Skip to main content

Second-Order Linear Equations

Push a child on a swing, strike a tuning fork, watch a car settle after a pothole, or measure the current sloshing through a radio circuit — and you are watching the same equation play out. A second-order linear differential equation is the mathematical machine that describes almost everything that oscillates, rings, bounces, or settles down. It is the single most important equation in classical physics and engineering, and once you can read it fluently, an enormous slice of the physical world becomes predictable.

The reason it earns "second-order" is that it involves the second derivative — an acceleration, a curvature, a rate of change of a rate of change. Nature is full of such laws because Newton's second law itself relates force to acceleration. Learning to solve these equations means learning to predict motion.

Learning Objectives

  • Recognize a second-order linear differential equation and identify its coefficients.
  • Distinguish homogeneous from non-homogeneous equations and understand why the distinction matters.
  • Form and solve the characteristic equation for constant-coefficient homogeneous equations.
  • Handle the three root cases: real distinct, repeated, and complex conjugate roots.
  • Model and interpret harmonic oscillators — undamped, underdamped, critically damped, and overdamped.
  • Build the general solution of a non-homogeneous equation from complementary and particular parts.

Quick Answer

A second-order linear ODE has the form a(x)y+b(x)y+c(x)y=g(x) a(x)y'' + b(x)y' + c(x)y = g(x). It is homogeneous when g(x)=0 g(x) = 0 and non-homogeneous otherwise. For constant coefficients, you solve the homogeneous version by guessing y=erx y = e^{rx}, which turns the differential equation into an algebraic characteristic equation ar2+br+c=0 ar^2 + br + c = 0. The two roots determine the solution's shape: real distinct roots give pure exponentials, a repeated root adds an xx factor, and complex roots produce oscillations. The full non-homogeneous solution is the homogeneous (complementary) solution plus one particular solution. This one framework describes every spring, pendulum, and simple electrical circuit.

Where It Came From

The story begins with springs and sound. In the early 1600s, natural philosophers wanted to understand vibration: why a plucked string produces a definite pitch, why a pendulum keeps near-perfect time. But the tools to describe changing rates of change did not yet exist.

Two developments forced the second-order equation into being. First, in 1660 Robert Hooke discovered that the restoring force of a spring is proportional to its stretch — force equals kx-kx. Second, in 1687 Isaac Newton published his second law, force equals mass times acceleration, and acceleration is the second derivative of position. Put Hooke and Newton together and you are compelled to write mx=kx m x'' = -kx. There was no way to state the physics of a spring without a second-order differential equation. The equation was not invented for its own sake; it was the unavoidable translation of a physical law.

Through the 1700s the Bernoulli family and especially Leonhard Euler developed the systematic solution method. Around 1739 Euler introduced the exponential substitution y=erx y = e^{rx} that reduces the whole problem to solving a quadratic — the characteristic equation. This was a stunning simplification: an infinite-dimensional problem about functions collapsed into ordinary high-school algebra. Jean le Rond d'Alembert applied these ideas to the vibrating string in 1747, launching the study of waves. When damping (friction, air resistance, electrical resistance) was added in the 1800s, the same equation described how vibrations die away, and it became the backbone of mechanical and electrical engineering.

The Building Block: Homogeneous Equations and the Characteristic Equation

A constant-coefficient homogeneous equation looks like

ay+by+cy=0. a y'' + b y' + c y = 0.

The key insight is that the exponential function is the only function whose derivatives are just scaled copies of itself. So we guess y=erx y = e^{rx}. Then y=rerx y' = re^{rx} and y=r2erx y'' = r^2 e^{rx}. Substituting:

ar2erx+brerx+cerx=0. a r^2 e^{rx} + b r e^{rx} + c e^{rx} = 0.

Since erx e^{rx} is never zero, we divide it out and are left with the characteristic equation:

ar2+br+c=0. a r^2 + b r + c = 0.

The differential equation is solved once we solve this quadratic. There are three cases, decided by the discriminant b24ac b^2 - 4ac.

Case 1: Two real distinct roots

Worked example. Solve y5y+6y=0 y'' - 5y' + 6y = 0.

