-space for any prefer to

. U s e of these notions is vital for rigor, but the reader m a y classical terms,

think of spectra in m o r e i>0.

restricting attention to V = R i for

Proposition 2.Z. (i) Q
CO

Let

X

he a

~0-space

and

E

a

~-spectrum. in the

X is a

~-spectrum

and is the free ~0-spaces

~-spectrum

generated by X

sense that a m a p of

f:X -~ E 0 of
%2

extends uniquely to a m a p

~'.Qoo X -~ E

~ -spectra such that

f~] = f.

249

(ii)

S

is a

~-spectrum

and the unit

e:S -~ E

is a

]J -map.

(iii) The m o n a d .~o-map, so t h a t

O in :Y restricts t o a m o n a d i n E 0 is a Q - a l g e b r a in ~0[~'].

~0[y] and ~:QE 0

~E

isa

T h e following analog of [IZ,IV. 1.10] is central to the definitions proposed by W a l d h a u s en. Example Z.3. For a ~-spaee X without zero, construct a ~-space X + with

zero by adjoining a disjoint basepoint dent way.
spaces.

0 to X a n d e x t e n d i n g t h e a c t i o n in the e v i and
~ =

Q

oo

X + is then a
is any A oo

]~ - s p e c t r u m
o p e r a d and

~ : X + - ~ Q X + is a m a p of ~' x ~ , then a ]~-f ~'-spaee

~0is a to

If ~I

-space via the projection be used in the present theory. A Go space X.

~ -~ ]~ ', while the projection Therefore Qoo X +

anows ~

is an Aco ring s p e c t r u m for any

Remarks

2.4.

F o r what it is worth, w e note that m u c h of the discussion of O n e first

orientation theory given in [IZ,IV ~3] r e m a i n s valid for Aoo ring spectra.

checks that commutativity of the underlying ring spectra is not essential to the general theory in[IZ,III]. Independently of this, one finds that the assertion of [IZ,IV.3.1] is valid for operads and }J-spectra E for Aoo operads ~ as well as for Eta /J-spaces

~ . T h e cited result gives a certain c o m m u t a t i v e d i a g r a m of

~ - m a p s , the middle r o w of w h i c h yields a fibration sequence G e FE T ~B(G;E) cl ~" B G Be ~ BFE

after passage one step to the right by use of the classifying space functor on -spaces. Here FE is the union of those c o m p o n e n t s of E 0 w h i c h are units in

the ring w0E0, G

is the infinite group or m o n o i d corresponding to s o m e theory of B(G;E) is the classifying

bundles or fibrations, such as O , U , Top, or F, and space for E-oriented G-bundles or fibrations. of orientation and the m a p s

The map

q corresponds to neglect

e and -r are interpreted in ~2,11I. 2.5]. T h e point is

250

that the notion of A

oo

ring s p e c t r u m is just strong enough to yield B e : B G

~

BFE,

w h i c h is the universal obstruction to the E-orientability of G-bundles; c o m p a r e

[IZ,lv. 3.z].

§3.

T h e recognition principle We first s h o w thatthe zero th space of an Aco ring s p e c t r u m is an Aco ring

space and then s h o w that the s p e c t r u m d e t e r m i n e d by the additive Eco structure of an A oo ring space is a n A (2o ring spectrum. W e also obtain c o m p a r i s o n s between co ring level version of the Barratt-

the two evident composite functors and give an A Quillen theorem.

All of this is in precise analogy with the corresponding develop-

m e n t for Eco ring spaces and spectra in [IZ,VII], and w e n e e d only point out the trivial changes of definition involved. Let pairs E (Do (~, ~ ) be an A co operad pair and suppose given a m a p of o p e r a d ~co ~ is the infinite little convex bodies are defined in [IZ,VII§ 1 and §Z] ~co co :K co acts naturally -~ Q . of m o n a d s With these

(It,p): (~', ~ ) -~ ( J~oo' 2), w h e r e Here ~ co and its action by

operad.

(and w e are suppressing technical p r o b l e m s handled there). on the zero in [~ th spaces of spectra, and there is a m o r p h i s m Ir induces a m o r p h i s m C -~ K oo =

Similarly,

of m o n a d s in ~

notations, the proof of [iZ,VII. Z.4 ] applies ve=batim to prove the following result, in w h i c h the second part follows f r o m the first via part (iii) of Proposition 2.2.

Theorem monads (ii) in Y

3.1.

(i) T h e m o r p h i s m s

Ir:C -" Kco in

and

~co:Kco -~ Q

of

restrict to m o r p h i s m s ~-spectrum,

of m o n a d s

~0 [ •]. E0 is a ( C , /~ )-space by

If E

is a

then its zero th space

pullback of its Q-action A n Eoo space Thomason X

6: Q E 0 -~ E 0 along C~coW. determines a s p e c t r u m _BX. T h a n k s to recent w o r k by

and myself [16], w e n o w k n o w that all infinite loop space m a c h i n e s yield

e q u i v a l e n t s p e c t r a w h e n a p p l i e d to X , b u t it is e s s e n t i a l to t h e p r e s e n t

251

m u l t i p l i c a t i v e l y e n r i c h e d t h e o r y t h a t we u s e the c o n s t r u c t i o n p r e s e n t e d in [1Z,VII §3]. We a s s u m e n o w t h a t t h e A m o p e r a d p a i r (w, p) is given by the projections, w h e r e ~'(j) is contractible (but not necessarily definiteness, one might think of the e x a m p l e (~,~) is ( ~' X ~co, ~ ' X ~ ) and that

(~ ', ~') is an operad pair such that each N.-free) and j ~' is an A co operad. The proofs of For

(~', ~') = ( ~ , ~ ) .

[IZ, Vll.4.1 and 4. Z] apply verbatim to yield the following results. Theorem the 3.P. If X is a ( ~ , ~ ) - s p a c e , ~-spectrum. X and the zero t~ space BoX is s u m m a r i z e d by thenBX ( f o r m e d w i t h r e s p e c t to

~ -space structure) is a

T h e relationship between a natural d i a g r a m X ~
J

B(C,C,X)
L

1. B ( Q , C , X )

~

BoX-

The first and third solid arrows are equivalences, canonicalhomotopy fore also inverse to the first arrow.

and

~ is obtained by use of a

The middle solid arrow, and there-

L, is a group completion

(see [12,p.168] or, for a full discussion,

[ii,51]).
T h e o r e m 3.3. spaces. T h e s o l i d a r r o w s in t h i s d i a g r a m a r e m a p s of ( ~ , ~ ) L is a m a p of /~0-spaces. ~ - m a p s , but t h i s is

The d o t t e d a r r o w

T h e c a n o n i c a l h o m o t o p y i n v e r s e , and L , a r e n o t of little significance.

T h e basic idea is that w e have group completed the additive

s t r u c t u r e of X w h i l e c a r r y i n g a l o n g the m u l t i p l i c a t i v e s t r u c t u r e . We h a v e the f o l l o w i n g c o n s i s t e n c y s t a t e m e n t s in s p e c i a l c a s e s , the p r o o f s being identical to those in [iZ,p.191-19Z]. m a p of spectra ~':B__E0 ~ E, and For a spectrum I~, there is a natural

~'~ is an equivalence if E is connective (that is,

i f w i E = 0 for i < 0).

252 Proposition 3.4. of ~ -spectra. For a ~0-space Y, CY is a ( ~ , ~)-space by D e f i n i t i o n 1.1 a n d Moreover, 3.1(i).
QY is

If E

is a

~ -spectrum,

then

~:B. ~ 0 --E

is a m a p

a ( ~ , ~ ) - s p a c e b y P r o p o s i t i o n 2.2(i) a n d T h e o r e m ~ c o ~ r : C Y -~ QY is a m a p of ( ~ , For a ~)-spaces ~0-space

3.1(ii).

by T h e o r e m

P r o p o s i t i o n 3.5.

Y, t h e c o m p o s i t e m a p of ~ - s p e c t r a

__BCY

--

>, _BBQY

co ~ Qooy v: Q
O0

is a strong deformation retraction. induced by the freeness of Q c o Y is thus a m a p of ]~ -spectra.

Its inverse inclusion ~o-map Y

Y -~ B C Y CY

is and

f r o m the

~

~ P BoCY

When

Y = CS 0,

CY = _LL C(j)/Y,j

:

i l K(Ej,1).

H e r e the l a s t r e s u l t theorem,
CO

i s a m u l t i p l i c a t i v e l y e n r i c h e d f o r m of the B a r r a t t - Q u i l l e n f o r m of w h i c h a p p e a r s on t h e E tions come fromA
OO co

the strongest

ring level.

Interesting A

ring level applica-

spaces,

s u c h as m o n o i d s ,

via Example 2.3. The

T h e p r e v i o u s r e s u l t c a n be r e l a t e d to t h e H u r e w i c z h o m o m o r p h i s m . monad N a s s o c i a t e d to ~ a s s i g n s to a s p a c e Y its infinite symmetric ~ a c t s on ~ .

product, There-

or free commutative topological monoid, and any operad fore then N restricts to a m o n a d in .~0 [ 7 ] If ~ : ~ -~ ~

is the augmentation, These observations

(~, I):(~,

/ ~ ) -~ ( ~ , ~ )

i s a m a p of operad p a i r s .

i m p l y the following result.

L e m m a 3.6. spaces, hence
Now

For a

/~0-space

Y,

~ : C Y -~NY -spectra.

is a map of ( ~ , ~ ) -

B E : B C Y -~ BNY

is a m a p of

forget all about the multiplicative structure on Y.

B y an oversight,

the f o l l o w i n g r e s u l t w a s o m i t t e d f r o m m y e a r l i e r w o r k s i n t h i s a r e a .

253

Proposition 3.7. to H . Y

F o r a based space

Y,

Ir,__BNY is naturally isomorphic

and h = B E o v: Qoo Y -~ B_NY

induces the stable Hurewicz h o m o m o r p h i s m

on passage to homotopy groups. Proof. The zero

In particular, h is a rational equivalence. h: Q Y -* B o N Y is obtained by passage to direct

~h

map

limits f r o m the top composite in the commutative diagram

n ~ ny

n

n

~ n 0-

~ ~nznNy 11

f2n~n~

. - ~n~EnBoNY -

n ~-n B NY
n

Y

NY

BoNY

BoNY

I

where

0n

is the iterated structure m a p of the spectrum
i% n

B_NY (and ~ is written

for the unit of both m o n a d s N and f~ ~ ). Hurewicz h o m o m o r p h i s m

Therefore

h will induce the stable induces the ordi-

on homotopy groups if L~]:Y ~ B o N Y If Y and

nary Hurewicz h o m o m o r p h i s m . equivalence while w,NY = H,Y

is connected, then L is a natural on w.

~] induces the Hurewicz h o m o m o r p h i s m

by results of Dold and Thorn [3]. spaces. and let Let N Y ~:Y -* N Y

Thus the problem is to account for non-connected

denote the free commutative topological group generated by Y and k : N Y -~ N Y denote the natural inclusions, so that

k~ = ~. morphism

D o l d a n d T h o r n give t h a t on ~,.

s,NY = H,Y

a n d "~ i n d u c e s t h e H u r e w i c z h o m o -

One c o u l d p r o v e t h a t

k is a group completion by direct homoNY --~ BONY, b u t we s h a l l

logical calculation and then deduce an equivalence
reverse this idea. Consider the following diagram.

254

~NEY

~ g2BNY

I" B o N Y

Y R e s u l t s of D o l d and Thorn i m p l y t h a t nected) k is a w e a k e q u i v a l e n c e ( s i n c e EY is con-

and t h a t t h e r e is a n a t u r a i w e a k e q u i v a l e n c e ~ s u c h t h a t

~@~ --~ f ~ ' o r I.

By a r e s u l t of M i i g r a m [17, p. Z45] ( s e e a l s o [9, 8 . 7 and 8 . 1 1 ] ) , t h e r e is a h o m e o morphism ~ : N E y -*. BNY s u c h t h a t ~2 ~ r e s t r i c t s on Y to r-,rl, w h e r e ~ i s the

s t a n d a r d m a p [9, 8 . 7 ] . a n d t h u s of N - s p a c e s ,

F i n a l l y , s i n c e the p r o d u c t on NY i s a m a p of m o n o i d s [1Z, 3.4] t o g e t h e r w i t h the p r o o f of [ 1 1 , 3 . 7 ( p . 75)] give a ~ such that ~
--~ ~ Since ( m o d u l o the use of w e a k equiva~. on the left is the H u r e w i c z of the

natural weak equivalence

l e n c e s with a r r o w s going the w r o n g way). homomorphism, s o i s (L~). on t h e r i g h t .

It is a standard c o n s e q u e n c e

finiteness of the stable h o m o t o p y rationals, the stable H u r e w i c z homology theories. We diagram

groups of spheres that, upon tensoring with the b e c o m e s an i s o m o r p h i s m of

homomorphism

record the following corollary of the proof a n d an elaboration of the

above w h i c h s h o w s that ~k

is homotopic to
Y, k:NY

f~(k~ -1 )~ : N Y
-~NY

-~ ~ N ~ Y .

Corollary 3.8.

F o r any space

is a group completion.

255 §4. M a t r i c e s w i t h e n t r i e s i n Aoo r i n g s p a c e s

Let sider the set ~-space,

X be a (C, ~)-space, MnX of n X n m a t r i c e s

where

(~,

~ ) is any operad pair, and conClearly ~ -space MnX is a

w i t h e n t r i e s i n X.

n a m e l y t h e n Z - f o l d G a r t e s i o n p r o d u c t of t h e

X with itself. MnX is not

We w i s h to s h o w t h a t if ( ~ , ~ )

is an Am operad pair, then, while

a n A m r i n g s p a c e , i t i s at l e a s t a ( m u l t i p l i c a t i v e ) Aco s p a c e . non-trivial, since MnX is n o t a c t u a l l y a ~ -space.

Even this much is

Since

MnX i s c e r t a i n I y n o t c o m m u t a t i v e ,

it is c o n v e n i e n t to f i r s t e I i m ~ (j); t h e s e s e r v e Thus

i n a t e t h e e x t r a n e o u s a c t i o n s by s y m m e t r i c

groups on the spaces

o n l y to h a n d l e c o m m u t a t i v i t y h o m o t o p i e s i n the g e n e r a l t h e o r y of o p e r a d s . assume now that B is anon-E operad, inthe s e n s e of [10,3.1Z].

By[10,3.131, for some

.th a t y p i c a l Aoo o p e r a d h a s t h e f o r m #J X ?7/ ( w i t h j space non-E operad ~ is given b y m a p s k : ]~(k) × ~ (jl) X . . . × ~(jk) with each ~ (j) contractible.

.~(j) X Z j )

A n a c t i o n of /~ on t h e o p e r a d

-

~(jl...jk )

with the properties specified in [IZ, VI. 1.6] (see R e m a r k I. 2), except that its extraneous equivariance condition (c) must be deleted. [10,3.13 ] shows that the non-E operad ~ acts on ~ A slight elaboration of if and only if the operad

×~

acts on C.

A(¢,fl)-spaeeisdefinedtobea

~-spaceand

0-space X
the notions of

such that the additive action G X -~ X
(~,fj)-space and

is a m a p of /~0-spaces, and
there-

(C,~ × Pl)-space are then equivalent,

fore the theory of Aoo ring spaces m a y as well be developed in terms of (~=,~)spaces for an E

0o

operad

~

and a non-Z operad ~ with each

~(j)

contractible.

However, the w o r k in the rest of this section requires only that ~ acted upon by a non-E operad ~ O.j: ~ ( j ) X X j -* X d e n o t e the a c t i o n s of ~ and that X ~j: on X. be a (~, ~)-space. ~(j) XX j -* X

be an operad Let

and ~

a n d of

256

We w a n t to u s e m a t r i x M X. n Obviously this entails

multiplication

to d e f i n e a n Aoo s p a c e

structure

on X.

u s e of b o t h t h e m u l t i p l i c a t i o n j-fold products, that is,

and the addition on the products

We a r e q u i t e h a p p y to u s e a r b i t r a r y arbitrary additions, actual e l e m e n t s of ~(j).

given by

W e a r e l e s s h a p p y to u s e a r b i t r a r y addition is present = ~

n j-l-fold c a s e of

but, since a canonical with ~

o n l y in t h e t r i v i a l Thus define

commutativity,

, we have no choice.

