Second-Order Linear Differential Equations

Homogeneous and non-homogeneous second-order linear differential equations for physical harmonic oscillators.

What it is

A second-order ordinary linear differential equation has the general form: $a(x) \frac{d^2 y}{dx^2} + b(x) \frac{dy}{dx} + c(x) y = g(x)$ Dividing by the leading non-zero coefficient $a(x)$ yields canonical standard form: $y'' + p(x) y' + q(x) y = f(x)$ If external source term $f(x) = 0$, it is homogeneous; if $f(x) \neq 0$, it is inhomogeneous (or driven/forced).

For the canonical case of constant coefficients ($y'' + a y' + b y = 0$), seeking trial exponential solutions $y(x) = e^{r x}$ leads to the characteristic polynomial: $r^2 + a r + b = 0, \quad \text{with roots } r_{1,2} = \frac{-a \pm \sqrt{a^2 - 4b}}{2}$ Three physical regimes arise depending on the discriminant $\Delta = a^2 - 4b$:

1. $\Delta > 0$ (Distinct real roots $r_1 \neq r_2$): Overdamped regime. $y_h(x) = C_1 e^{r_1 x} + C_2 e^{r_2 x}$
2. $\Delta = 0$ (Repeated real root $r_1 = r_2 = -a/2$): Critically damped regime. $y_h(x) = (C_1 + C_2 x) e^{r_1 x}$
3. $\Delta < 0$ (Complex conjugate roots $r = \alpha \pm i\beta$): Underdamped (oscillatory) regime. $y_h(x) = e^{\alpha x} \left(C_1 \cos(\beta x) + C_2 \sin(\beta x)\right) = A e^{\alpha x} \cos(\beta x - \phi)$

For the driven equation, the linear superposition principle establishes that the general solution is: $y(x) = y_h(x) + y_p(x)$ where $y_p(x)$ is any particular solution generated by driving forcing $f(x)$.

Why it appears in physics

Second-order equations dominate classical physics because Newton's second law couples applied forces to acceleration ($d^2 x / dt^2$).

• Simple harmonic oscillator: $m \ddot{x} + k x = 0$ (the universal parabolic minimum of any stable equilibrium potential well).
• Damped forced oscillator: $m \ddot{x} + b \dot{x} + k x = F_0 \cos(\omega t)$ (automotive shock absorbers, seismographs, aeroelastic flutter).
• Series RLC resonator: $L \ddot{q} + R \dot{q} + \frac{1}{C} q = V(t)$ (telecom frequency tuning).
• Special functions (Bessel, Legendre, Hermite): Arise naturally upon separating coordinates in wave, diffusion, or Schrödinger equations across curvilinear geometries.

Visual or geometric intuition

The solution manifold of the homogeneous ODE forms a two-dimensional vector space. Any arbitrary solution is an exact linear combination of two linearly independent functions ${y_1(x), y_2(x)}$, called a fundamental basis. Linear independence is verified via the Wronskian determinant: $W(y_1, y_2)(x) = \begin{vmatrix} y_1(x) & y_2(x) \\ y_1'(x) & y_2'(x) \end{vmatrix} = y_1 y_2' - y_1' y_2 \neq 0$ By Abel's identity, the Wronskian evolves according to: $W(x) = W(x_0) \exp\left(-\int_{x_0}^x p(\xi)\,d\xi\right)$ If $p(x) > 0$ (dissipative damping), the phase-space area spanned by trajectories in $(y, y')$ contracts exponentially toward the origin attractor $(0, 0)$.

