Efficient Coins Selection for UTXOs through Evolutionary and Random Draw Methods

Abstract

This study proposes and evaluates a novel hybrid optimization framework that integrates stochastic Random Draw sampling with Evolutionary Algorithms (EA) to address the multi-objective challenges of Unspent Transaction Output (UTXO) selection. In digital asset management, selecting an optimal subset of coins from fragmented wallet pools requires balancing transaction privacy, economic efficiency, and computational throughput-a task complicated by a combinatorial search space exceeding 10^41 possibilities. The proposed methodology utilizes a random-draw seeding mechanism to bypass initial combinatorial complexity, followed by iterative evolutionary refinement. Results demonstrate that this hybrid approach achieves a 97% success rate and a convergence speed 43% faster than standard optimization techniques, reaching optimal solutions in an average of 14 ms. Furthermore, the inclusion of mutation and crossover operators ensures high UTXO diversity, significantly enhancing privacy by mitigating identifiable transaction patterns common in deterministic heuristics. This research concludes that the hybrid model provides a robust, scalable solution for real-time wallet infrastructures, effectively reconciling the trade-offs between long-term wallet health and immediate transaction requirements.