Chapter 14 — Front Matter
Purpose and Overview
This chapter closes the book. It does not introduce a single new technique. Instead, it brings together everything that came before and asks a harder question than any individual chapter has asked: given a real situation, what do you do?
Throughout this book you have learned to separate variables, to find integrating factors, to solve characteristic equations, to sketch slope fields, to trace solution curves through phase planes, to apply Laplace transforms, to run Euler's method, to analyze stability, and to recognize what partial differential equations describe. Each of those skills lives in a chapter with its own context, its own worked examples, and its own practice problems. In that setting, you almost always knew what method to use, because the chapter told you.
Real problems are different. A real situation comes to you without a label. You must look at it, decide what kind of differential equation describes it, choose a method, carry out the analysis, check the result, and explain what it means. That is the work this chapter trains you to do.
This chapter treats the complete differential equation modeling process as its own discipline. Choosing variables, building rate equations, selecting methods, interpreting solutions, checking assumptions, estimating sensitivity, using technology, and communicating results are not decorations around the mathematics. They are the mathematics, applied.
Opening Question
A wildlife biologist is monitoring a fish population in a lake. The population grows naturally, but the lake can support only a limited number of fish. Each year, a fixed number of fish are harvested. The biologist wants to know: will the population stabilize, collapse, or grow without bound? What harvest rate is sustainable? How sensitive is the long-term outcome to uncertainty in the initial population count? This question involves modeling, classification, solution, qualitative analysis, stability, sensitivity, and interpretation. No single chapter technique answers it alone. That is exactly the kind of question this chapter prepares you to address.
15Sections
10Objectives
6Capstone Problems
10+Error Types Analyzed
Learning Objectives
- 1Describe the complete differential equation modeling process and apply it to new situations.
- 2Define appropriate variables, parameters, and units for a modeling problem.
- 3Translate verbal rate descriptions into differential equations.
- 4Classify a differential equation and select an appropriate solution or analysis method.
- 5Solve exactly when possible, and use numerical or qualitative methods when exact solutions are unavailable or impractical.
- 6Interpret the solution in context, including units, initial conditions, parameters, and long-term behavior.
- 7Check a model's assumptions and identify its limitations.
- 8Analyze sensitivity to parameter changes.
- 9Use technology to support modeling, computation, visualization, and interpretation.
- 10Communicate a dynamic model clearly in writing.
Sections 14.1 – 14.4
Building the Model: Process, Variables, Equations, and Method Selection
Ten-Step Process · Variables & Parameters · Rate Equations · Classification
Section 14.1
The Differential Equation Modeling Process
Mathematical modeling is the practice of using mathematical language to represent a real situation precisely enough to answer meaningful questions about it. Differential equation modeling represents situations where change is the central feature. The process has a structure — it is not a single step, and it is not just writing an equation.
The Ten-Step Modeling Process
- 1Understand the situation. Read carefully. Identify what is changing, what drives the change, what stays fixed, and what you are trying to find out.
- 2Define variables and parameters. Assign names to the quantities that change. Assign names to the quantities that are fixed but unknown. Specify units.
- 3Build the rate equation. Translate the description of how things change into a differential equation. Identify what the rate of change depends on.
- 4Classify the equation. Determine whether the equation is first-order or higher order, linear or nonlinear, autonomous or nonautonomous, separable, or part of a system.
- 5Choose a solution or analysis method. Based on classification, decide whether to solve exactly, solve numerically, or analyze qualitatively.
- 6Carry out the solution or analysis. Apply the chosen method carefully. Show work. Keep constants of integration visible. Apply initial conditions clearly.
- 7Check the result. Substitute back. Check units. Verify initial conditions are satisfied. Consider whether the solution's behavior is consistent with the situation.
- 8Interpret the result. Explain what the solution means for the system — parameters, equilibria, long-term behavior, and practical implications.
- 9Discuss assumptions and limitations. Identify where the model simplifies real complexity and assess whether those simplifications are acceptable.
- 10Communicate the result. Write a clear explanation that a reader who is not doing the computation can understand.
MGU Library Connection
- The modeling process described here is shared across MGU Library pages in Physics, Engineering Mechanics, Economics Modeling, Biology and Epidemiology, and Data Science Foundations
- The language of variables, parameters, rate equations, and interpretation is the common language of all quantitative applied disciplines
Section 14.2
Choosing Variables and Parameters
Before you write a differential equation, you must decide what the equation is about. A variable is a quantity that changes as the system evolves, typically a function of time. A parameter is a quantity that remains constant within a given model, even if it takes different values in different scenarios. Choosing well makes the rest of the modeling process much cleaner.