Minimal symbolic language

• Linear differential operator: $L[y] = \frac{d^2 y}{dx^2} + p(x) \frac{dy}{dx} + q(x) y$
• Abel's formula: $W'(x) + p(x) W(x) = 0 \implies W(x) = C e^{-\int p(x)\,dx}$
• Variation of parameters (universal particular solution formula): $y_p(x) = -y_1(x) \int \frac{y_2(x) f(x)}{W(y_1, y_2)(x)}\,dx + y_2(x) \int \frac{y_1(x) f(x)}{W(y_1, y_2)(x)}\,dx$
• Quality factor and resonance: $Q = \frac{\omega_0}{2 \gamma} = \frac{\sqrt{k m}}{b}$ At resonance with low damping ($Q \gg 1$), the steady-state amplitude is amplified by approximately $Q$: $A_{\text{res}} \approx Q \frac{F_0}{k}$.

What the result means

1. Transient vs. Steady-State: The homogeneous solution $y_h(t)$ represents the transient system response shaped by initial state conditions, fading to zero in dissipative media ($a > 0$). The particular solution $y_p(t)$ captures the steady-state rhythm imposed by external forcing.
2. Response phase lag: The driven output exhibits a frequency-dependent phase delay relative to the driver: near zero phase lag at low frequencies, exactly $\pi/2$ at resonance, and anti-phase $\pi$ at high frequencies.
3. Resonance: When driver frequency $\omega$ approaches natural frequency $\omega_0$, energy transfer maximizes. In undamped systems ($b = 0$), amplitude grows without bound ($t \sin(\omega_0 t)$).

When not to use it

• Large-amplitude oscillations: The small-angle approximation $\sin\theta \approx \theta$ fails for large amplitudes, requiring nonlinear Duffing or elliptic formulations.
• State-dependent coefficients: Equations such as Van der Pol's $\ddot{x} - \mu(1 - x^2)\dot{x} + x = 0$ break linear superposition.
• Vanishing Wronskians: Variation of parameters requires non-zero Wronskian denominators.

Operational summary

1. Normalize to $y'' + p y' + q y = f(x)$.
2. Solve the homogeneous equation via $r^2 + p r + q = 0$.
3. Identify damping regime: overdamped, critically damped, or underdamped. Assemble $y_h(x) = C_1 y_1 + C_2 y_2$.
4. Determine $y_p(x)$:
   • Use undetermined coefficients for polynomial, exponential, and sinusoidal source functions (apply modification rule if driving frequency hits an eigenvalue).
   • Use variation of parameters for general source terms (tangents, secants, arbitrary functions).
5. Assemble full general solution $y(x) = y_h(x) + y_p(x)$.
6. Apply initial conditions simultaneously on $y(t)$ and $y'(t)$ to isolate $C_1, C_2$.

Procedure: Solving Second-Order Linear ODEs

Before calculating

1. Normalize the ODE: Ensure the coefficient of $y''$ is unity: $y'' + p(x) y' + q(x) y = f(x)$
2. Inspect coefficients: Constant or variable? If variable of type $x^2 y'' + \alpha x y' + \beta y = 0$, use Cauchy-Euler substitution $y = x^r$.
3. Classify the source term: Choose between undetermined coefficients and variation of parameters.

Step 1: Identify the mathematical object

Formulate the characteristic polynomial for the homogeneous equation: $r^2 + p r + q = 0$ Compute roots: $r_{1,2} = \frac{-p \pm \sqrt{p^2 - 4q}}{2}$

Step 2: Choose the operation

2A. Homogeneous solution construction $y_h(x)$

• Case 1: Positive discriminant ($p^2 - 4q > 0$): Two distinct real roots $r_1 \neq r_2$: $y_h(x) = C_1 e^{r_1 x} + C_2 e^{r_2 x}$
• Case 2: Zero discriminant ($p^2 - 4q = 0$): Repeated real root $r = -p/2$: $y_h(x) = (C_1 + C_2 x) e^{r x}$
• Case 3: Negative discriminant ($p^2 - 4q < 0$): Complex conjugate pair $r = \alpha \pm i\beta$ (with $\alpha = -p/2$, $\beta = \frac{\sqrt{4q - p^2}}{2}$): $y_h(x) = e^{\alpha x}\left(C_1 \cos(\beta x) + C_2 \sin(\beta x)\right)$

