アフィリエイト広告を利用しています

2019年05月29日

数学: 位相空間上の層の圏としてのトポス

トポスは「有限極限と羃対象を持つ圏」として特徴付けられた. その結果として小さな圏 $\mathscr{C}$ から集合の圏 $\mathbf{Set}$ への関手の圏 $\mathrm{Func}(\mathscr{C}^{\mathrm{op}}, \mathbf{Set})$ がトポスになることが証明できた.†
† Michael Barr, Charles Wells, "Toposes, Triples and Theories", Chapter 2

本の中ではもう一つの仕方でトポスを特徴付けている. それは

景 (site) の上の層の圏

というものである. 景は位相空間の圏論的な一般化である.

ここでは $X$ を任意の位相空間として $X$ 上の層を定義する.

定義: 前層 (presheaf). 位相空間 $X$ 上の開集合の全体 $\mathscr{O}(X)$ は, 包含写像を射として小さい圏になる. このとき, すでに示した結果から $\mathscr{O}(X)$ から $\mathbf{Set}$ への反変関手の全体 $\mathrm{Func}(\mathscr{O}(X)^{\mathrm{op}},\mathbf{Set})$ はトポスになる. トポス $\mathrm{Func}(\mathscr{O}(X)^{\mathrm{op}},\mathbf{Set})$ の対象を $X$ 上の前層 (presheaf over $X$) と呼ぶ. $X$ 上の前層の圏を $\mathbf{Psh}(X)$ で表わす. つまり
\begin{equation*}
\newcommand{\Ar}[1]{\mathrm{Ar}(#1)}
\newcommand{\Arr}[1]{\mathrm{Arr}(#1)}
\newcommand{\Card}{\text{card}}
\newcommand{\Cdot}{\,\cdot^{\mathrm{op}}}
\newcommand{\Cocone}{\mathrm{Cocone}}
\newcommand{\Colim}{\mathrm{colim}\,}
\newcommand{\CommaCat}[2]{(#1/#2)}
\newcommand{\Cone}{\mathrm{Cone}}
\newcommand{\Eqclass}[4]{{#1#2#3}_{#4}}
\newcommand{\EqCls}[2]{{\left[#1\right]}_{#2}}
\newcommand{\Eqcls}[1]{\left[#1\right]}
\newcommand{\FnRest}[2]{{#1}|{#2}}
\newcommand{\Func}[2]{\mathrm{Func}(#1,#2)}
\newcommand{\g}{\varg}
\newcommand{\Hom}{\mathrm{Hom}}
\newcommand{\Id}[1]{\mathrm{id}_{#1}}
\newcommand{\In}{\mathrm{incl}}
\newcommand{\Inc}[2]{\mathrm{incl}\left(#1,#2\right)}
\newcommand{\Incl}[2]{\mathrm{incl}_{#1}^{#2}}
\newcommand{\Mb}[1]{\mathbf{#1}}
\newcommand{\Mr}[1]{\mathrm{#1}}
\newcommand{\Ms}[1]{\mathscr{#1}}
\newcommand{\Mt}[1]{\mathtt{#1}}
\newcommand{\Nat}{\mathrm{Nat}}
\newcommand{\Ob}[1]{\mathrm{Ob}(#1)}
\newcommand{\Opp}[1]{{#1}^{\mathrm{op}}}
\newcommand{\Prj}[2]{\mathrm{proj}\left(#1,#2\right)}
\newcommand{\Proj}[2]{\mathrm{proj}^{#1}_{#2})}
\newcommand{\q}{\hspace{1em}}
\newcommand{\qq}{\hspace{0.5em}}
\newcommand{\Rel}[1]{\langle{#1}\rangle}
\newcommand{\Rest}[2]{{#1}|{#2}}
\newcommand{\SkelCat}[1]{\mathrm{sk}(#1)}
\newcommand{\Slash}[1]{{\ooalign{\hfil/\hfil\crcr$#1$}}}
\newcommand{\SliCat}[2]{{#1}\,\big/\,{#2}}
\newcommand{\Src}{d^{0,\mathrm{op}}}
\newcommand{\Sub}{\mathrm{Sub}}