Principles for Choosing Variables
Choose the variable that represents the quantity whose change you care about. Ask: what is changing? Population? Temperature? Charge? Displacement? Concentration? Then choose the independent variable — usually time. Give each variable a symbol and state its meaning and units explicitly. Do not let symbols float without definition.
Principles for Choosing Parameters
Identify all rates, constants, capacities, and coefficients that appear in the rate description. Give each a name and state its units. Make sure units are consistent: if population is measured in individuals and time in years, then a growth rate should carry units of per year, so that the product of rate and population gives individuals per year — the correct unit of dP/dt.
Example 14.2.1
Defining variables and parameters — fish population with harvesting
- ProblemA fish population grows at a rate proportional to its current size, but fishers remove fish at a constant rate. Define variables and parameters.
- ThinkIdentify what changes (population), what drives change (growth and harvest), and what is fixed (rates).
- Define
Let P(t) = fish population (thousands of fish) at time t (years)
Let r > 0 = intrinsic growth rate (per year)
Let H ≥ 0 = constant harvesting rate (thousands of fish per year)
Initial condition: P(0) = P₀
- CheckEvery symbol has a name, a meaning, and a unit. This makes writing the equation straightforward.
Variables: P(t), t. Parameters: r, H. Initial condition: P(0) = P₀.
Section 14.3
Building Rate Equations
The rate equation expresses how fast the dependent variable is changing, in terms of the current state and the independent variable. To build it, translate the verbal description of change into mathematical language. Addition represents input to the system; subtraction represents output or removal.
| Verbal Description | Rate Equation |
|---|
| Changes at a rate proportional to its current value | dy/dt = ky |
| Grows logistically with carrying capacity K | dP/dt = rP(1 − P/K) |
| Cools at a rate proportional to difference from ambient temperature Ta | dT/dt = −k(T − Tₐ) |
| Changes due to inflow minus outflow | dA/dt = (rate in) − (rate out) |
| Accelerates in proportion to net force (Newton's second law) | m·x'' = F_net |
Example 14.3.1
Building the rate equation — fish population with harvesting
- ThinkThe population grows at rate rP (positive: increases population) and fishers remove fish at constant rate H (negative: decreases population). Net rate of change is the sum.
- Solve
dP/dt = rP − H
- CheckIs this first-order? Yes. Is it linear? Yes: P' − rP = −H matches y' + p(t)y = q(t). Is it separable? Yes, write dP/(rP − H) = dt. Is it autonomous? Yes, right side depends only on P.
dP/dt = rP − H, a first-order linear autonomous equation.
Student Warning — Signs
The most common error in building rate equations is getting the sign wrong on a term. If something increases the quantity, its contribution to dy/dt is positive. If something decreases the quantity, its contribution is negative. Reread the verbal description carefully and check each term's sign before proceeding.
Section 14.4
Choosing a Solution Method
Once you have a differential equation, the next question is: how do you solve or analyze it? Classification drives method selection. The type of equation determines the available approaches.
| Equation Type | Key Features | Primary Methods |
|---|
| First-order separable | Variables can be separated | Separate and integrate |
| First-order linear | Form y' + p(x)y = q(x) | Integrating factor |
| First-order autonomous | y' = f(y) only | Phase line, equilibria, qualitative analysis |
| First-order nonlinear | Cannot separate or linearize easily | Qualitative methods, numerical methods |
| Second-order, constant coefficients, homogeneous | ay'' + by' + cy = 0 | Characteristic equation |
| Second-order, constant coefficients, nonhomogeneous | ay'' + by' + cy = g(t) | Complementary + particular solution |
| Linear system | x' = Ax | Eigenvalue method, phase plane |
| Discontinuous input | Piecewise or switched forcing | Laplace transform |
| No tractable closed form | Any type | Euler's method, improved Euler, Runge-Kutta |
| Autonomous system or equation | y' = f(y) or x' = f(x) | Stability analysis, bifurcation analysis |
Decision Process — Ask in Order
- 1Is it first-order? → Ask: separable? linear? autonomous? Apply the matching method.
- 2Is it second-order, linear, constant coefficients? → Homogeneous: characteristic equation. Nonhomogeneous: complementary + particular.
- 3Does it involve multiple coupled quantities? → Write as a system. Use eigenvalue analysis for linear systems; phase plane and numerical methods for nonlinear.