34n(J) -- C(n)-l) x ~(j).
Let order that T(j) d e n o t e t h e s e t of a l l s e q u e n c e s lexicographically. u. = s. J Define , j : ~ n(j) × (MnX)J by the followh~g formula, where x(r,s) -~ MnX (r,s) th e n t r y of a m a t r i x x. Let U = (u 0 . . . . ,uj) with 1 K u . x< n a n d _ U such

T(j)

T(r,s,j)

d e n o t e t h e s u b s e t of t h o s e

u 0 = r and

denotes the

J
(1) qJj(c,g;x 1 ..... xj)(r,s) = @nj-l(C;ueT(r,s,j)× ~J(g; qX--1 X q ( U q - l ' U q ) ) ) iterated matrix "

A l l w e h a v e d o n e is to w r i t e d o w n o r d i n a r y

multiplication,

allowing for parametrized families of both multiplications and additions on the underlying "ring". The rest of this section will be devoted to the proof of the

following

result. Theorem 4.1. The ~ n(j) .th are the 3 spaces ~n on MnX. and ~0 i s t h e i n c l u s i o n of t h e 9 4 n ( I ). Clearly of a non-2 operad
)4 n '

and the maps Proof. identity matrix ~I(1)

~j

specify an action of By convention, I n in M X. n Let

~ n ( 0 ) = {*} 1 = (1,1)t We m u s t

~'(1) X ~ (1)= specify maps

is the identity map of M nX. ~7: ~ n (k) × ~ n ( J l ) × ...

× ~ n ( J k)

~)4n(J),

J = Jt + "'"

+ Jk '

with respect to which ~ mute, w h e r e

is an operad and which m a k e the following diagram c o m -

~ is the evident shuffle h o m e o m o r p h i s m :

2.57

~n(k) X,.%/n(jt ) X...× ~/n(Jk ) X (MnX)J -y X t

~n(j ) X (MnX)J

q~J •

MnX

(*)

i x i~ 1

Ck
x ~j ix...
x¢.

Jk
*"

~4n(k ) X j~n(jl ) X (MnX) ji X ... X ~n(Jk) X (MnX) jk

~/n(k)

X

(MZ) k

(Compare [10,1.5].)

We first chase the diagram to see how "~ must be defined
is an operad.
n

and then verify that, with this definition, ~ details are perfectly straightforward: However,

In principle, the

one does what one has to do and it works.

since I omitted all such routine verifications from [IZ] and since this one

is m u c h less intuitively obvious than most, 1 will try to give s o m e idea of the c o m binatorics involved. W e first calculate the composite around the bottom of the diagram (*).

By (1), we have

(z)

d#k(iX~jtX'''XCjk)(tXbt)(c'g;cl'g{

.....
X

c k ' g k ; x t . . . . . xj)(r,s) k
~k(g;
X

= @nk-i(C;u,T(r,s,k)

q=l Zq(Uq-i'

Uq))),

where, with eq = -Ji + "°" + J q ' (3) Zq = q~jq(Cq, gq; Xeq_i+i ,Xeq. t+ 2 . . . . . x eq) .
is a u~^-space' q.

If jq= O, then z q = In. Since X (4) Let (5)

k ~k(g;qX__iZq(Uq.l, Uq)) = 0 if jq= 0 and Uq_ I / Uq for any {ql .... ,qm}, ip = jqp , d m <k, = e

denote those q, in order, suchthat jq> 0 and set = c , fp gqp , and yp = Zq = for I<_ p < m .

p

%

, b

p

%

p

258 Let s(j i . . . . . th coordinate jk ) ~ ]~(£t) X . . . X ~J(~k) , Eq= 0 o r t, have q th t ~ ~ (1) if j q > 0 and q c o o r d i n a t e * ~ ~(0) if j q = O. Set f = y(g;s(j I ..... jk)) c ~ ( m ) . jq: 0, define V = (v0 ..... Vm) C T(m)

(6)

If U c T(k) satisfies Uq_ l = Uq whenever by deletion of the duplicated entries

u . Then, since I e X is the basepoint for q

the ~-action and Zq(Uq_i,Uq) = I if jq= 0, (7) k ~k(g;qX i Zq(Uq_t'Uq)) = m ~m(f; pX__ yp(Vp-l'Vp)) i

Note that each V E T(m) a r i s e s uniquely f r o m such a U ¢ T(k) and let t0l ..... 1 E ~(1) jk ) E ~ ( f f l ) X . . . X ~ ( ~
n

m_l) ,

6 r = 0 o r 1, have

r

th

coordinate

if j q = 0 implies

Uq_ 1 = Uq for the r th element Set

U ~ T(k) and r th

coordinate
(8) b =

* ~ ~ (0) otherwise. "y(c;t(j 1 . . . . .
jk ) ~

~(nm-i). ~ -action, (4)-(8) imply

Since 0 ~ X is the basepoint for the (9)

k × 0nk_l(c; U : T(r,s,k) ~k (g; q><=i z q(Uq_ I' Uq)))
m

: 0nm_ l(b; V, T ×r , s , m ) ~m(f;PX i Yp(Vp_ i, Vp))). ( Evaluating yp by (I), (3), and (5) and then using the definition, [IZ,VI. 1.8 and VI. I. i0], of a (~ ,~)-space, we find
m m

410)

~m(f;pX= lYp(Vp_l,Vp) ) = ~m(f;#=l 0nip_l(bp;

× y(Wp))) . WpE T(Vp_i, Vp, :p)

OnJ_m(k(f ; b I . . . . .

bin);

× H ~ S(n il "i , o . . , n

l) ~m(f ; YH (v))) mi
°

where if W p = (11)

(Wp, 0 ..... Wp,ip) c T(ip), then

P y(Wp) = ~i (fp; t=X1 Xd l+t(Wp, t - l ' W p , t )) p p-

259
i -1 1 -< h p < n p , then c o o r d i n a t e the h th P coordinate

andif

H= (hl,...,hm) YH(V) E Xm has

with p th

(lZ)

X . y(Wp) . of % ~ T(Vp.1, vp, Ip)

By (i0) a n d the d e f i n i t i o n of a

~-space

[ 1 0 , 1 . 3 o r 12,VI. 1.3],

(13)

X 8nm_ l(b; V¢ T ( r , s , m )

~m(f;p Xl =
);

m

yp(Vp_l,Vp)))

m-I @ nJ- i(~(b;k(f;bl ....bm)n

X
V~ T ( r , s , m ) HcS(n il-1

X
im-t ..... n

~m(f; YH(V))). )

Here

k(f;bl,...,bm) c C(nJ-m), since iI + ... + i m : j' hence application of ~(nJ'l). Further, by (Ii),(12),

N(b;-) to its n m'l st power yields an element of and the definition of a ~ -space,

(14) whe re

~m(f;YH(V))

=

~j(N(f;f 1 . . . . . uj) , T(j) . W z ' l.' . . .

J fm) ; qX I Xq(Uq_l,Uq)) ,

UH(V ) = (u 0 . . . . . (wi, 0, " ' ' ' w t ' i l

is the s e q u e n c e

wz,iz . . . . . . . . . . W m ' l
W P

Wm'im) of the o r d e r e d sets T(r,s,m) T ( r , s, j). and H r u n s Let ~ ~ ~ n J _ l be

o b t a i n e d by s p l i c i n g t o g e t h e r the h t___h e l e m e n t s P T(Vp_l,Vp,ip) for l<__p<__m.

As V r u n s t h r o u g h

t h r o u g h S(n i l - 1 . . . . .

n i m - i ) , UH(V ) r u n s t h r o u g h

that p e r m u t a t i o n which c h a n g e s the g i v e n l e x i c o g r a p h i c o r d e r i n g of T ( r , s , j ) o r d e r i n g s p e c i f i e d by UH(V ) < UH,(V' )

to the

if V < V' o r if V = V' a n d H < H'

(in the lexicographic ordering; see [IZ,VI.l.4]).

Substituting (14) into (13),

(13) into (9), and (9) into (2) and using the evident equivariance identification (to rearrange "addends"), we arrive at the formula

260 4t5) ~k(t X qJjlX... X ~jk)(t X ~)(c, g; c t, g t . . . . . m-i = e I_J _ l ( ~ ( b ; h ( f ; b _ n ..... bm)n. )~; X ~j('~(f;ft- ..... f ); JX x (u . u ))). U~T(r,s,j) m q=l q q - i , q Ck, gk; x t . . . . . x j ) ( r , s)

C o m p a r i n g (1) and (15), we see that the d i a g r a m (*) w i l l c o m m u t e p r o v i d e d that we define m-i (16) " ~ ( c , g ; c l , g 1. . . . . Ck' g k ) = ('Y(b;X ( f ; b l . . . . . b m )n )~''f(f;fl ..... fm ) ) "

Here, w h e n

jq = 0, (Cq, gq) = * and w e m a y think of gq as

* c fJ(0); (5), (6),

and the definition of an operad then imply

(l?)

"?(f; fi . . . . .

fm ) = ~/(g; gl . . . . .

gk)"

No s u c h r e i n t e r p r e t a t i o n of the f i r s t f a c t o r of (16) i s p o s s i b l e (as we would have to interpret c
We Certainly

q

as a n e l e m e n t of

~ ( n -1) to m a k e the n u m b e r s w o r k out).
~ n is a non-~, operad. (c,g) e Mn(k)' by

m u s t s h o w that, with this definition, = (c,g)

~/(i;c,g)

and ~/(c,g;l~ = (c,g) for

[IZ,VI. 1.6(b) and (b')]. for iteration of the m a p s

It remains to check the associativity formula [I2,VI. 1.e(a)] -f, and the reader w h o has followed the combinatorics so

far should not have too m u c h trouble carrying out the requisite verification for himself. and ~ The details involve use of the corresponding associativity formulas for , the equivariance formulas [iZ,Vl. 1.2(c) and VI. i.6(c')], the interi.Z, and a rather

action formulas [1Z,VI. 1.6(a) and (a')] as corrected in R e m a r k horrendous check that the permutations c o m e out right.

261

§ 5.

Strong homotopy ($', ~)-spaces
In this rather speculative

and m a t r i x r i n g s
(which will play little part in our later elaborato the

section

work),

we

make

aninitial definition in the direction of an up to homotopy of Aoo and Eoo ring spaces section. an up to homotopy

tion of the theories matrix "rings" Lada,

and explain its likely relevance

studied in the previous in [Z,V], has developed

generalization of the

theory

of

~-spaces.
monad

(See [13,~6] for a sketch.)
C, and the essential of the theory of (~,

His theory is based

on use of the
up to homoC

associated topy

starting point of the analogous ~)-spaces surely

generalization

must be the fact that

restricts to a m o n a d in the category of be appropriate. Definition 5.1. non-E operad). Let (~, ~ )

/~o-Spaces.

The following definition should

be an operad pair (where sh, (~ ,~ )-space

~

might be a -space

A strong homotopy, or 0 and a ~0-space

X is a

(X,@) with basepoint @:CX--~ X is an

(X, 6) with basepoint

i such that

sh G - m a p . defined and discussed in [Z,V§3]. It is

The notionof a.n sh G - m a p i s

required that the basic distributivity diagram GCX G@ ~- GX ,

CX

@

-~ X should

t h e c o m m u t a t i v i t y of w h i c h is the d e f i n i n g p r o p e r t y of a ( ~ , ~ ) - s p a c e ,

h o m o t o p y c o m m u t e and t h a t t h i s h o m o t o p y s h o u l d be t h e f i r s t of a n i n f i n i t e s e q u e n c e of h i g h e r c o h e r e n c e h o m o t o p i e s . M o r e general notions of sh (g,.~)-spaces, or only an sh with X only an sh g'-space

~ -space, surely also exist but would be m u c h m o r e complicated.

262
Unfortunately, the most general notion would presumably be essential to a fully h o m o t o p y invariant t h e o r y . I w o u l d hope t h a t if X is an ( ~,/~)-space passage from sh ( ~ , f l J ) - s p a c e , t h e n t h e r e is an a c t u a l

UX e q u i v a l e n t to X in an a p p r o p r i a t e l y s t r u c t u r e d way, so that the Aoo r i n g s p a c e s to Aoo r i n g s p e c t r a and f r o m Eao r i n g s p a c e s to Eoo

ring spectra directly generalizes to sh (~, ]J )-spaces for suitable pairs (~, •). T h i s w o u l d be in a n a l o g y to L a d a ' s one o p e r a d t h e o r y , and his c u b i c a l b a r c o n struction UX = B ( C , C , X ) of [Z,V §Z] would be the o b v i o u s c a n d i d a t e f o r UX.

H o w e v e r , I h a v e not a t t e m p t e d to p u r s u e t h e s e i d e a s . We r e t u r n to c o n s i d e r a t i o n of MnX f o r a ( ~ , ] J ) - s p a c e a non-2~ o p e r a d for c o n v e n i e n c e . X, with ~J b e i n g

As f o r m u l a e (4.16) and (4.17) m a k e c l e a r , p r o ~/n -* /1~ . H o w e v e r , MnX

j e c t i o n on the s e c o n d f a c t o r g i v e s a m o r p h i s m of non-~, o p e r a d s Therefore o'~n a c t s on ~ by p u l l b a c k of the a c t i o n of ~ on ~ .

is not a ( ~ , . ~ n ) - S p a c e o r any o t h e r a c t i o n of

b e c a u s e the d i s t r i b u t i v i t y d i a g r a m f a i l s to c o m m u t e (for this ~ n on ~ ) . As e x p l a i n e d in [12,VI §i and 2, p. 77], the d i a -

g r a m in q u e s t i o n r e s u l t s by p a s s a g e to d i s j o i n t unions and t h e n to quotients f r o m the following d i a g r a m s : H n ( k ) X ~ ( j l ) X (MnX) J1 X . . . 1X8. X . . . X 8. J1 Jk >

X ~'(jk) X ( M n X ) j k

Mn(k)

x

0

%x) k

ix F X --...~(il~X . . . X --..~'(Jk X (MnX)J 1× ]

... X (MnX)jk

Ck
MX
n

(*) ~n(k) j+l

AXtX5
X ~(jl ) X... X ~ ( j k ) X ((MnX)k) j

x x (¢k)j
2Nn(k ) X ~'(Jl)X . . . X ~ ( J k ) X (:~4n(k) X (MnX)k) j

[,

• ,, ~'(j) x (MnX)J

263
Here A is the i t e r a t e d d i a g o n a l , 2 ~k is as d e f i n e d in f o r m u l a (4.t), the @j give the additive a c t i o n on (MnX) = X n , and Jt Jk 5: (MnX) X . . . X (MnX) "*" ((MnX)k) j j = Ji" "" Jk ' the ~ a r e shuffle h o m e o m o r p h i s m s ,

is specified by

6(y i . . . . .

yk) =

X YI ' I~S(j I ..... Jk )
if

with

YI : (xti~ ..... Xk~)

yq= (xq~ ..... ~qjq).

T h e following result shows that, as far as the relationships between any

h o m o l o g i c a l and h o m o t o p i c a l i n v a r i a n t s of the m a p s

O. and ~k go, any r e s u l t s J

valid for Aoo ring spaces are also valid for matrix rings with entries in Aoo ring

spaces.

( C o m p a r e the a n a l y s i s of the h o m o l o g y of Eoo r i n g s p a c e s in [Z,II]; we

shall return to this point in section 9. )

P r o p o s i t i o n 5.Z.

If X

is a (~, ~ 3 ) - s p a c e , w h e r e C

is an Eco o p e r a d and

is a n y n o n - Z o p e r a d w h i c h a c t s on ~ ,

t h e n the d i a g r a m (*) is

~. X . . . X ~. -equivariantly h o m o t o p y commutative.

Jl

3k
Proof. O n the one hand, the corresponding d i a g r a m for the ( ~ , ~ ) - s p a c e is a E - s p a c e [i0,i.5]

X, formula (4.1), and the d i a g r a m which expresses that X imply

O)

~ k ( i X @ j t X . . . X @jk)(c,g; c t, Yt . . . . .
k-1 = e j n k _ i ( ~ ( c ; × ( g ; c 1 ..... % ~ );

Ck' Yk) X I~S(j i ..... jk )

X UE T ( r , s , k )

~k(g; YI(U))),

where

yi(U) = (xlil(u0,ul) ..... Xkik(Uk_1,Uk)). is a ~= -space imply

O n the other hand, formula (4.1)

and the fact that X

(z)

@j(kX ~k3)l~(A X i X 5)(t X i~)(c, g ; c t , Yt . . . . .

Ck' Yk ) ~k(g; Yl(U))).

