text/appendixes/moam.tex
author Scott Morrison <scott@tqft.net>
Tue, 21 Sep 2010 14:44:17 -0700
changeset 550 c9f41c18a96f
parent 501 fdb012a1c8fe
child 553 94cadcb4831f
permissions -rw-r--r--
deleting nn's
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
492
833bd74143a4 put in a stub appendix for MoAM, but I'm going to go do other things next
Scott Morrison <scott@tqft.net>
parents:
diff changeset
     1
%!TEX root = ../../blob1.tex
833bd74143a4 put in a stub appendix for MoAM, but I'm going to go do other things next
Scott Morrison <scott@tqft.net>
parents:
diff changeset
     2
833bd74143a4 put in a stub appendix for MoAM, but I'm going to go do other things next
Scott Morrison <scott@tqft.net>
parents:
diff changeset
     3
\section{The method of acyclic models}  \label{sec:moam}
499
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     4
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     5
Let $F_*$ and $G_*$ be chain complexes.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     6
Assume $F_k$ has a basis $\{x_{kj}\}$
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     7
(that is, $F_*$ is free and we have specified a basis).
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     8
(In our applications, $\{x_{kj}\}$ will typically be singular $k$-simplices or 
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
     9
$k$-blob diagrams.)
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    10
For each basis element $x_{kj}$ assume we have specified a ``target" $D^{kj}_*\sub G_*$.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    11
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    12
We say that a chain map $f:F_*\to G_*$ is {\it compatible} with the above data (basis and targets)
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    13
if $f(x_{kj})\in D^{kj}_*$ for all $k$ and $j$.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    14
Let $\Compat(D^\bullet_*)$ denote the subcomplex of maps from $F_*$ to $G_*$
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    15
such that the image of each higher homotopy applied to $x_{kj}$ lies in $D^{kj}_*$.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    16
501
Kevin Walker <kevin@canyon23.net>
parents: 500
diff changeset
    17
\begin{thm}[Acyclic models]  \label{moam-thm}
499
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    18
Suppose 
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    19
\begin{itemize}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    20
\item $D^{k-1,l}_* \sub D^{kj}_*$ whenever $x_{k-1,l}$ occurs in $\bd x_{kj}$
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    21
with non-zero coefficient;
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    22
\item $D^{0j}_0$ is non-empty for all $j$; and
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    23
\item $D^{kj}_*$ is $(k{-}1)$-acyclic (i.e.\ $H_{k-1}(D^{kj}_*) = 0$) for all $k,j$ .
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    24
\end{itemize}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    25
Then $\Compat(D^\bullet_*)$ is non-empty.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    26
If, in addition,
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    27
\begin{itemize}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    28
\item $D^{kj}_*$ is $m$-acyclic for $k\le m \le k+i$ and for all $k,j$,
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    29
\end{itemize}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    30
then $\Compat(D^\bullet_*)$ is $i$-connected.
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    31
\end{thm}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    32
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    33
\begin{proof}
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    34
(Sketch)
550
c9f41c18a96f deleting nn's
Scott Morrison <scott@tqft.net>
parents: 501
diff changeset
    35
This is a standard result; see, for example, \nn{need citations: Spanier}.
499
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    36
591265710e18 start on MOAM appendix
Kevin Walker <kevin@canyon23.net>
parents: 492
diff changeset
    37
We will build a chain map $f\in \Compat(D^\bullet_*)$ inductively.
500
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    38
Choose $f(x_{0j})\in D^{0j}_0$ for all $j$
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    39
(possible since $D^{0j}_0$ is non-empty).
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    40
Choose $f(x_{1j})\in D^{1j}_1$ such that $\bd f(x_{1j}) = f(\bd x_{1j})$
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    41
(possible since $D^{0l}_* \sub D^{1j}_*$ for each $x_{0l}$ in $\bd x_{1j}$
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    42
and $D^{1j}_*$ is 0-acyclic).
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    43
Continue in this way, choosing $f(x_{kj})\in D^{kj}_k$ such that $\bd f(x_{kj}) = f(\bd x_{kj})$
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    44
We have now constructed $f\in \Compat(D^\bullet_*)$, proving the first claim of the theorem.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    45
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    46
Now suppose that $D^{kj}_*$ is $k$-acyclic for all $k$ and $j$.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    47
Let $f$ and $f'$ be two chain maps (0-chains) in $\Compat(D^\bullet_*)$.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    48
Using a technique similar to above we can construct a homotopy (1-chain) in $\Compat(D^\bullet_*)$
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    49
between $f$ and $f'$.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    50
Thus $\Compat(D^\bullet_*)$ is 0-connected.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    51
Similarly, if $D^{kj}_*$ is $(k{+}i)$-acyclic then we can show that $\Compat(D^\bullet_*)$ is $i$-connected.
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    52
\end{proof}
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    53
5702ddb104dc temporarily done with MOAM
Kevin Walker <kevin@canyon23.net>
parents: 499
diff changeset
    54
\nn{do we also need some version of ``backwards" acyclic models?  probably}