Partial Differential Equations Interview Prep
PDE interview prep — heat, wave, and Laplace equations, separation of variables, Fourier methods, Green's functions, finite differences, and finite elements.
Quick answer
Partial Differential Equations (PDEs) describe how multivariate functions evolve under rules involving partial derivatives.
PDE questions test whether candidates can move fluently between the integral and differential forms of physical laws and between continuous models and their numerical discretizations.
Editorial review
Written by
CompoundLearn editorial team
Wireless / RF / hardware engineering
Reviewed by
CompoundLearn editorial team
Wireless / RF / hardware engineering
Last reviewed
Built from curated topic maps, editorial validation, and subject-matter review so the page stays aligned with the interview intent and the current content pipeline.
What it is
Partial Differential Equations (PDEs) describe how multivariate functions evolve under rules involving partial derivatives. Standard topics include the classification of second-order linear PDEs into elliptic (Laplace, Poisson), parabolic (heat, diffusion), and hyperbolic (wave, transport) types; well-posedness and boundary conditions (Dirichlet, Neumann, Robin); separation of variables and the resulting eigenvalue problems on rectangular, cylindrical, and spherical domains; Fourier series and integral transform methods (Fourier, Laplace) for solving linear PDEs on infinite domains; Green's functions and fundamental solutions; the maximum principle and uniqueness arguments for elliptic and parabolic equations; method of characteristics for first-order and hyperbolic equations; numerical methods including finite differences, finite volumes, and finite elements; the CFL stability condition for explicit time-marching schemes; and weak/variational formulations as the foundation for FEM. For engineering candidates, PDEs are the language of fields and continuous systems: Maxwell's equations are PDEs in the electric and magnetic fields, the heat equation governs every thermal-management problem, the Navier-Stokes equations govern fluid dynamics, the Schrodinger equation governs quantum systems including semiconductor band-structure calculations, and structural mechanics is governed by PDEs for displacement and stress.
Why interviewers ask
PDE questions test whether candidates can move fluently between the integral and differential forms of physical laws and between continuous models and their numerical discretizations. EM and antenna interviews probe whether you can derive the wave equation from Maxwell's curl-curl identity, separate variables on a waveguide cross-section, or apply Green's functions for radiation problems. Thermal-management interviews test the heat equation: separation of variables, similarity solutions, and numerical timestepping for transient problems. Fluid-dynamics interviews probe Navier-Stokes simplifications (Stokes flow, boundary-layer equations, potential flow) and the dimensional analysis (Reynolds number, Peclet number) that selects between regimes. Numerical-method interviews test discretization choices (finite difference vs FEM vs FV), stability conditions (CFL), and convergence rates. The strongest candidates connect each PDE to its physical interpretation, sketch boundary-condition setups for representative geometries, and quote standard fundamental solutions (Coulomb potential, Gaussian heat kernel, retarded potentials) without re-deriving them. Interviewers also use PDE questions to probe analytical maturity: candidates who can sketch a phase plane for a non-linear system, recognize a similarity variable in a heat-equation problem, or identify when separation of variables fails (irregular geometry, mixed BCs) signal years of applied-math work, not just exposure.
Common mistakes
The most common mistake is misidentifying the PDE type and applying inappropriate techniques — trying to march a Laplace equation in time, or using separation of variables on a non-separable geometry. A second mistake is mishandling boundary conditions: candidates apply Dirichlet conditions where Neumann are physically appropriate (e.g. insulated boundary in a heat problem), or forget that mixed (Robin) conditions appear in convective heat transfer. A third mistake is sloppy work in the separation-of-variables eigenvalue problem — candidates write the spatial ODE correctly but pick the wrong sign convention for the separation constant, which determines whether you get sines/cosines (oscillatory) or exponentials (decaying), and the wrong choice produces solutions that cannot satisfy the boundary conditions. Fourth, on Fourier methods, candidates compute coefficients on the wrong basis (sine series vs cosine series vs full Fourier series) for the boundary conditions of the problem, and the answer is off by a constant or a phase. Fifth, on Green's functions, weak candidates know the formula but cannot apply it because they cannot identify the right boundary conditions for G itself. Sixth, on numerical methods, candidates choose explicit time integration on stiff problems and watch the simulation blow up; or they violate CFL and report unstable nonsense; or they pick a non-conservative discretization for a conservation law and lose the wrong invariant. Seventh, on weak formulations, candidates do not understand why integration by parts shifts derivatives onto the test function — that is the whole technical content of FEM, and not understanding it means the candidate cannot reason about basis-function smoothness requirements. Finally, candidates often fail to non-dimensionalize, which is the single most useful preprocessing step on any PDE problem and surfaces the relevant dimensionless ratios (Reynolds, Peclet, Mach) that govern the regime.