= Ojnk" i(Y(k(c'g; c t . . . . . Ck); c J); Ic sO l ..... jk ) U¢ T ( rX s , k ) X ,

264

N o definition of k(c, g; c i ..... Ck) will m a k e the right sides of (I) and (Z) agree, and w e take the pullback definition k(g; cj, .... Ck). In view of the difference in T(r, s, k), the addends

order of appearance of the indexing sets S O l ..... jk ) and in (i) and (2) differ by a permutation v ,~ Ejnk_ I.

The m a p s

f,g: ~n(k) X ~(jl ) X ... × ~(jk ) -* ~(jn k-l) specified by
n

k-1 )

f(c,g;c t ..... and

c k) = ~ ( c ; k ( g ; c I . . . . .

c k)

g(c, g; c 1 . . . . .

%)

= -y(×(g; c 1 . . . . .

% ) ; cJ)v
~(jn k-i) is

are E. × ... × E. -equivariant, w h e r e the action of this group on Ji Jk determined by its tensorial e m b e d d i n g in Ej k-i of ~'2 in (~j)n is an E Co

[10,VI. 1.4], the diagonal e m b e d d i n g

, and the block s u m embedding of the latter in ~jnk_ I . Since

operad, the d o m a i n and c o d o m a i n of f and g are ~. )< ... X ~. -free 21 Jk The conclusion

and contractible, hence f and g are equivariantly homotopic . follows.

Of course, if w e had chosen to w o r k with permutations in our multiplicatire operads, then the d i a g r a m (*) would be h o m o t o p y commutative. If the homotopies of the proposition can be chosen with suitable cornpatibility as k and the j q vary, they will together yield the first of the infinite ~ k × ~n )< ... X ~ . -equivariantly 21 Jk

sequence of homotopies needed to verify the following assertion.

C o n j e c t u r e 5,3.

If X is a ( ~ , . ~ ) - s p a c e , then
~(j)

where ~ MX

is an ECO o p e r a d and (¢,~n)-Space

is any n o n - ~ o p e r a d w h i c h acts on ¢ ,
W i t h ~ a n ECO o p e r a d a n d e a c h

is an sh

contractible,

the notion of an sh ring space.
CO

(~o ]~)-space is an up to h o m o t o p y generalization of the notion of an A

265

T h e c o n j e c t u r e g i v e s the a p p r o p r i a t e

s e n s e in w h i c h i t m i g h t b e t r u e t h a t

MnX

is

a n Aoo r i n g s p a c e if X is a n Aoo r i n g s p a c e . T h e p r o o f of P r o p o s i t i o n which gives a similar operad. 5. Z i s p r e c i s e l y a n a l o g o u s to t h a t of [10, 1 . 9 ( i i ) ] , ~= - s p a c e w h e n ~ is a n E (3o

result about the product @ on a

L a d a [6] h a s s t u d i e d t h e p a s s a g e f r o m t h a t r e s u l t to t h e a s s e r t i o n t h a t and the problems he encountered there illustrate what would be

is a n s h C - m a p ,

i n v o l v e d i n a p r o o f of C o n j e c t u r e 5 . 3 .

§6.

The comparison

between

Mn X and

Mn+lX

As in section 4, l e t

X be a (~, ~)-space, where ~ i s a n y operad and
W e have exhibited a ~n(j) = ~-space structure and

is any non-E operad which acts on X. an :~n-Space structure on M n X, where

~ (nj-l) X ~(j), and have studied

the relationship between these actions. MnX and Mn+IX.

W e here study the relationship between

W e adopt the notations of section 4, but with an identifying Vn:MnX ~ Mn+l x denote the

subscript
natural

n where necessary for clarity. Let First consider the diagram

inclusion.

@
~ ( j ) X(MnX) j nj * MnX

IxvJ[ n
~(j)

I vn

X( M n + t X ) J

@n+l,,j

"

Mn+l x

H e r e @n a n d @n+t a r e d e t e r m i n e d

e n t r y w i s e f r o m t h e a c t i o n of ~ on X, a n d t h i s r <_ n a n d s <_ n.

d i a g r a m c e r t a i n l y c o m m u t e s on the ( r , s ) t h m a t r i x e n t r i e s f o r Similarly,

b o t h c o m p o s i t e s a l w a y s give ( r , s) t h m a t r i x e n t r y 0 if e i t h e r but n o t b o t h
However, for c c ~(j) and x i ~ M J , w e have

of r and s is n+J.

Vn0nj(C, x i . . . . .

xj)(n+t, n+t) = t
x j ) ( n + l , n + t ) = Oj(c; t j) .

but On+t,j(t X vJ)(c,x t .....

266

Thus

v

n

f a i l s to b e a ~ - m a p .

Indeed, as for discrete cases

rings,

1 a n d O . ( c l l j) l i e in j X is c o n t r a c t i b l e we have

different components

in a l l n o n - t r i v i a l

s i n c e , b y [1Z,p.140],

if 0 a n d 1 l i e in t h e s a m e c o m p o n e n t . the following result. Theorem is a map ~n+t-map, Tn: ~ n + t where Proof. Its h y p o t h e s i s Assume that ~n of n o n - ~ is a n

For the multiplicative

structures,

~(1) = {1} w i l l be d i s c u s s e d ~(t) operads i s t h e p o i n t 1. such that For

after the proof. each n~ t, there is an

6.i. -MnX

Vn:MnX -~Mn+t X Vn.

~n+t-space X ...

by pullback along

Let

tnj e ~ ( 6 t )

X ~"(~(n+i)j_t) ,

£i = 0 or t, have i th

coordinate * e ~ (0) if the i element th .th 1 coordinate
I , ~ (t) otherwise. U = (u i . . . . .

U ~ Tn+l(j-2) has any u

q

= n+l

and have
are

O b s e r v e t h a t if e l e m e n t s then U ~-~(r,u t ..... Tn+t(j-2)

of Tn+t(j-2)

written in the form

uj_t),

uj_t,s ) gives a and Tn+t(r ,s,j) for

bijective correspondence

between the orderedsets Define

each r and s between t and n+t.

. nJ.: ~4 n+t(j) : ~ ( ( n + 0 j-~) X 8(j) -~ C (n j - t ) X ~ (j) = ~4n(j)
by

Tnj(C, g)
By convention,

= ('7(e;

tnj), g).
tnl = 1 c ~ (t) argument, formulas so that Tnl is the identity map. the

Tn0(* ) = * a n d combinatorial

Another laborious associativity

which uses formula (4.16), ~

and equivariance

for the operad formulas

[1Z,VI. 1. Z(a) a n d (e)],

the interaction, a c t i o n of ~ on

unit, and equivariance ~ (see Remark of

[IZ,VI. 1 . 6 ( a ' ) , ( b ' ) , (c')1 f o r t h e of p e r m u t a t i o n s

i.Z), and a careful consideration

based on the description

~ given after formula

(4. 14) s h o w s t h a t t h e f o l l o w i n g

diagrams c o m m u t e .

~l~n+l(k ) X ~14n+l(ji ) X... X
Tnk X T . X . . . X T . nj 1 nJk ~n(k) X

,,~/n+l(Jk )

-~ ~+i(jl +... +

jk )

n ' ~ l +" " ° +Jk

~n(ji ) X

...

X

~n(jk )

>'

3~n(jI +... + Jk )

267

Thus

"r

n is a m a p of n o n - E o p e r a d s .
(MnX)J Tnj × i

Now consider the following diagrams. ~nj MnX

~ n + i (j) ×

~- .2~n(j) X (MnX)J

>

~ n + l ( j ) X (Mn+IX)J

L

~n+l, j

-~ M n + t X

1
~(1) ,

O u r c l a i m is t h a t t h e s e d i a g r a m s to be a point• Consider

commute,

and it is f o r t h i s that w e r e q u i r e

L~n+l,j(c'g; Yl . . . . .

y j ) ( r , s)

J = O(n÷i)j_ 1 (c~
where c E ~,' "t t n + t j" - i j , " ~ J

× u~ T+i(r,s,J)
yi= u

~J(g;q×i= yq(Uq - i ' u q ) ) )

g c ] J ( j ) , and q-i and

Vn(Xi). q

T h e U th f a c t o r on t h e r i g h t is If r ~ n and s <__n,

0 c X if e i t h e r but not b o t h of u

is n+i f o r any q.

it f o l l o w s t h a t the r i g h t s i d e is e q u a l to

J
X @nJ- i (X(c; tnj); Vc T ( r , s , j ) ~j(g! X X q ( U q _ l , U q ) ) ) q=l

=

qJj(Tnj(C , g ) ; x 1 . . . . . r a n d s is

x j ) ( r , s) . there

I f e i t h e r but not b o t h of exists

n+l, then, for any U ~ T+l(r,s,j), Uq is n+l, h e n c e

q s u c h t h a t e i t h e r but not b o t h of Uq_ t and y j ) ( r , s) = O. We t h e r e f o r e have

•j(c, g; Yi . . . . .

V C n j ( V n j X 1)(c, g; x i . . . . . as d e s i r e d ,
whereas

x j ) ( r , s) = ~ n + t , j ( l × Vn ) (c, g; x I . . . . . j Here
to

x j ) ( r , s) ,

unless

r = s = n+l.
side reduces

we find that the l e f t s i d e is 1 ~ X

the right

O(n÷t)j_l(c;O ..... whet e

O, 1)

=

O t ( y ( C ; S n j ), l) ~ X ,

Snj= (* ..... *,1)~ ~' (o)(n÷l)J-I - i X ~'(i).

268

Indeed, all elements of Tn+l(n+l, n+l, j) except the last have either but not both of Uq_ 1 and Uq equal to n+l for s o m e q, whereas the last U which yq(n+l,n+l)= l and ~j(g;q=l i)= i. j is (n+l,...,n+l) for ~(i)= {i} ensures

The assumption

that ~/(C;Snj) = I and therefore @i(-¢(c ;snj); i) = I. Unfortunately, it is not in general the case that ~(I) = {i}; for example, this fails for the canonical E oo operad ~ (9o used in section 3. W e could avoid The contractibility of ~(I) can be

this assumption by appealing to Lada's theory. used to prove that w is an sh H n + l-map. n

However, this solution (which I

w o r k e d out in detail in an earlier draft) leads to further complications in later sections. Our preferred solution is to prove that Aco ring spaces can be

functorially replaced by equivalent Aoo ring spaces with respect to a different Aco operad pair for which ~(1) is a point. This replacement process works equally

well in the Eco ring context. W e exploit the fact that the particular Eco operad (I) = { I}. (~ , ~ Eoo Moreover, as explained in [IZ,VI § Z and 4], ~ oo operad pair. Let (C ~) ~ o f [II,§4] has acts on itself and thus

) is an E ~

be any operad pair such that C is an ]By use of products

operad.

might be either an Aoo or an Eoo operad.

and projections, we then have operad pairs, and m a p s thereof, (Wl 'Wl )
" (¢

(¢,.,'.z)
Therefore D and C X D

x~,

,,2 x~)

(WZ 'I)

~ (IQ.,

,.2 x Q_).

( C , /~)-spaces are

(~ X ~ ,

~ X~)-spaces

by pullback, while both

are m o n a d s in the category of ( /~ X ~)0-spaces and ~rz:C X D -~ D

is a m o r p h i s m of m o n a d s in this category. W e proceed to construct a functor W (~ , ~X ~)-spaces. from ( ~ , ~ )-spaces to from

As explained in [IZ,vI. 2.7(iii)], there is a functor W

-spaces to ~ - s p a c e s specified in terms of the two-sided bar construction of [i0,§9] by W X = B ( D , C X D,X) and there is a natural diagram of ( C X ~ ) - s p a c e s

269

B= Z

X"~
Here 8

B(C X D , C × D , X )

~-B(D,C × D , X )

= WX ,
B w z is

is a h o m o t o p y equivalence with a natural h o m o t o p y inverse and Technically, w e should a s s u m e

also an equivalence.

or arrange (without loss of

structure by an elaboration of the a r g u m e n t s in [10,A.8 and A.II ]) that I ~ ~(I) and 0 e X are non-degenerate basepoints, so that the simplicial spaces used in W e have the following result. then WX is a and ~ and

our constructions are proper [10,11.2 and II,A.5]. Proposition 6.Z. (~, /~X~)-space, If X is a

(~= , ~ ) - s p a c e , is a (C X ~,

B ( C X D , C X D , X) ~×

~.~X ~)-space,

B w Z are m a p s Proof. DWX "~ W X

of ( ¢ × { ,

4)-spaces.

B y f o r m a l verifications f r o m [10,9.6 and 9.9], the action on W X , the action of C X D on B(C × D , C X D,X), and ~ and and are

of D

B ~ Z are all geometric realizations of m a p s of simplicial (.~ X ~ ) 0 - s p a c e s therefore m a p s of ( ~ X ~ ) 0 - s p a c e s by [10,12.Z]o

Clearly, w e m a y as well start our analysis of matrix rings of Aoo ring spaces by first replacing ~(I) = {I} X by W X . In particular, our assumption that This construction also handles a

results in no real loss of generality.

different technical problem, one that w e have heretofore ignored. Remarks = 6.3. ~'X A s explained in [IZ,VII§I and Z], ~to ~ and the product operads

oo

n e c e s s a r y to the proof of the recognition principle in section 3 are only partial m o n a d s .

are only partial operads and their associated m o n a d s However,

it is not hard to see that our replacement a r g u m e n t above w o r k s perfectly ~ . Thus, by use of the functor W, w e m a y a s s u m e without loss of

well for such

generality that all operads in sight are honest operads in the d e v e l o p m e n t of the present theory since ~ and all multiplicative operads (see [iZ,p.178]) are honest.

270

Remarks

6.4. By the method of proof of T h e o r e m and T' : ~.~p+q-~ ~ Pq q O: M P X X M X q

6.1, one can construct m a p s

: ~.~p+q -~ ~ p Pq

of non-~ operads such that the usual specifies an ~!p+q-map, where and Pq -r' Pq

block s u m of matrices M P X and M q X

~ M p + Xq

are regarded as M P X X M

~ p + q - S p a c e s by pullback along T X is given the product q

respectively and where From

~p+q-Structure [10,1.7].

this point of view, the problem with ~(I) {I} -~X = M I X is not an ~ l - m a p

above simply reflects the fact that unless ~=(i) = {i}. In order for

the inclusion

these s u m m a p s to be useful, one would have to understand their stabilization, that is, to analyze the diagrams

v ×v
M X × M X P q
O

P

q

rM p + i X X M q + i X

M P+q If X

X

Vp+q+ 1 o Vp+q ~ Mp+q ÷zX

is a ring, these composites obviously differ only by conjugation by a perIn general, the definition of such a conjugation entails an arbiA full

mutation matrix.

trary choice of product and yields only a homotopy commutative diagram.

analysis of the situaNon would presumably entail application of Lada's theory of strong homotopy maps.

271

7.

The Algebraic K-theo:ry of Aoo ring spaces Let (C , ~) be a n A o o operad pair. A ( C , #j)-space X will be saidto

be grouplike if w o e a semi-ring.

is a group under addition and therefore a ring rather than just is a grouplike Aoo ring space if and

U p to w e a k equivalence then, X th space of an A oo

only if it is the zero

ring spectrum. n X to be the pullback in the

F o r a grouplike ( ~, ~ )-space X, define F M following diagram, where

d denotes the discretization m a p . FMX >MX

n

n

GL(n, That is, FM n X is the space

w0X )

• Mn(=0X in M n X.

) The I% write notation is chosen GLn(X for ) FM n in X. If
FM X n

of unit components Z. 4; Waldhausen
GL(n,X);

analogy X

with that in Remarks

would
for general

is a discrete

ring then FM X = n

topological

rings,

is larger than GL(n,X). for non-triviality.

We

reiterate that 0 and i m u s t be in different c o m p o n e n t s vn m a p s FMnX into F M n + I X . If each Vn is a co-

Clearly

fibration, w e let F M c o X be their telescope. functorial delooping. Theorem connected
FM X. (3o

denote the union of the F M n X ;

otherwise w e let F M o o X has a

It is the purpose of this section to prove that F M o o X

7.1. spaces

There is a functor

T f r o m grouplike weak

(~, ~)-spaces to between ~ T X and

based

together i t h a natural w

equivalence

In particular,
subgroup. necessary, Replacing TX

WlTX = w0FMcoX
by a naturally

= GL(co,
weakly

~0 X)

has a perfect

commutator
if [Zl,

equivalent

CW-complex

we can take its plus construction

in t h e s e n s e o f Q u i l l e n ( s e e e . g .

§ I]). W e have the following definition of the algebraic K-theory of Aoo ring spaces.