\newcommand{\Tgt}{d^{1,\mathrm{op}}}
\newcommand{\TwArCat}[1]{\mathrm{Tw}(#1)}
\newcommand{\VectCat}[1]{#1 \mathchar`- \mathbf{Vect}}
\newcommand{\Grp}{\mathbf{Grp}}
\newcommand{\Mon}{\mathbf{Mon}}
\newcommand{\POs}{\mathbf{Pos}}
\newcommand{\Psh}{\mathbf{Psh}}
\newcommand{\Set}{\mathbf{Set}}
\newcommand{\Sh}{\mathbf{Sh}}
\newcommand{\Top}{\mathbf{Top}}
\newcommand{\A}{\mathscr{A}}
\newcommand{\B}{\mathscr{B}}
\newcommand{\C}{\mathscr{C}}
\newcommand{\D}{\mathscr{D}}
\newcommand{\E}{\mathscr{E}}
\newcommand{\F}{\mathscr{F}}
\newcommand{\sH}{\mathscr{H}}
\newcommand{\I}{\mathscr{I}}
\newcommand{\J}{\mathscr{J}}
\newcommand{\K}{\mathscr{K}}
\newcommand{\sL}{\mathscr{L}}
\newcommand{\M}{\mathscr{M}}
\newcommand{\N}{\mathscr{N}}
\newcommand{\sO}{\mathscr{O}}
\newcommand{\sP}{\mathscr{P}}
\newcommand{\R}{\mathscr{R}}
\newcommand{\sS}{\mathscr{S}}
\newcommand{\T}{\mathscr{T}}
\newcommand{\U}{\mathscr{U}}
\newcommand{\V}{\mathscr{V}}
\newcommand{\W}{\mathscr{W}}
\newcommand{\X}{\mathscr{X}}
\newcommand{\Y}{\mathscr{Y}}
\newcommand{\Z}{\mathscr{Z}}
\newcommand{\Pw}{\mathbf{P}}
\newcommand{\Rn}[1]{\Real^{#1}}
\newcommand{\Pw}{\mathbf{P}}
\newcommand{\Rn}[1]{\Real^{#1}}
\Psh(X)=\Func{\Opp{\sO(X)}}{\Set}
\end{equation*} である.

$U$ を $X$ の任意の開集合 (つまり $X$ の開集合の圏 $\sO(X)$ の対象), $\left\{ U_i \right\}_{i \in I}$ を $U$ の任意の開被覆とする. 簡単のために, 以下で混乱が無いときには添数集合 $I$ を省略して書く.

各 $U_i$ に対して, $FU_i$ の元の族 $\left\{ x_i \in FU_i \right\}$ を考える. 便宜上各 $i,j$ に対して包含写像 (つまり圏 $\sO(X)$ の射) $\Incl{U_i \cap U_j}{U_i} : {U_i \cap U_j} \hookrightarrow U_i$ に対して
\begin{equation*}
\Rest{x_i}{U_i \cap U_j} = F\!\left(\Incl{U_i \cap U_j}{U_i}\right)x_i
\end{equation*} と書き, $x_i$ の $U_i \cap U_j$ への 制限 (restriction) と呼ぶ.

定義: 層 (sheaf). $F : \sO(X) \rightarrow \Set$ を $X$ 上の前層とし, $U$ を $X$ の任意の開集合 ${U_i}$ を $U$ の任意の開被覆とする. 各 $FU_i$ の元の族 $\left\{ x_i \right\}$ が, 各々の $i, j$ に対して条件

\begin{equation*}
\Rest{x_i}{U_i \cap U_j}=\Rest{x_j}{U_i \cap U_j}
\end{equation*} を満たしているとき, ある元 $x \in FU$ で各 $i$ について
\begin{equation*}
\Rest{x}{U_i} = x_i
\end{equation*} となるものが一意的に存在するならば, $F$ を $X$ 上の 層 (sheaf) と呼ぶ. $X$ 上の層の全体は $X$ 上の前層の圏 $\Psh(X)$ の充満な部分圏をなす (前層の圏における射, すなわち関手間の自然変換はそのまま層の圏の射となる). この圏を $\Sh(X)$ と表わす.