2B. Particular solution determination $y_p(x)$

• Undetermined coefficients:
  • If $f(x) = P_n(x)$: test $y_p(x) = x^s (A_n x^n + \dots + A_0)$.
  • If $f(x) = e^{a x}$: test $y_p(x) = A x^s e^{a x}$.
  • If $f(x) = \cos(\omega x)$ or $\sin(\omega x)$: test $y_p(x) = x^s (A \cos(\omega x) + B \sin(\omega x))$. (Where $s$ is the smallest non-negative integer preventing duplication with $y_h$).
• Variation of parameters:
  1. Compute Wronskian: $W(x) = y_1 y_2' - y_1' y_2$.
  2. Compute integrals: $u_1(x) = -\int \frac{y_2 f}{W}\,dx, \quad u_2(x) = \int \frac{y_1 f}{W}\,dx$
  3. Form $y_p(x) = u_1(x) y_1(x) + u_2(x) y_2(x)$.

Step 3: Conserve units and signs

• Note frequency units $\text{rad}/\text{s}$.
• Carefully chain-rule angular frequencies and alternate signs when differentiating sinusoids.

Step 4: Calculate

1. Assemble complete solution: $y(t) = y_h(t) + y_p(t)$
2. Differentiate analytically to obtain $y'(t)$.
3. Apply initial conditions at $t = 0$ to solve the algebraic system for $C_1, C_2$.

Step 5: Interpret the result

1. Check for resonance ($s > 0$).
2. Confirm that damping causes $y_h(t) \to 0$ as $t \to \infty$, leaving the steady-state solution $y_p(t)$.

Self-check checklist

• Did I evaluate the discriminant of the characteristic polynomial accurately?
• Did I invoke the resonance boost factor $x^s$ when the driver matches an eigenvalue?
• Did I fit initial conditions to $(y_h + y_p)$ rather than $y_h$ alone?
• Is the Wronskian non-zero throughout the entire domain?
• Are frequency and amplitude units dimensionally consistent?

Worked Examples: Second-Order Linear Equations

Step-by-step guided example

Problem

A driven damped mechanical oscillator comprises a mass $m = 1\,\text{kg}$, a viscous dashpot with damping $b = 2\,\text{N}\cdot\text{s}/\text{m}$, and a spring of stiffness $k = 5\,\text{N}/\text{m}$. The system is excited by an external oscillatory driving force $F(t) = 10 \cos(t)\,\text{N}$. At $t = 0$, the mass is at rest at the equilibrium origin: $x(0) = 0\,\text{m}$, $\dot{x}(0) = 0\,\text{m}/\text{s}$.

1. Formulate the normalized equation of motion.
2. Determine the homogeneous solution $x_h(t)$ and identify the damping regime.
3. Find the steady-state particular solution $x_p(t)$ via undetermined coefficients.
4. Construct the complete particular solution satisfying initial conditions.

Detailed solution

Step 1: Formulation and normalization From Newton's second law: $m \ddot{x} + b \dot{x} + k x = F(t) \implies \ddot{x} + 2 \dot{x} + 5 x = 10 \cos(t)$ Normalized with $p = 2$, $q = 5$, and $f(t) = 10 \cos(t)$.

Step 2: Homogeneous solution $x_h(t)$ Characteristic polynomial: $r^2 + 2 r + 5 = 0 \implies r = \frac{-2 \pm \sqrt{4 - 20}}{2} = -1 \pm 2 i$ Since roots are complex conjugates with $\alpha = -1$ and $\beta = 2$, the system is underdamped. $x_h(t) = e^{-t}\left(C_1 \cos(2 t) + C_2 \sin(2 t)\right)$