- 4Are initial conditions given with a possibly discontinuous input? → Laplace transforms handle this most cleanly.
- 5Is exact solution difficult or unavailable? → Use numerical methods (Euler, improved Euler, Runge-Kutta).
- 6Is qualitative behavior the primary question? → Use phase lines, phase planes, stability analysis regardless of exact solvability.
Applying Method Selection — Harvesting Model
The equation dP/dt = rP − H is first-order, separable (write dP/(rP−H) = dt), linear (P' − rP = −H), and autonomous. All three approaches — separation, integrating factor, and qualitative analysis — are available. A complete treatment uses all three at different levels of depth.
Sections 14.5 – 14.10
Solution, Verification, Stability, Assumptions, and Sensitivity
Interpreting Solutions · Checking · Numerical Comparison · Stability · Assumptions · Sensitivity
Section 14.5
Interpreting Solutions
A solution to a differential equation is a function. Interpretation means explaining what that function says about the real system — connecting the mathematical output back to the variables, parameters, and situation you defined at the beginning of the modeling process.
What to Interpret
Consider: the form of the solution (does it grow, decay, oscillate, or approach a limit?); the parameters in the solution (what does each constant control?); the initial condition (how does the starting value affect long-term behavior?); the equilibrium behavior (does the solution approach a steady state?); and the units (do the values make sense for the real quantities involved?).
Example 14.5.1
Solving and interpreting the harvesting model
- ProblemSolve dP/dt = rP − H with P(0) = P₀. Interpret the result.
- ThinkSeparate variables: dP/(rP − H) = dt. Integrate both sides.
- Solve
∫ dP/(rP − H) = ∫ dt
(1/r) ln|rP − H| = t + C₁
rP − H = Ce^(rt)
P(t) = H/r + (C/r)e^(rt)
Apply P(0) = P₀:
P₀ = H/r + C/r → C = rP₀ − H
Therefore:
P(t) = H/r + (P₀ − H/r)e^(rt)
- InterpretThe equilibrium is P* = H/r. If P₀ = H/r, the population is constant. If P₀ > H/r, the exponential term is positive and P grows without bound. If P₀ < H/r, the exponential term is negative and P collapses to zero. The equilibrium is unstable.
- AnswerThe equilibrium H/r is an unstable threshold. Constant harvesting at rate H creates no naturally stable population level.
P(t) = H/r + (P₀ − H/r)e^(rt). Unstable equilibrium at P* = H/r.
Section 14.6
Checking Units and Initial Conditions
Checking is not optional. A mathematically valid manipulation can produce a wrong answer if you misread the problem, misidentify a parameter, or drop a sign. Checking catches errors before they propagate into interpretation.
Checking the Initial Condition
P(0) = H/r + (P₀ − H/r)e⁰ = H/r + P₀ − H/r = P₀ ✓
Checking by Substitution into the ODE
P'(t) = r(P₀ − H/r)e^(rt)
rP(t) − H = r[H/r + (P₀ − H/r)e^(rt)] − H
= H + r(P₀ − H/r)e^(rt) − H
= r(P₀ − H/r)e^(rt) = P'(t) ✓
Checking Units
In the harvesting equation, every term must carry units of thousands of fish per year. The term rP has units of (per year)(thousands of fish) = thousands of fish per year. The term H has units of thousands of fish per year. All terms match.
Student Warning — Units First
Units are a model's first line of defense. If terms in your differential equation carry different units, the equation is wrong, and no amount of correct algebra will fix it. Always check units when you build the equation, not just at the end.
Section 14.7
Comparing Exact and Numerical Solutions
When an exact solution exists, numerical methods verify it and visualize it at specific points. When an exact solution does not exist, numerical methods are the primary tool. Comparing the two builds intuition about accuracy, step size, and the limitations of numerical approximation.
Example 14.7.1
Euler's method vs. exact solution — harvesting model
- Setupr = 0.2, H = 0.5, P₀ = 3. Exact solution: P(t) = 2.5 + 0.5e^(0.2t). Euler update: P_{n+1} = Pₙ + h(0.2Pₙ − 0.5), h = 1 year.