272

Definition 7. Z. i induces an i s o m o r p h i s m

Let

i: T X -~ K X

be the plus construction of T X , EX

so that

on h o m o l o g y and K X is a simple space. For q > 0, let K q

is called K X q is

the connected algebraic K - s p a c e of X. called the qth algebraic K - g r o u p of X.

X = Wq(I<X).

Remarks M X = Xn n

7.3. Z

A s a space, F M

n

X

is just the union of s o m e of the c o m p o n e n t s of is equivalent to t h e c o m p o n e n t SFM X n as an

and each of these

components FMnX

of the identity matrix. H-space, component to S F M n X

Indeed,

is equivalent,

a l t h o u g h n o t in g e n e r a l

XGL(n'w0X)

[Z,I.4.6].

Further, S F M n X N'InX 0 w h e r e

is equivalent to the X 0 is the comIt

of the zero matrix,

and the latter is just and X oo

ponent of zero in X (and MnX follows that, for q > 0, w F M q ~J; of course, 7.4. i. maps

MnX 0 are additive

infinite loop spaces).

is the direct s u m of infinitely m a n y copies of to Kq+lX. , the proof of T h e o r e m 7.1 will

this group naturally

Remarks

B y restriction tothe spaces UT

SFMnX

yield a functor

f r o m grouplike ( ~ , /J)-spaces to simply connected based f2UTX and

spaces together with a natural w e a k equivalence between SFMooX = Tel S F M n X and a natural m a p UTX UTX -~ T X

compatible with the w e a k

equivalences.

Thus, homotopically,

will be the universal cover of T X . 7. i. Let We 4

T h e rest of this section will be devoted to the proof of T h e o r e m begin by reviewing the basic theory of classifying spaces of Aoo spaces. be an Aoo operad. T h e r e is a functor V from

/~-spaces to topological m o n o i d s

specified in t e r m s of the two-sided bar construction of [I0,§9] by V X = B ( M , G , X ) . Here tion M denotes the free monoid, or J a m e s induces a m a p of m o n a d s construction, m o n a d . The augmenta-

6: ~ -~ ~

6: G -~ M , and w e obtain a natural

d i a g r a m of
(i)

~-maps
X , £ B(G,G,X) B6 ~ B(M,G,X) = VX .

273

Here

% is a h o m o t o p y equivalence with a natural h o m o t o p y inverse.

The m a p

B6

is also a h o m o t o p y equivalence (by [ll,A.g(ii)], tion of [I0,13.5] is removed). If X

in which the connectivity a s s u m p ~-space

is itself a m o n o i d considered as a

by pullback along 6, there is a natural composite
(2) VX = B(M,G,X) B8 > . B ( M , M , X ) and a homotopy 8 ) X If X = GY for a based

which is both a map of monoids

equivalence.

space

Y, there is a natural equivalence of m o n o i d s of Aco operads.

V X -~ M Y .

Moreover,

all of

this is natural with respect to m a p s We can therefore deloop

See [10,13.5] for details.

2~ -spaces by applying the standard productB of [17 or 9,§7-8] to V X . F o r our ~ :X -~ f~BX

preserving classifying space functor purposes,

the crucial property of B is that there is a natural m a p X such that ~ is a w e a k equivalence if ~0 X

for monoids

is a group (e.g.[9,8.7]). and

Turning to the proof of T h e o r e m construct Aco operads ~n

7.1, fix an Aoo operad pair ( ~ , ~ ) to obtain

as in section 4 (either crossing with 7 ~

actual operads, with permutations, n o n - E operads). ing X C(1)= by WX

or rephrasing the a r g u m e n t s above in t e r m s of

B y use of the a r g u m e n t at the end of the previous section, replacif necessary, we may a s s u m e without loss of generality that

{I}. Clearly T h e o r e m s 4. 1 and 6. 1 imply that F M FMnX n X is a sub ~)~n-Space of FMnX is V

MnX an

and that

Wn

restricts to an 5~n+ l - m a p

-~ F M n + I X, w h e r e Write Vn

~)~n+l-space by pullback along

-rn" ~/n+l -~ ~ n " ~n"

for the functor

above defined with respect to the operad

Consider the following m a p s ,

where

the notation on the left abbreviates that on the right.

274

Vn F M n X Tn

B ( M , Hn, F M n X )

V n + 1F M n X

T

B ( M , Hn+ 1 , F M n X )

T

B(1, Tn, l)

Vn+IVn I Vn+IFIVin+l x

] B(l'l'v n) B(M, Hn+ 1, FIVIn+I X )

Again, in order to ensure that all simplicial spaces involved in our constructions are proper [I0, ii.2 and II,A. 5], we should a s s u m e or arrange that In~ F M n X and 1 c ~{n(1) are nondegenerate basepoints before applying the With this precaution, w e have the

functors Vn and B; see [10, p. 127 and 167-171]. following result. Lemma
Therefore B'r n

7.5.

"rn:V n + I F M n X

-~ V n F M n X

is a homotopy equivalence.

is a l s o a h o m o t o p y e q u i v a l e n c e .

Proof. any ~n-Space

B y [ll,A.Z(ii)], ~n:Hn+IY -~ H n Y Y.

is a homotopy equivalence for

B y an argument just like the proof of [1 i, A. 4 (see 16, 5.5 and is a homotopy equivalence for any homotopy equivalence B(I,Tn, I) is a homotopy equivalence by [ll,A.4(ii)], and

5.6)], l~f:NiX -~ M X ' f:X -bX,. Therefore

the conclusion for B"r follows from the s a m e result; see [9, p.32]. n -i At this point, w e could choose a homotopy inverse (B~rn) to BTn
let TX b e t h e t e l e s c o p e of t h e s p a c e s Certainly BV F M X. n n (B'rn)-1

and

H o w e v e r , we w o u l d t h e n r u n

into a naturality problem.

is n a t u r a l up to h o m o t o p y s i n c e

B~rn is natural (by a trivial formal argument), but functoriality up to homotopy of
the telescope would require (B'rn)'l to be n a t u r a l up to n a t u r a l h o m o t o p y . In f a c t , (BTn)-1

t r a c i n g v e r y c a r e f u l l y t h r o u g h the p r o o f s c i t e d a b o v e , one m a y c h e c k t h a t

c a n b e so c h o s e n , b u t t h e r e i s a m u c h s i m p l e r a n d m o r e p r e c i s e s o l u t i o n to t h e

problem.

275

Definition 7.6. double m a p p i n g cylinder BVFMnX

Construct

TX

as follows.

Let

T X n

be the reduced

U
BT n

.(BVn+IFMnX

A I+)- l 1 BV+IW n

BV

n+l

FM

n+l

X

and let T X BVFMnX

be obtained f r o m the disjoint union of the T X n Tn_IX with the bottom, BVFMnX, from of TX

by identifying the top, for n > 1. Clearly based T

, of

is then the object function of a funetor spaces. Visibly, TX is homotopy

( ~ , /~)-spaces

to c o n n e c t e d

equivalent

to t h e t e l e s c o p e

of the BV FM X n n

with

respect to composites

( B V n +i Vn) (gTn) - i , any questions of naturality being thrown T h e properties of T stated in

irrelevantly onto the choice of equivalence. Theorem above.

7. i are i m m e d i a t e f r o m the definition and the general theory discussed

Remarks

7.7.

O n e w o u l d like to construct a product

TX

X T X -~ T X

(not an

H - s p a c e structure of course) by use of block s u m of matrices so as to be able to obtain an H - s p a c e structure on case w h e n X is a discrete ring. I~X by m i m i c r y of W a g o n e r ' s proof [Zl,§l] in the 6.4.

T h e m a i n obstruction is explained in R e m a r k s

§8.

Monomial

matrices and Q 0 ( B F K

II {0})

A s before, let ( ~ , ]-~ ) be an Aoo o p e r a d pair and let X be a grouplike ( ~ , ~ )-space. SFX Let Let FX = FM1X of I c X. be the space of unit components Then FX and SFX are sub of X and let of X.

be the component V

~-spaces

be the functor of the previous BFX = BVFX as a .~ - s p a c e

section,

d e f i n e d w i t h r e s p e c t t o 1/ , a n d If F X is a topological space monoid

abbreviate regarded

and similarly by pullback,

for SFX.

then this classifying

a g r e e s u p to

276

natural equivalence with the standard one ably for the union of a space

Write

Z II {0}

or

Z +

interchange-

Z and a disjoint basepoint 0. Let Q0 Y

( W e agree n e v e r to use

the + notation for the plus construction.) in the space QYo

denote the c o m p o n e n t of zero

T h e purpose of this section is to construct a natural m a p

~:Oo(Brxll~o))

-

Kx

and thus a natural transformation f r o m the stable h o m o t o p y groups of B F X + to the algebraic K - g r o u p s of X. the stable s t e m .s = S s 0 " q q Of course, ~ S ( B F X ÷ ) q When is the direct s u m of w S B F x q and

X = Q S 0' such a m a p

w a s asserted to exist by

W a l d h a u s e n [22, § Z]. T h e construction is b a s e d on the use of m o n o m i a l Definition 8.1. of the m o n o m i a l Let F X n matrices. X w h i c h consists

denote the subspace of F M

n

matrices with entries in F X ,

n a m e l y those matrices with pre-

cisely one n o n - z e r o entry in each r o w and c o l u m n and all n o n - z e r o entries in F X . Let maps F X denote the telescope of the F X with respect to the restrictions of the n Similarly, let S F X n and let S F co denote the space of m o n o m i a l m a t r i c e s with

oo

v n.

entries in S F X

X = Tel S F X. n it is i m m e d i a t e f r o m f o r m u l a (4.1) that X. F X n

A s will b e c o m e and S F X n are sub ~

clear below, -spaces of F M

n

n

T h e a r g u m e n t s of the previous section X replaced by F X. n T h e r e results the

can be carried out w o r d for w o r d with F M following analog of T h e o r e m Theorem 8.2. 7.1.

n

T h e r e is a functor

P

f r o m grouplike ( ~ ,

/J )-spaces to ~PX

connected b a s e d spaces together with a natural w e a k equivalence b e t w e e n and F X. Moreover, there is a natural m a p P X -* T X F

oo

the loop of w h i c h agrees X.

under the w e a k equivalences withthe inclusion

co

X -b F M

co

277

In p a r t i c u l a r ,
apparent that Ir0FnX ~rlPX

~ I P X = W o % X = c o l i r n woFnX.
is the wreath product ~nf ~r0FX.

It w i l l s o o n b e c o m e
It follows easily that

has a perfect c o m m u t a t o r

subgroup (see [7, 1,2]), and this will also drop out isomorphism from PX to

of our a r g u m e n t s below since they will give a h o m o l o g y the simple space Q0(BFX II {0}).

This h o m o l o g y i s o m o r p h i s m

will immediately

imply the following theorem. Theorem Q0(BFXJI{0}), 8.3. T h e plus construction on P X -* T X PX is naturally equivalent to

hence the m a p
{0}) -*

induces a natural (up to homotopy) m a p

~:~o(B~Xli
Remarks 8.4.

Kx.
FnX ~2SPX by SF X n SF oo throughout, w e obtain X, and a natural m a p SPX, a natural comis

Replacing

w e a k equivalence between

and

S P X ~- P X

patible with the w e a k equivalences. naturally equivalent to Q 0 ( B S F X KX agrees with the restriction of 8.5.

Moreover,

the plus construction on S P X

~{0}) ~.

and the resulting m a p

f r o m this space to

Remarks

There is a natural inclusion

7,FX -- B F X

(adjbint to Via the

: F X -- f~BFX), hence ~ restricts to a natural m a p basepoint of ~ F X , remarks

Q0(~,FX J-I{0}) ~" K X .

there is a further natural restriction Q~S0 -* K X . replaced by SF.

Similar

hold with F

T h e rest of this section will be devoted to the proof of T h e o r e m

8.3,

We

begin with a well-known observation about the classifying spaces of wreath products and an equally well-known consequence of the Barratt-Quillen theorem. Lemma m o r p h i c to E ~
n

8. 6. X n

F o r a topological m o n o i d (BX) n.

X,

B(~nfX

) is naturally h o m e o -

278

Proof.

Our conventions on wreath productsare in [2, p.51].

B

and E are

obtained

as geometric

realizations homeomorphism ¢, : E,(Gn~X

of certain

simplicial spaces

spaces

[9, p. 31].

We

define

a EjX-equivariant

of simplicial

) -~ E,52 n X (E,X) n

by the formula

Oq[(o~, h) ..... (O-q, y~)l(¢q+~, yq+t) : ([¢1 .....
where

Cq]°q+l' ~) '

cr ~ ~n' Yi = (xii..... Xin) , X n , and z , ( E X ) n has "th coo rdinate i I -I 4. is given by

[xl,0.Z... ~q+i(i )..... Xq, trq+i(i)]Xq+i,i;

~qi([~

i .....

O-q]O-q+i, [Xi i . . . . .

Xqt ]Xq+i,i

.....

[Xin .....

Xqn]Xq+i,n)

= [(o-:, y:) ..... (o-q, yq)](:q+:, yq+1) ,
where Yi = (x -i -i -t i, 0-q+i 0-q • • • 0-i+i(I ) . . . .x . -i -t -i ) " i, Uq+lUq • • . 0-i+I (n)

The conclusion

follows since realization c o m m u t e s with products. For a based space Y, the inclusion of ~ n fixing the last letter and the inclusion of y n inclusion E E n X E n y n in yn+i in ~n+l as yn as the subgroup X {*} induce an

-- E E n + i × Y n + i Y n + i •

Proposition 8.7. isomorphism Tel E>nnX ~

For connected spaces

Y, there is a natural homology

Proof. is precisely

y n -~ Q0(yjj_{0}). n For the E operad ~ of [10,§4 and IZ, §Z and 4], the space m Y + oo E Y n X z yn. n Let C = ~ X ~oo (or ~ X ~oo if one prefers

il n>0

to avoid partial operads) as in section 3, and note that the projection C Y +-* D Y + is a homotopy equivalence b y [li,A.Z(ii)]. By Proposition 3.5, with multiplicative structure ignored, Q Y + follows as in [2, I. 5. I0]. is naturally a group completion of G Y +. The conclusion

279

W e shall reduce Theorem 8.3 to an application of the previous two results. For this purpose, we require an understanding of ~ n f X
rather

when X

is a ~-space

than

a

monoid.
For a ~ - s p a c e X, d e f i n e a ~ - s p a c e

D e f i n i t i o n 8.8.

mnYXa s

follows.

As a space, Z n ; X = Z n X X n. The action ~n of ~J on ~ J X

is given by the maps

~nj:
specified for g ¢ ~(j),
~nj(g;et, x I .....

~(J) X(Z n x x n ) j -

zn x x n

0-q~ Y'n' and Xq = (Xq,1..... Xq,n) ~ X n by n
¢j,xj)

j

= (0-t -o.0.,3 i=1 ~J(g;q><=i Xq, 0.q+ t ' ' ' ° - j ( i ) ). × ~/ is t a k e n as a n o n - ~ o p e r a d ;

( T e c h n i c a l l y , this f o r m u l a is a p p r o p r i a t e when c o m p a r e s e c t i o n 4.) parametrized

T h i s is j u s t the o r d i n a r y i t e r a t e d w r e a t h p r o d u c t , but w i t h a Let Wn: Z n ~ X ~ ~ , n + t f X jt~ - s p a c e s . The f o l l o w i n g denote the

f a m i l y of m u l t i p l i c a t i o n s on X. u n

n a t u r a l i n c l u s i o n , and o b s e r v e that

is a m a p of

We w i s h to c o m m u t e the f u n c t o r

V past wreath products.

r a t h e r e l a b o r a t e f o r m a l a r g u m e n t b a s e d on the m a p s d i s p l a y e d in (7. t) and (7.2) suffices. Lemma 8.9. For ~ J - s p a c e s X, the h o r i z o n t a l a r r o w s a r e h o m o t o p y and /~ -maps,

e q u i v a l e n c e s in the f o l l o w i n g c o m m u t a t i v e d i a g r a m of ~ - s p a c e s where UX = B ( G , G , X ) .

Sa
v(z Sx) , B8
VVn]

U(ZnfX)
Wv n

fUX n n Vn

~nfB6 ~ ~ S V X n IVn ~n+l sB6

V(~n+I/X ) . B~

Zn+I ] u x

* Zn+~VX

Application of the functor V to this diagram gives a commutative diagram which can be extended to the left and right by the commutative diagrams

280

V(~'n] X)*

V nl

~°B6

VV(~]JX)

and

V(~Zn~VX )

£0B6

-""~.

j[VX

1

n" %p n ~ ~n+ifV X

V(Zn+fSX) ~

VV(Zn+IfX)

goB8 V(Zn+IJVX)

The resulting composite diagram is a commutative diagram of m a p s of topological
monoids in which all horizontal arrows are homotopy equivalences.

