What Counts as a Solution?

A solution is a function, not a number

In the first two guides you met the idea of a differential equation and learned to read its order and whether it is linear or nonlinear. Now we ask the question the whole subject hangs on: when has the equation actually been answered? The mental jump is this. In ordinary algebra, solving 'x^2 = 9' hunts for a number. In differential equations the unknown is a whole function, so a solution is an entire curve — a rule that supplies an output value at every input at once.

Picture the equation as posting an instruction at every point of the plane: 'here your slope must be exactly this much.' A solution is then a curve that, faithfully and everywhere along an interval, runs with precisely the slope the rule demands — never fighting it. For y' = 2y, the curve y = e^(2x) qualifies because at each point its own slope, 2 e^(2x), equals 2 times its height, 2 times e^(2x). The curve and the rule agree at every single x. That, and only that, is what 'being a solution' means.

Verify by substitution — the one test that always works

Here is the most reassuring fact in this rung: you never have to trust a solution on faith. Because the unknown was a function, you can simply plug a candidate function and its derivatives back into the equation and check, mechanically, that both sides agree for every x in the interval. This is verifying a solution by substitution, and it works even when you had no idea how the answer was found — handed a guess, a textbook formula, or a friend's claim, you can confirm or refute it without solving anything.

1. Claim: y = sin(x) solves y'' + y = 0. Compute the derivatives you need: y' = cos(x), then y'' = -sin(x).
2. Substitute into the left side: y'' + y = (-sin(x)) + sin(x).
3. Simplify: -sin(x) + sin(x) = 0, which equals the right side for every x. Verified.
4. Same recipe refutes a wrong guess: y = x fails y'' + y = 0 because y'' + y = 0 + x = x, which is not 0.

Substitution is also how you check an implicit solution, one given as a relation tying x and y together rather than y served up alone. For x^2 + y^2 = C, you differentiate implicitly: 2x + 2y y' = 0, which rearranges to y' = -x/y. So that relation is a genuine solution of y' = -x/y, even though y is never isolated. An implicit answer is a complete answer; you only solve for y when you actually need numerical values, and then you may have to choose the right branch of the curve.

One solution, or a whole family?

Solve a differential equation and you rarely get a lone curve — you get a fleet of them. The general solution is the master formula that produces every member by dialing adjustable knobs called arbitrary constants. The count of knobs matches the order: a first-order equation carries one, a second-order equation two, an nth-order equation n. For y' = k y the general solution is y = C e^(kx), and as the single constant C sweeps through every real value the curves stack up and fill the plane — a one-parameter family of solutions.

A particular solution is one chosen member of that family, with no leftover free constants. You obtain it by feeding the family a starting fact. Begin with y = C e^(kx); told that y(0) = 5, set x = 0 to get 5 = C e^0 = C, so C = 5 and the particular solution is y = 5 e^(kx). The general-to-particular move — write the family, then apply conditions to pin the constants — is the standard final step of nearly every method you will learn, and it is exactly what the next guide develops into the initial value problem.

equation order n  ->  general solution has n arbitrary constants

  y' = k y            y = C e^(kx)              (1 constant)
  y'' + y = 0         y = C1 cos x + C2 sin x   (2 constants)
  y''' = 0            y = C1 + C2 x + C3 x^2    (3 constants)

The honest traps: when "general" misses solutions

It is tempting to read 'general solution' as 'literally every solution.' For linear equations that is true. For nonlinear ones, stay alert — there are stragglers the constant-bearing family never catches. The first kind appears when you solve a separable equation by dividing through by y. Dividing assumes y is not zero, so the step quietly discards any constant solution where y = 0. That discarded curve is a lost constant solution: a perfectly valid answer your algebra threw away. Always check separately whether a constant function satisfies the equation.

The deeper kind is a singular solution: a genuine solution that lies entirely outside the general family, obtainable from no value of the constants whatsoever. The classic case is Clairaut's equation, whose family is a set of straight lines while a curved envelope — tangent to all of them — also solves it, yet corresponds to no constant. Such an envelope is the geometric signature of a singular solution. The take-home: 'general' means 'the main family,' not a guarantee of completeness, and for nonlinear equations you must look for what the family leaves out.

When no formula exists — and why that is fine

Here is the most sobering honesty of the whole subject. The tidy closed-form answers of a first course — e^(kx), sin and cos, neat implicit relations — are a small, special collection. The large majority of differential equations have NO solution expressible by elementary formulas. Even something as innocent-looking as y' = x^2 + y^2 cannot be solved by any combination of the functions you know. A solution still exists; you just cannot write it down with a finite formula.

Far from being a dead end, this is why the rest of the subject exists. When no formula is available, two doors open. Numerical methods march a solution forward in tiny steps and hand you an approximate curve of numbers. Qualitative methods describe the solution's behaviour — does it rise, fall, level off at an equilibrium, or blow up — without ever producing a formula. The very last guide in this rung opens that second door with the slope field, a way to SEE the family of solution curves before you compute a single one.