- Solve
t=0: Exact 3.000 | Euler 3.000 | Error 0.000
t=1: Exact 3.112 | Euler 3.100 | Error 0.012
t=2: Exact 3.246 | Euler 3.220 | Error 0.026
t=3: Exact 3.405 | Euler 3.364 | Error 0.041
t=4: Exact 3.595 | Euler 3.537 | Error 0.058
t=5: Exact 3.859 | Euler 3.744 | Error 0.115
- InterpretEuler's method tracks the exact solution closely at first, but error grows over time. The error at t = 5 is more than eight times the error at t = 1 — characteristic of Euler's method with a large step size. Reducing h or using Runge-Kutta would substantially improve accuracy. Both solutions confirm that the population grows, consistent with starting above the unstable equilibrium P* = 2.5.
Section 14.8
Stability and Long-Term Behavior
Long-term behavior is often the most important question in a dynamic model. Engineers want to know whether a control system stabilizes. Biologists want to know whether a population survives. Epidemiologists want to know whether an epidemic fades. Long-term behavior is determined by equilibria and their stability.
Equilibrium Analysis — Harvesting Model
Set f(P) = rP − H = 0. The unique equilibrium is P* = H/r. Since f'(P) = r > 0, the equilibrium is unstable. The derivative criterion confirms what the solution showed directly.
Phase Line — Harvesting Model
Draw a horizontal axis representing population P. Mark the equilibrium P* = H/r with an open circle (unstable). For P > P*: f(P) > 0, arrow points right (population increases). For P < P*: f(P) < 0, arrow points left (population decreases). Arrows point away from P* on both sides, confirming instability.
Long-Term Behavior in Phase Planes
| Eigenvalue Type | Equilibrium Classification | Long-Term Behavior |
|---|
| Real, both negative | Stable node | All trajectories approach equilibrium |
| Real, both positive | Unstable node | All trajectories flee equilibrium |
| Real, opposite signs | Saddle point | Unstable; some approach, some flee |
| Complex, negative real part | Stable spiral | Oscillate inward to equilibrium |
| Complex, positive real part | Unstable spiral | Oscillate outward from equilibrium |
| Pure imaginary | Center | Closed orbits, neither approach nor flee |
Section 14.9
Model Assumptions and Limitations
No differential equation model is a perfect description of reality. Every model rests on assumptions, some explicit and some implicit. A responsible modeler identifies those assumptions, understands which are critical, and communicates the model's limitations clearly.
| Common Assumption | What It Simplifies | When It May Break Down |
|---|
| Continuous change | Real births, deaths, and harvests are discrete events | Small populations where individual events matter |
| Constant parameters | Growth rates and harvesting rates may vary seasonally or with density | Seasonal fisheries; density-dependent competition |
| Spatial homogeneity | Real populations are distributed across space | Migratory species; spatially structured habitats |
| Independence | Single-equation models ignore predators, food sources, habitat | Complex ecosystems with multi-species interaction |
| Known initial conditions | P₀ may be estimated with measurement uncertainty | Near unstable equilibria, small errors amplify rapidly |
Student Tip — Limitations Show Maturity
Stating model limitations is not a sign of weakness. It is a sign of mathematical maturity. A modeler who understands and communicates the limits of a model is more trustworthy than one who presents results without qualification.
Section 14.10
Sensitivity to Parameters
Sensitivity analysis asks: how much does the model's output change when a parameter changes slightly? Parameters are rarely known exactly — growth rates are estimated from data, physical constants have measurement uncertainty, economic coefficients depend on behavioral assumptions. If a model's conclusions change dramatically with a small parameter change, those conclusions should be treated with caution.
Example 14.10.1
Sensitivity analysis — harvesting model equilibrium
- Equilibrium
P* = H/r
- Sensitivity to H
∂P*/∂H = 1/r
With r = 0.2: increasing H by 1 (one thousand fish per year)
decreases P* by 1/0.2 = 5 thousand fish.
- Sensitivity to r
∂P*/∂r = −H/r²
With H = 0.5, r = 0.2: ∂P*/∂r = −0.5/0.04 = −12.5
A small increase in r substantially shifts P*.
- Relative sensitivity to H
(∂P*/∂H) · (H/P*) = (1/r) · (H / (H/r)) = 1
A 1% increase in H produces a 1% increase in P*.
- InterpretBecause the equilibrium is unstable, a parameter error that shifts P* could place the initial population on the wrong side of the threshold — reversing the prediction from survival to collapse. This is a strong argument for precise parameter estimation and safety margins in fishery management policy.
Sections 14.11 – 14.13
Technology, Communication, and Capstone Problems
Technology-Assisted Modeling · Communicating Models · Six Complete Applied Problems
Section 14.11
Technology-Assisted Modeling
Modern applied mathematics is practiced with technology. Computational tools handle arithmetic, produce visualizations, solve intractable systems, and allow efficient exploration of parameter spaces. Technology does not replace understanding — it multiplies it. A modeler who does not know what Euler's method is cannot interpret an ODE solver's output. A modeler who does not understand eigenvalue analysis cannot read a phase portrait. The techniques in this book provide the conceptual foundation. Technology extends it.
