diff options
| -rw-r--r-- | lic_malinka.pdf | bin | 483340 -> 483862 bytes | |||
| -rw-r--r-- | sections/conj_classes.tex | 122 | ||||
| -rw-r--r-- | sections/fraisse_classes.tex | 21 | ||||
| -rw-r--r-- | sections/preliminaries.tex | 2 | ||||
| -rw-r--r-- | uwagi_29_06_22.txt | 4 |
5 files changed, 90 insertions, 59 deletions
diff --git a/lic_malinka.pdf b/lic_malinka.pdf Binary files differindex 03b45e9..671b753 100644 --- a/lic_malinka.pdf +++ b/lic_malinka.pdf diff --git a/sections/conj_classes.tex b/sections/conj_classes.tex index 96522f5..9620220 100644 --- a/sections/conj_classes.tex +++ b/sections/conj_classes.tex @@ -82,7 +82,7 @@ \begin{theorem} \label{theorem:generic_aut_general} - Let $\cC$ be a Fraïssé class of finite $L$-structures. + Let $\cC$ be a Fraïssé class of finitely generated $L$-structures. Let $\cD$ be the class of structures from $\cC$ with additional unary function symbol interpreted as an automorphism of the structure. If $\cC$ has the weak Hrushovski property @@ -90,11 +90,25 @@ automorphism group of the $\Flim(\cC)$. \end{theorem} + Before we get to the proof, let us establish some notions. If + $g\colon\Gamma\to\Gamma$ is a finite injective function, then we say that + $g$ is \emph{good} if it gives (in a natural way) an isomorphism between + $\langle \dom(g)\rangle$ and $\langle\rng(g)\rangle$, i.e. substructures + generated by $\dom(g)$ and $\rng(g)$ respectively. Of course, in our + case, $g$ is good + if and only if $[g]_{\Aut(\Gamma)} \neq\emptyset$ (becuase of ultrahomogeneity + of $\Gamma$. + + Also it is important to mention that an isomorphism between two finitely + generated structures is uniquely given by a map from generators of one structure + to the other. This allow us to treat a finite function as an isomorphism + of finitely generated structure. + \begin{proof} - Let $\Gamma = \Flim(\cC)$ and $(\Pi, \sigma) = \Flim(\cD)$. Let $G = \Aut(\Gamma)$, - i.e. $G$ is the automorphism group of $\Gamma$. First, by the Theorem + Let $\Gamma = \Flim(\cC)$ and $(\Pi, \sigma) = \Flim(\cD)$. First, by the Theorem \ref{theorem:isomorphic_fr_lims}, we may assume without the loss of generality - that $\Pi = \Gamma$. + that $\Pi = \Gamma$. Let $G = \Aut(\Gamma)$, + i.e. $G$ is the automorphism group of $\Gamma$. We will construct a strategy for the second player in the Banach-Mazur game on the topological space $G$. This strategy will give us a subset $A\subseteq G$ and as we will see a subset of the $\sigma$'s conjugacy class. @@ -105,8 +119,8 @@ sets $\{g\in G\mid g\upharpoonright_A = g_0\upharpoonright_A\}$ for some finite set $A\subseteq \Gamma$ and some automorphism $g_0\in G$. In other words, a basic open set is a set of all extensions of some finite partial - isomorphism $g_0$ of $\Gamma$. By $B_{g}\subseteq G$ we denote a basic - open subset given by a finite partial isomorphism $g$. Note that $B_g$ + automorphism $g_0$ of $\Gamma$. By $B_{g}\subseteq G$ we denote a basic + open subset given by a finite partial isomorphism $g$. Again, Note that $B_g$ is nonemty because of ultrahomogeneity of $\Gamma$. With the use of Corollary \ref{corollary:banach-mazur-basis} we can consider @@ -150,20 +164,32 @@ For technical reasons, let $g_{-1} = \emptyset$ and $X_{-1} = \emptyset$. Suppose that player \textit{I} in the $n$-th move chooses a finite partial - isomorphism $f_n$. We will construct a finite partial isomorphism $g_n\supseteq f_n$ - and a set $X_n\subseteq\bN^2$ - such that following properties hold: + automorphism $f_n$. We will construct a finite partial automorphism + $g_n\supseteq f_n$ together with a finitely generated substrucutre + $\Gamma_n \subseteq \Gamma$ and a set $X_n\subseteq\bN^2$ + such that the following properties hold: \begin{enumerate}[label=(\roman*)] - \item $g_n$ is an automorphism of the induced substructure $\Gamma_n$, + \item $g_n$ is good and + $\dom(g_n)\cup\rng(g_n)\subseteq\langle\dom(g_n)\rangle = \Gamma_n$, + i.e. $g_n$ gives an automorphism of a finitely generated + substructure $\Gamma_n$ \item $g_n(v_n)$ and $g_n^{-1}(v_n)$ are defined, - \item let - $\{\langle (A_{n,k}, \alpha_{n, k}), (B_{n,k}, \beta_{n,k}), f_{n, k}\rangle\}_{k\in\bN}$ - be the enumeration of all pairs of finite structures of $T$ with automorphism - such that the first is a substructure of the second, i.e. - $(A_{n,k}, \alpha_{n,k})\subseteq (B_{n,k}, \beta_{n,k})$, and $f_{n,k}$ - is an embedding of $(A_{n,k}, \alpha_{n,k})$ in the $\FrGr_{n-1}$ (which - is the substructure induced by $g_{n-1}$). Let + + \end{enumerate} + Before we give the third point, suppose recursively that $g_{n-1}$ already + satisfy all those properties. Let us enumerate + $\{\langle (A_{n,k}, \alpha_{n, k}), (B_{n,k}, \beta_{n,k}), f_{n, k}\rangle\}_{k\in\bN}$ + all pairs of finitely generated structures with automorphisms such + that the first substructure embed into the second by inclusion, i.e. + $(A_{n,k}, \alpha_{n,k})\subseteq (B_{n,k}, \beta_{n,k})$, and $f_{n,k}$ + is an embedding of $(A_{n,k}, \alpha_{n,k})$ in the $(\FrGr_{n-1}, g_{n-1})$. + We allow $A_{n,k}$ to be empty. Although $g_{n-1}$ is a finite function, + we may treat it as an automorphism as we have said before. + + \begin{enumerate}[resume, label=(\roman*)] + \item + Let $(i, j) = \min\{(\{0, 1, \ldots\} \times \bN) \setminus X_{n-1}\}$ (with the order induced by $\gamma$). Then $X_n = X_{n-1}\cup\{(i,j)\}$ and $(B_{n,k}, \beta_{n,k})$ embeds in $(\FrGr_n, g_n)$ so that this diagram @@ -178,13 +204,15 @@ \end{center} \end{enumerate} - First item makes sure that no infinite orbit will be present in $g$. The - second item together with the first one are necessary for $g$ to be an - automorphism of $\Gamma$. The third item is the one that gives weak - ultrahomogeneity. Now we will show that indeed such $g_n$ may be constructed. + % First item makes sure that no infinite orbit will be present in $g$. The + % second item together with the first one are necessary for $g$ to be an + % automorphism of $\Gamma$. The third item is the one that gives weak + % ultrahomogeneity. Now we will show that indeed such $g_n$ may be constructed. - First, we will suffice the item (iii). Namely, we will construct $\Gamma'_n, g'_n$ - such that $g_{n-1}\subseteq g'_n$ and $f_{i,j}$ extends to an embedding of + First, we will satisfy the item (iii). Namely, we will construct $\Gamma'_n, g'_n$ + such that $g_{n-1}\subseteq g'_n$, $\Gamma_{n-1}\subseteq\Gamma'_n$, + $g'_n$ gives an automorphism of $\Gamma'_n$ + and $f_{i,j}$ extends to an embedding of $(B_{i,j}, \beta_{i,j})$ to $(\Gamma'_n, g'_n)$. But this can be easily done by the fact, that $\cD$ has the amalgamation property. Moreover, without the loss of generality we can assume