ODE: Method of Undetermined Coefficients

Solving a nonhomogeneous ordinary differential equation (ODE) often feels like a two-part job: first, solving the complementary homogeneous equation, and second, finding a single particular solution to the nonhomogeneous one. The Method of Undetermined Coefficients is a powerful, systematic technique for finding that particular solution when the nonhomogeneous term, or forcing function, has a specific, predictable form common in engineering systems: polynomials, exponentials, sines, cosines, and sums/products thereof. Mastering this method is essential for modeling forced mechanical vibrations, electrical circuits with applied voltage, and other dynamic systems where an external drive dictates a specific response.

The Core Principle: Educated Guessing

The method is fundamentally an exercise in "educated guessing." Its power lies in a simple observation: the derivative of an exponential is an exponential, the derivative of a polynomial is another polynomial, and the derivatives of sines and cosines cycle among themselves. Therefore, if the forcing function $f(x)$ is one of these types or a combination, a particular solution $y_p(x)$ will likely be of the same general family, but with undetermined coefficients that we can solve for.

The process follows a clear sequence:

1. Solve the associated homogeneous equation to find the complementary solution $y_c(x)$.
2. Based on the form of $f(x)$, propose a trial particular solution $y_p(x)$ with undetermined coefficients (e.g., $A$, $B$, $C$).
3. Substitute $y_p(x)$ and its necessary derivatives into the original nonhomogeneous ODE.
4. Collect like terms and equate coefficients to create a system of algebraic equations.
5. Solve the system for the undetermined coefficients.
6. The general solution is then $y(x)=y_c(x)+y_p(x)$.

The true art—and where errors are often made—lies in correctly proposing the initial form of $y_p(x)$.

Crafting Trial Solutions for Standard Forcing Functions

The choice of trial solution is not arbitrary; it is dictated by the form of $f(x)$. Here are the standard rules for common forcing functions, assuming no conflict with the complementary solution (a critical exception covered next).

• For Polynomials: If $f(x)$ is a polynomial of degree $n$, the trial solution is a general polynomial of the same degree.
• Example: For $f(x)=3x^2-1$, propose $y_p=A{x}^2+Bx+C$.

• For Exponentials: If $f(x)$ is an exponential $C{e}^{rx}$, the trial solution is a constant multiple of the same exponential.
• Example: For $f(x)=5e^{-2x}$, propose $y_p=A{e}^{-2x}$.

• For Sine/Cosine: If $f(x)$ is a combination like $C_1\cos (\omega x)+C_2\sin (\omega x)$, the trial solution must include both sine and cosine terms of the same frequency, even if one coefficient is zero in $f(x)$.
• Example: For $f(x)=4\sin (3x)$, propose $y_p=A\cos (3x)+B\sin (3x)$. The presence of the cosine term is necessary because its derivative will produce a sine term.

• For Products: If $f(x)$ is a product of the above forms, the trial solution is the product of the corresponding trial forms.
• Example: For $f(x)=x{e}^x$, propose $y_p=(Ax+B)e^x$.
• Example: For $f(x)=x^2\cos (5x)$, propose $y_p=(A{x}^2+Bx+C)\cos (5x)+(D{x}^2+Ex+F)\sin (5x)$.

The Modification Rule: Handling Resonance Cases

The standard rules fail when the proposed trial solution is already part of the complementary solution $y_c(x)$. This situation, called resonance in engineering contexts, occurs because the homogeneous equation already produces a function of that form, making the trial solution linearly dependent and therefore invalid when substituted into the ODE. The Modification Rule resolves this: if any term in the proposed $y_p$ is a solution to the homogeneous equation, multiply the entire trial solution by $x$ (or $x^s$, where $s$ is the smallest positive integer that removes all duplication).

Worked Example (Resonance): Solve $y^{\prime \prime }-4y^{\prime }+4y=2e^{2x}$.

1. Solve homogeneous: $r^2-4r+4=0$ gives $r=2$ (double root). So $y_c=C_1{e}^{2x}+C_{2}x{e}^{2x}$.
2. Standard rule: For $f(x)=2e^{2x}$, we'd propose $y_p=A{e}^{2x}$.
3. Conflict: $A{e}^{2x}$ is already present in $y_c$. Multiplying by $x$ gives $Ax{e}^{2x}$, but this is also present in $y_c$.
4. Apply modification: Multiply by $x^2$ (the smallest power that eliminates duplication). The correct trial is $y_p=A{x}^2{e}^{2x}$.
5. Substitute, solve, and find $A=1$. Thus, $y_p=x^2{e}^{2x}$.