What Technology Can Do
| Tool Type | Examples | What It Supports |
|---|
| Numerical ODE solvers | scipy.integrate.solve_ivp, MATLAB ode45 | Accurate solutions to complex equations without closed forms |
| Slope field/phase plane plotters | Desmos, GeoGebra, Python matplotlib | Qualitative visualization without integration |
| Symbolic computation | Mathematica, Maple, SymPy | Exact solutions, Laplace transforms, factoring |
| Graphing utilities | Graphing calculators, Jupyter notebooks | Rapid visualization; comparing solutions across initial conditions |
| Parameter exploration | Interactive sliders in Desmos or Python | Intuition about sensitivity and bifurcation |
Python Example — Harvesting Model
import numpy as np
from scipy.integrate import solve_ivp
import matplotlib.pyplot as plt
r = 0.2
H = 0.5
P0 = 3.0
def dP_dt(t, P):
return r * P[0] - H
t_span = (0, 10)
t_eval = np.linspace(0, 10, 200)
sol = solve_ivp(dP_dt, t_span, [P0], t_eval=t_eval)
plt.plot(sol.t, sol.y[0], label='Numerical solution')
plt.axhline(y=H/r, color='r', linestyle='--', label='Equilibrium P* = H/r')
plt.xlabel('Time (years)')
plt.ylabel('Population (thousands)')
plt.title('Fish Population with Constant Harvesting')
plt.legend()
plt.show()
What This Plot Shows
The numerical solution rises from P₀ = 3 above the dashed equilibrium line at P* = 2.5, growing as t increases. A second simulation starting at P₀ = 2 (below P*) would show P(t) decreasing toward zero. Both confirm the unstable equilibrium analysis.
Technology Ethics and Responsibility
Never accept a computational result that contradicts your understanding of the model without investigating the discrepancy. The discrepancy either reveals an error in your analysis or a problem with the computation. Technology produces results quickly; quick results can be misread, misinterpreted, or misused.
MGU Library Connection
- MGU Scientific Computing pages include Python notebooks for ODE solution, phase plane plotting, and parameter exploration
- MGU Data Science Foundations pages use numerical ODE methods for dynamic systems in biology, economics, and epidemiology
Section 14.12
Communicating Dynamic Models
A dynamic model that cannot be communicated clearly has limited value. Mathematical results must be translated into language that serves the audience, whether that audience is a fellow mathematician, an engineer, a policy maker, a biologist, or a general reader.
Elements of a Clear Model Communication
- 1State the situation. Describe the physical, biological, economic, or social context before any mathematics.
- 2Define all symbols. Every variable, parameter, function, and constant should be defined explicitly, with units.
- 3Explain the equation. After writing the differential equation, explain it in words. Do not let an equation stand alone without commentary.
- 4Show key steps. Show enough work that a reader can follow the derivation. Skip routine arithmetic; do not skip conceptual steps.
- 5State the result. Give the solution, equilibria, stability classification, or numerical approximation, clearly labeled.
- 6Interpret the result. What happens in the short term? What is the long-term behavior? What do the parameters control?
- 7Acknowledge limitations. State the key assumptions and identify where the model is most likely to break down.
- 8Connect to the original question. Return to the motivating question and answer it directly based on the model's results.
Example 14.12.1
Model communication — fish population with harvesting
- Report
We model a fish population P(t), measured in thousands of fish, subject to natural growth at rate r = 0.2 per year and constant harvesting at rate H thousand fish per year. The model is:
dP/dt = rP − H, P(0) = P₀
This equation says that the net rate of change of the population equals natural growth minus harvest. The unique equilibrium is P* = H/r. Stability analysis shows this equilibrium is unstable: a population starting above P* grows without bound, while one starting below P* declines to zero.
For r = 0.2 and H = 0.5, the equilibrium is P* = 2.5 thousand fish. A fishery starting with more than 2,500 fish under constant harvest will grow; one with fewer will collapse. A constant-harvest policy creates no naturally stable population level. Fishery management should consider proportional harvesting, which creates a stable equilibrium.
This model assumes constant r and H, a spatially uniform and continuously reproducing population, and no predator or environmental interactions. In practice, r and H may vary seasonally, and environmental factors should be incorporated in a more complete model.
