Exponential Growth and Decay Models — Exponential Models...

SECTION 1

Historical Context & Motivation

The idea that a quantity's rate of change is proportional to its current size is one of the oldest and most consequential insights in mathematical modeling. Long before differential equations had formal notation, scholars noticed that populations, investments, and natural processes exhibit a strikingly self-similar pattern: the larger something becomes, the faster it grows—or the more material remains, the faster it decays. Translating this observation into the language of calculus produced the exponential differential equation, dy/dt = ky, a deceptively simple first-order ODE whose solution underpins fields from nuclear physics to epidemiology. Understanding the historical arc of this equation reveals why it remains a cornerstone of applied mathematics.

1683

Jacob Bernoulli and Compound Interest

While investigating continuously compounded interest, Jacob Bernoulli discovered the limiting value (1 + 1/n)ⁿ as n → ∞, arriving at the constant we now call e ≈ 2.71828. This laid the algebraic groundwork for exponential functions.

1798

Malthus and Population Growth

Thomas Malthus published An Essay on the Principle of Population, arguing that unchecked populations grow geometrically (exponentially). His qualitative model later became the differential equation dP/dt = kP, the prototypical growth model.

1902

Rutherford–Soddy Law of Radioactive Decay

Ernest Rutherford and Frederick Soddy demonstrated that the rate at which a radioactive substance transforms is proportional to the amount present: dN/dt = −λN. This confirmed the exponential decay model experimentally and introduced the concept of half-life.

1920s–1930s

Logistic Refinements

Pearl, Reed, and Lotka extended the simple exponential model to account for limited resources, yielding the logistic equation. The pure exponential ODE remained important as the linearized approximation near the origin and as the foundation upon which more complex models are built.

The central question that ties all these milestones together is remarkably concise: if the instantaneous rate of change of a quantity is proportional to the quantity itself, what function satisfies that condition, and how do initial data and the proportionality constant shape its behavior? Answering this question is the goal of the present lesson.

SECTION 2

Core Principles & Definitions

Before we solve anything, it is essential to fix the key ideas that govern exponential models. Every concept below connects directly to the differential equation dy/dt = ky and its solution y(t) = y₀e^kt. Internalize these principles and the rest of the lesson—worked examples, applications, and extensions—follows naturally.

Proportional Rate of Change

The defining property of an exponential model is that the derivative dy/dt is directly proportional to y itself. This single statement encodes the ODE dy/dt = ky, where k is the proportionality constant.

Sign of k Determines Behavior

When k > 0, the quantity grows without bound—exponential growth. When k < 0, the quantity decreases toward zero—exponential decay. The magnitude |k| controls the speed.

Initial Condition y₀

The general solution of dy/dt = ky is y = Ce^kt. Imposing y(0) = y₀ fixes the constant C = y₀, yielding the unique particular solution y(t) = y₀e^kt.

Separation of Variables

Because both sides of the ODE can be algebraically rearranged so that y appears only on one side and t on the other, the equation is separable. This technique—separation of variables—provides the standard solution pathway.

Half-Life and Doubling Time

For decay (k < 0), the half-life t₁/₂ = ln 2 / |k|. For growth (k > 0), the doubling time t_d = ln 2 / k. Both are constant, independent of the current amount.

✦ KEY TAKEAWAY

Think of an exponential process like a snowball rolling downhill: the bigger it gets, the more surface area it has to pick up snow, so it grows faster and faster. If instead the snowball were evaporating, the larger it is the more surface area loses material, so it shrinks—but ever more slowly. In both cases the rate is proportional to the current size, and the mathematical description is exactly dy/dt = ky.

SECTION 3

Visual Explanation — Growth vs. Decay Curves

The graph below contrasts exponential growth and exponential decay on the same set of axes. Both curves share the same initial value y₀ and differ only in the sign of k. Observe how the growth curve accelerates away from the initial value while the decay curve asymptotically approaches zero. The slope field arrows indicate the direction and magnitude of dy/dt at representative points, reinforcing that the rate at each location is proportional to y.

