5.2 Exponential models

Exponential models describe situations where a quantity changes by a constant factor over equal intervals of time (or space). They show up everywhere: population growth, compound interest, radioactive decay, the spread of diseases. Understanding how they work gives you a powerful tool for analyzing systems that grow or shrink faster and faster over time.

Definition of exponential models

An exponential model captures any process where the amount you have gets multiplied by the same factor during each time step. If a bacterial colony doubles every hour, that's exponential. If a car loses 15% of its value every year, that's also exponential (decay).

This is fundamentally different from adding the same amount each time (which would be linear). The multiplicative pattern is what makes exponential behavior so dramatic over longer periods.

Key components

• Initial value ($a$): The starting quantity at time zero.
• Growth/decay factor ($b$): The multiplier applied each interval. If $b > 1$, the quantity grows. If $0 < b < 1$, it decays.
• Independent variable ($x$ or $t$): Usually represents time or the number of intervals that have passed.
• Exponent: Shows how many times the growth factor has been applied. After 5 intervals, you've multiplied by $b$ five times.

Exponential growth vs decay

Growth happens when the factor $b > 1$. Each step, the quantity gets bigger, and because you're multiplying a bigger number by the same factor, the amount of increase accelerates. Population growth under ideal conditions and compound interest are classic examples.

Decay happens when $0 < b < 1$. Each step, the quantity shrinks, but it never quite reaches zero. Radioactive decay and asset depreciation follow this pattern.

Mathematical representation

Base and exponent

The base ($b$) is the growth or decay factor, and the exponent ($x$) tells you how many times to multiply:

$b^x = b \times b \times b \times \ldots \text{ (x times)}$

Some common bases you'll encounter:

• 2 for doubling scenarios
• 10 for orders-of-magnitude problems
• $e \approx 2.71828$ for continuous growth models (more on this later)

The exponent can be positive (growth), negative (decay), or even fractional (which corresponds to roots).

General form of equation

The standard exponential equation is:

$y = ab^x$

• $y$ = the output (dependent variable)
• $a$ = the initial value (the value of $y$ when $x = 0$, since $b^0 = 1$)
• $b$ = the base (growth or decay factor)
• $x$ = the input (independent variable, often time)

A common variation for compound interest is $y = a(1 + r)^x$, where $r$ is the rate of change per period. If something grows by 5% per year, $r = 0.05$ and $b = 1.05$.

Properties of exponential functions

Domain and range

• Domain: All real numbers. You can plug in any value of $x$.
• Range: $(0, \infty)$ when $a > 0$. The output is always positive and never equals zero.
• The y-intercept is always $a$ (the initial value), since $y = ab^0 = a$.
• There is no x-intercept for a standard exponential function because the curve never touches the x-axis.

Asymptotic behavior

Exponential functions have a horizontal asymptote at $y = 0$. For decay functions, the curve gets closer and closer to zero but never actually reaches it. For growth functions, the curve shoots upward with no upper bound, while still never crossing below zero.

Rate of change

The rate of change of an exponential function is proportional to the function's current value. That's what makes it exponential: the bigger it is, the faster it grows (or the more it has left, the more it loses per interval).

• The derivative of $ab^x$ is $ab^x \ln(b)$.
• The percent rate of change stays constant across the entire domain. If something grows by 8% per year, it's 8% per year whether the quantity is 100 or 10,000.

Applications in real-world

Population growth

Under ideal conditions (unlimited food, space, no predators), populations grow exponentially:

$P(t) = P_0 e^{rt}$

Here $P_0$ is the initial population, $r$ is the growth rate, and $t$ is time. For example, a bacteria culture starting with 500 cells and growing at a rate of $r = 0.03$ per minute would have $P(60) = 500e^{0.03 \times 60} = 500e^{1.8} \approx 3,025$ cells after one hour.

This model assumes unlimited resources, which is realistic only for short-term predictions or early-stage growth.

Compound interest

The compound interest formula is:

$A = P\left(1 + \frac{r}{n}\right)^{nt}$

• $A$ = final amount
• $P$ = principal (initial investment)
• $r$ = annual interest rate (as a decimal)
• $n$ = number of times interest compounds per year
• $t$ = time in years

If you invest $1,000 at 6% annual interest compounded monthly, after 10 years you'd have:

$A = 1000\left(1 + \frac{0.06}{12}\right)^{12 \times 10} = 1000(1.005)^{120} \approx \$1,819.40$

Radioactive decay

Radioactive materials decrease over time following:

$N(t) = N_0 e^{-\lambda t}$

Here $N_0$ is the initial amount, $\lambda$ (lambda) is the decay constant, and $t$ is time. The half-life is the time it takes for half the material to decay. Carbon-14 has a half-life of about 5,730 years, which is why carbon dating works for archaeological artifacts up to roughly 50,000 years old.

Solving exponential equations

Logarithmic approach

When the unknown variable is in the exponent, logarithms are your main tool. They "undo" exponentiation.

The key property: $\log_a(x) = y$ means $a^y = x$.