Frequently asked questions
- What are the three classical PDE types and how do they differ?
- The three canonical second-order linear PDEs are: elliptic (Laplace, Poisson) — describes equilibrium states with no time dependence, e.g. steady-state heat distribution or electrostatic potential; parabolic (heat equation) — describes diffusive processes that smooth over time; hyperbolic (wave equation) — describes propagation with finite speed and conserved energy. Their solutions have qualitatively different character: elliptic solutions are smooth interior of the domain (maximum principle), parabolic solutions become smoother over time (irreversible), hyperbolic solutions preserve discontinuities and propagate them along characteristics.
- What is separation of variables and when does it work?
- Separation of variables looks for solutions u(x, t) = X(x)·T(t) and substitutes into the PDE; if the equation separates into one piece in x and one in t equal to a separation constant, you get two ODEs to solve. It works for linear, homogeneous PDEs on regions with separable boundary conditions (typically rectangular, cylindrical, or spherical domains aligned with the coordinate system). The technique reduces the PDE to ODE eigenvalue problems whose solutions become Fourier series, Bessel functions, Legendre polynomials, or spherical harmonics depending on the geometry.
- How do Maxwell's equations reduce to wave equations?
- Take the curl of Faraday's law (∇ × E = -∂B/∂t) and substitute Ampere's law in the curl-curl identity ∇ × (∇ × E) = ∇(∇ · E) - ∇²E. In free space (no charges), ∇ · E = 0, so the equation collapses to ∇²E = με·∂²E/∂t² — a vector wave equation with propagation speed 1/√(με) = c. The same derivation yields a wave equation for B. This is exactly how Maxwell predicted electromagnetic waves in 1865, and it's the canonical EM-interview derivation.
- What is Green's function and why is it useful?
- Green's function G(x, x') solves the linear PDE with a delta-function source at x': L·G = δ(x - x'). For an arbitrary source f(x), the solution is u(x) = ∫ G(x, x') f(x') dx' — the convolution of G with f. Green's functions for the Laplacian in free space are 1/(4π|x - x'|) in 3D (Coulomb potential) and 1/(2π) ln(|x - x'|) in 2D, which is why electrostatics and 2D potential flow look identical. The technique applies broadly: every fundamental solution in physics — retarded potentials, heat kernels, wave-equation propagators — is a Green's function in disguise.
- How do finite-difference and finite-element methods discretize a PDE?
- Finite differences (FD) replace derivatives with difference quotients on a regular grid: ∂²u/∂x² ≈ (u_{i+1} - 2u_i + u_{i-1})/h². Implementation is straightforward but it struggles with complex geometries and boundary conditions. Finite-element methods (FEM) discretize the domain into elements (triangles, tetrahedra), expand the unknown as a sum of local basis functions, and apply a weak (variational) form of the PDE. FEM handles arbitrary geometries elegantly and is the basis of every commercial EM and structural simulator (HFSS, ANSYS, COMSOL).
- What is the Courant-Friedrichs-Lewy (CFL) condition?
- For an explicit time-marching scheme on a hyperbolic PDE, the CFL condition restricts the time step relative to the spatial step: Δt ≤ Δx/c, where c is the maximum wave speed in the system. Geometrically, the numerical domain of dependence at a point must include the physical domain of dependence (the cone of points that could have influenced it via wave propagation). Violating CFL produces unbounded numerical instability — the simulation explodes within a few time steps. CFL is the canonical example of how PDE structure constrains numerical algorithm design.
Related topics
Essential AI-Native Skills for Partial Differential Equations
Modern engineering work increasingly uses AI tools for design and code review, debugging, documentation, test and testbench generation, and workflow automation. The goal is not to let AI replace engineering judgment — it is to move faster while keeping verification discipline.
- Use AI to explain unfamiliar code, logs, waveforms, datasheets, or test failures.
- Break large problems into small, reviewable steps you can verify independently.
- Ask AI for hypotheses, then validate them against tests, measurements, simulations, or lab data.
- Version-control your analysis scripts, testbenches, and configs — keep changes small and reviewable.
- Document your assumptions, design tradeoffs, and debugging decisions.
- Verify AI output before trusting it: run the checks that fit the domain — unit tests, linters, simulations, or bench/lab measurements.
- Review AI output for correctness, edge cases, and real-world consequences.
Partial Differential Equations — coming to the question bank
The adaptive practice engine is already live for core wireless, RF, and ML systems. Partial Differential Equations isn't covered in the question bank yet — get notified when it's added.