Characteristic equation: r25r+6=0 r^2 - 5r + 6 = 0, which factors as (r2)(r3)=0(r-2)(r-3) = 0, so r=2 r = 2 and r=3 r = 3. Each root gives an independent solution, and the general solution is their combination:

y=C1e2x+C2e3x. y = C_1 e^{2x} + C_2 e^{3x}.

Check. Take y=e2x y = e^{2x}: then y5y+6y=4e2x10e2x+6e2x=0 y'' - 5y' + 6y = 4e^{2x} - 10e^{2x} + 6e^{2x} = 0. Correct.

Case 2: A repeated real root

When the discriminant is zero, the quadratic gives one root rr twice. A single erx e^{rx} is not enough — a second-order equation always needs two independent solutions. The second one turns out to be xerx x e^{rx}.

Worked example. Solve y4y+4y=0 y'' - 4y' + 4y = 0.

Characteristic equation: r24r+4=(r2)2=0 r^2 - 4r + 4 = (r-2)^2 = 0, so r=2 r = 2 (repeated). The general solution is

y=C1e2x+C2xe2x. y = C_1 e^{2x} + C_2 x e^{2x}.

Case 3: Complex conjugate roots

When b24ac<0 b^2 - 4ac < 0, the roots are r=α±βi r = \alpha \pm \beta i. Using Euler's formula eiθ=cosθ+isinθ e^{i\theta} = \cos\theta + i\sin\theta, the complex exponentials repackage into real oscillations:

y=eαx(C1cosβx+C2sinβx). y = e^{\alpha x}\left(C_1 \cos\beta x + C_2 \sin\beta x\right).

Worked example. Solve y+4y=0 y'' + 4y = 0.

Characteristic equation: r2+4=0 r^2 + 4 = 0, so r=±2i r = \pm 2i. Here α=0\alpha = 0 and β=2\beta = 2:

y=C1cos2x+C2sin2x. y = C_1 \cos 2x + C_2 \sin 2x.

This is pure oscillation with angular frequency 2 2 — a vibration that never dies. That is exactly the undamped spring.

Harmonic Oscillators and Damping

The physical star of this topic is the damped harmonic oscillator. A mass mm on a spring of stiffness kk, dragged by friction with damping coefficient cc, obeys

mx+cx+kx=0. m x'' + c x' + k x = 0.

The three algebraic cases above become three physical behaviors, decided by how damping compares to stiffness. Define the undamped angular frequency ω0=k/m\omega_0 = \sqrt{k/m}.

Discriminant signRootsPhysical nameBehavior
c24mk<0 c^2 - 4mk < 0complexUnderdampedOscillates, amplitude decaying
c24mk=0 c^2 - 4mk = 0repeated realCritically dampedFastest return, no oscillation
c24mk>0 c^2 - 4mk > 0two realOverdampedSlow return, no oscillation

Worked example (underdamped). A mass with m=1 m = 1, c=2 c = 2, k=5 k = 5: solve x+2x+5x=0 x'' + 2x' + 5x = 0.

Characteristic equation: r2+2r+5=0 r^2 + 2r + 5 = 0. Discriminant =420=16<0 = 4 - 20 = -16 < 0, so r=2±162=1±2i. r = \frac{-2 \pm \sqrt{-16}}{2} = -1 \pm 2i.

Thus α=1\alpha = -1, β=2\beta = 2 and

x=extime (C1cos2t+C2sin2t),i.e.x(t)=et(C1cos2t+C2sin2t). x = e^{-x_{\text{time}}}\ (C_1 \cos 2t + C_2 \sin 2t),\qquad \text{i.e.}\qquad x(t) = e^{-t}(C_1\cos 2t + C_2 \sin 2t).

The cos2t \cos 2t and sin2t\sin 2t are the ringing; the envelope et e^{-t} is friction bleeding energy away. This is a plucked guitar string, a struck bell, or a car suspension after a bump — an oscillation that fades.

Engineers choose critical damping for things that must settle fast without wobbling: a door closer, a car shock absorber, or a measuring-instrument needle. A little less damping and the door slams past and bounces (underdamped); a lot more and it crawls shut (overdamped).