Section 14.13
Capstone Modeling Problems
The following problems are complete applied modeling exercises. Each requires the full modeling process: define variables, build the equation, classify, solve or analyze, check, interpret, and communicate. The method is not told to you. That is intentional.
Capstone Problem 1
Drug Clearance
A patient receives a single intravenous dose of a medication. The drug is cleared by the kidneys at a rate proportional to the current concentration. The half-life is 6 hours. The initial concentration is 200 mg/L.
- Define variables and write the differential equation for drug concentration over time.
- Classify the equation and choose a solution method.
- Solve the initial value problem.
- Find the concentration at t = 12 hours and at t = 24 hours.
- Find the time at which the concentration falls below 10 mg/L.
- Check your solution by substitution.
- Interpret the solution in the context of medication dosing. What does the half-life mean for clinical practice?
- Discuss at least two assumptions of this model and their practical limitations.
Capstone Problem 2
A Damped Spring
A mass of 0.5 kg is attached to a spring with constant k = 8 N/m. The damping coefficient is b = 2 N·s/m. The mass starts at rest, displaced 0.3 m from equilibrium.
- Write the second-order initial value problem for the displacement of the mass.
- Find the characteristic equation and its roots. Classify the damping type.
- Solve the initial value problem.
- Describe the long-term behavior of the displacement.
- At what time does the displacement first reach zero? Estimate numerically if necessary.
- Verify your solution by substitution.
- What would happen to the motion if b were reduced to 0? Describe the qualitative change.
Capstone Problem 3
Competing Populations
Two species compete for the same food source. Let x(t) and y(t) be populations (thousands) with the system:
dx/dt = 2x − x² − xy
dy/dt = 3y − 2y² − xy
- Find all equilibrium points of this system.
- Classify the ecological meaning of each equilibrium point.
- Determine the stability of each equilibrium by computing the Jacobian and evaluating its eigenvalues.
- Describe the long-term behavior of the system for initial conditions near each equilibrium.
- Interpret: can the two species coexist? Under what conditions does one drive the other to extinction?
- Identify two major assumptions of this model.
Capstone Problem 4
A Heating System with Discontinuous Input
An office heating system satisfies T' + 0.1T = H(t), where T(t) is interior temperature (°C) and H(t) = 5 for 0 ≤ t < 60 min, H(t) = 0 for t ≥ 60 min. Initial temperature T(0) = 15°C.
- Express H(t) in terms of unit step functions.
- Take the Laplace transform of the differential equation.
- Solve for the Laplace transform of T.
- Invert the transform to find T(t) for all t ≥ 0.
- Find the maximum temperature and when it occurs.
- Find the temperature at t = 120 minutes.
- Interpret the model: what does the heating schedule do to the interior temperature over time?
Capstone Problem 5
Epidemic Modeling — SIR System
A simple SIR model tracks susceptible S(t), infected I(t), and recovered R(t) individuals in a fixed population N:
dS/dt = −βSI
dI/dt = βSI − γI
dR/dt = γI
- Verify that d/dt(S + I + R) = 0, confirming conservation of total population.
- Find the equilibrium points. Interpret each in epidemiological terms.
- The basic reproduction number is ℛ₀ = βN/γ. Explain what ℛ₀ > 1 means for the epidemic.
- Use Euler's method with h = 1 day, β = 0.0003, γ = 0.1, N = 1000, S(0) = 990, I(0) = 10, R(0) = 0, to approximate the system for t = 0, 1, 2, 3, 4, 5 days.
- Describe the qualitative behavior of the epidemic: does I(t) grow or decay in the first few days? What determines this?
- Discuss two assumptions of the SIR model and how they might limit its accuracy.
Capstone Problem 6
Logistic Growth with Proportional Harvesting
Replace constant harvesting H with proportional harvesting hP, giving:
dP/dt = rP(1 − P/K) − hP
- Simplify the equation by factoring out P.
- Find the equilibrium points in terms of r, K, and h.
- Determine the conditions under which a positive equilibrium exists.
- Classify the stability of each equilibrium using the derivative criterion.
- Compare this model to the constant-harvesting model. Which produces a stable positive equilibrium? Explain the biological significance.
- With r = 0.4, K = 500, h = 0.1, find the positive equilibrium and verify its stability.
- Draw the phase line for these numerical values. Describe long-term behavior for P₀ = 100, P₀ = 350, and P₀ = 600.
