A second order differential equation is an equation that involves the second derivative of an unknown function but no higher derivatives. It connects a function to its rate of change and its rate of change of that rate of change. If a first order equation tells you how fast something is changing, a second order equation tells you how the speed of that change is itself changing. This makes it the natural language for describing anything that accelerates, oscillates, or vibrates.
The General Form
In its most common linear form, a second order differential equation looks like this:
a(x) y” + b(x) y’ + c(x) y = f(x)
Here, y is the unknown function you’re trying to find, y’ is its first derivative (how fast y changes), and y” is its second derivative (how fast the rate of change itself changes). The functions a(x), b(x), and c(x) are coefficients that can depend on x, and f(x) is a “forcing function” representing some outside influence on the system. The coefficient a(x) can’t be zero everywhere, or the equation would drop down to first order.
In many textbook problems, the coefficients are just constants, giving you the simpler form: ay” + by’ + cy = f(t). This constant-coefficient version covers a surprisingly large number of real-world problems.
Linear vs. Nonlinear
A second order equation is linear if the unknown function y and its derivatives appear only to the first power and aren’t multiplied together. The general form above is linear. An equation like y” + y² = 0 is nonlinear because of that squared term. The distinction matters enormously: linear equations have well-developed, systematic solution methods, while nonlinear second order equations are often impossible to solve with a neat formula and require numerical or approximate approaches.
Homogeneous vs. Nonhomogeneous
A linear second order equation splits into two categories based on the right-hand side. When f(x) = 0, the equation is called homogeneous:
ay” + by’ + cy = 0
This describes a system left to its own devices, with no outside push. A guitar string vibrating after being plucked, for instance. When f(x) is something other than zero, the equation is nonhomogeneous, meaning an external force is driving the system.
Solving a nonhomogeneous equation involves two pieces: the “complementary solution” (the solution to the homogeneous version) and a “particular solution” (any single solution that satisfies the full equation with the forcing function included). The general solution is the sum of both.
How to Solve Constant-Coefficient Equations
For the homogeneous equation ay” + by’ + cy = 0, the standard approach is to guess that the solution looks like y = e^(rx), plug it in, and see what r has to be. This produces the “characteristic equation,” a quadratic in r. The three possible outcomes for the roots of that quadratic each give a different type of solution.
Two distinct real roots r₁ and r₂: the solution is y = Ae^(r₁x) + Be^(r₂x). This typically represents a system that decays or grows without oscillating.
Complex roots r ± si: the solution is y = e^(rx)(C cos(sx) + D sin(sx)). The cosine and sine terms produce oscillation, while the e^(rx) part controls whether the oscillations grow, shrink, or stay steady. This is the case that describes vibrating and wave-like behavior.
One repeated root r: the solution is y = Ae^(rx) + Bxe^(rx). The extra factor of x in the second term is needed because with only one root, you’d otherwise have just one independent solution instead of the two that a second order equation requires.
In each case, A and B (or C and D) are arbitrary constants determined by your specific starting conditions.
Solving Nonhomogeneous Equations
When the right-hand side isn’t zero, you need a particular solution on top of the complementary one. Two main techniques handle this. The method of undetermined coefficients works by guessing a form for the particular solution based on what f(x) looks like (polynomials, exponentials, sines, cosines) and solving for the unknown coefficients. It’s straightforward but limited to a fairly small class of forcing functions.
Variation of parameters is more general and works for a wider range of forcing functions, though the algebra can get heavier. It builds a particular solution directly from the complementary solutions you’ve already found.
Initial Conditions and Boundary Conditions
Because the general solution to a second order equation contains two arbitrary constants, you need two additional pieces of information to pin down a specific solution. How you supply those two pieces defines two different types of problems.
In an initial value problem, both conditions are given at the same point: the value of y and the value of y’ at some starting time t₀. Think of launching a ball: you know where it starts and how fast it’s initially moving. Under mild continuity requirements on the coefficients, an initial value problem is guaranteed to have exactly one unique solution.
In a boundary value problem, conditions are specified at two different points, such as the value of y at the left end and the right end of an interval. A vibrating string fixed at both endpoints is a classic example. Boundary value problems are trickier: even for well-behaved equations, you might get no solution, exactly one, or infinitely many, depending on the specific boundary values.
Why Second Order Equations Appear Everywhere
Newton’s second law, force equals mass times acceleration, is itself a second order differential equation. Acceleration is the second derivative of position with respect to time. This single fact is why second order equations dominate physics and engineering.
The mass-spring-damper system is the classic mechanical example. A mass m attached to a spring with stiffness k, subject to friction (damping) c, satisfies: m·x” + c·x’ + k·x = f(t). Here x is displacement, x’ is velocity, x” is acceleration, and f(t) is any external force applied to the system. The coefficient of x” (mass) controls inertia, the coefficient of x’ (damping) controls how fast energy is lost to friction, and the coefficient of x (stiffness) controls the restoring pull of the spring. The natural frequency of this system depends on the ratio of stiffness to mass, while the damping ratio depends on how the friction coefficient relates to both mass and natural frequency.
Electrical circuits follow the same math. An RLC circuit, one containing a resistor, inductor, and capacitor, produces a second order equation where inductance plays the role of mass, resistance plays the role of damping, and capacitance plays the role of the spring’s stiffness. The natural frequency depends on the product of inductance and capacitance. This deep parallel means that anything you learn about solving mechanical vibration problems transfers directly to analyzing electrical circuits, and vice versa.
Beyond these textbook cases, second order differential equations describe the motion of planets, the bending of beams, the propagation of sound waves, the behavior of simple pendulums, and the dynamics of population models with acceleration-like feedback. Any system where the current state depends not just on how things are changing but on how fast that change is speeding up or slowing down will naturally involve a second derivative, landing you squarely in second order territory.