Functional Differential Geometry by Gerald Jay Sussman, Jack Wisdom, and Will Farr presents differential geometry as a computational and conceptual language for physics, particularly general relativity and classical mechanics. The authors aim to make mathematical ideas precise by combining geometric reasoning with functional programming. Rather than relying heavily on traditional index notation, they emphasize vector fields, one-forms, differential forms, and operations that have clear semantic meaning. 

The book develops the basic structure of manifolds and then introduces vector fields, one-form fields, coordinate bases, integral curves, and changes of basis. It proceeds to integration in higher dimensions, the exterior derivative, and Stokes's theorem, showing how these concepts provide a unified framework for classical vector calculus. A central theme is that mathematical objects should be defined independently of particular coordinates whenever possible, making their geometric meaning easier to understand. 

A major portion of the text concerns derivatives on manifolds. The authors introduce the Lie derivative, covariant derivative, parallel transport, and geodesics, followed by curvature, torsion, geodesic deviation, and the Bianchi identities. Metrics are subsequently introduced to connect geometry with physical quantities, including the Lagrange equations and applications to relativity. The later chapters apply the framework to the Hodge star, electrodynamics, the wave equation, and special relativity, including Lorentz transformations and the twin paradox.

One of the book's distinctive ideas is that programming can function as a tool for mathematical understanding, rather than merely numerical computation. A computer requires formulas to be sufficiently precise and unambiguous, so translating mathematics into executable functional programs can expose hidden assumptions, ambiguities, and errors. For example, the authors reinterpret the Lagrange equation $ \frac{d}{dt}\frac{\partial L}{\partial \dot q}-\frac{\partial L}{\partial q}=0 $ by explicitly specifying the path $q=w(t)$ and its derivative. Their broader philosophy is that formalization, computation, and debugging can reinforce mathematical understanding.