linear differential operator

Imagine a machine that you feed a function into and that hands you back another function, built by differentiating, scaling, and adding. A linear differential operator is exactly such a machine. The most famous one is just d/dx, but you can bolt several together: take a second derivative, add five times a first derivative, add six times the function itself. Writing the whole differential equation as one tidy object — operator acting on y equals input — is the move that turns a messy equation into a clean piece of algebra.

Formally, for an n-th order linear equation we write L[y] = y^(n) + p_{n-1}(x) y^(n-1) + ... + p_1(x) y' + p_0(x) y. Here L takes a function y and returns the combination on the right. The defining property is linearity: L[a y_1 + b y_2] = a L[y_1] + b L[y_2] for any constants a, b. So L respects addition and scaling — the same structural property that matrices have. The original equation L[y] = g(x) is homogeneous when g = 0 and nonhomogeneous when g is some forcing term.

This single idea organizes the entire theory of linear ODEs. Because L is linear, the solutions of L[y] = 0 form a vector space, the difference of any two solutions of L[y] = g is a solution of L[y] = 0, and the general solution splits cleanly into a homogeneous part plus one particular part. The same operator viewpoint reappears across applied mathematics — in linear PDEs, in the Laplace and Fourier transforms (which diagonalize constant-coefficient operators), and in quantum mechanics, where operators are the central object.