Via the homotopy invariance properties of the classifying space functor B [9,7.3(ii)] and the telescope, the previous result implies a chain of natural equivalences of telescopes which, together with the first two results above, leads to the following conclusion. T h e o r e m 8.10. F o r .~ -spaces X,there is a natural homotopy equivalence ~ Tel E ~ n × ~
n

Tel B V ( ~ n S X ) ----Tel B ( ) n J V X )

(BVX)n "

Therefore there is a natural homology isomorphism

Tel BV(Zn~X) -- ~0(BVX il{0})
W e can n o w prove T h e o r e m 8.3. B y application of the previous theorem to the ~ -space F X , with B F X = B V F X by notational convention, it suffices to

prove the following result. The proof again m a k e s strong use of the assumption ~(i) = {i} justified at the end of section 6. T h e o r e m 8.11. F o r grouplike ( C , ~ ) - s p a c e s X, there is a natural h o m o topy equivalence P X -~ Tel BV(SZnIFX ). Proof. Define a h o m e o m o r p h i s m matrix x to (0-,x(1) ..... x(n)), where x(i,j) / 0 a n d w h e r e o~ :F X "~ ~ ] F X n n n n by sending a m o n o m i a l

~, Z

is specified by o-(i)= j if be the evident

x(j)= x(i,j)for this i. Let ~n" ~)~n -~ ~

projection of operads and regard ~ n ~ F X W e claim first that a n is an ~ n - m a p

as an ~ n - S p a c e by pullback along ~n"

or, equivalently, that the actions

281

a n- i ~nWnHn~n and ~n c o i n c i d e on F X. n i~.i~j,

R e c a l l f o r m u l a (4. t). a unique If this

Given x. c F X f o r 1 n

t h e r e is f o r e a c h r b e t w e e n t a n d n such that

s b e t w e e n t and n a n d a U is the k TM e l e m e n t of

unique U c T ( r , s , j )
T(r,s,j) and if

Xq(Uq_t,Uq) ~ 0.

Sn,j,k = (* ..... *,i,* ..... *) E C(O) k-I X C(1)× then, f o r c c ~(n j-t) and g ~ ~(j),

(01nj-l - k

J
C n , j ( c , g ; x I. . . . . x j ) ( r , s ) = Ol(~(c ; s n , j , k ) ; ~ j ( g ; X
• q=l

Xq(Uq_i,Uq)))

J
= ~i(g; × X q ( U q _ i , U q ) ) , q=i

the last equality holding since ~(C;Sn,j,k) = i ( ~ ( i ) follows by comparison with Definition 8.8. B V F nX n
B'rn l BVn+ 1an BVn+IFnX B%+ iVn ~" BVn+I(Zn~FX )

by assumption.

The claim

Now consider the following diagram. Bw

BV ~ n n. .

BVn(~nfFX ) BT n

n

.BV(~ni[X

)

BWn+ i

> BV(Zn~FX ) ] BVUn m/rn+l -"mv(152~n+ ~X) l

IBVn+iWn B V + I °In+l~"B V + I (Zn÷l~ FX)

BVn+iFn+IX

The left horizontal arrows are homeomorphisms and the right horizontal arrows are homotopy equivalences by [ll,A.2(ii) and A.4(ii)]. The diagram commutes by
n a t u r a l i t y and the f a c t s that natural equivalence w n Tn = Wn+l and Vna n = an+lVn . The r e q u i r e d

PX ~ T e l B V ( E n S F X ) f o l l o w s by p a s s a g e to r e d u c e d double

mapping cylinders and then to unions as in Definition 7.6.

282

§ 9.

S o m e homotopical and homolo~ical properties of K X . Again, let ( ~ , /J) be an Aoo operad pair and let X be a grouplike

( ~,

~)-space.

From

the point of view of analysis of its invariants, the

sophisticated functors V n (and W ) w h i c h entered into the construction of K X are of no significance. T h e y simply replace a given structured space by a h o m o t o p y Thus, up to homotopy, only the classiThese

equivalent space with different structure.

fying space functor, the telescope, and the plus construction are involved. facts and R e m a r k s

7.3 imply that the analysis of K X is considerably less refractory

than the complicated theory n e c e s s a r y for its construction w o u l d suggest. W e begin with two elementary h o m o t o p y i n v a r i a n c e properties, w h i c h will

be seen later to be simultaneous generalizations of Waldhausen's assertions [2Z, i.I and 2.3] and [ZZ, 1.3 and Z.4]. Recall that a m a p isomorphism for i < n f: X -* Y is said to be an n-equivalence if ir.f is an
i

and an e p i m o r p h i s m

for i = n for all choices of basepoint

in X (and analogously for m a p s Proposition 9.1. n-equivalence, then Proof. Theorem also Kf

of pairs and for n - h o m o l o g y equivalences ). is a m a p of grouplike ( ~, ~ )-spaces and an

If f:X ~ Y

Kf" K X -* K Y

is an (n+l)-equivalence. -~ F M 0 o Y is an n-equivalence. Therefore Tf By

By Remarks

7.3, F M o o f : F M o o X

7.i, T f : T X -* T Y

is thus an (n+l)-equivalence. Since

and thus

are (n+l)-homology equivalences.

K X and K Y

are simple spaces,

the conclusion follows by the W h i t e h e a d theorem. W e next want the relative version of this result, and w e n e e d s o m e preliminaries in o r d e r to take account of the non-existence of an unstable relative Whitehead t h e o r e m and to handle s o m e technical points ubiquitously ignored in the literature. Consider a h o m o t o p y c o m m u t a t i v e d i a g r a m of spaces

283

X

f

>Y

(~)
Z
Definitions 9. Z. is an m - e q u i v a l e n c e and g The diagram

f,

,,, ~ W
if f

(*) is said to be an (m, n)-equivalence It is said to be q - h o m o t o p y

is an n-equivalence. of triads

Cartesion if there exists a m a p

"f : (f;f',f):(Mg;Z,X) such that the map of pairs and Mg' f - : ( M g , X) ~ ( M g ' , Y )

-~ ( M g ' ; W , Y ) is a q - e q u i v a l e n c e , If (*) c o m m u t e s , where Mg

denote the mapping cylinders required,

of g and g'.

w i t h no

homotopy

w e i n s i s t t h a t t h i s c o n d i t i o n b e s a t i s f i e d w i t h T(x, t) = (fx, t) o n that the natural map Fg ~ Fg' (by t h e of

the cylinder,

a n d it is t h e n e q u i v a l e n t to r e q u i r e

homotopy fibres be a (q+l)-equivatence standard verification

f o r e a c h c h o i c e of b a s e p o i n t in X

t h a t t h e t w o d e f i n i t i o n s of t h e r e l a t i v e h o m o t o p y

groups of a

map agree), Remarks triads 9.3. In t h e g e n e r a l case, with based spaces and maps, the map of

f- i n d u c e s a m a p
f:Fg = X X g PZ
f

-* Y X g , P W

= Fg'

via

f'~(zt)
"~(x,~) = (fx, ca), where ~o(t) = rF(x, Zt-1)
with then ~ to m e , r: M g ' -~ W

if

OKt~
1

I

if ~ < _ t ~ l
If T is a q-equivalence of pairs

b e i n g the canonical retraction.

is a (q+l)-equivalence.

A converse construction is not i m m e d i a t e l y obvious Clearly a

and the definition has been given in the fo~rm w e w i s h to use. h:f'g --~ g'f induces a m a p

homotopy

of triads of the sort specified, via

284

I

h ( x , Zt)

if

1 0<_t<7
1

'

T(~, t) =

(fx, Zt-1)

if 7 Kt<_ 1

but whether or not f- is a q-equivalence of pairs really does depend on the choice of homotopy. Lemma n >__i. For 9.4. Assume that (*) is an (m,n)-equivalence, where m > 0 and

q < m+n,

(*) is q - h o m o t o p y Cartesian if and only if there exists a -~ W suchthat Ck ~--f' and @j ~ g', w h e r e and M(g,f) is are

q-equivalence

@:M(g,f)

the double m a p p i n g cylinder of g and the natural inclusions. If (*) c o m m u t e s ,

f and

k'Z-~M(g,f)

j:Y-~ M(g,f) factoring

¢ m u s t be the natural m a p

through the quotient m a p to the pushout of f and g. Proof. T h e last statement will be a consequence of the conventions in the natural m a p

Definitions 9.2. B y the h o m o t o p y excision theorem,

(Mg, X) -- (M(g,f),Y)
Clearly m a p s

is an (m+n)-equivalence.

(The range is misstated in [2Z].)

~ : M g -~ Mg'

as in Definitions 9. Z factor uniquely through m a p s ~ (Mg';W,Y). @k = f' and #j = g. Conversely,

¢ = (¢; f',l ): (M(g, f); Z, Y) If r:Mg' --~W given is the retraction and let

r@ = @, then

@ as in the statement, Since

@ be the composite of @ and the inclusion is a cofibration, @ is homotopic to a m a p is a of

i: W -~ Mg'.

y If Z -~ Mg'

triads ~ as displayed.

B y t h e five l e m m a , ~: M(g~f) -~ Mg'

~:(M(g,f),Y) -~ (Mg',Y) is a q-equivalence.

q-equivalence if and only if follows. Proposition 9.5.

T h e conclusion

If ($) is a (strictly) c o m m u t a t i v e

d i a g r a m of grouplike with

( ~ ' , ,~ )-spaces which is a (q-l)-homotopy Cartesian (m-l,n-l)-equivalence
m >__Z, n > Z, and q<m+n, then

285

KX

Kf

>

KY

KZ is
a q - h o m o t o p y Cartesian (m, n)-equivalence. Proof. By Remarks

>

KW

7.3, application of F M
oo

to (*) gives a (q-l)7.1 and a little standard

h o m o t o p y Cartesian (m-l,n-l)-equivalence.

By Theorem

a r g u m e n t with h o m o t o p y fibres, the (strictly) c o m m u t a t i v e d i a g r a m TX Tg Tf ~ TY

i
TZ

[Tg' Tf'
> TW

is a q - h o m o t o p y Cartesian (m,n)-equivalence. i
x

Consider the following d i a g r a m

Kf y i K

TX Tf

~ - T y ~

/
Tf) / m ,

M(Tg,

i

I /

Kf'

"~

Kz
TZ

___~ KW

/
•- T W "-------

i

Tf'

w

Breaking the cylinder of M ( T g , Tf) into three parts, m a p p i n g the middle third T X X [ i/3,Z/3] to K X X [0, i] via i
x

and expansion, and using homotopies

i oTg --~ K g . i
z a map

x

and Kfoi

x

-- i ~Tf on TX X [0,1/3] and TX X [2/3,1], we obtain
y such that ij = jly and ik = ki z on the bases. By

i:M(Tg, Tf) -- M ( K g , Kf)

the van K a m p e n on wli ~rI since m

t h e o r e m and the fact that Tf, Tg, Kf, and K g > 2 and n>_ 2, ~rlM(Tg, Tf) = ~rlTX,

induce i s o m o r p h i s m s

~rlM(Kg, Kf) = HI T X , and i is

is Abelianization.

B y the Mayer-Vietoris sequence and the five l e m m a ,

286

a homology isomorphism. M(Tg, Tf).

Therefore

i is equivalent to the plus construction on B y the universal property unique up to homotopy, Since

In particular, M(Kg, Kf) is a simple space.

of the plus construction, there is a m a p such that #i ~ iw~0, w h e r e ~jiy 9J -----Kg' =

9: M ( K g , Kf) -~ K W ,

<~ : M(Tg, Tf) -* T W ~ij

is the natural map. -- K g ' o i y ,

-- i w ~ j = iwTg' Similarly

by the universal property.

9k ~ kf'. Since ~ is a Therefore 9 is a The

q-equivalence (by the l e m m a ) ,

it is a q - h o m o l o g y equivalence.

q - h o m o l o g y equivalence and thus a q-equivalence by the Whitehead theorem. conclusion follows f r o m the l e m m a . T h e proofs above have the following useful consequence. Lemma 9.6. If (~) is a c o m m u t a t i v e

d i a g r a m of grouplike (~,/J)-spaces, ~: F K g --~ F K g ' which m a k e s

then there is a canonical h o m o t o p y class of m a p s the following d i a g r a m h o m o t o p y c o m m u t a t i v e w h e n (e) is regarded as a m o r p h i s m grouplike (~ , ~ ) - s p a c e s

and which is functorial up to h o m o t o p y of

(f,f'):g -~ g' in the category of m a p s

and is natural up to h o m o t o p y with respect to m o r p h i s m s

of such d i a g r a m s (*).
~KZ ~Kf' FKg ~ KX Kg KZ

f2KW

FKg'

J, I<Y

Kg' ~ K W

H e r e the unlabeled a r r o w s are the natural m a p s sequences. Proof. Of course, Barratt-Puppe

of the displayed fibration

sequence arguments

give a m a p

~, not

uniquely determined up to homotopy. as in the previous proof, d e f o r m

W e ignore this. Construct -~ Mg'

9: M ( K g , Kf) -~ K W

i9, i: K W

to a m a p of triads

= ( ~ ; K f ' , 1):(M(Kg, Kf);KZ, KY) -" (Mg'~KW, KY)

287

as in the proof of L e m m a

9.4, let f" be the composite of ~ and the natural m a p as in R e m a r k s 9.3. It is simple to check (by

M K g -* M(Kg, Kf), and construct

standard cofibration arguments f)r the passage f r o m ¢ to ~) that ~ is a w e l l d e f i n e d h o m o t o p y c l a s s w h i c h m a k e s the d i s p l a y e d d i a g r a m h o m o t o p y c o m m u t e . I t s f u n e t o r i a l i t y a n d n a t u r a l i t y a r e t h e n e a s i l y v e r i f i e d by t h e s a m e s o r t s o f h o m o topical arguments as those above.

T u r n i n g to h o m o l o g y , we f i r s t r e c o r d t h e f o r m t a k e n in o u r c o n t e x t b y t h e

standard spectral sequence for t h e calculation of the homology of classifying spaces of topological monoids. Since Vn: F M n X -* F M n + I X only c o m m u t e s up to homotopy with the multian H-space structure.

plications, there is a slight ambiguity in giving F M c o X

There is no ambiguity in its Pontryagin product, however, and the spectral sequences of the filtered spaces result. (See e.g. [9, 13. i0].) Proposition 9.7. Take homology with coefficients in a field k. {ERE} There is BV FM X n n pass to limits to give the following

then a natural spectral sequence H , F Moo X from EZx = Tor

of differential coalgebras which converges

(k,k) to H , K X .

Field coefficients serve only to yield a conceptual description of EZx.
Since

i: T X "4" E X

induces an i s o m o r p h i s m on k, for any connective homology

theory k. (by the Atiyah-Hirzebruch spectral sequence), w e obtain a spectral sequence {ErX} converging to k , K X with E 1 X = kq((FMcoX)P ) for any such k,, Pq

d I being induced by application of k, to the alternating s u m of the standard bar construction face maps. (See [i0, II.14] for details.)

The proposition focuses attention on the problem of computing the P0ntryagin algebras H , F M n X . It is clear f r o m previous experience what

288

procedures one should adopt: one should w o r k in all of H , M n X and exploit the Z d i a g r a m of Proposition 5.2. Since M n X = X n is an (additive) infinite loop space, all of the m a c h i n e r y of h o m o l o g y operations explained in [g,I] is available. Assuming coproduct that H . M n X is understood additively (as a Hopf algebra with product X, augmentation ~ , h o m o l o g y operations QS and **

~ , conjugation

Steenrod operations

p r with ,,

P, r = Sq, r at the prime 2), w e can study its products

by use of the following two results, the proofs of which are exactly the s a m e as those in [2, p. 79-81 ]. Proposition 9.8. x,y,z , H , MnX, Take h o m o l o g y with coefficients in any field. Let

n >__ i, and let [0],[i], and [-i] be the classes of the zero matrix, In, and any matrix in the component additively inverse to that

the identity matrix of I . n (i) (ii)

[O]x = (£x)[O],

[l]x

= x,

and

[-l]x=×x. , where %bz = ~ z ' • z".

(x ~' y)z = ~

( _ l ) d e g y d e g z' xz'* yz"

Proposition 9.9. any prime.

