Learning wave propagation theory is harder than most professors admit
The math looks clean on paper. The wave equation is deceptively simple: ²u/t² = c²²u. You write it down, you feel clever, then you try to solve anything with actual boundaries and your confidence evaporates. Most people bounce off this subject because the textbooks skip the parts that actually matter in practice. They show you the ideal case, the infinite domain, the perfect initial conditions, and then wonder why nobody can model signal decay in a real medium. The best lectures online tend to come from university departments that actually use these models in research. MIT OpenCourseWare has Walter Greiner's treatments that stay closer to applied physics than the typical pure math derivation. Stanford's computional electromagnetics lectures by Jeffrey Pendry cover dispersion and scattering in ways that help you understand what the equations are actually describing. The University of Colorado Boulder has solid wave mechanics lectures by Steve Simon that include practical problem-solving rather than just theorem-stating. YouTube channels like Michel van Biezen work through example problems step by step, which helps when you're stuck on the calculus parts. For something more engineering-oriented, check out NPTEL's courses from IIT Bombay on electromagnetic wave theory. I spent weeks going through different lecture series before settling on a combination. The problem with most single-source approaches is that they assume a background you don't have yet. You need someone who can derive the Helmholtz equation from Maxwell's equations in one video and then show you why you'd use it for antenna design in the next.
Here's what most lecture series won't tell you: you need to work through problems alongside the theory. Watching someone derive the reflection coefficient at an interface teaches you nothing if you haven't tried setting up the boundary conditions yourself first. I always worked through a derivation, paused the lecture, and then checked my work. The gap between thinking you understand and actually being able to produce the result on a blank page is where most students fall behind.
The actual sticking points nobody mentions
Group velocity versus phase velocity trips everyone up at first. The definitions are straightforward, but understanding when they diverge and why that divergence matters for signal integrity is where things get real. In a dispersive medium, the group velocity can actually drop below c without violating relativity, and the phase velocity can exceed c entirely. This isn't a paradox, it's just that information travels at neither of those velocities in general cases. Another thing that catches people off guard: the difference between evanescent waves and propagating waves. Near-field effects in waveguides and antennas rely entirely on evanescent modes, which decay exponentially. These don't carry power in the traditional sense, but they dominate the behavior of structures at sub-wavelength scales. If you're only comfortable with traveling wave solutions, you'll fail any practical problem involving apertures, probes, or coupling structures. I ran into this specific issue when trying to model ultrasound transducer arrays for a medical imaging project. The simulation kept producing unrealistic results near the transducer face, and it took me two days to realize the mesh was too coarse to resolve the evanescent field decay. The near-field region extends roughly d²/ from the aperture, and within that region, standard far-field approximations break down completely. Switching to a finer mesh and applying the correct near-field Green's function instead of the plane wave assumption fixed it. That's the kind of detail most lecture courses gloss over.
Get the Full Details
Building actual understanding
You need a foundation in partial differential equations and complex analysis before wave propagation theory makes sense. If you can't comfortably manipulate Fourier transforms or solve second-order linear ODEs with constant coefficients, the jump to PDEs will be rough. Spend a week or two on those prerequisites if you're rusty. It saves months of frustration later. When studying the material, organize your notes around solution methods rather than topics. The separation of variables approach applies to rectangular, cylindrical, and spherical geometries, and seeing those three derivations side by side reveals patterns that isolated topic-based notes hide. The same idea recurs in eigenfunction expansions, Green's functions, and integral transform methods. Recognizing the structural similarity between these approaches is what turns memorization into understanding. Simulation tools help but they also create a false sense of comprehension. Running a finite element simulation in COMSOL or Ansys HFSS and getting a pretty color plot doesn't mean you understand what's happening. I always verify my simulation results against hand calculations for at least one simple geometry. If the simulation and the analytic solution disagree, something is wrong with the model setup, not the math.
One more thing that helps: work through at least one problem where you have to derive the dispersion relation from scratch. Start with a physical system—maybe a string with discrete masses, or a photonic crystal, or even a simple LC ladder network—and build up to the continuum limit. This connects the abstract math to something tangible and makes the concept of band gaps and cutoff frequencies much less mysterious.
When the standard approach fails
Wave propagation theory has real limitations. The linear wave equation assumes small perturbations. Once amplitudes get large enough for nonlinear effects to matter—shock waves, harmonic generation, soliton formation—the whole framework changes. Geometric optics breaks down at diffraction limits. Time-domain and frequency-domain approaches each have regimes where one is dramatically more efficient than the other, and picking the wrong one can turn a thirty-minute calculation into a hours-long exercise in numerical instability. If you're working with highly irregular geometries or heterogeneous media where analytical solutions don't exist, numerical methods become necessary. But numerical methods introduce their own failure modes: dispersion errors in FDTD schemes, boundary reflection artifacts if your PML layers aren't set up correctly, mesh convergence issues that require systematic refinement studies. These are the problems that separate people who can do homework from people who can produce reliable results in real work. The most useful skill you can develop is knowing which approximation is valid in which regime. That comes from solving enough problems across different contexts that you start recognizing the underlying structure rather than treating each problem as unique. The lectures give you the framework. The practice gives you the judgment.
