Solving linear programming with constraints unknown

DOI10.1007/978-3-662-47672-7_11zbMATH Open1440.90020arXiv1304.1247OpenAlexW2163020700MaRDI QIDQ3448779FDOQ3448779

Authors: Xiaohui Bei, Ning Chen, Shengyu Zhang 

Publication date: 27 October 2015

Published in: Automata, Languages, and Programming (Search for Journal in Brave)

Abstract: What is the value of input information in solving linear programming? The celebrated ellipsoid algorithm tells us that the full information of input constraints is not necessary; the algorithm works as long as there exists an oracle that, on a proposed candidate solution, returns a violation in the format of a separating hyperplane. Can linear programming still be efficiently solved if the returned violation is in other formats? We study this question in a trial-and-error framework: there is an oracle that, upon a proposed solution, returns the index of a violated constraint (with the content of the constraint still hidden). When more than one constraint is violated, two variants in the model are investigated. (1) The oracle returns the index of a "most violated" constraint, measured by the Euclidean distance of the proposed solution and the half-spaces defined by the constraints. In this case, the LP can be efficiently solved. (2) The oracle returns the index of an arbitrary (i.e., worst-case) violated constraint. In this case, we give an algorithm with running time exponential in the number of variables. We then show that the exponential dependence on n is unfortunately necessary even for the query complexity. These results put together shed light on the amount of information that one needs in order to solve a linear program efficiently. The proofs of the results employ a variety of geometric techniques, including McMullen's Upper Bound Theorem, the weighted spherical Voronoi diagram, and the furthest Voronoi diagram. In addition, we give an alternative proof to a conjecture of L'aszl'o Fejes T'oth on bounding the number of disconnected components formed by the union of m convex bodies in R^n. Our proof, inspired by the Gauss-Bonnet Theorem in global differential geometry, is independent of the known and reveals more clear insights into the problem and the bound.

Full work available at URL: https://arxiv.org/abs/1304.1247

Recommendations

• Linear programming using limited-precision oracles
• A combinatorial bound for linear programming and related problems
• Linear Programming in O([n3/ln n]L) Operations
• scientific article; zbMATH DE number 1099078
• How partial knowledge helps to solve linear programs

Mathematics Subject Classification ID

Linear programming (90C05) Approximation methods and heuristics in mathematical programming (90C59) Analysis of algorithms and problem complexity (68Q25) Computer graphics; computational geometry (digital and algorithmic aspects) (68U05) Abstract computational complexity for mathematical programming problems (90C60) Computational aspects related to convexity (52B55)

Cites Work

• Geometric algorithms and combinatorial optimization
• Title not available (Why is that?)
• Title not available (Why is that?)
• Linear Programming in Linear Time When the Dimension Is Fixed
• Las Vegas algorithms for linear and integer programming when the dimension is small
• A subexponential bound for linear programming
• On convex cones
• Information collection for linear programs with uncertain objective coefficients
• Title not available (Why is that?)
• Linear programming without the matrix
• On a Multidimensional Search Technique and Its Application to the Euclidean One-Centre Problem
• Title not available (Why is that?)
• Recognizing Dirichlet tesselations
• Recognizing Voronoi Diagrams with Linear Programming
• Solving linear programming with constraints unknown
• A property of convex sets and its application
• On the complexity of trial and error
• On the complexity of trial and error for constraint satisfaction problems

Cited In (6)

• Unrestricted variables in linear programming
• Solving nonlinear programming problems with very many constraints
• Solving linear optimization over arithmetic constraint formula
• On the complexity of trial and error for constraint satisfaction problems
• Title not available (Why is that?)
• Solving linear programming with constraints unknown

This page was built for publication: Solving linear programming with constraints unknown

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3448779)