Dori Medini | Searcharxiv

arXiv2021

Hitting Sets and Reconstruction for Dense Orbits in $\text{VP}_e$ and $ΣΠΣ$ Circuits

In this paper we study polynomials in $\text{VP}_e$ (polynomial-sized formulas) and in $ΣΠΣ$ (polynomial-size depth-$3$ circuits) whose orbits, under the action of the affine group $\text{GL}_n^{\text{aff}}(\mathbb{F})$, are $\mathit{dense}$ in their ambient class. We construct hitting sets and interpolating sets for these orbits as well as give reconstruction algorithms. As $\text{VP}=\text{VNC}^2$, our results for $\text{VP}_e$ translate immediately to $\text{VP}$ with a quasipolynomial blow up in parameters. If any of our hitting or interpolating sets could be made $\mathit{robust}$ then this would immediately yield a hitting set for the superclass in which the relevant class is dense, and as a consequence also a lower bound for the superclass. Unfortunately, we also prove that the kind of constructions that we have found (which are defined in terms of $k$-independent polynomial maps) do not necessarily yield robust hitting sets.

cs.CC↗

arXiv2018

Ranked Enumeration of Minimal Triangulations

A tree decomposition of a graph facilitates computations by grouping vertices into bags that are interconnected in an acyclic structure, hence their importance in a plethora of problems such as query evaluation over databases and inference over probabilistic graphical models. The relative benefit from different tree decompositions is measured by diverse (sometime complex) cost functions that vary from one application to another. For generic cost functions like width and fill-in, an optimal tree decomposition can be efficiently computed in some cases, notably when the number of minimal separators is bounded by a polynomial (due to Bouchitte and Todinca), we refer to this assumption as "poly-MS." To cover the variety of cost functions in need, it has recently been proposed to devise algorithms for enumerating many decomposition candidates for applications to choose from using specialized, or even machine-learned, cost functions. We explore the ability to produce a large collection of "high quality" tree decompositions. We present the first algorithm for ranked enumeration of the proper (non-redundant) tree decompositions, or equivalently minimal triangulations, under a wide class of cost functions that substantially generalizes the above generic ones. On the theoretical side, we establish the guarantee of polynomial delay if poly-MS is assumed, or if we are interested in tree decompositions of a width bounded by a constant. We describe an experimental evaluation on graphs of various domains (including join queries, Bayesian networks, treewidth benchmarks and random), and explore both the applicability of the poly-MS assumption and the performance of our algorithm relative to the state of the art.

cs.DS↗