Steps to solve an exponential equation:

1. Isolate the exponential expression on one side of the equation.
2. Take the logarithm of both sides (use ln for base $e$, or log for base 10).
3. Use the power rule: $\ln(b^x) = x \ln(b)$ to bring the exponent down.
4. Solve for the unknown variable.

Example: Solve $3^x = 20$

1. Take ln of both sides: $\ln(3^x) = \ln(20)$
2. Bring down the exponent: $x \ln(3) = \ln(20)$
3. Solve: $x = \frac{\ln(20)}{\ln(3)} \approx \frac{3.00}{1.10} \approx 2.73$

Graphical methods

You can also solve exponential equations by graphing both sides as separate functions and finding where they intersect. The x-coordinate of the intersection point is your solution. This approach is especially useful for getting a visual sense of how many solutions exist and their approximate values.

Exponential vs linear models

Differences in growth rates

Identifying appropriate contexts

• Use a linear model when the quantity changes by a fixed amount each period (e.g., driving at a constant speed, filling a pool at a steady rate).
• Use an exponential model when the quantity changes by a fixed percentage each period (e.g., 5% annual return on investment, 10% annual depreciation).

To figure out which model fits your data, look at the differences between consecutive values. If the differences are roughly constant, it's linear. If the ratios between consecutive values are roughly constant, it's exponential.

Data analysis and curve fitting

Exponential regression

When you have real data that looks exponential, you can fit a model to it:

1. Take the natural log (ln) of each y-value in your data set.
2. Plot ln(y) against x. If the data is truly exponential, this plot should look roughly linear.
3. Perform linear regression on the transformed data to get a line $\ln(y) = mx + c$.
4. Convert back: $a = e^c$ and $b = e^m$, giving you $y = ab^x$.

Assess how well the model fits using $R^2$ (R-squared). Values closer to 1 indicate a better fit.

Interpreting model parameters

Once you have your model $y = ab^x$:

• $a$ tells you the starting value (at $x = 0$).
• $b$ tells you the factor per unit of $x$. If $b = 1.12$, the quantity grows by 12% per unit. If $b = 0.85$, it decays by 15% per unit.
• Doubling time (for growth): $t = \frac{\ln(2)}{\ln(b)}$
• Half-life (for decay): $t = \frac{\ln(2)}{\ln(1/b)}$

Always check whether predictions from your model make sense in context. Extrapolating too far beyond your data range can give unrealistic results.

Limitations and assumptions

Carrying capacity considerations

Pure exponential growth can't continue forever in real systems. Resources run out, space fills up, competition increases. The logistic growth model accounts for this by introducing a carrying capacity $K$:

$P(t) = \frac{K}{1 + \left(\frac{K}{P_0} - 1\right)e^{-rt}}$

This produces an S-shaped curve: growth starts exponentially, then slows as the population approaches $K$. You'll see this pattern in ecology, technology adoption, and market saturation.

Short-term vs long-term predictions

Exponential models tend to be reliable for short-term predictions but can break down over longer time horizons. Factors that limit long-term accuracy include:

• Resource depletion
• Environmental changes
• Competition or market saturation
• Regulatory or technological shifts

The takeaway: use exponential models as a starting point, but be skeptical of projections that extend far beyond your data. Updating your model as new data comes in is always a good practice.

Exponential models in other fields

Economics and finance

• Moore's Law observed that the number of transistors on a chip roughly doubles every two years, a pattern that held for decades.
• Inflation models use exponential functions to project how prices increase over time.
• Asset depreciation (cars, equipment) often follows exponential decay.

Biology and ecology

• Early-stage spread of invasive species and infectious diseases often follows exponential patterns.
• Bacterial growth in lab conditions is a textbook example of exponential growth.
• Predator-prey dynamics and enzyme kinetics also involve exponential components, though these systems are more complex overall.

Advanced topics

Continuous vs discrete models

Discrete models use whole-number exponents: $a^n$ where $n = 0, 1, 2, 3, \ldots$ These apply when changes happen at specific intervals (yearly interest payments, generational population counts).

Continuous models allow any real-number exponent: $a^x$ where $x \in \mathbb{R}$. These apply when change happens constantly (radioactive decay, continuously compounded interest).

The two are connected by the identity $a^x = e^{x \ln(a)}$. Continuous exponential growth is described by the differential equation:

$\frac{dy}{dx} = ky$

where $k$ is the growth constant. The solution to this equation is $y = y_0 e^{kx}$.

Exponential models with base e

The number $e \approx 2.71828$ is the "natural" base for exponential functions. What makes it special:

• The derivative of $e^x$ is just $e^x$ again. No other base has this property.
• The integral of $e^x$ is $e^x + C$.
• It arises naturally from the limit $e = \lim_{n \to \infty}\left(1 + \frac{1}{n}\right)^n$.

Because of these properties, $e$ is the preferred base in calculus, physics, and engineering. Any exponential function $ab^x$ can be rewritten as $ae^{x \ln(b)}$, which is often more convenient for differentiation and integration.