Sections 14.14 – 14.15
Common Mistakes and Readiness for Advanced Mathematics
Ten Modeling Errors · A Map of What Lies Ahead · A Final Note
Section 14.14
Common Modeling and Interpretation Mistakes
Modeling errors are often more subtle than computational errors. A student can follow every algebraic step correctly and still produce a wrong answer if the model was built incorrectly or interpreted carelessly.
Mistake 1 — Writing the equation before defining variables and units
Writing dP/dt = rP − H with no definition of P, t, r, or H
Correct: Before writing any equation, list every variable, every parameter, and the units of each.
Mistake 2 — Getting the sign wrong on a rate term
Modeling removal by writing +H instead of −H, producing growth instead of loss
Correct: For each term, ask — does this increase or decrease the quantity? Assign the sign accordingly.
Mistake 3 — Confusing equilibrium with the general solution
Reporting P* = H/r as "the solution" to the harvesting model
Correct: An equilibrium is a particular constant solution. The general solution is a family of functions. Analyze stability separately.
Mistake 4 — Not checking the initial condition
Stopping after finding the general solution without verifying y(0) = y₀
Correct: Always substitute t = 0 into your solution and confirm it equals the given initial value.
Mistake 5 — Declaring an unstable equilibrium stable because solutions approach from one side
Classifying P* as stable because the population approaches it from above
Correct: A stable equilibrium requires attraction from both sides. Check f(y) on both sides independently.
Mistake 6 — Interpreting a formula without units or context
Presenting P(t) = 2.5 + 0.5e^(0.2t) without explaining what P, t, or the constants represent
Correct: Always state what the solution means: what is P, what are its units, what does the behavior say about the real system?
Mistake 7 — Confusing parameter and variable
Treating the growth rate r as something that changes over time within the model
Correct: A parameter is fixed within the model. A variable changes as the system evolves. Distinguish these explicitly in setup.
Mistake 8 — Assuming the model is perfectly accurate
Presenting model results as exact descriptions of reality without acknowledging assumptions
Correct: Every model is an approximation. Always acknowledge assumptions and state what aspects of reality the model does not capture.
Mistake 9 — Using a method that does not apply to the equation type
Applying undetermined coefficients to a forcing function not in the standard table
Correct: Classify the equation carefully before choosing a method. Use the decision framework in Section 14.4.
Mistake 10 — Reporting numerical results without assessing accuracy
Running Euler's method with h = 2 for a rapidly oscillating equation and reporting results without noting the step size
Correct: When using numerical methods, state the step size. Compare with smaller step sizes or an exact solution when possible.
Section 14.15
Readiness for Advanced Differential Equations and Applied Mathematics
You have now completed this book. The question that remains is: where do you go from here? Differential equations is not the end of applied mathematics. It is the beginning of a much larger practice.
What This Book Has Given You
| Skill | Chapters |
|---|
| Recognize and classify differential equations | 1–2 |
| Solve first-order equations by separation, integrating factor, and qualitative analysis | 3–5 |
| Solve second-order linear equations with constant coefficients | 6 |
| Analyze mechanical vibrations and oscillations | 7 |
| Solve nonhomogeneous equations by undetermined coefficients and variation of parameters | 8 |
| Analyze systems using matrix methods and phase planes | 9 |
| Apply Laplace transforms to initial value problems and discontinuous inputs | 10 |
| Apply Euler's method, improved Euler, and Runge-Kutta to approximate solutions | 11 |
| Analyze nonlinear equations and systems using phase lines, phase planes, equilibria, and stability | 12 |
| Build, solve, interpret, and communicate complete applied models | 5, 14 |
What Comes Next
Advanced ODE. Power series methods, Sturm-Liouville theory, existence and uniqueness theorems, and bifurcation theory extend the foundation built here into deeper mathematical territory.
Partial differential equations. Fourier series, Fourier transforms, separation of variables in complex geometries, and the method of characteristics bring PDE methods to heat flow, wave motion, diffusion, electromagnetics, and financial mathematics.
Numerical methods. Stiff equations, boundary value problems, finite difference methods, and finite element methods for PDEs extend the numerical ideas from Chapter 11 into a rigorous discipline.
Dynamical systems and control. Invariant manifolds, limit cycles, Lyapunov stability, global bifurcations, feedback control, and transfer functions build on the stability and transform ideas from Chapters 10 and 12.
Mathematical modeling in the sciences. Biology, epidemiology, pharmacokinetics, ecology, engineering, economics, finance, and environmental science each have rich traditions of differential equation modeling that follow directly from what you have learned here.
