Put $P=[0,1]$. Is there a compact subset $L$ of the hyper space of $P$ such that the pair $(P, L)$ satisfies the following axioms of projective geometry. Furthermore the obvious maps from the configuration space of $P$ to $L$ and configuration space of $L$ to $P$ would be continuous?
-
1$\begingroup$ I like this question while I still do not follow the second part, about configuration spaces and about the natural mappings. I am not even sure about the definition of the hyperspace (Different papers apply different definitions which are not equivalent). $\endgroup$– Włodzimierz HolsztyńskiCommented Mar 16, 2017 at 0:27
-
1$\begingroup$ I am quite sure that the hyperspace is defined as the space of compact subsets (possibly non-empty compact subsets) with the Hausdorff metric. (?) Then my guess would be no if line $\ p(a\ b)\ $ depends continuously on $\ \{a\ b\}\ $ for $\ a\ne b$. $\endgroup$– Włodzimierz HolsztyńskiCommented Mar 16, 2017 at 0:49
-
1$\begingroup$ @WłodzimierzHolsztyński Thank you for your comments on my question. By Hyperspace of $P$,, as you pointed out, I mean the space of all non empty compact subsets of $P$ which is a compact space with Hausdorff metric. The obvious map sends the pair $(a,b)\in P\times P\;\;\; a\neq b$, the configuration space, to the unique line $l(a,b)$. $\endgroup$– Ali TaghaviCommented Mar 19, 2017 at 16:27
-
$\begingroup$ (To do LaTeX on MO is hard :) ). $\endgroup$– Włodzimierz HolsztyńskiCommented Mar 19, 2017 at 16:29
-
1$\begingroup$ @WłodzimierzHolsztyński Dear Prof. Holsztynski My apology I did not recognized you. I have deep admiration to you. I have admiration also to the mathematical heritage of Prof. Borsuk. Thanks again for your attention to my question. $\endgroup$– Ali TaghaviCommented Mar 19, 2017 at 22:07
2 Answers
If $(P,L)$ is an abstract projective plane, then for any point $p\in P$ and any line $\ell\in L$ not incident to $p$ there is a bijection between the set of points incident to $\ell$ and the set of lines incident to $p$. Under a reasonable definition of "topological projective plane" this bijection should be a homeomorphism.
If $P=[0,1]$, then the space of all lines incident to $0$ has a continuous surjection from the connected space $(0,1]$. It follows that for each line $\ell$ the space of all points incident to $\ell$ is connected. But one of these lines has both $0$ and $1$ in it, and the only connected subset of an interval containing both endpoints is the whole interval. Contradiction.
-
$\begingroup$ Prof. GoodWillie thank you very much for your interesting answer. I am thinking to replacement of other spaces by the interval for example circle, spheres,.etc. $\endgroup$ Commented Mar 21, 2017 at 12:46
-
$\begingroup$ It's rather clear that the topological projective planes should be topologically homogenous. In the compact case they cannot be 1-dimensional. $\endgroup$ Commented Mar 21, 2017 at 23:33
-
$\begingroup$ Assuming these are Hausdorff spaces, the plane must be locally homeomorphic to the product of two of its lines, so it seems that it must be even-dimensional. $\endgroup$ Commented Mar 22, 2017 at 11:59
As a useful (I hope) step toward solving Ali Taghavi's problem, let me propose the following general definition of a topological projective plane:
DEFINITION A topological projective plane is an ordered quadruple $\ \mathbf P\ :=\ (\,P\ L\ \vee \wedge\,)\ $ such that $\ P\ $ and $\ L\ $ are topological spaces, and $\ \vee:\binom P2\rightarrow L\ $ and $\ \wedge:\binom L2\rightarrow P\ $ are continuous functions, and $\ \mathbf P\ $ is a projective plane in the usual (non-topological, just abstract) sense, i.e. the following axioms hold:
$\ \forall_{\{a\ b\ c\}\in\binom P3}\ (\,a\vee b = b\vee c\ \Rightarrow\ a\vee c=a\vee b\,) $
$\ \forall_{\{A\ B\ C\}\in\binom L3}\ (\,A\wedge B = B\wedge C\ \Rightarrow\ A\wedge C = A\wedge B\,) $
$\ \forall_{\{A\ B\ C\}\in\binom L3}\ \left( A\wedge B=A\wedge C\quad or \quad (A\wedge B)\vee(A\wedge C) = A \right) $
$\ \forall_{\{a\ b\ c\}\in\binom P3}\ \left(a\vee b=a\vee c\quad or \quad (a\vee b)\wedge(a\vee c) = a \right) $
$\ \exists_{E\in\binom P4}\ \left|\left\{x\vee y: \{x\ y\}\in\binom E2\right\}\right| = \binom 42 $
REMARK Axioms 1 and 2 can be written in the style of axioms 2 and 3 as follows:
1'. $\ \forall_{T\in\binom P3}\ \left|\left\{x\vee y: \{x\ y\}\in\binom T2\right\}\right| \ =\ 1\ \text{or}\ \ 3 $
2'. $\ \forall_{t\in\binom L3}\ \left|\left\{X\wedge Y: \{X\ Y\}\in\binom t2\right\}\right|\ =\ 1\ \text{or}\ 3 $
Now one can impose additional constraints, perhaps topological, on the notion of the topological projective plane to obtain more specialized (narrower) classes.
Acknowledgment The definition here is a simplification and generalization of the definition given by Ali Taghavi from the Question above.
**ADDITIONAL DEFINITIONS**
Now we may define the induced linear sets and pencils
$$ \forall_{A\in L}\ \ _\{A_\}\ :=\ \bigcup \vee^{-1}(A)\ =\ \bigcup \left\{ \{a\ b\}\in \binom P2: a\vee b = A \right\} $$ and $$ \forall_{a\in P}\ \ ^\{p^\}\ :=\ \bigcup \wedge^{-1}(a)\ =\ \bigcup \left\{ \{A\ B\}\in \binom L2: A\wedge B = a \right\} $$
The (default) topology in $\ \binom X2\ $ is induced by the canonical map $\ X\times X\setminus\{(x\ x):x\in X\}\rightarrow\binom X2\ $ given by $\ (x\ y)\mapsto\{x\ y\},\ $ where $\ X\ $ is $\ P\ $ or $\ L$.
-
$\begingroup$ Ali, thank you for correcting my typos--I even similarly replaced the LaTeX \vee by \wedge in axiom 2'. $\endgroup$ Commented Mar 20, 2017 at 21:42