Why does variation of parameters introduce functions u₁(x) and u₂(x) rather than directly guessing the form of the particular solution?
ABecause u₁ and u₂ are always easier to compute than a direct guess
BTo avoid using the Wronskian, which can be difficult to compute
CBecause undetermined coefficients requires f(x) to have a specific form, but variation of parameters makes no assumption about f(x) — it works for any continuous right-hand side
DBecause u₁ and u₂ represent the homogeneous solution, not the particular solution
The whole point of variation of parameters is generality. Undetermined coefficients works by guessing the form of the particular solution — which requires f(x) to be a polynomial, exponential, sine, cosine, or combination thereof. If f(x) = sec(x), ln(x), or any function that doesn't generate a finite family of derivatives, undetermined coefficients fails. Variation of parameters sidesteps this by assuming only that y_p = u₁y₁ + u₂y₂ and then solving for u₁ and u₂ through integration — no assumption about the form of f(x) is required.
Question 2 Multiple Choice
A student is solving y'' + y = sec(x). They try undetermined coefficients but get stuck. Which statement correctly explains why variation of parameters succeeds where undetermined coefficients fails?
AVariation of parameters only works for constant-coefficient equations, which is why it handles this case
Bsec(x) and its derivatives form an infinite, non-repeating family (sec, sec·tan, sec³+sec·tan², ...) so no finite trial solution exists; variation of parameters integrates u₁' and u₂' directly without needing to guess the form
Csec(x) is not continuous, so undetermined coefficients fails on continuity grounds
DUndetermined coefficients requires a nonhomogeneous term, which sec(x) is not
Undetermined coefficients requires f(x) to generate a finite family of linearly independent derivatives so you can build a trial solution from them. sec(x) does not: each successive derivative introduces new functions (sec·tan, sec³, ...), so no finite trial solution exists. Variation of parameters avoids this entirely — it only requires integrating u₁' = −y₂f/W and u₂' = y₁f/W, and ∫sec(x)·cos(x)/W dx simplifies to something integrable. The method's lack of restriction on f(x) is precisely its advantage.
Question 3 True / False
The simplifying constraint u₁'y₁ + u₂'y₂ = 0 imposed in variation of parameters is an arbitrary choice that could be replaced by any other condition — it is just one of many equally valid approaches.
TTrue
FFalse
Answer: False
False — while the constraint is a choice (not forced by the algebra), it is far from arbitrary: it is the natural choice that simultaneously eliminates the highest-derivative cross-terms from the substitution and produces the cleanest possible 2×2 system in u₁' and u₂'. Without this constraint, the algebra does not reduce to a solvable system of two equations. Other constraints would produce more complex and less tractable systems. The constraint is conventional but deeply motivated.
Question 4 True / False
The Wronskian W = y₁y₂' − y₂y₁' must be nonzero for variation of parameters to succeed, because a zero Wronskian means y₁ and y₂ are linearly dependent and do not form a fundamental solution set.
TTrue
FFalse
Answer: True
True. The Wronskian appears in the denominator of the formulas u₁' = −y₂f/W and u₂' = y₁f/W. If W = 0, division by zero means the method breaks down. Crucially, W = 0 precisely when y₁ and y₂ are linearly dependent. Linearly dependent solutions do not span the solution space of the homogeneous equation, so the foundation of the method — using y₁ and y₂ as a basis — fails. This is why you need a fundamental solution set before starting.
Question 5 Short Answer
Explain the role that 'promoting constants to functions' plays in the logic of variation of parameters. Why does the method begin with the homogeneous solution's structure?
Think about your answer, then reveal below.
Model answer: The homogeneous solution y_h = c₁y₁ + c₂y₂ spans all solutions when f(x) = 0. Variation of parameters asks: what if the constants c₁ and c₂ were actually functions of x? This transforms the homogeneous structure into a flexible ansatz that can absorb the effect of f(x). By starting with y₁ and y₂ — functions already guaranteed to satisfy the homogeneous equation — the method reduces the problem to finding two scalar functions u₁ and u₂ via integration. The homogeneous solutions do the structural work; u₁ and u₂ provide the degrees of freedom needed to match f(x).
Instead of guessing a new function, you recycle the homogeneous solution's building blocks and 'vary the parameters' (the constants). Because y₁ and y₂ already satisfy the homogeneous ODE, substituting u₁y₁ + u₂y₂ into the full ODE with the simplifying constraint produces a system involving only u₁' and u₂' — no second derivatives of the unknown functions appear. This keeps the system algebraically tractable and reduces a second-order problem to two first-order integrations.