Non-Homogeneous Equations: Adding a Driving Force

When there is an external push — a periodic force on the spring, a battery driving a circuit — the equation gains a right-hand side:

ay+by+cy=g(x). a y'' + b y' + c y = g(x).

The full solution is y=yc+yp y = y_c + y_p, where yc y_c is the complementary solution (the homogeneous solution from the characteristic equation) and yp y_p is any single particular solution that produces g(x) g(x). The complementary part carries the two arbitrary constants and describes the system's natural response; the particular part is the system's forced response.

Worked example. Solve y5y+6y=12 y'' - 5y' + 6y = 12.

Complementary part comes from Case 1 above: yc=C1e2x+C2e3x y_c = C_1 e^{2x} + C_2 e^{3x}. For the particular part, since the right side is a constant, guess a constant yp=A y_p = A. Then yp=yp=0 y_p'' = y_p' = 0, so the equation becomes 6A=12 6A = 12, giving A=2 A = 2. The general solution is

y=C1e2x+C2e3x+2. y = C_1 e^{2x} + C_2 e^{3x} + 2.

The method of guessing a form for ypy_p that matches g(x)g(x) (a polynomial for a polynomial, ekx e^{kx} for an exponential, sines and cosines for a sinusoid) is called the method of undetermined coefficients.

Real-World Applications

  • Vehicle suspension. Springs plus shock absorbers are designed near critical damping so the car settles quickly after a bump without bouncing passengers.
  • Building and bridge engineering. Structures have natural frequencies from exactly this equation; designers add damping to prevent resonance from wind or earthquakes (the Tacoma Narrows collapse is the cautionary tale).
  • Electrical RLC circuits. An inductor–resistor–capacitor loop obeys Lq+Rq+1Cq=V(t) L q'' + R q' + \frac{1}{C} q = V(t), mathematically identical to the spring. Radio tuning exploits its resonance.
  • Clocks and timekeeping. Pendulum and quartz oscillators rely on stable, lightly damped simple harmonic motion.
  • Seismographs and instruments. Damped oscillators convert ground motion or signals into readable, quickly-settling measurements.

Common Mistakes

Mistake 1: Forgetting the second solution for a repeated root. Students write y=C1e2x y = C_1 e^{2x} for (r2)2=0(r-2)^2 = 0 and stop. This is wrong because a second-order equation must have two independent solutions to satisfy two initial conditions. Correction: always append the xerx x e^{rx} term: y=C1e2x+C2xe2x y = C_1 e^{2x} + C_2 x e^{2x}.

Mistake 2: Leaving complex exponentials in the answer. Writing y=C1e(1+2i)x+C2e(12i)x y = C_1 e^{(1+2i)x} + C_2 e^{(1-2i)x} is technically a solution but hides that the motion is real oscillation. Correction: use Euler's formula to convert to eαx(C1cosβx+C2sinβx) e^{\alpha x}(C_1\cos\beta x + C_2\sin\beta x), which is real-valued and physically transparent.

Mistake 3: Confusing homogeneous and non-homogeneous solution roles. Students try to make the complementary solution alone satisfy g(x) g(x), or drop the constants once they find yp y_p. Correction: the complementary solution (yc y_c, with the constants) and the particular solution (yp y_p) are added; only yc y_c carries the arbitrary constants, and you apply initial conditions to the full y=yc+yp y = y_c + y_p at the very end.

Comparison and Connections

The whole subject is a dictionary between algebra and physics. The quadratic you solve is the physics of the system:

Characteristic rootsSolution formOscillator meaning
Real, distinct, both negativeC1er1x+C2er2x C_1 e^{r_1 x} + C_2 e^{r_2 x}Overdamped, settles slowly
Real, repeated, negative(C1+C2x)erx(C_1 + C_2 x)e^{rx}Critically damped, settles fastest
Complex, negative real parteαx(C1cosβx+C2sinβx) e^{\alpha x}(C_1\cos\beta x + C_2\sin\beta x)Underdamped, decaying ringing
Pure imaginaryC1cosβx+C2sinβx C_1\cos\beta x + C_2\sin\beta xUndamped, eternal oscillation

Second-order equations generalize the first-order linear equations you met earlier (one root, one exponential, exponential growth or decay), and they are the gateway to systems of equations and to partial differential equations like the wave and heat equations, where the same characteristic-equation thinking reappears in higher dimensions.