Both curves start at the same initial value y₀ (yellow dot). The green growth curve (k > 0) rises ever more steeply, while the red decay curve (k < 0) falls toward zero asymptotically. Short line segments represent the slope dy/dt at selected points, visually confirming that the slope scales with the height of the curve.

Several features of this diagram merit emphasis. First, the growth curve is concave up: its second derivative is also positive, since d²y/dt² = k(dy/dt) = k²y > 0. Second, the decay curve is concave up as well—it bends upward toward the horizontal asymptote y = 0 rather than crossing it. Third, neither curve ever reaches zero or infinity in finite time; the exponential function is defined for all real t. These qualitative properties are immediate consequences of the ODE and do not require computing any particular solution.

SECTION 4

Mathematical Framework — Solving dy/dt = ky

We now derive the general solution rigorously using separation of variables. The technique works because both y and t can be isolated on opposite sides of the equation, allowing us to integrate each side independently.

THE EXPONENTIAL ODE

dy/dt = ky

y = y(t) is the unknown function, t is the independent variable (usually time), and k ∈ ℝ is the growth rate constant. When k > 0 the model describes growth; when k < 0 it describes decay.

Derivation by Separation of Variables

Assuming y ≠ 0, divide both sides by y and multiply both sides by dt to obtain (1/y) dy = k dt. Integrating the left side with respect to y and the right side with respect to t gives ∫(1/y) dy = ∫k dt, which yields ln |y| = kt + C₁ where C₁ is an arbitrary constant of integration. Exponentiating both sides produces |y| = e^{kt + C₁} = e^C₁·e^kt. Absorbing the sign and the constant e^C₁ into a single constant C, we write y = Ce^kt, where C may be any real number (including zero, which recovers the trivial solution y ≡ 0 that we temporarily excluded).

GENERAL SOLUTION

y(t) = Ce^(kt)

C is an arbitrary constant determined by initial or boundary data. This is a one-parameter family of solutions.

PARTICULAR SOLUTION (IVP)

y(t) = y₀ e^(kt), where y(0) = y₀

Setting t = 0 in the general solution gives y(0) = C·e⁰ = C, so C = y₀. This yields the unique solution to the initial value problem (IVP).

HALF-LIFE / DOUBLING TIME

t₁/₂ = ln 2 / |k| (decay); t_d = ln 2 / k (growth)

Setting y(t) = y₀/2 (decay) or y(t) = 2y₀ (growth) and solving for t yields these closed-form expressions. Note that both are independent of y₀—a hallmark of exponential processes.

ℹ️ Existence and Uniqueness

By the Picard–Lindelöf theorem, since f(t, y) = ky is Lipschitz continuous in y on all of ℝ, the IVP dy/dt = ky, y(0) = y₀ has a unique solution on (−∞, ∞). There is no possibility of blow-up in finite time or non-uniqueness. This theoretical guarantee is part of what makes the exponential model so clean and tractable.

SECTION 5

Applications & Classification of Exponential Models

The same ODE dy/dt = ky appears across remarkably different disciplines. In each application, the variable y, the meaning of k, and the relevant units change, but the mathematical structure and its solution remain identical. The table below catalogs the most common instances, and the diagram that follows illustrates how the rate constant k maps onto qualitatively different solution behaviors.

