DuckDB features a sophisticated optimizer that significantly enhances its query performance, particularly in OLAP scenarios. The optimizer's structure is clear and extensible, housing numerous optimization rules that developers can learn from. The article zeroes in on explaining these rules rather than detailing the source code, emphasizing the underlying algorithms that drive the optimization process. A notable aspect is the EXPRESSION_REWRITER rule, which receives a source-code-level analysis. The Optimizer class encapsulates DuckDB's optimization logic and includes key components such as a binder for adding operators, an expression rewriter for managing optimization rules, and methods for running built-in optimizers and verifying logical plans. The core function, Optimize, takes an unoptimized logical plan and processes it through a series of built-in and custom optimizers. The process consists of three main parts: executing pre-optimization functions from any extensions, running DuckDB's built-in optimizers, and then applying additional optimization functions. DuckDB's optimizer supports 26 built-in rules like EXPRESSION_REWRITER, JOIN_ORDER, and FILTER_PULLUP. Some of these rules are complex, particularly those related to join reordering and subquery handling. Developers can also create custom optimization rules by inheriting from the OptimizerExtension class, which allows for greater flexibility in enhancing the optimizer's capabilities. The article references a demo file that illustrates how to implement a custom rule, inviting readers to explore further.