Step 3: Particular solution $x_p(t)$ Driver frequency is $\omega = 1\,\text{rad}/\text{s}$ (non-resonant with $\beta = 2$, hence $s = 0$). Trial form: $x_p(t) = A \cos(t) + B \sin(t)$ Differentiating: $\dot{x}_p(t) = -A \sin(t) + B \cos(t)$ $\ddot{x}_p(t) = -A \cos(t) - B \sin(t)$ Substitute into $\ddot{x}_p + 2 \dot{x}_p + 5 x_p = 10 \cos(t)$: $(-A + 2 B + 5 A) \cos(t) + (-B - 2 A + 5 B) \sin(t) = 10 \cos(t)$ $(4 A + 2 B) \cos(t) + (-2 A + 4 B) \sin(t) = 10 \cos(t)$ Matching terms: $\begin{cases} 4 A + 2 B = 10 \\ -2 A + 4 B = 0 \implies A = 2 B \end{cases} \implies 10 B = 10 \implies B = 1, \quad A = 2$ Thus: $x_p(t) = 2 \cos(t) + \sin(t)\,\text{m}$

Step 4: Complete initial value solution $x(t) = e^{-t}\left(C_1 \cos(2 t) + C_2 \sin(2 t)\right) + 2 \cos(t) + \sin(t)$ Differentiating: $\dot{x}(t) = -e^{-t}\left(C_1 \cos(2 t) + C_2 \sin(2 t)\right) + e^{-t}\left(-2 C_1 \sin(2 t) + 2 C_2 \cos(2 t)\right) - 2 \sin(t) + \cos(t)$ Applying initial conditions at $t = 0$:

1. $x(0) = 0 \implies C_1 + 2 = 0 \implies C_1 = -2$
2. $\dot{x}(0) = 0 \implies -C_1 + 2 C_2 + 1 = 0 \implies 2 + 2 C_2 + 1 = 0 \implies C_2 = -1.5$ Final solution: $x(t) = e^{-t}\left(-2 \cos(2 t) - 1.5 \sin(2 t)\right) + 2 \cos(t) + \sin(t)\,\text{m}$

Inverse example

Problem

The unforced transient vibration of a unit-mass elastic beam follows: $y(t) = (3 + 4 t) e^{-5 t}$

1. Identify the damping regime and reconstruct the governing second-order linear homogeneous ODE.
2. Determine equivalent spring constant $k$ and damping dashpot coefficient $b$.

Detailed solution

Step 1: Identifying characteristic roots The presence of $(C_1 + C_2 t) e^{r t}$ with $r = -5$ indicates a repeated real root of multiplicity 2: $r_1 = r_2 = -5$ This signifies critical damping.

Step 2: Reconstructing the ODE $(r + 5)^2 = r^2 + 10 r + 25 = 0 \implies \ddot{y} + 10 \dot{y} + 25 y = 0$

Step 3: Extracting physical parameters Comparing with $m \ddot{y} + b \dot{y} + k y = 0$ for $m = 1\,\text{kg}$: $b = 10\,\text{N}\cdot\text{s}/\text{m}, \quad k = 25\,\text{N}/\text{m}$ Discriminant check: $b^2 - 4 m k = 100 - 100 = 0$ Confirms exact critical damping.

Common Errors: Second-Order Linear Equations

Error 1: Ignoring resonance in undetermined coefficients

Attempting $y_p = A \cos(\omega x) + B \sin(\omega x)$ when $\omega$ matches the natural frequency of an undamped system.

• Consequence: Substitution yields $0 = \cos(\omega x)$ (incompatible system). You must multiply by $x$ to account for secular resonance.

Error 2: Fitting initial conditions to the homogeneous solution alone

Evaluating $C_1, C_2$ from $y_h(0) = y_0$ and $y_h'(0) = v_0$ prior to computing and adding $y_p$.

• Correction: Initial conditions apply to the total combined motion: $y(0) = y_h(0) + y_p(0) = y_0, \quad y'(0) = y_h'(0) + y_p'(0) = v_0$

Error 3: Dropping the factor $x$ in repeated characteristic roots

Writing $y_h = C_1 e^{r x} + C_2 e^{r x}$ for a double root $r$.

