References & Further Reading

References

1. L. Ramalho, Fluent Python, O'Reilly Media, 2nd ed., 2022

   Chapters 7-9 on closures, decorators, and design patterns with first-class functions are the definitive Python-specific treatment. The coverage of `functools.wraps`, decorator stacking, and parameterized decorators is particularly thorough.

2. D. Beazley and B. K. Jones, Python Cookbook, O'Reilly Media, 3rd ed., 2020

   Chapter 9 covers metaprogramming with decorators, including decorators that can work with or without arguments, class decorators, and decorators that preserve type signatures.

3. C. Hattingh, Using Asyncio in Python, O'Reilly Media, 2020

   Covers context managers in the async context (`async with`), which extends the patterns from this chapter to concurrent scientific computing.

4. K. Smith and J. Montanaro, PEP 318 — Decorators for Functions and Methods, 2003. [Link]

   The original proposal for decorator syntax in Python. The "Design Goals" and "Current Syntax" sections explain why the `@` symbol was chosen.

5. G. van Rossum and N. Coghlan, PEP 343 — The 'with' Statement, 2005. [Link]

   Defines the context manager protocol and the `with` statement. Section on `contextlib` utilities is particularly relevant.

6. Python Software Foundation, Python 3.12 Documentation — functools, 2024. [Link]

   Official documentation for `functools.wraps`, `partial`, `lru_cache`, `cache`, `reduce`, and `singledispatch`.

7. Python Software Foundation, Python 3.12 Documentation — contextlib, 2024. [Link]

   Official documentation for `contextmanager`, `suppress`, `ExitStack`, `redirect_stdout`, and `nullcontext`.

Further Reading

• Advanced decorator patterns

  G. van Rossum, 'Five-Minute Multimethods in Python' (blog post, 2005)

  Demonstrates how decorators can implement multiple dispatch (function overloading based on argument types). This idea evolved into `functools.singledispatch` in Python 3.4.

• Context managers in scientific libraries

  PyTorch documentation — torch.no_grad, torch.cuda.amp.autocast

  Shows how production ML frameworks use context managers for gradient control and mixed-precision training. These are the patterns you will use when scaling to GPU computing in Part 3.

• Functional programming in Python

  M. Mertz, *Functional Python Programming*, Packt, 3rd ed., 2022

  Comprehensive coverage of functional patterns in Python: higher-order functions, closures, immutability, and lazy evaluation. Goes deeper than this chapter on reduce, partial application, and function composition.

• Property-based testing for decorated functions

  Hypothesis library documentation (https://hypothesis.readthedocs.io/)

  When writing decorators that modify function behavior, property-based testing can verify that invariants are preserved. Hypothesis generates random inputs to find edge cases your unit tests miss.