that all embeddings are inclusions. @@ -197,34 +225,48 @@ \end{tikzcd} \end{center} - By the weak ultrahomogeneity we may assume that $\Gamma'_n\subseteq \Gamma$: - - \begin{center} - \begin{tikzcd} - (B_{i,j}\cup\Gamma_{n-1}, \beta_{i,j}\cup g_{n-1}) \arrow[d, "\subseteq"'] \arrow[r, "\subseteq"] & \Gamma \\ - (\Gamma'_{n}, g'_n)\arrow[ur, dashed, "f"'] - \end{tikzcd} - \end{center} - - Now, by the WHP of $\cK$ we can extend the graph $\Gamma'_n\cup\{v_n\}$ together - with its partial isomorphism $g'_n$ to a graph $\Gamma_n$ together with its - automorphism $g_n\supseteq g'_n$ and without the loss of generality we + It is important to note that $g'_n$ should be a finite function and once + again, as it is an automorphism of a finitely generated structure, we may + think it is simply a map from one generators of $\Gamma'_n$ to the + others. By the weak ultrahomogeneity of $\Gamma$, we may assume that + $\Gamma'_n\subseteq \Gamma$. + + % \begin{center} + % \begin{tikzcd} + % B_{i,j}\cup\Gamma_{n-1} \arrow[d, "\subseteq"'] \arrow[r, "\subseteq"] & \Gamma \\ + % \Gamma'_{n}\arrow[ur, dashed, "f"'] + % \end{tikzcd} + % \end{center} + + Now, by the WHP of $\cK$ we can extend $\langle\Gamma'_n\cup\{v_n\}\rangle$ together + with its partial isomorphism $g'_n$ to a finitely generated structure $\Gamma_n$ + together with its + automorphism $g_n\supseteq g'_n$ and (again by weak ultrahomogeneity) + without the loss of generality we may assume that $\Gamma_n\subseteq\Gamma$. This way we've constructed $g_n$ that has all desired properties. Now we need to see that $g = \bigcap^{\infty}_{n=0}g_n$ is indeed an automorphism of $\Gamma$ such that $(\Gamma, g)$ has the age $\cH$ and is weakly ultrahomogeneous. It is of course an automorphism of $\Gamma$ as it is defined for every $v\in\Gamma$ - and is a sum of increasing chain of finite automorphisms. + and is an union of an increasing chain of automorphisms of finitely generated + substructures. - Take any finite structure of $T$ with automorphism $(B, \beta)$. Then, there are + Take any $(B, \beta)\in\cD$. Then, there are $i, j$ such that $(B, \beta) = (B_{i, j}, \beta_{i,j})$ and $A_{i,j}=\emptyset$. By the bookkeeping there was $n$ such that - $(i, j) = \min\{\{0,1,\ldots\}\times\bN\setminus X_{n-1}\}$. + $(i, j) = \min\{\{0,1,\ldots n-1\}\times\bN\setminus X_{n-1}\}$. This means that $(B, \beta)$ embeds into $(\Gamma_n, g_n)$, hence it embeds - into $(\Gamma, g)$, thus it has age $\cH$. - With a similar argument we can see that $(\Gamma, g)$ is weakly ultrahomogeneous. + into $(\Gamma, g)$. Hence, the age of $(\Gamma, g)$ is $\cH$. + + It is also weakly ultrahomogeneous. Having $(A,\alpha)\subseteq(B,\beta)$, + and an embedding $f\colon(A,\alpha)\to(\Gamma,g)$, we may find $n\in\bN$ + such that $(i,j) = \min\{\{0,1,\ldots n-1\}\times X_{n-1}\}$ and + $(A,\alpha) = (A_{i,j},\alpha_{i,j}), (B,\beta)=(B_{i,j},\beta_{i,j})$ and + $f = f_{i,j}$. This means that there is a compatbile embedding of $(B,\beta)$ into + $(\Gamma_n, g_n)$, which means we can also embed it into $(\Gamma, g)$. + Hence, $(\Gamma,g)\cong(\Gamma,\sigma)$. By this we know that $g$ and $\sigma$ conjugate in $G$. As we stated in the beginning of the proof, the set $A$ of possible outcomes of the game (i.e. possible $g$'s we end up with) is comeagre in $G$, thus $\sigma^G$ is also diff --git