Integer Linear-Exponential Programming in NP by Quantifier Elimination

Dmitry Chistikov,Alessio Mansutti,Mikhail R. Starchak

DOI: https://doi.org/10.4230/LIPIcs.ICALP.2024.132

2024-07-10

Abstract:This paper provides an NP procedure that decides whether a linear-exponential system of constraints has an integer solution. Linear-exponential systems extend standard integer linear programs with exponential terms $2^x$ and remainder terms ${(x \bmod 2^y)}$. Our result implies that the existential theory of the structure $(\mathbb{N},0,1,+,2^{(\cdot)},V_2(\cdot,\cdot),\leq)$ has an NP-complete satisfiability problem, thus improving upon a recent EXPSPACE upper bound. This theory extends the existential fragment of Presburger arithmetic with the exponentiation function $x \mapsto 2^x$ and the binary predicate $V_2(x,y)$ that is true whenever $y \geq 1$ is the largest power of $2$ dividing $x$. Our procedure for solving linear-exponential systems uses the method of quantifier elimination. As a by-product, we modify the classical Gaussian variable elimination into a non-deterministic polynomial-time procedure for integer linear programming (or: existential Presburger arithmetic).

Logic in Computer Science

What problem does this paper attempt to address?

This paper aims to solve a specific computational problem, that is, how to determine whether a linear - exponential system has an integer solution. The linear - exponential system extends the standard integer linear programming by adding exponential terms of the form \(2^x\) and remainder terms of the form \(x \mod 2^y\). Specifically, the main contributions of the paper can be summarized as follows:

1. **Problem Definition**: The paper studies a special class of constraint systems - linear - exponential systems. In addition to linear inequalities, these systems also contain constraint conditions of the form \(y = 2^x\) and \(z=(x \mod 2^y)\).

2. **Main Result**: The author provides an algorithm in the NP complexity class for determining whether a linear - exponential system has an integer solution. This result implies that the existential theory of the structure \((\mathbb{N}, 0, 1, +, 2(\cdot), V_2(\cdot, \cdot), \leq)\) has an NP - complete satisfiability problem, thus improving a recent ExpSpace upper bound.

3. **Methodology**: The paper uses the method of quantifier elimination to solve the problem of linear - exponential systems. As a by - product, the author modifies the classical Gaussian variable elimination method into a non - deterministic polynomial - time process for solving integer linear programming problems (or called existential Presburger arithmetic).

4. **Theoretical Background**: The paper discusses the logical theory related to Buchi - Semenov arithmetic, which is a theory that extends linear arithmetic by adding the exponential function \(x\mapsto2^x\) and the binary predicate \(V_2(x, y)\), where the latter represents the largest power of 2 that can divide \(x\).

5. **Application Example**: The paper demonstrates the powerful capabilities of linear - exponential systems through the problem of computing the discrete logarithm to the base 2. Given non - negative integers \(m, r\in\mathbb{N}\), find an \(x\in\mathbb{N}\) such that \(2^x - r\) is divisible by \(m\). This problem can be solved by checking the feasibility of at most \(\log m\) linear - exponential systems.

6. **Technical Details**: The paper describes in detail the specific implementation of the algorithm, including how to solve linear - exponential systems through a non - deterministic polynomial - time process, and provides the pseudo - code of the algorithm and the introduction of key concepts, such as quotient systems, delayed substitution, and the original linear - exponential system.

In conclusion, through the introduction of new algorithms and techniques, this paper successfully classifies the solvability problem of linear - exponential systems into the NP class, providing an important theoretical basis and practical tool for research in related fields.