• Consequence: Collapses the solution space into a single degree of freedom. The valid fundamental set is ${e^{r x}, x e^{r x}}$.

Error 4: Sign errors in damping exponents

Writing $e^{+\alpha t}$ instead of $e^{-\alpha t}$ when $\alpha = p/2 > 0$.

• Consequence: Converts a dissipative physical system into an explosive runaway process violating energy conservation.

Error 5: Omitting normalization in variation of parameters

In $a(x) y'' + b(x) y' + c(x) y = g(x)$, using $f(x) = g(x)$ directly in Wronskian formulas without dividing by $a(x)$.

• Correction: Variation of parameters strictly requires monic form $y'' + p y' + q y = f(x)$ with $f(x) = g(x)/a(x)$.

Quick control rule

For damped systems ($b > 0$): as $t \to \infty$, the transient component $y_h(t)$ must vanish identically, leaving exclusively the driven steady-state response $y_p(t)$.

Physical Applications: Second-Order Linear Equations

1. Structural earthquake engineering and Tuned Mass Dampers (TMD)

Skyscrapers such as Taipei 101 suspend massive second-order resonant pendulums near their crowns to mitigate wind and seismic excitations: $M \ddot{X} + (B + b) \dot{X} + (K + k) X - b \dot{x} - k x = F_{\text{wind}}(t)$ Tuning the TMD natural frequency to the building's fundamental resonance diverts destructive kinetic energy into the hydraulic damping fluid.

2. Automotive suspension dynamics and quarter-car models

A quarter-car model evaluates chassis vertical displacement ($m$) riding over road roughness profile $z_r(t)$: $m \ddot{z} + b \dot{z} + k z = b \dot{z}_r(t) + k z_r(t)$ Automotive engineers design a damping ratio $\zeta = \frac{b}{2\sqrt{km}} \approx 0.7$ (subcritical) to balance passenger comfort with road-tire traction.

3. Telecommunications: RF resonant filters and RLC tuners

Signal selection in wireless communications relies on series $RLC$ circuit resonance: $L \ddot{q} + R \dot{q} + \frac{1}{C} q = V_0 \cos(\omega t)$ Peak circulating current occurs at resonance $\omega_0 = 1/\sqrt{LC}$. The bandwidth $\Delta \omega = R/L = \omega_0 / Q$ establishes filter selectivity for channel separation.

4. Atomic Force Microscopy (AFM)

In dynamic tapping mode, a microscopic silicon cantilever vibrates near resonance: $m^* \ddot{z} + \frac{m^* \omega_0}{Q} \dot{z} + k^* z = F_0 \cos(\omega t) + F_{\text{tip-sample}}(z)$ Nanoscale shifts in vibrational amplitude and phase caused by Van der Waals forces map single-molecule surface topography.

5. Architectural acoustics and Helmholtz resonators

A Helmholtz resonator (e.g. bass traps in acoustic studios) acts as a second-order acoustic mass-spring system: $m_{\text{neck}} \ddot{\xi} + R_{\text{rad}} \dot{\xi} + K_{\text{cavity}} \xi = p_{\text{ext}}(t) S$ Air in the neck provides oscillating inertia while cavity volume acts as a pneumatic spring, absorbing targeted acoustic frequencies.

Interdisciplinary connection

Second-order linear ODEs form the backbone of closed-loop control stability (Routh-Hurwitz criterion), aeroelastic flutter analysis in aeronautics, and circadian biological limit-cycle pacemakers.

Guided Practice: Second-Order Linear Equations

Exercise 1: Pure undamped resonance

An oscillator of mass $m = 2\,\text{kg}$ and stiffness $k = 32\,\text{N}/\text{m}$ is driven by an exact resonant force: $F(t) = 8 \cos(4 t)\,\text{N}$. The system starts from rest: $x(0) = 0\,\text{m}$, $\dot{x}(0) = 0\,\text{m}/\text{s}$.

