Non-Homogeneous Systems: Undetermined Coefficients
What you'll learn: How to solve systems of linear ODEs with polynomial, exponential, or sinusoidal forcing using educated guesses based on the forcing function's form.
The Core Idea
When you have a system of linear ODEs like x' = A x + f(t), where f(t) is a forcing vector (the non-homogeneous part), the method of undetermined coefficients extends naturally from single equations to systems. Just as you did with second-order ODEs, you guess a particular solution x_p whose form mimics the forcing function, substitute it into the system, and solve for the unknown coefficients.
How It Works
Polynomial forcing: If f(t) = [t², 3t]ᵀ, guess x_p = at² + bt + c (vectors of constants).
Exponential forcing: If f(t) = [2e³ᵗ, -e³ᵗ]ᵀ, guess x_p = ae³ᵗ.
Sinusoidal forcing: If f(t) = [cos(2t), sin(2t)]ᵀ, guess x_p = acos(2t) + bsin(2t).
The resonance rule still applies: If your guess duplicates a solution to the homogeneous system (i.e., e³ᵗ is an eigenvalue solution), multiply your guess by t.
After guessing, differentiate x_p, substitute both x_p' and x_p into the system, equate coefficients of like terms (t², t¹, constants, or exponentials/trig functions), and solve the resulting algebraic system for your coefficient vectors.
Why It Matters
This method is significantly faster than variation of parameters when the forcing matches these standard forms. You avoid computing Wronskians and messy integrals—just algebra with vectors and matrices.
Key Takeaway: For systems with polynomial, exponential, or sinusoidal forcing, guess a particular solution matching the forcing's form (with vector coefficients), substitute into the system, and solve algebraically—but watch for resonance requiring multiplication by t.