Superposition for Combined Forcing Functions

The principle of superposition for linear ODEs simplifies problems with combined forcing terms. If the nonhomogeneous term is a sum, $f(x)=f_1(x)+f_2(x)$, you can find particular solutions $y_{p1}$ for $f_1$ and $y_{p2}$ for $f_2$ independently. The particular solution for the full equation is then the sum: $y_p=y_{p1}+y_{p2}$. This allows you to break down complex forcing functions into manageable pieces.

Worked Example (Superposition): Find a trial solution for $y^{\prime \prime }+y=x+\sin (x)$.

1. Break $f(x)$ into $f_1(x)=x$ and $f_2(x)=\sin (x)$.
2. For $f_1(x)=x$: Propose $y_{p1}=Ax+B$.
3. For $f_2(x)=\sin (x)$: Solve $y_c$ for homogeneous $y^{\prime \prime }+y=0$ gives $y_c=C_1\cos (x)+C_2\sin (x)$. A standard proposal of $D\cos (x)+E\sin (x)$ conflicts with $y_c$.
4. Apply modification: Multiply by $x$. Propose $y_{p2}=x(D\cos (x)+E\sin (x))$.
5. By superposition, the full trial is $y_p=y_{p1}+y_{p2}=Ax+B+Dx\cos (x)+Ex\sin (x)$.

A Systematic Decision Framework

To consistently choose the correct trial solution, follow this decision tree:

1. Identify $f(x)$. Write it as a sum of terms from distinct families (polynomial, exponential, trig).
2. Solve for $y_c(x)$. This is non-negotiable; you must know the homogeneous solution.
3. Propose initial $y_p$. For each distinct family term in $f(x)$, write its standard trial solution (using product rules if needed). Sum them per superposition.
4. Check for conflicts. Compare each term in your proposed $y_p$ against each term in $y_c$.

• If no term in $y_p$ matches any term in $y_c$, your proposal is final.
• If any term matches, apply the Modification Rule to the entire family group in $y_p$ that contains the conflicting term. Multiply that group's trial form by $x^s$ to eliminate all conflicts.

1. Write the final $y_p$ form with undetermined coefficients and proceed with substitution.

Common Pitfalls

1. Insufficient Trial for Trig Functions: Proposing only $A\sin (\omega x)$ for $f(x)=\sin (\omega x)$ is a classic error. You must include the cosine term: $A\cos (\omega x)+B\sin (\omega x)$. Their derivatives are linked, so both are needed to match coefficients upon substitution.
2. Misapplying the Modification Rule: The rule applies to the entire family/group of terms from a single component of $f(x)$, not just the offending term. In the example $y^{\prime \prime }-4y^{\prime }+4y=x{e}^{2x}$, the trial starts as $(Ax+B)e^{2x}$. Since $e^{2x}$ and $x{e}^{2x}$ are in $y_c$, you multiply the entire polynomial-exponential product by $x^2$: $y_p=x^2(Ax+B)e^{2x}$.
3. Ignoring Superposition: Trying to create a single, convoluted trial for a sum like $x+\sin (x)$ is inefficient and error-prone. Always break it into parts, find the trial for each, and add them at the end, applying modification checks to each part independently.
4. Forgetting to Solve for $y_c$ First: Attempting to guess $y_p$ without knowing $y_c$ is like navigating without a map. You cannot identify resonance conflicts, which guarantees an unsolvable system of equations when you attempt to equate coefficients.

Summary

• The Method of Undetermined Coefficients is a systematic algebraic method for finding particular solutions to linear nonhomogeneous ODEs with constant coefficients when the forcing function is a polynomial, exponential, sine, cosine, or a sum/product of these.
• The initial trial solution $y_p$ is chosen based on the form of the forcing function $f(x)$, using standard rules for each functional family.
• The critical Modification Rule (multiplying by $x^s$) must be applied when any part of the proposed trial solution is already present in the complementary solution $y_c$, a condition known as resonance.
• The principle of superposition allows you to handle sums of forcing functions by finding trial solutions for each component separately and adding them to form the complete $y_p$.
• A successful application requires a strict sequence: 1) Find $y_c$, 2) Propose initial $y_p$ based on $f(x)$, 3) Modify for resonance, 4) Substitute and solve for coefficients. Skipping or misordering these steps is the most common source of error.