Practice Questions

Recall

State the characteristic equation of 2y+3y5y=0 2y'' + 3y' - 5y = 0. Answer: 2r2+3r5=0 2r^2 + 3r - 5 = 0.

Understanding

Explain why a repeated root r r requires the extra solution xerx x e^{rx} rather than a second copy of erx e^{rx}. Guidance: Two copies of erx e^{rx} are not independent — they collapse into one arbitrary constant, leaving you unable to satisfy two initial conditions. The function xerx x e^{rx} is independent and can be verified to satisfy the equation, restoring the needed second degree of freedom.

Application

Solve y+6y+9y=0 y'' + 6y' + 9y = 0. Answer: r2+6r+9=(r+3)2=0 r^2 + 6r + 9 = (r+3)^2 = 0, repeated root r=3 r = -3, so y=(C1+C2x)e3x y = (C_1 + C_2 x)e^{-3x}. This is critically damped.

Analysis

For the spring equation x+cx+4x=0 x'' + c x' + 4x = 0, find the value of cc that produces critical damping, and describe what happens for slightly smaller and slightly larger cc. Answer: Critical damping needs discriminant zero: c24(1)(4)=0 c^2 - 4(1)(4) = 0, so c=4 c = 4. For c<4 c < 4 the system is underdamped and rings while decaying; for c>4 c > 4 it is overdamped and returns to rest slowly without oscillating.

FAQ

Why do we guess erx e^{rx}? Isn't guessing unrigorous? It is a rigorous, justified guess: exponentials are the eigenfunctions of differentiation, so they are the natural candidates. Existence-and-uniqueness theorems guarantee that once you find two independent solutions, you have found all solutions, so the guess is complete, not lucky.

What do the two constants C1 C_1 and C2 C_2 physically mean? They encode the two initial conditions a second-order system needs: typically the starting position and the starting velocity. Given those, the constants are determined uniquely.

How is an RLC circuit "the same" as a spring? The equations have identical form: charge plays the role of position, inductance the role of mass, resistance the role of damping, and inverse capacitance the role of spring stiffness. Any result you prove for one applies to the other — this is called an analogy or duality.

When is undamped motion (c=0 c = 0) physically real? Never exactly, since some friction always exists, but it is an excellent idealization over short times — a tuning fork or quartz crystal loses energy so slowly it is nearly undamped, which is precisely why they make good frequency standards.

Does the method work if the coefficients change with xx? The characteristic-equation shortcut needs constant coefficients. Variable-coefficient equations (like Euler–Cauchy or Bessel's equation) need other techniques such as power series, but the homogeneous-plus-particular structure still holds.

Quick Revision

  • Form: ay+by+cy=g(x) a y'' + b y' + c y = g(x); homogeneous when g=0 g = 0.
  • Homogeneous solution: guess erx e^{rx}, solve ar2+br+c=0 a r^2 + br + c = 0.
  • Real distinct roots: y=C1er1x+C2er2x y = C_1 e^{r_1 x} + C_2 e^{r_2 x}.
  • Repeated root: y=(C1+C2x)erx y = (C_1 + C_2 x)e^{rx}.
  • Complex roots α±βi\alpha \pm \beta i: y=eαx(C1cosβx+C2sinβx) y = e^{\alpha x}(C_1\cos\beta x + C_2\sin\beta x).
  • Non-homogeneous: y=yc+yp y = y_c + y_p; find yp y_p by undetermined coefficients.
  • Damping test on mx+cx+kx=0 m x'' + c x' + k x = 0: sign of c24mk c^2 - 4mk gives over- / critical- / underdamped.

Prerequisites

  • Euler's formula and complex numbers
  • Simple harmonic motion (Physics)
  • RLC circuits (Electrical engineering)

Next Topics

  • Systems of differential equations
  • The wave and heat equations (partial differential equations)
  • Laplace transforms for driven and discontinuous forcing