Take h o m o l o g y with m o d p coefficients, Then

where

p is

Let x,y ~ H , M n X , n > i.

(Qsx>y
(~C~%)y

=

~
i

Q

s+i. i

(xP.y)

and, if p > Z ,

= 7 . ~ Q s+i (xP, y) - ~ ( - l ) deg x Q s+ii i (xP,~y). i i

Remarks 9.10. is an A algebra oo

A p p l i e d to the Aoo r i n g s p a c e s

QX + of E x a m p l e s 2.3, w h e r e

X

s p a c e , the p r o p o s i t i o n s above c o m p l e t e l y d e t e r m i n e the P o n t r y a g i n H , F Q X + in t e r m s of the a d d i t i v e s t r u c t u r e of H , QX +. Compare [2,II§4],

w h e r e the a n a l o g o u s a s s e r t i o n for the Eco ring s p a c e space X is e x p l a i n e d in d e t a i l . To d e t e r m i n e

QX + o b t a i n e d f r o m an Eoo

H , F M n Q X + f o r n > 1, one would

289

require additional formulas to explain the effect on h o m o l o g y of translations by more general matrices than in Proposition 9.8(i).

§I0.

E x a m p l e s , n a t u r a l m a p s , and f o r m a l p r o p e r t i e s of KX
Recall f r o m section I that any Aoo operad pair (~ ,/J) admits an a u g m e n -

tation (~, ~ ) -~ ( ~ , ~

A n (~ ,]~ )-space is precisely a topological semi-ring is precisely a topological ring. If R Therefore a topois discrete, then

and a grouplike (~ , ~ ) - s p a c e logical ring R FM n

is a grouplike (~ , ~)-space by pullback.

R = GZ(n, R), and the following e x a m p l e is i m m e d i a t e f r o m the constructions

of section 7. Example I0.i. If R is a discrete ring, then K R BGL(oo, R). Therefore KqR, is naturally equivalent to the q > 0, is Quillen's qth alge-

plus construction on

braic K - g r o u p of R [5, 18]. F o r general topological rings, our theory reduces to the topological

v e r s i o n of W a l d h u a s e n ' s [22, § i ].
Example KR,. 10.2 . F o r simplicial rings R,, W a l d h a u s e n defined a certain functor

It would be i m m e d i a t e f r o m the definitions that his K R , is naturally IR, I is the geometric realization of R,, w e r e it

equivalent to our K I R , I, w h e r e

not that he has chosen to throw in a discrete factor Z and w e have not, so that his KR. is our K I R . I X Z.

Remarks

10.3. R

T h e r e is one vital distinction to be m a d e between and KX for an arbitrary grouplike Aoo ring space

KR X.

for a topoAs is well-

logical ring

k n o w n in the discrete case [20; 21; 12] and will be proven in general at the end of this section, K R that K X is an infinite loop space. In contrast, Isee little reason to think

is an infinite loop space and have not yet been able to prove that it is even although I believe this to be the case ( c o m p a r e Ren~arks 7.7).

an H-space,

290

It is immediate from the definitions that, for any ( ~ , ~ )-space X, the discretization m a p d:X -~ ~0 X is a m a p of (~, .~)-spaces. W h e n X is grouplike,

this has the following important consequence. proposition 10.4. F o r grouplike ( ~ , ~ )-spaces X, there is a natural augmentation Kd: K X -~ KIr0X. This suggests the following reduced variant of KX. Definition 10.5. Define l~fx 9.6, K I is then a functor of X mutative for any m a p f:X -~ Y
to

be the homotopy fibre of Kd°

By Lemma

such that the following diagram is homotopy c o m of grouplike (~ , ~)-spaces. Kd

~KTr0X

K'X

~'KX

> KTr0X

~KWoY

~'K'Y

~" K Y

Kd

~-KTr0Y

Of course, K'X

is trivial if X

is discrete.

W e can m i m i c this construction after restriction to m o n o m i a l matrices. The resulting functor m a y be described, up to equivalence, as follows. Definition i0.6. Define Q' (BFX+), 0 B F X + = B F X J_L {0}, to be the h o m o Fw0X is the g r o u p of

topy fibre of Q0(Bd+): Q0(BFX+) -~ Q0(BF~0X+), where
u n i t s in t h e r i n g =oX.

A s i n t h e p r e v i o u s d e f i n i t i o n , L e m m a 9 . 6 (or r a t h e r i t s

m o n o m i a l matrix analog) implies that this is the object function of a functor of X. Since SFIr0X = {I}, the corresponding functor obtained with F X is equivalent to Q o B S F X . At the risk of belaboring the obvious, w e combine the definitions above with T h e o r e m 8.3, R e m a r k s 8.5, and further applications of L e m m a logs thereof) in the following theorem. 9.6 (and anareplaced by S F X

291

Theorem

10.7.

F o r grouplike (C,

}j)-spaces

X, the following is a

h o m o t o p y commutative

d i a g r a m and is natural up to homotopy.

QoZSFX

-....
QoBSFX
b

Q" ( B F-X + ) 0

~ K'X

Qo(ZSFX+)

TI \ "

O0(ZFX +)

Q0(BSFX+) %(Bd ÷)
Q0 SO ~ .... _ ... .

,

Q0(BFX+)

~

~- K X

-~ Q0(BF~0 x÷)

~ ,_ K(,~0X)

T h e dotted a r r o w s denote the presence of evident sections, and the c o l u m n s are fibrations; ~' and w are obtained as in L e m m a 9.6 and the remaining horizontal

a r r o w s are inclusions.

T h e d i a g r a m c o u l d b e e x p a n d e d f u r t h e r (by u s e of Q0(EFw0X+), e t c . we desist. to t h e c a s e I n v i e w of i t s i m p o r t a n c e i n t h e a p p l i c a t i o n s , X = QS 0, w h e r e FX and SFX

),

but

we c a l l s p e c i a l a t t e n t i o n F a n d SF

are generally denoted F~r0X = Z 2.

( o r G a n d SG) a n d w h e r e

w0X = Z a n d t h u s

(Up to n o t a t i o n , t h e

r e s u l t i n g d i a g r a m is a n e l a b o r a t i o n of o n e c l a i m e d b y W a l d h a u s e n [22, a b o v e 2 . 3 ] . ) Here the bottom composite

Q0 s°
is induced by the inclusions standard m a p ~

~ Q0(BZ;)
n

~ ~Kz

-~ GL(n, Z) of permutation matrices and is thus the

studied by Quillen [19] (see also [12, Vl § 5 and Vlll. 3.6]); the m o n o ~t has been studied in [~Z, VI. 5.9, Vll. 4. 6 and VIII. 3.6] and also m a k e clear, this e x a m p l e is universal:

mial matrix m a p

in L o d a y [7]. As the following r e m a r k s

292

the d i a g r a m for
space X. i0.8.

QS 0 maps

naturally to that for X

for any grouplike A

Co

ring

Remarks

By Proposition

Z. Z a n d T h e o r e m

3.1, the zero

th

m a p of t h e u n i t of .Let 3

a n y Aeo r i n g s p e c t r u m us write from of A FX co

E is a map th

e : QS 0 ~ IEO of g r o u p l i k e ring spectrum

Aco r i n g s p a c e s .

for the zero ring space

space of the A If X

Co

obtained as in section weak equivalence we may

an A co

X.

is grouplike,

we have a natural 3.3.

ring spaces e

between X and FX by Theorem C~S0 ~ X .

By a slight abuse,

thus regard

as a unit map

Proposition 2. Z and T h e o r e m ring space for any A obtain Waldhausen's

3.1 also s h o w that We

QY

is a grouplike A

OD

Co

space with zero Y.

shall specialize this e x a m p l e to W e note here

algebraic K - t h e o r y of spaces in the next section.

that the following generalized version of Waldhausen's immediate Theorem consequence 7.1, R e m a r k s of Propositions 7.3,

assertion [22, 2. 2] is an 3.6 together with

3.5 and 3.7 and L e m m a

and Proposition 9.7. F o r any A space with zero Y, the H u r e w i c z map

Proposition I0.9. h: Q Y -~ F N Y Therefore is a m a p Kh: K Q Y

oo

of grouplike A

oo

ring spaces and a rational equivalence.

-~ K F N Y

is also a rational equivalence.

Remarks hence

I0. i0.

NS 0 and

is precisely the additive m o n o i d of non-negative integers, d: F N S 0 -- Z and therefore also Kd: K F N S 0 -~ K Z are KZ.

w0FNS 0 = Z

equivalences.

T h u s w e m a y v i e w the H u r e w i c z

m a p of K Q S 0 as having target

The deepest

source

of

oxam:ples

is the theory of E

co

ring spaces

and E

co

ring spectra. spectra. A co

B y neglect of structure, these are A the zero th

oo

ring spaces and A

o3

ring

F o r example,

spaces of all of the various T h o m

spectra are

ring spaces [12, IV §2].

Of greater interest are the e x a m p l e s

c o m i n g out of the

chain of functors constructed in [12, VI and VII]:

293

symmetric

bimonoidal categories

~
h

blow

up

bipe rmutative categories J[ classifying space E oo nng" spaces
li

V

][ recognition principle E
O0

rlng' spectra zero space

V

grouplike

oo

ring spaces bimonoidal category ~OR of finitely

F o r instance, starting with the s y m m e t r i c

generated projective m o d u l e s over a c o m m u t a t i v e at an associated grouplike E oo ring space, denoted

topological ring R, w e arrive FB~0R, in which addition c o m e s W e inter-

f r o m t h e direct s u m and multiplication c o m e s ject the following note ( c o m p a r e [ZZ, § 1 ]). Remarks i0. ii.

f r o m the tensor product.

W e are here faced with a conflict of definitions and notations. R (not necessarily commutative), I wrote K q R for K q R for

F o r a topological ring q FB~°R

in [10,VIII §l].

F o r discrete R, this agrees with the present I suggest writing KtR

q > 0. F o r general R, it is quite different. construction on B G L R and writing

for the plus

K t R = w KtR, thinking of this as a topological q q for example). In

K-theory (which it is w h e n R is the topologized c o m p l e x n u m b e r s practice, the functors of R. K.R and 14t, R

tend to be of interest for different choices

In fact, the theory sketched above applies equally well to both generalizations of Quillen's theory f r o m discrete rings to topological rings. ~R To see this, let as

denote the permutative category of finitely generated free R-modules, Define J~R

described explicitly in [IZ, VI. 5. Z].

to be the category defined in

294

precisely the s a m e way, except that the space of m o r p h i s m s cal m o n o i d FM n R.

n -* n is the topologi-~R

Then, with structure as specified in the cited definition,

is a permutative category and is a bipermutative category if R Moreover, the inclusion ~ZR -* ~ R

is commutative. T h e argu-

preserves all structure in sight.

m e n t used to prove that F o B ~ R

is equivalent to the plus construction on B G L R is equivalent to the plus [II,

in [IZ, VlIl §i] applies equally well to prove that F o B g I ~ construction on B F M C O R p. 85]). (and of course I~0B~2R

is equivalent to F o B ~ R

W e have proven the following result. Proposition 10.1Z. F o r a topological ring FB~R If R co and R, KtR and K R are the zero

components

of infinite loop spaces i:FB~R-~ ring spaces and

FB~R,

and there is a natural then I~B~2~R and

infinite loop m a p FB~R are E co

FBg~R.

is commutative, ring map.

i is an E

H e r e the additive infinite loop space structures associated to permutative categories are uniquely determined by the a x i o m s in [14]; in particular, Segal's m a c h i n e [20] and m i n e give equivalent spectra. At the m o m e n t , braic K-theories K.FB nothing is k n o w n about the resulting "second order" alge~R and K . F B ~ R of c o m m u t a t i v e topological rings. K,R or

T h e y do not appear to be very closely related to the "first order" theories Kt.R. Since w 0 F B ~.Z'R = w 0 F B ~ R KFB~fR

= Z, Proposition I0.4 and naturality give m a p s -~ K F B g ~ R -~ K Z -* K t R -* KR. Z, w e could

If w e had followed W a l d h a u s e n and crossed everything with consistently write KKtR = KFB~R and KKR = KFB9%R,

hence our view of these as second order theories.

295

§11.

The algebraic K-theory of spaces Finally, w e specialize our theory to m a k e rigorous the algebraic K-theory

of spaces suggested by Waldhausen [2Z]. In this section, X

will denote a non-

degenerately based GW-space with basepoint 1. There are at least three ways that
f~X can be interpreted as an Aoo-space. (I) The ordinary loop space of X is a ~l-space' where Aoo operad of [ i0, §4]. (2) (3) The M o o r e loop space of X is a topological monoid, or ~2-space. ~i is the little l-cubes

IGSXI, the geometric realization of the K a n loop group of the total singular complex of X is also an ")?~-space.
We lean towards

the first choice, but the theory

works

equally

well with

any choice.

As in E x a m p l e 2.3, w e construct from and thus a grouplike Aoo ring space

fiX an A Q(~X+).

co

space with zero

f2X + = ~x_[J.{0} space

F o r any based

Y,
~0QY = colim wn~n G n y = ¢olim~n~nY = Toy, It on passage to components.

this i s o m o r p h i s m being realized by h : Q Y -- F N Y follows easily Z[WlX].

that, as a

ring, w0Q(~X+ ) = w 0 F N ( ~ X + ) is the integral group ring

In reading the following definitions, it will be useful to keep in m i n d the Go ring spaces (see Propositions 3.7, i0.4, and

following commutative diagram of A 10.9 and R e m a r k s I0. I0). Qs o n

, Q ( ~ x +)

e

~, QS o

hi
FNS 0 9

[h
~ FN(f2X +) ~

lh
y I~NS 0

Z

q

~ ZDrlX]

e

>

Z

F o r uniformity of notation, we write ~q for m a p s induced by inclusions of basepoints of spaces or trivial subgroups of groups and write £ for the corresponding

296

projections,

so that

gN = i. d.

The vertical composites in the d i a g r a m are again N,E, d, and h upon application

disc retization m a p s of the functor K.

W e shall continue to write

Definitions 11.1. be KQ(~X+). Further,

Define

AX,

the connected algebraic K - s p a c e of X, to

define the following reduced variants. ~ A{*}), A{*} = K Q S 0 : K'Q(g2X +)
A'{*} : K ' ~ °

]~X = fibre (E : A X

A ' X = fibre (d " a m -~ KZ[WlX])
~ , x : fibre (~':A'X--A'{*}),

F o r q > 0, define

A X, the qth algebraic K - g r o u p of X, to be q

~ AX, q

and introduce

similar notation for the reduced variants. W e have an analogous algebraic K-theory of X with coefficients in Z. Definition II.Z. variants "~(X;Z) = fibre ( Z : A ( X ; Z ) -~ A(*;Z)), A(*;Z) : K Z : K'I~N(~X+). Define Define A(X; Z) : KFN(CiX +) and define the reduced

A'(X; Z) = fibre ( ~':A(X; Z) -~ KZ[~IX]) Here A'(*; Z) - - - -{*}, hence w e set ~'(X; Z) -- A'(X;Z),

Aq(X; Z) = w e ( X ;

Z) and similarly for the reduced variants. ring R, write

F o r a (discrete) group ~ and c o m m u t a t i v e KR[=] = fibre (~ :KR[~] By Lemma -" KR).

9.6, these are all well-defined functors of their variables such
We s u m m a r i z e

that the various canonical m a p s in sight are natural. ships between these functors in the following result.

the relation-

297

Theorem

II.3.

T h e r o w s and c o l u m n s are fibration sequences in the

following natural h o m o t o p y c o m m u t a t i v e diagram.
N

~'X

w

~x

d. . . . .

KZ[WlX ]

A'X

--

L AX

a'{*}

--

, a{*}

l 11

d .........

, KZ[wlX

]

1
sends this

d
-~ K z

T h e dotted a r r o w s denote the presence of sections.

The Hurewicz map

d i a g r a m naturally, via a rational equivalence, to the corresponding natural 3 X 3 d i a g r a m of fibration sequences with its variants. Proof. In view of our earlier results, only the construction of w and the This w o u l d be AX and its variants replaced by A ( X ; Z ) and

verification that it is equivalent to the fibre of ~ are needed. obvious enough if w e k n e w that A X w e r e an H-space.

