Most methods courses teach you to solve the problems that can be solved, then fall silent about the far larger class that cannot. This book does not.
Its first thirteen chapters develop the exact methods without apology, because they are beautiful and they work: contour integration, transforms, eigenfunction expansions, Green's functions, and tensors. Its final three concede what every practising scientist discovers, that real equations are nonlinear, or contain a small parameter, and almost never possess closed form solutions.
What then? You approximate, honestly, with a stated error and a way to check it.
WHAT'S INSIDE
Complex Analysis (Chapters 1 to 4). Analyticity, Cauchy's theorem, Laurent series, residues, and conformal mapping. Evaluate real integrals beyond the reach of real variable methods.Transforms (Chapters 5 to 6). Fourier and Laplace transforms with contour inversion. Differential equations become algebra.Differential Equations (Chapters 7 to 11). Series solutions, Bessel and Legendre functions, Sturm, Liouville theory, the Fredholm alternative, partial differential equations, and Green's functions.Variational and Tensor Methods (Chapters 12 to 13). Deriving physical laws from minimisation and expressing them in forms valid in every reference frame.Asymptotics and Perturbation Theory (Chapters 14 to 16). Divergent asymptotic series, WKB methods, secular terms, limit cycles, and Prandtl's boundary layer.TWO COMMITMENTSEvery number was computed, not quoted. Every derivation was checked symbolically. Every worked example was evaluated independently. Whenever predicted error scaling was testable numerically, the result appears exactly as obtained. Quoted results are identified as such.
The failures remain. A method that works is instructive. A method that fails is more instructive when its limits are explained. You will find a WKB calculation that fails. You will watch a perturbation solution leave its domain of validity and miss by a factor of three. You will read plainly that Prandtl's theory collapses at separation and offers no warning of its own breakdown.
BUILT FOR ACTUAL STUDY
- Worked examples with every step, unit, and intermediate value shown.
- Hundreds of practice problems with complete worked solutions and reasoning.
- Eight reference appendices covering transform tables, residues, special functions, vector and tensor identities, Green's functions, and asymptotic results. Every formula was verified computationally.
- A final appendix, Ten Ways to Get It Wrong, collecting common mistakes that produce convincing but incorrect answers.
- Complete glossary, symbol list, and index.
WHO IT IS FOR
Advanced undergraduates, graduate students, and practising engineers and physicists. Multivariable calculus, linear algebra, and a first course in differential equations are assumed. No prior background in complex analysis, transforms, or asymptotics is required.
The habit this book aims to leave behind is not the ability to evaluate a contour integral. It is the habit of treating failure as evidence. Notice when an approximation breaks down. Ask what its failure reveals. Read the answer.
A divergent series announces that it is blind to an exponentially small effect. A secular term announces that a frequency is shifting. A boundary condition that cannot be satisfied announces that the solution has retreated into a thin region.