Decentralized combinatorial optimization

Abstract

Combinatorial optimization is a widely-studied class of computational problems with many theoretical and real-world applications. Optimization problems are typically tackled using hardware and software controlled by the user. Optimization can be competitive where problems are solved by competing agents in isolation, or by groups sharing hardware and software in a distributed manner. Blockchain technology enables decentralized applications (DApps). Optimization as a DApp would be run in a trustless manner where participation in the system is voluntary and problem-solving is incentivized with bitcoin, ether, or other fungible tokens. Using a purpose-built blockchain introduces the problem of bootstrapping robust immutability and token value. This is solved by building a DApp as a smart-contract on top of an existing Turing-complete blockchain platform such as Ethereum. We propose a means of using Ethereum Virtual Machine smart contracts to automate the payout of cryptocurrency rewards for market-based voluntary participation in the solution of combinatorial optimization problems without trusted intermediaries. We suggest use of this method for optimization-as-a-service, automation of contests, and long-term recording of best-known solutions.