Bloch's higher Chow group explained

In algebraic geometry, Bloch's higher Chow groups, a generalization of Chow group, is a precursor and a basic example of motivic cohomology (for smooth varieties). It was introduced by Spencer Bloch and the basic theory has been developed by Bloch and Marc Levine.

In more precise terms, a theorem of Voevodsky[1] implies: for a smooth scheme X over a field and integers p, q, there is a natural isomorphism

\operatorname{H}p(X;Z(q))\simeq\operatorname{CH}q(X,2q-p)

between motivic cohomology groups and higher Chow groups.

Motivation

One of the motivations for higher Chow groups comes from homotopy theory. In particular, if

\alpha,\beta\inZ*(X)

are algebraic cycles in

X

which are rationally equivalent via a cycle

\gamma\inZ*(X x \Delta1)

, then

\gamma

can be thought of as a path between

\alpha

and

\beta

, and the higher Chow groups are meant to encode the information of higher homotopy coherence. For example,

CH*(X,0)

can be thought of as the homotopy classes of cycles while

CH*(X,1)

can be thought of as the homotopy classes of homotopies of cycles.

Definition

Let X be a quasi-projective algebraic scheme over a field (“algebraic” means separated and of finite type).

For each integer

q\ge0

, define

\Deltaq=\operatorname{Spec}(Z[t0,...,tq]/(t0+...+tq-1)),

which is an algebraic analog of a standard q-simplex. For each sequence

0\lei1<i2<<ir\leq

, the closed subscheme
t
i1

=

t
i2

==

t
ir

=0

, which is isomorphic to

\Deltaq-r

, is called a face of

\Deltaq

.

For each i, there is the embedding

\partialq,:\Deltaq-1\overset{\sim}\to\{ti=0\}\subset\Deltaq.

We write

Zi(X)

for the group of algebraic i-cycles on X and

zr(X,q)\subsetZr+q(X x \Deltaq)

for the subgroup generated by closed subvarieties that intersect properly with

X x F

for each face F of

\Deltaq

.

Since

\partialX,=\operatorname{id}X x \partialq,:X x \Deltaq-1\hookrightarrowX x \Deltaq

is an effective Cartier divisor, there is the Gysin homomorphism:
*:
\partial
X,q,i

zr(X,q)\tozr(X,q-1)

,that (by definition) maps a subvariety V to the intersection

(X x \{ti=0\})\capV.

Define the boundary operator

dq=

q
\sum
i=0

(-1)i

*
\partial
X,q,i
which yields the chain complex

\tozr(X,q)\overset{dq}\tozr(X,q-1)\overset{dq-1

}\to \cdots \overset\to z_r(X, 0).

Finally, the q-th higher Chow group of X is defined as the q-th homology of the above complex:

\operatorname{CH}r(X,q):=\operatorname{H}q(zr(X,)).

(More simply, since

zr(X,)

is naturally a simplicial abelian group, in view of the Dold–Kan correspondence, higher Chow groups can also be defined as homotopy groups

\operatorname{CH}r(X,q):=\piqzr(X,)

.)

For example, if

V\subsetX x \Delta1

[2] is a closed subvariety such that the intersections

V(0),V(infty)

with the faces

0,infty

are proper, then

d1(V)=V(0)-V(infty)

and this means, by Proposition 1.6. in Fulton’s intersection theory, that the image of

d1

is precisely the group of cycles rationally equivalent to zero; that is,

\operatorname{CH}r(X,0)=

the r-th Chow group of X.

Properties

Functoriality

Proper maps

f:X\toY

are covariant between the higher chow groups while flat maps are contravariant. Also, whenever

Y

is smooth, any map to

Y

is contravariant.

Homotopy invariance

If

E\toX

is an algebraic vector bundle, then there is the homotopy equivalence

CH*(X,n)\congCH*(E,n)

Localization

Given a closed equidimensional subscheme

Y\subsetX

there is a localization long exact sequence

\begin{align} \\ CH*-d(Y,2)\toCH*(X,2)\toCH*(U,2)\to&\\ CH*-d(Y,1)\toCH*(X,1)\toCH*(U,1)\to&\\ CH*-d(Y,0)\toCH*(X,0)\toCH*(U,0)\to&0 \end{align}

where

U=X-Y

. In particular, this shows the higher chow groups naturally extend the exact sequence of chow groups.

Localization theorem

showed that, given an open subset

U\subsetX

, for

Y=X-U

,

z(X,)/z(Y,)\toz(U,)

is a homotopy equivalence. In particular, if

Y

has pure codimension, then it yields the long exact sequence for higher Chow groups (called the localization sequence).

References

Notes and References

  1. Book: Lecture Notes on Motivic Cohomology. Clay Math Monographs. 159.
  2. Here, we identify

    \Delta1

    with a subscheme of

    P1

    and then, without loss of generality, assume one vertex is the origin 0 and the other is ∞.