Exponential growth assumes nothing ever runs out. Logistic growth is what you write down when something does.

The framework builds the model from a sentence rather than from a formula: the rate of change of a quantity is jointly proportional to the size of the quantity and the difference between the quantity and the carrying capacity. Translated,

\[\frac{dy}{dt} = ky(a-y).\]

Two factors, each doing a job. The \(y\) makes growth proportional to what is already there, which is the exponential part. The \((a-y)\) is the brake, and it tightens as \(y\) approaches \(a\).

From the sentence to the equation

“Jointly proportional to” is the phrase carrying the structure, and it means proportional to a product, not to a sum. So the rate is \(k\) times \(y\) times \((a - y)\), and the two misreadings both come from mishandling that phrase.

The first is writing \(\tfrac{dy}{dt} = ky + (a-y)\), which is proportional to nothing in particular and does not have \(y = a\) as an equilibrium. The second is writing \(\tfrac{dy}{dt} = k(a-y)\), dropping the size factor — that is a perfectly good model, but it is Newton’s law of cooling rather than logistic growth, and its graph has no S-shape because nothing slows it down at the start.

The test for whether the equation is right is whether it has two constant solutions. Logistic growth stalls at the bottom as well as at the top, and both stalls have to be visible in the algebra.

Everything without solving

The framework states, as its own piece of essential knowledge, that the logistic differential equation and initial conditions can be interpreted without solving the differential equation. That is the method, not a fallback. The exam does not ask for \(y(t)\), and the two quantities it does ask for are both properties of the right-hand side.

The carrying capacity. Growth stops when \(\tfrac{dy}{dt} = 0\), and \(ky(a-y) = 0\) exactly when \(y = 0\) or \(y = a\). Those are the two constant solutions. Starting anywhere strictly between them, the rate is positive, so \(y\) increases and cannot pass \(a\) without the rate changing sign first. The limiting value is \(a\), read straight off the factored form.

Where growth is fastest. As a function of \(y\), the rate \(ky(a-y)\) is a downward parabola with roots at \(0\) and \(a\), so it peaks halfway between them. The fastest growth happens at

\[y = \frac{a}{2}, \qquad \text{with rate} \quad \frac{ka^2}{4}.\]

No calculus was needed for that, only the symmetry of a parabola. Differentiating the rate with respect to \(y\) gives \(k(a-2y)\), which agrees.

What that means for the shape

The value \(y = \tfrac{a}{2}\) is also where the solution curve changes concavity, and the reason is the chain rule applied to the equation itself:

\[\frac{d^2y}{dt^2} = \frac{d}{dt}\Big[ky(a-y)\Big] = k(a-2y)\frac{dy}{dt} = k^2\,y\,(a-y)(a-2y).\]

For \(0 < y < a\) the first two factors are positive, so the sign is the sign of \(a - 2y\): concave up below half capacity, concave down above it. That is the S-shape, and it was derived without ever writing down a solution.

The carrying capacity is fixed at 100. Short strokes give the slope field for dy/dt = ky(100 − y); the pale curves are solutions from a spread of starting values and the dark one is the solution from the chosen y₀. Every curve is computed by stepping the differential equation numerically, never from a formula for y — which is the point of the topic. The two dashed lines are the equilibria at 0 and 100; the dotted line at 50 is where every curve has its steepest point and its inflection, whichever side it starts on.

Every curve is produced by stepping the differential equation forward numerically, in the same spirit as Euler’s method but with a more accurate step. That is deliberate: nothing in the picture comes from a formula for \(y\), because the topic does not have one.

Three things are worth watching. Every curve starting strictly between 0 and 100 ends at 100, and so does every curve starting above it — the approach is from below or from above, and the destination does not care. The two flat lines are genuine solutions, not artifacts. And \(k\) changes how fast the whole thing happens without moving the carrying capacity or the half-capacity line by a single unit.

That last point is the one to hold onto. The framework asks for the carrying capacity and for the value where the quantity is changing fastest, and neither depends on \(k\) or on where you started.

Reasoning about a differential equation’s long-run behaviour without solving it has a name in the courses that follow this one, and it is most of what they do. The logistic model is where it starts.

Saying it in context

The suggested skill for this topic is to explain the meaning of mathematical solutions in context, which for a logistic model means three sentences and no formulas.

Take a fish population modelled by \(\tfrac{dP}{dt} = 0.02P(100-P)\), with \(P\) in thousands and \(t\) in years, starting at 10 thousand. The population approaches 100 thousand in the long run, because that is the value making the rate zero from a start below it. It is growing fastest when it reaches 50 thousand, half the carrying capacity. At that moment it is growing at \(\tfrac{ka^2}{4} = 50\) thousand per year.

Every one of those came from the equation. The one thing they do not tell you is when the population reaches 50 thousand, and that is the question the exam does not ask, because answering it needs the solution.

A quick sanity check on any logistic answer: the carrying capacity is the number in the parentheses when the equation is written as \(\tfrac{dy}{dt} = ky(a-y)\), and half of it is where growth peaks. If the equation arrives in the expanded form \(\tfrac{dy}{dt} = 0.5y - 0.001y^2\), factor first — that is \(0.001y(500-y)\), so the capacity is 500 and not 0.5 or 0.001. Reading a capacity off an unfactored equation is the most common way to lose the point.