Common applications of the exponential ODE dy/dt = ky

Application	Variable y	Constant k	Key Formula
Population Growth	P(t), population size	Birth rate − death rate (k > 0)	P(t) = P₀eᵏᵗ
Radioactive Decay	N(t), number of atoms	−λ (decay constant, k < 0)	N(t) = N₀e^(−λt)
Compound Interest	A(t), account balance	r, annual interest rate (k > 0)	A(t) = A₀eʳᵗ
Newton's Cooling	T(t) − Tₐ, temp difference	−h (cooling constant, k < 0)	T(t) = Tₐ + (T₀ − Tₐ)e^(−ht)
Drug Elimination	C(t), drug concentration	−k_el (elimination rate, k < 0)	C(t) = C₀e^(−k_el · t)

The number line of k values partitions all exponential models into qualitative regimes. Rapid decay (k ≪ 0) depletes material quickly; equilibrium (k = 0) holds y constant; and rapid growth (k ≫ 0) causes explosive increase. Each card shows a mini curve and a real-world example.

An important observation from the classification above is that Newton's law of cooling—while not literally about growth or decay of a single quantity—reduces to the same ODE once we define y = T(t) − Tₐ. This substitution trick of shifting by a constant is a recurring motif in differential equations: if the equilibrium is not at zero, translate the variable so that it is, solve the resulting exponential ODE, and then shift back.

SECTION 6

Worked Example — Carbon-14 Dating

Archaeologists discover a wooden artifact and measure that it retains 72% of its original carbon-14 content. The half-life of carbon-14 is 5730 years. Determine (a) the decay constant k, and (b) the approximate age of the artifact.

Carbon-14 Radioactive Decay

Step 1 — Write the ODE and General SolutionLet N(t) be the amount of carbon-14 at time t (in years). The decay model gives dN/dt = kN with k < 0. The general solution is N(t) = N₀e^kt, where N₀ = N(0) is the initial amount.

Step 2 — Find k from the Half-LifeAt t = t₁/₂ = 5730 years, N = N₀/2. Substituting: N₀/2 = N₀e^k(5730). Dividing by N₀ and taking the natural log: ln(1/2) = 5730k, so k = −ln 2 / 5730.

k ≈ −1.2097 × 10⁻⁴ per year

Step 3 — Set Up the Age EquationThe artifact retains 72% of its original C-14, so N(t) = 0.72N₀. Substitute into the particular solution: 0.72N₀ = N₀e^kt. Cancel N₀ to obtain 0.72 = e^kt.

Step 4 — Solve for tTake the natural logarithm of both sides: ln(0.72) = kt. Therefore t = ln(0.72) / k = ln(0.72) / (−ln 2 / 5730). Computing: ln(0.72) ≈ −0.3285, and −ln 2 ≈ −0.6931, so t = (−0.3285) / (−0.6931/5730) = (−0.3285)(5730) / (−0.6931).

t ≈ 2715 years

Step 5 — Interpret the ResultThe artifact is approximately 2715 years old. Notice that we never needed to know the actual initial quantity N₀—only the fraction remaining. This is a direct consequence of the exponential solution's multiplicative structure: the ratio N(t)/N₀ = e^kt depends only on k and t, not on the initial amount.

SECTION 7

Strengths and Limitations of the Exponential Model

The exponential model's elegance comes at a cost: its assumptions are restrictive. In many real-world scenarios the proportionality constant k is not truly constant—it may depend on time, on the quantity itself, or on external factors. The table below contrasts the model's genuine strengths with its well-known limitations.

Strengths vs. limitations of dy/dt = ky

Strengths	Limitations
Closed-form solution y = y₀eᵏᵗ — no numerical methods needed.	Unbounded growth: y → ∞ as t → ∞ when k > 0, which is unrealistic for populations with limited resources.
Only one parameter (k) to estimate from data, making the model parsimonious and easy to fit.	Assumes constant relative rate k — does not capture seasonal variation, feedback, or saturation effects.
Excellent short-term approximation near t = 0 for many nonlinear systems (linearization principle).	Ignores interactions: in ecology, competition, predation, and cooperation all violate the simple proportionality assumption.
Half-life / doubling time formulas provide immediate physical insight without solving the full IVP.	Cannot model quantities that oscillate, saturate, or change sign — the solution is strictly monotone.