a/sections/fraisse_classes.tex b/sections/fraisse_classes.tex index 0254280..993ca73 100644 --- a/sections/fraisse_classes.tex +++ b/sections/fraisse_classes.tex @@ -277,13 +277,13 @@ $A, B, C\in\cK$ such that $C = A\cap B$ the following diagram commutes: \begin{center} \begin{tikzcd} - & A\sqcup B & \\ + & A\sqcup_C B & \\ A \ar[ur, hook] & & B \ar[ul, hook'] \\ & C \ar[ur, hook] \ar[ul, hook'] & \end{tikzcd} \end{center} - $A\sqcup B$ here is an $L$-strcuture with domain $A\cup B$ such that + $A\sqcup_C B$ here is an $L$-strcuture with domain $A\cup B$ such that for every $n$-ary symbol $R$ from $L$, $n$-tuple $\bar{a}\subseteq A\cup B$, we have that $A\sqcup B\models R(\bar{a})$ if and only if [$\bar{a}\subseteq A$ and $A\models R(\bar{a})$] or [$\bar{a}\subseteq B$ and $B\models R(\bar{a})$]. @@ -291,8 +291,7 @@ Actually we did already implicitly worked with free amalgamation in the Proposition \ref{proposition:finite-graphs-fraisse-class}, showing that - the class of finite strcuture is indeed a Fraïssé class. - + the class of finite graphs is indeed a Fraïssé class. \subsection{Canonical amalgamation} @@ -377,19 +376,9 @@ \end{center} Then, by the Fact \ref{fact:functor_iso}, $\otimes_C(\eta)$ is an automorphism - of the pushout diagram: + of the pushout diagram that looks exaclty like the diagram in the second + point of the Definition \ref{definition:canonical_amalgamation}. - \begin{center} - \begin{tikzcd} - & A\otimes_C B \ar[loop above, "\delta"] & \\ - A \ar[ur] \ar[loop left, "\alpha"]& & B \ar[ul] \ar[loop right, "\beta"]\\ - & C \ar[ur] \ar[ul] \ar[loop below, "\gamma"] & - \end{tikzcd} - \end{center} - - TODO: ten diagram nie jest do końca taki jak trzeba, trzeba w zasadzie skopiować - ten z definicji kanonicznej amalgamcji. Czy to nie będzie wyglądać źle? - This means that the morphism $\delta\colon A\otimes_C B\to A\otimes_C B$ has to be automorphism. Thus, by the fact that the diagram commutes, we have the amalgamation of $(A, \alpha)$ and $(B, \beta)$ over $(C,\gamma)$ diff --git a/sections/preliminaries.tex b/sections/preliminaries.tex index 05aa3ed..2bf254c 100644 --- a/sections/preliminaries.tex +++ b/sections/preliminaries.tex @@ -363,7 +363,7 @@ \end{fact} \begin{proof} - Suppose that $\eta_(A)$ is an isomorphism for every $A\in\cC$, where + Suppose that $\eta_{A}$ is an isomorphism for every $A\in\cC$, where $\eta_{A}\colon F(A)\to G(A)$ is the morphism of the natural transformation coresponding to $A$. Then $\eta^{-1}$ is simply given by the morphisms $\eta^{-1}_A$. diff --git a/uwagi_29_06_22.txt b/uwagi_29_06_22.txt index 805fb21..0f1ef2c 100644 --- a/uwagi_29_06_22.txt +++ b/uwagi_29_06_22.txt @@ -144,7 +144,7 @@ n and without the loss of generality we may assume that - [x] Po 4.4 powinien być wniosek, że kanoniczna amalgamacja+whp dają konkluzję 4.4, a potem z tego, że wolna amalgamacja daje 4.4
-- [ ] W sekcji 4.3 brakuje założeń.
+- [x] W sekcji 4.3 brakuje założeń.
- [x] Jak przekształcenie naturalne jest izomorfizmem, to ten składowe też są izomorfizmami (w dwie strony)
@@ -156,7 +156,7 @@ n and without the loss of generality we may assume that - [x] Dodać uwagę, że jak piszę (\Pi, \sigma) to chodzi mi o co innego niż jak piszę \Pi
-- [ ] "Odmętnić" początek dowódu 3.23
+- [x] "Odmętnić" początek dowódu 3.23
- [x] Poprawić wielkości liter przy theorem, facts itd
|