2. Graph Theory Motivated Questions
-
Thin Tree Conjecture
If you have a graph $G$, and it is $k$-edge connected for some large $k$. That is, for every partition of the vertices into 2 parts $A \subseteq V(G)$ and $A^c$ there exists at least $k$ edges going from $A$ to $A^c$. A spanning tree is $\alpha$-thin with respect to $G$ if for every cut $A \cup A^c$, the number of edges in $T$ going from $A$ to $A^c$ is at most $\alpha E(A,A^c)$, where $E(A,A^c)$ is the number of edges from $A$ to $A^c$ in $G$.Conjecture 2.1.
There is a $k_0$ (independent of $|V|$) so that every $k_0$-edge connected graph has a $\frac{1}{2}$-thin spanning tree. -
Negative correlation for random forests
Problem 2.2.
For what $\epsilon$ is it true that $\Pr(e \in F \text{ and }f \in F) \le (1+\epsilon)\Pr(e \in F)\Pr(f \in F)$ for $F$ a uniformly random spanning forest? -
Problem 2.3.
Let $G = (V,E)$ be a graph of maximum degree $\Delta$. Let $\chi(t)$ be the chromatic polynomial (i.e. $\chi(q)$ is the number of ways of properly coloring the graph $G$ with $q$ colors). Then, is $\chi(t) \neq 0$ for any $t \in \C$ so that $|t| > 2\Delta$. Also if the real part of $\text{Re }t > \Delta + 1$, then do we have $\chi(t) \neq 0$?
Cite this as: AimPL: The geometry of polynomials in combinatorics and sampling, available at http://aimpl.org/geompolysampling.