✦ KEY TAKEAWAY

Think of the exponential model as the "free-body diagram" of differential equations: it strips away all complicating factors—friction, drag, resource limits—to reveal the purest form of self-referential change. Just as a free-body diagram is the first step in any statics or dynamics problem before adding complexity, the exponential ODE is the first model to try before layering on nonlinear terms (logistic, Lotka–Volterra, SIR, etc.).

SECTION 8

Connection to Advanced Models

The exponential ODE is the simplest member of a family of increasingly realistic models. In practice, one often begins with dy/dt = ky and then refines the model to include capacity constraints, time-varying rates, or coupled variables. The table below shows how the exponential model connects upward to three important extensions you will encounter in further coursework.

From exponential to advanced ODE models

Feature	Exponential Model	Advanced Extension
ODE	dy/dt = ky	Logistic: dy/dt = ky(1 − y/K)
Carrying capacity	None — unbounded growth	K — the population saturates at K
Rate constant	Constant k	Non-autonomous: dy/dt = k(t)y allows time-varying rates
Number of species	Single variable	Coupled systems (Lotka–Volterra, SIR) with two or more interacting variables
Linearity	Linear in y	Nonlinear — generally no closed-form solution; qualitative and numerical methods required

Crucially, the exponential model is the linearization of the logistic equation near y = 0. When a population is small relative to the carrying capacity K, the factor (1 − y/K) ≈ 1, and the logistic ODE reduces to dy/dt ≈ ky—the exponential model. This means that understanding exponential behavior is not just a stepping stone; it is the tangent-line approximation to every logistic-type system in its early phase. Similarly, the matrix exponential e^At that solves linear systems dx/dt = Ax generalizes the scalar solution e^kt to multiple dimensions, making the one-dimensional exponential ODE foundational for all of linear ODE theory.

SECTION 9

Practice Problems

PROBLEM 1 — CONCEPTUAL

Explain why the half-life of an exponentially decaying substance is independent of the initial amount. Reference the solution y(t) = y₀e^kt in your argument.

PROBLEM 2 — BASIC CALCULATION

A bacterial culture grows according to dP/dt = 0.35P, with P measured in thousands and t in hours. If P(0) = 2 (thousand bacteria), find P(4).

PROBLEM 3 — INTERMEDIATE

A substance decays from 120 mg to 75 mg in 10 hours. (a) Find the decay constant k. (b) How long will it take for the substance to reach 20 mg?

PROBLEM 4 — APPLIED

An investment of $5000 earns interest compounded continuously at an annual rate r. After 8 years the account holds $7450. Determine r and find how long it will take for the investment to triple from its original value.

PROBLEM 5 — CRITICAL THINKING

Consider the ODE dy/dt = k(t) · y, where k(t) = k₀ cos(ωt) is a periodically varying rate constant. (a) Show that this ODE is still separable and find y(t) in terms of k₀, ω, and y₀. (b) Is the resulting solution periodic? Does the long-term behavior resemble exponential growth, decay, or neither? Justify your answer.

SUMMARY

Lesson Summary

The exponential differential equation dy/dt = ky states that a quantity's instantaneous rate of change is proportional to its current value. Solving this separable ODE yields the general solution y(t) = Ce^kt, and imposing the initial condition y(0) = y₀ fixes C = y₀. The sign of the rate constant k determines whether the solution exhibits exponential growth (k > 0) or exponential decay (k < 0), while its magnitude governs the half-life or doubling time via t₁/₂ = ln 2 / |k|.

This single model unifies phenomena across population dynamics, radioactive decay, continuously compounded interest, Newton's law of cooling, and pharmacokinetics. While its assumption of a constant proportionality rate limits long-term accuracy for bounded or interacting systems, the exponential ODE serves as the foundational linearization from which more complex models—logistic, Lotka–Volterra, SIR—are built. Mastering this equation provides the essential toolkit for all subsequent work in ordinary differential equations.