Foundations of General Second-Order Linear Ordinary Differential Equations

1Introduction

This lecture distinguishes general second-order linear equations from their constant-coefficient special case and separates structural results from specialized solution methods.

2Equations Under Consideration

The standard form of a general second-order linear equation is

Let $$I$$I be an interval, let $$x_0\in I$$x_0\in I, and assume that $$P,Q,R:I\to \mathbb{K}$$P,Q,R:I\to\mathbb K are continuous, where $$\mathbb{K}$$\mathbb K denotes either $$\mathbb{R}$$\mathbb R or $$\mathbb{C}$$\mathbb C. For every $$a,b\in \mathbb{K}$$a,b\in\mathbb K, these assumptions guarantee a unique solution on all of $$I$$I satisfying $$y\left(x_0\right)=a$$y(x_0)=a and $$y^{\prime }\left(x_0\right)=b$$y'(x_0)=b.

3General Theory and Specialized Methods

The characteristic equation is not a general method for every second-order linear equation. It applies when the coefficients are constant, because differentiation preserves the form of $$e^{rx}$$e^{rx}. This distinction prevents the erroneous application of a characteristic polynomial to a variable-coefficient equation.

4Structure of the Linear Theory

Define

The operator $$L$$L maps functions to functions and is linear: for twice differentiable functions $$u,v$$u,v and constants $$\alpha ,\beta \in \mathbb{K}$$\alpha,\beta\in\mathbb K,

The solution set of the homogeneous equation $$L\left[y\right]=0$$L[y]=0 is a vector space. Closure under addition and scalar multiplication alone does not establish its dimension. Fixing $$x_0\in I$$x_0\in I, define

Existence and uniqueness for the initial-value problem show that $$\mathrm{\Phi }:\ker L\to K^2$$\Phi:\ker L\to\mathbb K^2 is a linear isomorphism. Consequently, $$\ker L$$\ker L is two-dimensional. If $$y_1,y_2$$y_1,y_2 are linearly independent solutions, every homogeneous solution has the form

For the nonhomogeneous equation $$L\left[y\right]=R$$L[y]=R, suppose that one particular solution $$y_p$$y_p is known. If $$h\in \ker L$$h\in\ker L, then

Conversely, if $$L\left[y\right]=R$$L[y]=R, then $$L[y-y_p]=0$$L[y-y_p]=0. Hence the full solution set is the affine space

This identity is the precise meaning of the familiar decomposition $$y=y_h+y_p$$y=y_h+y_p.

5First-Order System Formulation

Set $$x_1=y$$x_1=y and $$x_2=y^{\prime }$$x_2=y'. The equation is equivalent to

Continuity of the coefficient matrix and forcing term yields existence and uniqueness on $$I$$I by the standard theory of linear first-order systems. Two initial conditions are required because the state has the two components $$y$$y and $$y^{\prime }$$y'. In the constant-coefficient case, the matrix is constant, and its eigenvalues encode the same information as the roots of the characteristic polynomial.

The subsequent lecture on first-order systems develops this conversion, the matrix exponential, and the correspondence with eigenvalues in detail.

6Wronskian and Linear Independence

For two homogeneous solutions, define the Wronskian by

If $$W(y_1,y_2)\left(x_0\right)\ne 0$$W(y_1,y_2)(x_0)\ne0 at one point of $$I$$I, then $$y_1,y_2$$y_1,y_2 form a fundamental system on $$I$$I. Indeed, Abel's identity gives

so the Wronskian cannot vanish elsewhere on the interval. Conversely, linear dependence forces the Wronskian to vanish identically. Thus the Wronskian tests whether two solutions are linearly independent; together with the previously established two-dimensionality of the homogeneous solution space, this test determines whether they form a fundamental system.

7Criteria for Selecting a Solution Method

8Examples

The equation

is second-order and linear but does not have constant coefficients. Substitution of $$e^{rx}$$e^{rx} leaves dependence on $$x$$x and therefore does not produce an algebraic equation; an expression such as $$r^2+xr+1=0$$r^2+xr+1=0 is not a characteristic equation.

By contrast, $$y^{\prime }\text{'}-3y^{\prime }+2y=0$$y''-3y'+2y=0 has constant coefficients. Substitution of $$e^{rx}$$e^{rx} gives $$r^2-3r+2=0$$r^2-3r+2=0. Thus, applicability of the characteristic-equation method depends on constant coefficients, not merely on the order of the equation.

9Scope and Limitations

The vector-space and superposition results depend on linearity. They do not generally hold for nonlinear second-order equations. The global assertion on $$I$$I also depends on continuity of the normalized coefficients $$P,Q$$P,Q and forcing term $$R$$R throughout that interval; if the leading coefficient of an unnormalized equation vanishes, the equation must first be restricted to intervals on which normalization is valid.

10Exercises

11Related Lectures

12Previous Lecture

13Next Lecture