Mathematics
The Method of Separation of Variables for Partial Differential Equations
Quick fact
Separation of variables was pioneered by d'Alembert, Euler, and Fourier in the 18th century, and it is the reason Fourier series were invented.
Why this is interesting
You have a complicated equation that describes how heat spreads through a rod, or how a drumhead vibrates. Wouldn't it be great if you could split it into simpler pieces and solve each one separately?
Read the full explanation
Understanding The Method of Separation of Variables for Partial Differential Equations
Imagine you have a partial differential equation (PDE) that involves derivatives with respect to several variables, say time 't' and position 'x'. The method of separation of variables is based on a bold assumption: suppose the solution can be written as a product of two functions, one depending only on 'x' and the other only on 't' — that is, u(x,t) = X(x)T(t). If you plug this into the PDE, you can often rearrange the equation so that one side depends only on 'x' and the other only on 't'. Since 'x' and 't' are independent, both sides must be equal to a constant, called the separation constant. This single PDE then splits into two ordinary differential equations (ODEs), each depending on a single variable. These ODEs are typically much easier to solve. The boundary conditions of the original problem (e.g., the temperature fixed at the ends of a rod) turn into conditions on the X function, while the initial condition (e.g., the temperature at time zero) will later help determine the coefficients in a final sum of solutions.
A deeper explanation
The method works because linear PDEs obey the superposition principle: if u₁ and u₂ are solutions, then any linear combination c₁u₁ + c₂u₂ is also a solution. When we separate variables, we find infinitely many special solutions, each corresponding to a different value of the separation constant (the eigenvalue). The boundary conditions restrict these constants to a discrete set (the eigenvalues). Each eigenvalue yields an eigenfunction in the spatial variable. To satisfy an arbitrary initial condition, we express that condition as an infinite sum (a series) of these eigenfunctions. The coefficients of the series are determined using orthogonality, which is a property of the eigenfunctions (e.g., sines and cosines are orthogonal on an interval). So separation of variables reduces a PDE to two easier problems: an eigenvalue problem in space and a time evolution ODE. This is not just a trick; it reveals the underlying structure of the solution space: the solution is a superposition of modes, each evolving independently in time. For example, in the heat equation, each mode decays exponentially, with higher spatial frequencies decaying faster—this is why a hot spot smooths out quickly. The method is limited to linear PDEs with simple geometries (like rectangles, circles, or spheres) where separation works, but it is a cornerstone of mathematical physics.