In the absence of this, a

technical a r g u m e n t with Barratt-Puppe sequences, w h i c h w e defer to the appendix, is required. Of course, T h e o r e m 10.7 applies naturally to all Aoo ring spaces in sight In particular, w e have the m a p s F ~- A X ,

and produces a m a z e of c o m m u t a t i v e diagrams. O0 x + --~ O 0 ( B a X +) QO (Bn+)

Q o ( B F Q ( ~ X + ) ÷)

w h e r e X l is the c o m p o n e n t of the basepoint of X, X I --~B~2X by [9,14.3 or 15.4], N : ~ X -~ F Q ( ~ X +) is the natural A
CO

m a p of E x a m p l e

2.3, and ~

is the m o n o m i a l

matrix m a p of section 8.

This agrees with assertions of W a l d h a u s e n [Z2, §Z].

It iB desirable to have algebraic K-theories of spaces with coefficients in arbitrary c o m m u t a t i v e rings R. (For Z
n

For many

rings, this can he obtained topologically.

, compare

Dold and Thorn [3] to the material at the end of section 3. ) H o w -

ever it is m o s t natural to follow W a l d h a u s e n [ZZ,§l] and do this simplicially.

298

Definitions 4_1.4.

Define

A(X;R)

=

KIR[GSX] I,

w h e r e the group ring of

a simplicial group is f o r m e d degreewise.

Clearly

=0]R[OSX]l

= =oR[GSX] :

R[=oC, SX ] :

R [ = t X ].

Define reduced variants and algebraic K - g r o u p s as in Definitions ii.Z.

W e n o w have two definitions of A ( X ; Z ) above for

in sight.

With interpretation (3)

~X, the following result implies that these definitions agree up to natural

e quivalenc e. Proposition 11.5. equivalence of grouplike A Proof. F o r simplicial groups oo ring spaces b e t w e e n G, there is a natural w e a k r N I G I + and IZ[G]I.

N ] G I + is the free topological Abelian m o n o i d generated by IGI T h e inclusion G ~ Z[G] induces

(the disjoint basepoint being identified to zero). +

]GI

~ Pz[G]l

and thus, by f r e e n e s s ,

NIG I

-Iz[GII.

Moreover, this map
IGI, and it is

clearly extends over the free Abelian group ~ I G I + generated by easy to verify f r o m the fact that realization c o m m u t e s N IGJ + ~ IZtG]i is actually a h o m e o m o r p h i s m .

with products that

The natural m a p

× : N ] G I + -- N'IGI + ~
is clearly a m a p ( C , ~ ). Theorem

IZ[G]I
co operad pair By

of semi-rings, hence of ( ~ , .~ )-spaces for any A

B y Corollary 3.8 (recall t~at r = B 0 ), r × 3.3 and the fact that ]Z[G]I oo

is a w e a k equivalence.

is grouplike, F IZ[G]I

is naturally w e a k l y

equivalent ~s an A

ring space to IZ[G]I. Z -~ R, all of our d i a g r a m s r e m a i n present with as well as X. Z

In view of the unit m a p replaced by

R, and the resulting d i a g r a m s are natural in R

T h e sirnplicial approach has the advantage of giving us infinite loop space structures on our algebraic K-spaces, a simplicial group thus have G. When KR G and by Proposition 10.1Z applied to KIR[G]] IR[G]] = R. We for

is the trivial simplicial group, ~]:KR ~ K [ R [ G ] ] . Write

£..KIR[G]I-

KIR[G]I

= fibre

299
and give it the induced infinite loop structure. by a glance at h o m o t o p y groups. Proposition 1 1.6. the composite F o r simplicial groups G and c o m m u t a t i v e rings R, The following result is i m m e d i a t e

~bR[G]p ×KR
where

P× ~KIR[G]I ×KIR[G]I

~ -KIR[G]J,

p is the c a n o n i c a l m a p and

~ is the p r o d u c t , is a n a t u r a l e q u i v a l e n c e of

infinite loop s p a c e s .
T h e following is a special case. Corollary 11.7. A(X; R) is naturally equivalent as an infinite loop space

to ~ ( X ; R ) × KR. All of t h e s e a l g e b r a i c K - t h e o r i e s on s p a c e s a d m i t s t a b i l i z a t i o n s to generalized homology theories. T h e o r e m 11.8. There are natural homomorphisms
then A,

~ n X -~ ~n+l>~X

such

that if ~ S X q = colim ~ q + r E r X . are analogous theories

us

is a reduced h o m o l o g y theory.

There

~s A,(X; R)

defined and natural on R. and there is a H u r e w i c z

homomorphism

~s

h.

: A , X ~ ~ (X; Z) w h i c h is a r a t i o n a l e q u i v a l e n c e . The r e s t w i l i follow f r o m e v i d e n t

~s

We n e e d o n l y c o n s t r u c t the t h e o r i e s . naturality arguments. and c o m p a t i b l y f o r
Corollary A.3). Corollary 11.9. F o r unbased spaces

Of c o u r s e , A , X

m a p s n a t u r a l l y to

us A.X, and

similarly

A'.(X, R).

The following c o n s e q u e n c e is i m m e d i a t e (see

X, the definitions

s X A , X = K s ,( +. )

and a,(X; ~) : ~:(x+~ R)

specify unreduced h o m o l o g y theories.

N o t e h e r e t h a t ~2(X+) is a point, h e n c e

A~,(X +) = Z . ( S 0 ) .

The corollary

w a s asserted by W a l d h a u s e n [ZZ, l.4 and Z.8] (but with a rather misleading sketch proof). We shall prove a general result, T h e o r e m A.Z, on the stabilization of

3OO

functors to h o m o l o g y theories.

Theorem

ii.8 will be an i m m e d i a t e special case in

view of the following two results. Proposition ii.i0. T h e functors AX and A ( X ; R ) preserve m >_ 2, n >_ Z,

n-equivalences and q - h o m o t o p y Cartesian (m,n)-equivalences with and q <__m + n

provided that, for the latter, the d o m a i n square is strictly c o m -

mutative. Proof. passage to K W e lose a d i m e n s i o n upon appliction of ~ and gain it back upon and quotation of Propositions 9.1 and 9.5. W e n e e d only verify that

the intervening functors perties, w h e r e generated by C. R[C] Since

Q, F N , or R[?] have the appropriate preservation prois the free simplicial R - m o d u l e

for a simplicial set C ~ , F N ( X +) = H , X

and, by[8,§22], ~,R[C] = H,(C;R), the Since w,QX + = w,(X +) is the unreduced

conclusions are obvious in these cases.

h o m o l o g y theory associated to stable homotopy, of the Atiyah-Hirzebruch spectral sequence. Lemrna l l.ll. T h e functors AX and

the conclusion here follows by use

A(X;R)

f r o m based C W - s p a c e s

to

the h o m o t o p y category of based C W - s p a c e s Proof. This is not obvious.

are h o m o t o p y preserving. f r o m based spaces -* F ( T X , T Y ) X -~ Y is a m a p on

Recall that a functor T

to b a s e d spaces is said to be continuous if the function T : F ( X , Y ) function spaces is continuous. Since a h o m o t o p y between m a p s

I -* F(X, Y), continuous functors are h o m o t o p y preserving.

T h e various m o n a d s ,

bar constructions, and telescopes w h i c h entered into the construction of the functor T on (~ , ~ )-spaces are all continuous. (In particular, this uses the fact that

geometric realization of simplicial spaces is a continuous functor [12,p. 21].) W e pass f r o m T to K by first converting to C W - c o m p l e x e s by applying geometric

realization on the total singular complex,

this composite being h o m o t o p y preserv-

ing although not continuous, and then applying the plus construction (which is a h o m o t o p y functor by definition). T h e functors Q,FN, and also ~ w h e n interpreted

301

a s in (1) o r (2) a r e c o n t i n u o u s , a n d t h e f u n c t o r s

R[GSX]

are homotopy preserving

by standard facts on the relationship between simplicial and topological homotopy
theory (e.g. [8,§16 and §g6]).

§12.

Notes on W a l d h a u s e n ' s Since this w o r k

work

started with an attempt to understand Waldhausen~s,

a r u n d o w n of those things in [2g, § 1 and g] not considered above m a y not be taken

amiss. T h e r e a r e two c a l c u l a t i o n a t r e s u l t s in [ g g , § l ] c o n c e r n i n g s i m p l i c i a l ( o r topological) rings. O u r t h e o r y a d d s n o t h i n g n e w to the f o u n d a t i o n s h e r e e x c e p t f o r o n KR of P r o p o s i t i o n l 0 . 1 g and the concomitant

the i n f i n i t e loop s p a c e s t r u c t u r e s p l i t t i n g s of P r o p o s i t i o n 1 1 . 6 .

S i n c e t h e H - s p a c e l e v e l of t h e s e a d d i t i o n s p r o arguments, I shall run through the details request).
of

v i d e s s o m e c l a r i f i c a t i o n of W a l d h a u s e n ' s (modulo the relevant algebra;

t h e s e d e t a i l s a r e i n c l u d e d at R o t h e n b e r g ' s
Let f:R ~ R'

Proposition 12.1 ([22, 1.2]). topological rings, w h e r e n > 2. Let

be an (n-l)-equivalence fibres in the

F and F'

be the h o m o t o p y

following d i a g r a m and choose mute.

F ~ F'

which m a k e s

the top square h o m o t o p y

corn-

F

~ F'

BFM

1
oo

R

~ KR

l

BF M°°fi BFM R~ (3o

I Kf * KR t

F o r any Abelian group A, the d i a g r a m induces an i s o m o r p h i s m

H0(GL(w0R');Mco(lrn_l) @A) where

~ H0(BFMcoR' ;Hn(F ;A))

-~ H n ( F ' ; A ) ,

~rn_ 1 is t h e ( n - l ) st h o m o t o p y g r o u p of t h e f i b r e of f.

302

Proof.

Waldhausen asserts further that the left side is clearly iso-

morphic, via the trace, to the Hochschild homology group H0(w0R';Wn_ l ~ A ) ; I haven't checked the algebra. versal coefficients, H n ( F ; A ) ~ B y the Hurewicz theorem, R e m a r k s 7.3, and uniMoo(irn_l)(~A. The first isomorphism follows

from the definition of local coefficients, the fact that ~rlBFMooR' ~ GL(~0R' ), and a check that the action on Hn(F; A) agrees with the natural action on Meo(Wn_l)~A.

Let {E r} --~{'E r} be the m a p of Serre spectral sequences (with coefficients in A) induced by the diagram. The essential topological fact is that, since Kf is a m a p

of connected H-spaces which induces an e p i m o r p h i s m on fundamental groups, it has trivial local coefficients. morphism. Since Our assertion is thus that E On -~ ,E0n Z 2 is an iso-

E*0Z = H , ( B G L R ' ; A ) -~ H,(KR';A) : ,E,0Z and H , ( B G L R ; A ) r are isomorphisms and E . q
=

-~

I-I,(KR;A)
and

, r E.q

=

0 for 0 < < n, q

EZ nO

oo = En, 0

oo co 'EZn0 = 'E n~ , hence the five l e m m a gives that E0n-" 'E0n is an isomorphism.

A diagram chase f r o m the edge h o m o m o r p h i s m

gives that E c° -~ 'E~0 is an iso*0

m o r p h i s m , and another five l e m m a argument (involving the transgression dn+l :E Z Z n+l,0 -~ E 0 n ) gives the conclusion.

Propo~.!tion IZ.Z

([ZZ, i. 5]) Q if i = O

0

if i > 0

Proof.

Specialize the diagram of the previous result to

303

F

.',- F '

: E(sn; Z)

,

BGLZ

, ~Z d:R nZ
and

= A({*};Z) is the diseretization. F' is (n-1)(Q coefficients

whe re

R

n

= IZ[GSznsO] I and

connected. understood). where q F0

W e claim that H F' = Q n

H F' = 0 for n < q < gn-2 q

B y the W h i t e h e a d t h e o r e m applied to a w - i s o m o r p h i s m F 0 -~ K(Q,n), n : Q and

is the rationalization of F', it will follow that W n F ' ( ~ Q By Remarks

F'(~ Q = 0 for n < q<__ Zn-Z, hence the conclusion. q >_ Z and n > i we = WqBFMooRn Wq_iM R oo n have

7.3 and

[8, § ZZ], for WqF

= Wq-iFMeoRn = M ~ ooq-i R n

= ~q-iSFMcoRn = M H oo q-i (f~sn;z),

w h i c h is zero for

q<

Zn-Z and

q / n.

B y the Whitehead t h e o r e m applied to a

Wn-iSomorphism

F - ~ K(MooZ, n),

H n ( F ; Z ) = McoZ and

Hq(F; Z) = 0 o t h e r w i s e ,

0 < q <__2n-Z, h e n c e s i m i l a r l y w i t h Z r e p l a c e d by Q,
Farrell and Hsiang [4] and based on work to H.(BGLZ; of Borel

The key a l g e b r a i c f a c t , due to
) is is

[i], is that H.(BGLZ;McoQ via the trace,

isomorphic irrelevant
sequences.

Q); the definition of the isomorphism, here. Consider the rational homology

to the argument

Serre spectral
(say b y

Certainly E Z is finite-dimensional in each degree < Zn-Z [i]) and. in this range, g (E*0 = H.BGLZ, Eg

Borel's c a l c u l a t i o n s of H , ( B G L Z ) E2

=

E2

*0 ~ E 2 *n
Z Q -- E 0 n

*n ~

H, BGLZ).
B y Corollary 1 1.7,

B y the previous result, A(Sn; Z) ---~ A(sn; Z) X K Z

-~ ,E On Z

is an i s o m o r p h i s m .

and therefore

'E z

= 'E °° : H , K Z ~ H , F '

Z (,E , 0

:

z H , K Z , 'E *n

~ H , KZ).

304

Since

{E r}

and

{'E r)

converge to isomorphic homologies,

H F' = 0 for q

n < q<__ Zn-Z

by a trivial c o m p a r i s o n of dimensions. consequence, by T h e o r e m ii.8.

T h e following is an i m m e d i a t e Corollary 1Z. 3

A~{*)

~

Q

= {~

ifi> 0 ifi=

T h e results claimed about Postnikov systems in [ZZ, § Z] s e e m m u c h problematical (and are fortunately m u c h Remarks set R . 12.4.

more

less essential to the overall program).

T h e n th t e r m R~ n) of the natural Postnikov s y s t e m of a simplicial n x ~ y for q-simplices x and y if all of their iterated R(. n) is a simplicial ring

is R./( n ), w h e r e

faces of dimension

< n are equal [8,§8].

Visibly each

if R .

is so, and the n a t u r a l m a p s

R~n ) ~ R ( f ) f o r n > m a r e m a p s of

simplicial rings.

As W a l d h a u s e n s t a t e s [ Z Z , § l ] , t h e r e r e s u l t s a s p e c t r a l s e q u e n c e

the E Z - t e r m of which is given by the h o m o t o p y groups of the f i b r e s of the m a p s KIR%n) I -~ K I R ( n - l ) ] and w h i c h converges to

K, IR, I.

He a s s e r t s

f u r t h e r [ZZ,

Z. 5 and sequel] that the s a m e conclusions hold with R , replaced by an arbitrary ring up to homotopy, space X. that is, in our terminology, by an arbitrary grouplike A difficult to prove. (DO ring oo ring

If true, anything like this would be e n o r m o u s l y X (n) of X

Certainly, the coskeleta spaces of s o m e

could at best be strong h o m o t o p y A

sort (with m o r e homotopies

in sight than in Defimition 5.1; see the

discussion following that notion).

H e also asserts [22, Z.6 and sequel] that the coof which is given by the for any space X.

skeleta of QS 0 give rise to a spectral sequence the E Z - t e r m homotopy

groups of certain fibres and which converges to A , X

H e r e he thinks of QS 0 as the "coefficient ring" of AX, R in Definitions 11.4.

in analogy with the role of rigor s e e m s still

Since this is at best only a metaphor, splitting

further away.

The infinite loop space

Q(~2X +) --~ QS 0 X QCLX

does not s e e m

305

relevant. spectral

E v e n if t h e y do e x i s t , sequences

there

seems

to b e l i t t l e r e a s o n to b e l i e v e t h a t s u c h

would help much with explicit calculations.