1. Compute natural frequency $\omega_0$ and state the homogeneous solution.
2. Apply the modification rule to obtain the secular particular solution $x_p(t)$.
3. State full trajectory $x(t)$ and describe the envelope growth.

Solution

1. Natural frequency: $\omega_0 = \sqrt{\frac{32}{2}} = 4\,\text{rad}/\text{s}$ Normalized ODE: $\ddot{x} + 16 x = 4 \cos(4 t)$. Homogeneous solution: $x_h(t) = C_1 \cos(4 t) + C_2 \sin(4 t)$
2. Resonant driving frequency $\omega = \omega_0 = 4$ ($s = 1$): $x_p(t) = t(A \cos(4 t) + B \sin(4 t))$ Substituting into $\ddot{x}_p + 16 x_p = 4 \cos(4 t)$: $-8 A \sin(4 t) + 8 B \cos(4 t) = 4 \cos(4 t) \implies A = 0, \quad B = 0.5$ $x_p(t) = 0.5 t \sin(4 t)\,\text{m}$
3. With $x(0) = 0, \dot{x}(0) = 0 \implies C_1 = 0, C_2 = 0$. $x(t) = 0.5 t \sin(4 t)\,\text{m}$ Linearly growing envelope leads to mechanical fatigue rupture.

Exercise 2: Underdamped RLC circuit under DC step voltage

A series $RLC$ circuit with $L = 0.1\,\text{H}$, $R = 20\,\Omega$, and $C = 100\,\mu\text{F} = 10^{-4}\,\text{F}$ is energized by a DC battery $V_0 = 12\,\text{V}$ at $t = 0$. Uncharged initial state: $q(0) = 0, i(0) = 0$.

1. Formulate the charge ODE.
2. Determine attenuation $\alpha$ and damped pseudofrequency $\omega_d$.
3. Compute the peak charge during the initial overshoot.

Solution

1. Dynamic ODE: $\ddot{q} + 200 \dot{q} + 10^5 q = 120$
2. Characteristic roots: $r^2 + 200 r + 10^5 = 0 \implies r = -100 \pm 300 i$ $\alpha = 100\,\text{s}^{-1}$, $\omega_d = 300\,\text{rad}/\text{s}$, $q_p = 1.2\,\text{mC}$.
3. General charge: $q(t) = e^{-100 t}(C_1 \cos(300 t) + C_2 \sin(300 t)) + 1.2 \times 10^{-3}$ Applying $q(0) = 0, \dot{q}(0) = 0 \implies C_1 = -1.2 \times 10^{-3}, C_2 = -0.4 \times 10^{-3}$. Peak occurs at half-period $300 t = \pi \implies t_1 = \frac{\pi}{300}\,\text{s}$: $q_{\text{max}} = 1.2 \times 10^{-3}(1 + e^{-\pi/3}) \approx 1.62\,\text{mC} \quad (35\% \text{ overshoot})$

Exercise 3: Variation of parameters for secant forcing

Solve the inhomogeneous ODE: $y'' + 9 y = 4 \sec(3 x), \quad x \in \left(-\frac{\pi}{6}, \frac{\pi}{6}\right)$

Solution

1. Homogeneous basis: $y_1 = \cos(3 x), \quad y_2 = \sin(3 x), \quad W = 3$
2. Variation of parameters: $u_1 = -\int \frac{\sin(3 x) \cdot 4 \sec(3 x)}{3}\,dx = -\frac{4}{3} \int \tan(3 x)\,dx = \frac{4}{9} \ln|\cos(3 x)|$ $u_2 = \int \frac{\cos(3 x) \cdot 4 \sec(3 x)}{3}\,dx = \frac{4}{3} x$
3. Full solution: $y(x) = C_1 \cos(3 x) + C_2 \sin(3 x) + \frac{4}{9} \cos(3 x) \ln(\cos(3 x)) + \frac{4}{3} x \sin(3 x)$