4.7 Exponential and Logarithmic Models

Exponential and Logarithmic Models

Exponential and logarithmic models describe how quantities change over time in predictable patterns. They show up constantly in science, finance, and engineering because so many real-world processes follow these mathematical behaviors. This section covers how to use these models, when to choose one over another, and how the natural base $e$ ties everything together.

more resources to help you study

practice questions

Exponential and Logarithmic Models

Applications of exponential models

Exponential growth describes quantities that increase by a constant percentage over equal time intervals.

• General form: $y = a(1 + r)^t$
  • $a$ = initial amount
  • $r$ = growth rate (as a decimal)
  • $t$ = time
• Example: A bacteria colony starts at 500 and doubles every hour. With $r = 1.0$ (100% growth per hour), after 3 hours you'd have $500(1 + 1)^3 = 4000$ bacteria.

Exponential decay describes quantities that decrease by a fixed percentage over equal time intervals.

• General form: $y = a(1 - r)^t$
  • $a$ = initial amount
  • $r$ = decay rate (as a decimal)
  • $t$ = time
• Example: A car worth $20,000 depreciates 15% per year. After 4 years: $20000(1 - 0.15)^4 = 20000(0.85)^4 \approx \$10,455$.

The key difference between growth and decay is the base of the exponential: if $(1 + r) > 1$, the function grows; if $(1 - r) < 1$, it decays.

Real-world applications span many fields:

• Biology: tumor growth, spread of infectious diseases
• Physics: cooling of objects, capacitor discharge
• Chemistry: reaction rates, the pH scale (which is logarithmic)
• Finance: compound interest, asset depreciation, inflation

Newton's Law of Cooling problems

Newton's Law of Cooling models how an object's temperature changes as it approaches the temperature of its surroundings. The closer the object's temperature gets to the ambient temperature, the slower it changes.

Formula: $T(t) = T_s + (T_0 - T_s)e^{-kt}$

• $T(t)$ = temperature of the object at time $t$
• $T_s$ = surrounding (ambient) temperature
• $T_0$ = initial temperature of the object
• $k$ = positive cooling constant (depends on the object and environment)
• $e^{-kt}$ ensures the difference between the object and surroundings shrinks over time

Solving a typical problem:

1. Identify your known values: $T_s$, $T_0$, and usually one data point (a temperature at a specific time).
2. Substitute the data point into the formula to solve for $k$.
3. Once you have $k$, use the complete equation to find the unknown (usually a time or a temperature).

Example: A cup of coffee at 190°F is placed in a 70°F room. After 10 minutes it's 150°F. To find when it reaches 100°F:

• First, solve for $k$ using $150 = 70 + (190 - 70)e^{-10k}$, which gives $e^{-10k} = \frac{80}{120}$, so $k = \frac{-\ln(2/3)}{10} \approx 0.0405$.
• Then solve $100 = 70 + 120e^{-0.0405t}$ for $t$.

Applications include forensic science (estimating time of death from body temperature), food safety, and engineering (cooling of electronic components).

Logistic growth in populations

Logistic growth models what happens when a population can't grow forever because resources are limited. Unlike pure exponential growth, logistic growth has a built-in ceiling.

Formula: $P(t) = \frac{KP_0}{P_0 + (K - P_0)e^{-rt}}$

• $P(t)$ = population at time $t$
• $P_0$ = initial population
• $K$ = carrying capacity (the maximum population the environment can sustain)
• $r$ = growth rate

The behavior unfolds in three phases:

• Early on, when $P$ is much smaller than $K$, growth looks nearly exponential.
• Mid-phase, growth slows as resources become scarcer.
• Long-term, the population levels off and approaches $K$ asymptotically.

This produces the characteristic S-shaped (sigmoid) curve.

How the parameters affect the model:

• A higher $K$ raises the ceiling the population approaches.
• A larger $r$ makes the population reach that ceiling faster (steeper middle section of the S-curve).
• $P_0$ determines where on the curve you start.

Applications include animal populations in ecosystems, bacteria growing in a petri dish with limited nutrients, and modeling the spread of diseases through a finite population.

Model Selection and Analysis

Model selection for data sets

Choosing the right model depends on the pattern in your data and the real-world context.

How to decide:

1. Look at the data pattern. Calculate the ratio between consecutive values. If the ratio is roughly constant, that suggests exponential behavior. If growth starts fast then clearly levels off, think logistic.
2. Consider the context. Are there limiting factors (finite resources, maximum capacity)? If yes, logistic is more appropriate. If growth or decay appears unrestricted over the relevant time frame, exponential fits better.
3. Check your fit. Use regression tools (exponential regression, logistic regression) on your calculator or software to see which model fits the data more closely.

Know the limitations of each model:

• Exponential models assume unlimited growth or decay, which is rarely realistic over long time periods. They work well for short-to-medium term predictions.
• Logistic models are more realistic for populations but simplify complex dynamics (they don't account for predator-prey cycles, seasonal effects, etc.).
• Logarithmic models (covered below) fit data that increases quickly at first then grows more and more slowly, without a strict upper bound.

Always justify your choice by connecting the math to the situation. A model that fits the numbers but doesn't make sense in context isn't a good model.

Natural base e in exponential models

You can convert any exponential model into a form using the natural base $e$, which is especially useful for calculus and for comparing different growth/decay processes on equal footing.

Converting between forms:

To go from $y = a(1 + r)^t$ to $y = ae^{kt}$, set $k = \ln(1 + r)$.

For decay, $y = a(1 - r)^t$ becomes $y = ae^{kt}$ where $k = \ln(1 - r)$, and $k$ will be negative.

Interpreting the continuous form $y = ae^{kt}$:

• $a$ = initial amount (same as before)
• $k$ = continuous rate: positive means growth, negative means decay
• $t$ = time

Why use the $e$ form?

1. It simplifies calculus operations (derivatives and integrals are cleaner).
2. Natural log properties make solving for $t$ straightforward: just take $\ln$ of both sides.
3. It provides a universal format, so you can directly compare rates across different models.

Common models already written in this form:

• Continuously compounded interest: $A = Pe^{rt}$
• Newton's Law of Cooling: $T(t) = T_s + (T_0 - T_s)e^{-kt}$
• Radioactive decay: $N(t) = N_0 e^{-kt}$

Analyzing exponential and logarithmic models

Exponential models ($y = ae^{kt}$ or $y = a(1+r)^t$):

• Doubling time: the time for a growing quantity to double. Set $2a = ae^{kt}$ and solve: $t_{double} = \frac{\ln 2}{k}$. For example, if $k = 0.05$, the doubling time is $\frac{0.693}{0.05} \approx 13.86$ time units.
• Half-life: the time for a decaying quantity to drop to half. Set $\frac{a}{2} = ae^{kt}$ and solve: $t_{half} = \frac{\ln 2}{|k|}$. The formula is the same structure because doubling and halving are inverse processes.
• Horizontal asymptote at $y = 0$ for decay models (the quantity approaches zero but never reaches it). Growth models have no upper bound.

Logarithmic models ($y = a\ln(x) + b$):

• These are the inverse of exponential functions. They grow quickly for small $x$ values, then increase more and more slowly.
• They have a vertical asymptote at $x = 0$ (the function is undefined for $x \leq 0$), not a horizontal one.
• Useful for modeling phenomena like the Richter scale (earthquakes) or decibel scale (sound intensity), where each unit increase represents a multiplicative change in the underlying quantity.

Don't confuse the asymptote types: exponential functions have horizontal asymptotes, while logarithmic functions have vertical asymptotes. This distinction comes directly from the fact that they're inverses of each other.