Of course, it is conceivable that there is a simplicial analog of our theory for w h i c h this difficulty disappears, but I a m skeptical (and certain that other tech-

nical difficulties w o u l d appear in any such approach). It is time to discuss the m a i n issue. W a l d h a u s e n p r o p o s e d our A X as a

nice description of what he wanted, if it w e r e to exist, but he gave an alternative definition in t e r m s of w h i c h the proofs w e r e all to proceed. Waldhausen's functor (or rather its connected version). group of (11.3), then W X colimit over W e write WX for

If G S X

is the simplicial

is the plus construction on the classifying space of the
n

n and k of certain categories

(h~GSX)k

with objects simplicial
n

G S X - s e t s suitably related to the w e d g e of X and k copies of S . In the absence of any indications of proof, I for one find it hard to see h o w analogs for W X o f s o m e the results above for A X are to be m a d e rigorous f r o m this definition. T h e techof

nical details, for e x a m p l e of the rational equivalence required for Corollary iZ.3, m u s t surely be considerable. It w o u l d s e e m preferable to c o m p a r e A X and W X .

W a l d h a u s e n asserts (without proof, [ZZ, Z. i]) that the loop of the classifying space of the colimit over n of the categories ( h ~ G S X ) k n is equivalent to F M k Q ( I G S X I+).

While this certainly s e e m s plausible, his further claim fhat the equivalence is one of H - s p a c e s s e e m s m u c h m o r e difficult, and this in turn is n o w h e r e near strong

enough to prove the following assertion. Conjecture 12.5. A X and W X are naturally equivalent. s e e m farther apart, one might

Except that the definitions of A X and W X

v i e w this as analogous to the equivalence b e t w e e n his two definitions that w a s the pivotal result in Quillen's development of algebraic K - t h e o r y [5]. T h e point is that it is A X w h i c h is m o s t naturally connected with Quillen's algebraic K-theory~ but

306

it is W X

and its various equivalents in [22] w h i c h W a l d h a u s e n ' s a r g u m e n t s concordance.

relate

to the W h i t e h e a d groups for stable P L

307

Appendix.

Stabilizations of. functors t o h o m o l o g y theories n e e d e d to complete the proof of T h e o r e m
(presumably p a r t of t h e f o l k l o r e ) o n t h e

W e first give the technical l e m m a
1 1.3 a n d t h e n g i v e a v e r y g e n e r a l t h e o r e m

stabilization of h o m o t o p y functors to generalized h o m o l o g y theories. W e w o r k in the category ~/ of nondegenerately b a s e d spaces of the h o m o topy type of a C W - c o m p l e x and in its h o m o t o p y category h~/ . T h e proofs b e l o w

use w e l l - k n o w n facts about fibration sequences but, annoyingly, I k n o w of no published source w h i c h contains everything w e need; such details will appear in [15,

z~l].
Lemma A.I. Consider the following d i a g r a m in ~ , in w h i c h i and p are

written generically for the canonical m a p s parts of the d i a g r a m h o m o t o p y c o m m u t e ,
arrows erased also homotopy commute, (£0 ~ 1, e t c . ) .

of fibration sequences, the solid a r r o w the bottom squares with solid vertical

and the dotted arrows

~,~], a n d O a r e

homotopy sections

flF~ ..... -~ ..... ~-F~

.....

~ .....

~F5

-.- ....

e_ . . . .

~-Fe

,. ....

¼,
D

*Fe

fly

1

~f~

~-X

...

f

>Y

5

n
g

a',0

Lp I
¼
Fe such that

flW There exist maps e, ~-, a n d

~

Fg

~Z

~ W such that the top three

u , u n i q u e up t o h o m o t o p y ,

squares h o m o t o p y c o m m u t e
pot ~ ~r,

and there exists an equivalence

~:F~-*

Proof.

T h e h o m o t o p y commutativity of the lower three squares implies e is unique since two such m a p s
on

the existence of e, ir and L . H e r e
a c t i o n of a m a p F6-~ ~W

differ by the

a n d t h e a c t i o n of [ F 6 , ~ W ]

[FS,F~ ] is trivial since

3O8

i : ~ W ~- F £ unique.

is null h o m o t o p i c pew

(because of the section).

Similarly pe

= and ~ are (by the

Since

is null h o m o t o p i c by the diagram, T h u s there exists

is null h o m o t o p i e ~1:F.f -~ F e

exact sequence of the right column). p~' ~ w.

such that groups

T h e top r o w clearly induces a long exact sequence of h o m o t o p y

m a p p i n g onto a direct s u m m a n d sion that ~'

of that of the middle row, and the desired concluf r o m the five l e m m a if ~'~ w e r e

is an equivalence w o u l d be i m m e d i a t e However, there s e e m s 6:Fe-- Ff

h o m o t o p i c to i. so chosen.

to be no reason to suppose that a' can be 6i ~ii2p and pp = P6. Again, a'

Choose a map

such that

w o u l d be an equivalence if ~0~' w e r e h o m o t o p i c to p:F~/ -+ Ff. room for m a n e u v e r . ~2F£ X ~ W 6o~' = p ~ Since -~ ~Y. for s o m e ~Y is an H-space, Let ~ : ~ Y -~ ~ F E the s u m of~p

H e r e w e have m o r e and ~@ is an

equivalence Certainly

be an inverse projection. pM is given by the right action

g : F N -~ i~Y, w h e r e

of

[F~,~Y]

on

[F'7, F f ]

coming

from

Ff×~Y :

~

Ff.

Let

a=

a'¢,(-~).

Then

[5o~ = 6o(~e'~,(-I-t))
Since

= (I8ooe')(fap),qJ,(-b~) = ¢,(la) - ~,(la)

P(bt(faP),qJ,(--~)).

~,(la(fap),t~,(-p.))

= O. w e h a v e

~(~p),%(- ~) = (~e),(~)
for s o m e p,:w,F-{ w ~ [F~, f~W] -* w~Ff and thus ~ o ~ = p(~@),(w). It follows that 6,o ~, and

become

equal w h e n one passes f r o m the long exact h o m o t o p y

sequence of the middle r o w to its quotient by the the b o t t o m row. an i s o m o r p h i s m Here p, and ~

long

exact h o m o t o p y

sequence of ~, is

are i s o m o r p h i s m s , is an equivalence.

and w e conclude that

a n d thus that ~

T h e naturality of w prove the uniqueness of e

in T h e o r e m

1 1.3 follows f r o m the a r g u m e n t

used to

in the l e m m a . theories, w e define a

Turning to the desired construction of h o m o l o g y stab~ity sequence an {an)

to be a strictly increasing sequence of positive integers

such that a n - n

tends to infinity.

300
Theorem properties. A.Z. Let k: h~/ -~ hO/ he a functor with the following

O)

Application of k to an n-equivalence yields a bn-equivalence, w h e r e a stability sequence.

{Zbn}

is

(2)

Application of k

to a strictly c o m m u t a t i v e

2 n - h o m o t o p y Cartesian (n, n)-

equivalence yields an a -homotopy Cartesian square, w h e r e {an} is a n stability sequence. Let e : k X -~ k{*} be induced by X -~ {*} ~ :~Q( -~ f 2 ~ X and let ~ be the fibre of £. Then

there exist natural m a p s scope of the spaces and if %SXq

such that if y~Sx

is defined to be the tele-

~n~nx

with respect to the m a p s wq%SX, then k.
Ms

n ~ n f 2 ~ n + 1X, ~ ~, ~:kS X -~

is defined to be

is a reduced h o m o l o g y theory which

satisfies the w e d g e axiom. The following is a standard consequence.

C_.oroltary A.3.
pairs (X,A), define on A. Then
s

On unbased spaces

X, define k ~ X = ~ ( X + ) .
CA

O n unbased

k~(X,A)

= k ((Xu)CA) +) w h e r e

is the (unreduced) cone

s k~ is a generalized h o m o l o g y theory in the classical sense.

Returning to based spaces, w e first discuss the statement of the theorem. It will turn out that property (I) is only needed for the w e d g e axiom, for m a p s complexes. grams, maps, and then only

X -~ {*}, hence m a y be omitted in obtaining a h o m o l o g y theory on finite Property (Z) will also only be needed for a few simple types of diaSince £ is only given as a h o m o t o p y class of The first part of

to be displayed in the proof.

w e m u s t choose a representative before constructing ~ . A. 1 gives the following result. For

the proof of L e m m a Lemma "kf:~X -~'kY

A.4.

f:X -- Y, there is a unique h o m o t o p y class in h~/ .

such that the following d i a g r a m c o m m u t e s

310

kX It follows that ~ natural.

kf

I
~ kY . such that ~ - ~ k is

is a well-defined functor h q / -~ h~/

We also need the following analog. Lemma A.5 . Let For N:k{*} -~kX be i n d u c e d b y {*} - ~ X andlet ~kX kf:kX - ~ k Y be such

the fibre of ~.

f:X -- Y, there is a unique h o m o t o p y class commutes ~kX in h%/ . ~kf ~ ~kY

t h a t t h e following d i a g r a m

%x

,

%Y .

Proof. coordinate

T h e m a p f2kX ~ ~2k{*} X ~X w i t h f i r s t c o o r d i n a t e f~£ a n d s e c o n d

the canonical map is an equivalence.
A

It f o l l o w s t h a t k i s a f u n c t o r a n d ~ k - ' ~ Lemma Proof. canonical m a p A.6.

is natural.

T h e c o m p o s i t e ~kX -- ~2kX -- ~uX is a n a t u r a l e q u i v a l e n c e .

T h e m a p ~ k { * } X ~kX -~ f~kX g i v e n b y t h e s u m of ~ N a n d t h e is also an equivalence. suffice for the construction of o-. ~ : ~ X -* ~ E X such that 0- is an

T h e s e observations Lemma A. 7.

T h e r e is a natural m a p if X is n-connected.

(an+ l - l)-equivalence Proof. We

define 0 to be the top c o m p o s i t e in the d i a g r a m ~ .... -~ k ~ X ~'~X

kX

~

)" k { * )

k{*}

n -~kZX

H e r e w e have the tautological strict equality ~3£ -- ~16 , a n d the dotted a r r o w is canonical; its naturality up to h o m o t o p y X is n-connected, then the c o m m u t a t i v e is easily c h e c k e d by direct inspection. square If

311

x

~{*}

L
{*} • zx

is a (Zn+Z)-homotopy theorem (compare

Cartesian

(n+l, n+l)-equivalence

by the homotopy property

excision (2) i m p l i e s

Definition 9.2).

By a mild interpretation,

the result. The spaces with telescopes, "ksX o f T h e o r e m A.2 are now defined. equivalence
= ~ks~X .

Since 2 commutes

there is an evident homotopy
-~ Tel n~0

~s -r:k X = Tel [ 2 n ~ n x

~n+l~n+ix

n>__0

While .~s need not be a functor and might well be present, "c : ~ S x _ ~ s . . ~ X q q q*1 X = zr ~ X , q q

T n e e d not be natural,

since

liraI t e r m s

they induce functors k s X q groups.

and natural i s o m o r p h i s m s Alternatively, with

on passage to h o m o t o p y

w e could equally well define

~Sx = c o l i m g
q a n d not bother with the telescopes, F o r reduced h o m o l o g y

q+n

ZnX
T q then being evident.

the i s o m o r p h i s m s

theories,

excision reduces to the suspension a x i o m

just verified on trivial f o r m a l grounds,

without use of properties (I) and (2). T h e T h e following l e m m a veri-

things to be proven are exactness and the w e d g e axiom.

ties the appropriate
Lemma

exactness
If

axiom.
is a cofibration, then the sequence

A . 8. kSA q

~: A ~ X

ks ~ q

; ksX q

kSur q

~" k S ( x / A ) q

i s e x a c t f o r a l l q, w h e r e Proof.

w is the quotient map. n commute with cofibration sequences, a

Since the functors colimit

glance at the relevant

systems

s h o w s t h a t it s u f f i c e s

to p r o v e

"k A q

q

g X . q

q and

......g q ( X / A ) X/A are n-connected. By the

to b e e x a c t i n a s u i t a b l e

range when A,X,

312

homotopy

excision t h e o r e m again, the square A
L

,~X

{*}

x/A
Consider the following d i a g r a m

is a 2 n - h o m o t o p y Cartesian (n, n) -equivalence.

~ A ......... -~ Fk~r ......... -~F~ir

kA

~ kX

~,

kX

k{*}

i

~

=k(X/A) *

F
*k{*}

(X/A) ~

i

Verdier's a x i o m for fibration sequences applied to the triangle

kX

k(X/A)

gives an equivalence mutes.

Fklr -~ Fklr

such that the upper right square h o m o t o p y c o m 9.3 give an (an+l)-equivalence ~ A By Lemma -~ Fklr

Property (Z) and R e m a r k s

such that the upper left square h o m o t o p y c o m m u t e s . posite ~ A -~X in the d i a g r a m is "kb.

A.4, the corn-

T h e conclusion follows f r o m the long

exact sequence of h o m o t o p y groups of the right column. It r e m a i n s only to verify the w e d g e axiom. Lemma A.9. F o r any set of spaces {Xi}, the natural m a p

@~Sx. ~ ~ s (Vxi){ i q 1 q -

is an i s o m o r p h i s m Proof.

for all q. ~ n commute with wedges, a glance at the

Since the functors

relevant colimit s y s t e m s s h o w s that it suffices to prove

313

(9 kq(Xi) -~ kq( V x i) i i
to be an i s o m o r p h i s m in a suitable range w h e n each X. is n-connected.
i

If X

is

n-connected, then X -- {$} is an (n+l)-equivalence, hence bn+l-equivalence

~:kX-~ k{*}

is a

by property (I), hence "kX is (hn+ I- l)-connected. Therefore

the inclusion of V~_xi in the w e a k direct product of the kX.i (all but finitely m a n y i coordinates at the basepoint) is a (2bn+ I- l)-equivalence, and the conclusion follows.

314

Bibliography.
t.

A. Borel.

Stable real cohomology of arithmetic groups.

Ann. Sci. Ecole

N o r m a l e Sup. 4e serie t.7 (1974), 235-272.
2.

F.Cohen, T. Lada, and J.P. May.

The H o m o l o g y of Iterated Loop Spaces.

Springer Lecture Notes in Mathematics, Voi. 533, i976.
3.

A. Dold and R. Thorn. dukte.

Quasifaserungen und unendliche symmetrische pro-

Annals of Math. 67 (1958), Z39-Z81. Proc. A m e r . Math. Soc. S u m m e r Institute.

4.

F.T.Farrell a n d W . C . H s i a n g . Stanford, 1976. To appear.

5.

D. Grayson.

Higher algebraic K-theory:If (after D. Quillen). Springer

Lecture Notes in Mathematics. Vol. 551,2J6-Z40, 1976.
6.

T. Zada.

A n operad action on infinite loop space multiplication.

Canadian

J. Math. To appear.
7.

J.L. Loday.

Les matrices monomiales et le groupe de Whitehead W h z.

Springer Lecture Notes in Mathematics, Voi. 551, 155-163, 1976.
8.

J. P. May. J. P. May. 155, 1975.

Simplicial Objects in Algebraic Topology. Classifying Spaces and Fibrations.

D. van Nostrand, 1967.

9.

M e m o i r s A m e r . Math. Soc.

i0.

J. P. May.

The G e o m e t r y of Iterated Loop Spaces.

Springer Lecture Notes

in Mathematics, Voi. 271, 1972. II. J.P. May. Eco spaces, group completions, and permutative categories.

London Math. Soc. Lecture Note Series II, 61-94, 1974. 12. J.P. M a y (with contributions by N. Ray, F. Quinn~ and J. Tornehave). E
CO

Ring Spaces and E

CO

Ring Spectra.

S p r i n g e r L e c t u r e Notes in M a t h e -

matics, Voi. 577, 1977. 13. J.P. May. 456-494. 14. 15. J.P. May. J.P. May.
Press.

Infinite loop space theory.

Bull. A m e r . Math. Soe. 83(1977),

The spectra associated to permutative categories.

To appear. Academic

The Homotopical Foundations of Algebraic Topology.

In preparation.

315

16.

J.P. M a y and R. Thomason. To appear.

The uniqueness of infinite loop space machines.

17.

R.I. Milgram.

The bar construction and Abelian H-spaces.

Illinois $.

Math. 11(1957), 242-250. 18. D. Quillen. Mathematics, 19. m. Quillen. Higher algebraic K-theory I. Springer Lecture Notes in Voi, 341, 85-147, 1973. Letter from Quillen to Milnor on Im(Iri0 J> s
1

Ki z

Springer Lecture Notes in Mathematics, Vol. 551, 182-188, 1976. Z0. Z1. G. Segal. Categories and cohomologytheories. Topology 13(1974), 293-312.

J.B.Wagoner.

Delooping classifying spaces in algebraic K-theory.

Topology 11(197Z), 349-370. ZZ. F.Waldhausen. Math, Soc, Algebraic K-theory of topological spaces, I. Proc. A m e r . Institute. Stanford, 1976. To appear.

Summer