English

Data-Driven Solution Portfolios

Data Structures and Algorithms 2024-12-03 v1

Abstract

In this paper, we consider a new problem of portfolio optimization using stochastic information. In a setting where there is some uncertainty, we ask how to best select kk potential solutions, with the goal of optimizing the value of the best solution. More formally, given a combinatorial problem Π\Pi, a set of value functions VV over the solutions of Π\Pi, and a distribution DD over VV, our goal is to select kk solutions of Π\Pi that maximize or minimize the expected value of the {\em best} of those solutions. For a simple example, consider the classic knapsack problem: given a universe of elements each with unit weight and a positive value, the task is to select rr elements maximizing the total value. Now suppose that each element's weight comes from a (known) distribution. How should we select kk different solutions so that one of them is likely to yield a high value? In this work, we tackle this basic problem, and generalize it to the setting where the underlying set system forms a matroid. On the technical side, it is clear that the candidate solutions we select must be diverse and anti-correlated; however, it is not clear how to do so efficiently. Our main result is a polynomial-time algorithm that constructs a portfolio within a constant factor of the optimal.

Keywords

Cite

@article{arxiv.2412.00717,
  title  = {Data-Driven Solution Portfolios},
  author = {Marina Drygala and Silvio Lattanzi and Andreas Maggiori and Miltiadis Stouras and Ola Svensson and Sergei Vassilvitskii},
  journal= {arXiv preprint arXiv:2412.00717},
  year   = {2024}
}

Comments

Accepted at ITCS 2025

R2 v1 2026-06-28T20:18:25.338Z