層の例.
(i) 位相空間 $X$ の各々の開集合 $U$ に対して $FU=C(U, \mathbb{R})$ ($U$ 上の実数値連続関数全体の集合) と定義すると $F$ は $X$ 上の層となる.
実際に, 開集合 $U$ と $U$ の開被覆 $\left\{ U_i \right\}_{i \in I}$ に対して, 連続関数の族 $\left\{ x_i : U_i \rightarrow \mathbb{R} \right\}_{i \in I}$ で各 $i,j \in I$ について
\begin{equation*}
\Rest{x_i}{U_i \cap U_j}=\Rest{x_j}{U_i \cap U_j}
\end{equation*} が成り立つものが与えられているとする. 任意の $u \in U$ に対して, $u \in U_i$ となる $U_i$ を一つ選び,
\begin{equation*}
x(u)=x_i(u)
\end{equation*} と定義する. 族 $\left\{ x_i \right\}$ に関する仮定より, 関数 $x : U \rightarrow \mathbb{R}$ が一意的に定義される. $x$ は $U$ 上の連続関数となり, $\Rest{x}{U_i}=x_i \, (i \in I)$ が成り立つ. したがって $F=C(-,\mathbb{R})$ は層である.
(ii) $X$, $Y$ を位相空間, $p : Y \rightarrow X$ を連続写像とする. $X$ の開集合 $U$ に対して, $U$ 上の連続写像 $s : U \rightarrow Y$ で $p \circ s$ が $U$ の $X$ への包含写像, つまり $p \circ s = \Incl{U}{X} : U \hookrightarrow X$ となるものの全体を $\Gamma(U,Y)$ とおく. $\Gamma(U,Y)$ の各元を, $p$ の 断面 (section) と呼ぶ. このとき $\Gamma(-,Y) : \Opp{\sO(X)} \rightarrow \Set$ は層となる. この層を $p$ の断面の層 (sheaf of sections of $p$) と呼ぶ.

層はイコライザー図式によっても定義でき, 次の命題の形に述べられる. これは, 与えられた前層が層であることを証明するための有用な命題である.

命題. $F : \Opp{\sO(X)} \rightarrow \Set$ を $X$ 上の前層とする. $F$ が層となるための必要十分条件は, $X$ の任意の開集合 $U$ と $U$ の任意の開被覆 $\left\{ U_i \right\}$ に対して図式
\begin{equation*}
\begin{xy}
\xymatrix@=48pt {
FU \ar[r]^-{r} & {\prod FU_i} \ar@<1ex>[r]^-{d_0} \ar@<-1ex>[r]_-{d_1} & {\prod F(U_i \cap U_j)}
}
\end{xy}
\end{equation*} がイコライザーになることである. ここで $r$ は積 $\prod FU_i$ の普遍性から導かれる, 図式
\begin{equation*}
\begin{xy}
\xymatrix@=48pt {
FU \ar[r]^{r} \ar[dr]_{\Rest{-}{U_i}} & {\prod FU_i} \ar[d]^{p_i} \\
~ & FU_i
}
\end{xy}
\end{equation*} を可換にする一意的な射である. $d_0, d_1$ は積 $\prod F(U_i \cap U_j)$ の普遍性から導かれる, それぞれ図式
\begin{equation*}
\begin{xy}
\xymatrix@=48pt {
{\prod FU_i} \ar[r]^-{d_0} \ar[d]_{p_i} & {\prod F(U_i \cap U_j)} \ar[d]^{p_{ij}} & {\prod FU_i} \ar[r]^-{d_1} \ar[d]_{p_j} & {\prod F(U_i \cap U_j)} \ar[d]^{p_{ij}} \\
FU_i \ar[r]_-{\Rest{-}{U_i \cap U_j}} & F(U_i \cap U_j) & FU_j \ar[r]_-{\Rest{-}{U_i \cap U_j}} & F(U_i \cap U_j)
}
\end{xy}
\end{equation*} を可換にする一意的な射である.
posted by 底彦 at 23:30 | Comment(0) | TrackBack(0) |
この記事へのコメント
コメントを書く

お名前:

メールアドレス:


ホームページアドレス:

コメント:

この記事へのトラックバックURL
https://fanblogs.jp/tb/8847463

この記事へのトラックバック
ファン
検索
<< 2024年10月 >>
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    
最新記事
最新コメント
眼科の定期検査 〜 散歩 by コトタマ (02/15)
眼科の定期検査 by 三文字寄れば文殊のヒフミヨ (09/21)
本を読んで過ごす by 底彦 (12/13)
本を読んで過ごす by ねこ (12/12)
数学の計算をする by 底彦 (12/04)
タグクラウド
カテゴリアーカイブ
仕事(59)
社会復帰(22)
(43)
コンピューター(211)
(1417)
借金(8)
勉強(13)
(12)
数学(97)
運動(8)
日常生活(1387)
(203)
健康(36)
読書(20)
プロフィール

ブログランキング・にほんブログ村へ
にほんブログ村
にほんブログ村 メンタルヘルスブログ うつ病(鬱病)へ
にほんブログ村
にほんブログ村 科学ブログ 数学へ
にほんブログ村
にほんブログ村 IT技術ブログ プログラム・プログラマーへ
にほんブログ村