SearcharxivSearch

arXiv subjects

Rogozin Alexander

Publications and source records attributed to Rogozin Alexander.

2 recordsLinked to original sources

On the Vertices of Delta-modular Polyhedra

Let $P$ be a polytope defined by the system $A x \leq b$, where $A \in R^{m \times n}$, $b \in R^m$, and $\text{rank}(A) = n$. We give a short geometric proof of the following tight upper bound on the number of vertices of $P$: $$ n! \cdot \frac{\Delta}{\Delta_{\text{average}}} \cdot \text{vol}(B_2) \sim \frac{1}{\sqrt{\pi n}} \cdot \left(\frac{2 \pi}{e}\right)^{n/2} \cdot n^{n/2} \cdot \frac{\Delta}{\Delta_{\text{average}}}, $$ where $\Delta$ is the maximum absolute value of $n \times n$ subdeterminants of $A$, and $\Delta_{\text{average}}$ is the average absolute value of subdeterminants of $A$ corresponding to a triangulation of $P$'s normal fan. Assuming that $A$ is integer, such polyhedra are called $\Delta$-modular polyhedra. Note that in the integer case, the bound can be simplified via the inequality $\Delta_{\text{average}} \geq \Delta_{\min} \geq 1$, where $\Delta_{\min}$ is the minimum absolute value of subdeterminants of $A$ corresponding to feasible bases of $A x \leq b$. For this, we prove and use a symmetric variant of Macbeath's theorem. Additionally, we give a direct argument based on prior results in the field, showing that the graph diameter of $P$ is bounded by $O\bigl(n^3 \cdot \frac{\Delta}{\Delta_{\min}} \cdot \ln (n \frac{\Delta}{\Delta_{\min}}) \bigr)$. Thus, both characteristic of $P$ are linear in $\Delta/\Delta_{\min}$. From an algorithmic perspective, we demonstrate that: Given $A \in Q^{m \times n}$, $b \in Q^m$, and an initial feasible solution to $A x \leq b$, the convex hull of $P$ can be constructed in $O(n)^{n/2} \cdot m^2 \cdot \frac{\Delta}{\Delta_{\text{average}}}$ operations. For simple polyhedra, the dependence on $m$ reduces to linear; Given $A \in Z^{m \times n}$ and $b \in Q^m$, the number $|P \cap Z^n|$ can be computed in $O(n)^n \cdot \frac{\Delta^4}{\Delta_{\text{average}}}$ arithmetic operations.

math.CO

An acceleration of decentralized SGD under general assumptions with low stochastic noise

Distributed optimization methods are actively researched by optimization community. Due to applications in distributed machine learning, modern research directions include stochastic objectives, reducing communication frequency, and time-varying communication network topology. Recently, an analysis unifying several centralized and decentralized approaches to stochastic distributed optimization was developed in Koloskova et al. (2020). In this work, we employ a Catalyst framework and accelerate the rates of Koloskova et al. (2020) in the case of low stochastic noise.

math.OC