Software Evolution

Software evolution studies how software systems change over time, aiming to understand and improve the processes of software development, maintenance, and adaptation. Current research focuses on leveraging large language models (LLMs) for automated code generation, improvement, and bug fixing, often employing multi-agent frameworks and retrieval-augmented generation techniques. These advancements are impacting various areas, including automated software model completion, efficient vision transformer acceleration, and even the evolution of code in adversarial environments, ultimately improving software quality, development speed, and security. The broader impact lies in enabling more efficient and robust software systems through automated processes and improved understanding of software's evolutionary dynamics.