Data science and scientific computing. Parameter estimation from time-series data, neural ODEs, and state-space models in machine learning are active areas where differential equations meet modern data science.
A Final Note
The Mathematics of Change
Mathematics is cumulative. Separation of variables is integration. Integrating factors use the chain rule in reverse. Characteristic equations use algebra to predict calculus behavior. Laplace transforms are integrals that reorganize the problem. Euler's method is local linear approximation iterated forward. Phase lines are a way of reading the derivative without integrating.
The subject of differential equations is, at its core, the mathematical study of change. That is the most universal subject in quantitative science. Everything that changes over time, everything that interacts, everything that feeds back on itself — can be approached with the tools you have built here. You are ready to go further.
Chapter 14 — Back Matter
Chapter Summary
- Sec 14.1The modeling process has ten steps: understand, define variables, build the rate equation, classify, choose a method, carry out the analysis, check, interpret, discuss assumptions, and communicate. Each step is necessary; none can be skipped.
- Sec 14.2Variables change as the system evolves; parameters are fixed within a given model. Both require explicit names and units. Careful setup makes every later step cleaner and more reliable.
- Sec 14.3Rate equations translate verbal descriptions of change into differential equations. Addition represents input; subtraction represents output or removal. Common translation patterns cover proportional growth, logistic growth, Newton's cooling, mixing, and Newton's second law.
- Sec 14.4Classification drives method selection. First-order equations may be handled by separation, integrating factor, or qualitative analysis. Second-order equations use characteristic equations or complementary and particular solutions. Systems use eigenvalue methods. Discontinuous inputs call for Laplace transforms. When exact solutions are unavailable, numerical methods apply.
- Sec 14.5Interpretation explains what the solution means for the system — its form, parameters, initial conditions, equilibria, units, and long-term behavior. For the harvesting model, the solution reveals an unstable equilibrium and the practical consequence that constant harvesting creates no self-regulating population level.
- Sec 14.6Checking requires substituting the solution back into the equation, verifying the initial condition, confirming consistent units, and assessing whether behavior is contextually reasonable. Checking catches errors before they propagate into interpretation.
- Sec 14.7Comparing exact and numerical solutions reveals where numerical methods are accurate and where error accumulates. Euler's method with large step size produces growing error. Reducing step size or using higher-order methods improves accuracy substantially.
- Sec 14.8Long-term behavior is determined by equilibria and stability. The derivative criterion classifies equilibria of autonomous equations. Phase plane analysis classifies equilibria of systems using eigenvalue type: stable node, unstable node, saddle, stable spiral, unstable spiral, or center.
- Sec 14.9Every model rests on assumptions: continuous change, constant parameters, spatial homogeneity, independence of interactions, and known initial conditions. Identifying and communicating limitations is a mark of mathematical maturity, not weakness.
- Sec 14.10Sensitivity analysis measures how model outputs change with parameter changes, using partial derivatives of equilibria with respect to parameters. Near unstable equilibria, parameter uncertainty can reverse predictions from survival to collapse, motivating precise estimation and safety margins.
- Sec 14.11Technology extends the analytical foundation into larger and more complex problems. Numerical ODE solvers, phase plane plotters, symbolic computation systems, and interactive parameter exploration tools all require analytical understanding to use correctly and interpret responsibly.
- Sec 14.12A well-communicated model states the situation, defines all symbols, explains the equation in words, shows key steps, states the result clearly, interprets it in context, acknowledges limitations, and returns to the original question with a direct answer.
- Sec 14.13Six capstone problems require complete modeling exercises: drug clearance (first-order decay), a damped spring (second-order), competing populations (nonlinear system), a heated building with discontinuous input (Laplace), an SIR epidemic (system with Euler approximation), and logistic harvesting (stability comparison).
- Sec 14.14Ten recurring modeling errors: equations without defined variables; wrong signs; equilibrium mistaken for general solution; unchecked initial conditions; one-sided stability conclusions; uninterpreted formulas; parameters treated as variables; unacknowledged assumptions; misapplied methods; and numerical results reported without accuracy assessment.
- Sec 14.15This book provides functional fluency in the language of dynamic systems. Advanced courses in ODE, PDE, numerical methods, dynamical systems, control theory, and mathematical modeling in the sciences build directly on this foundation.
Key Terms Review
modeling process
The structured ten-step sequence from problem understanding through communication of results.
variable
A quantity that changes as the system evolves; typically a function of the independent variable, usually time.
parameter
A quantity that remains fixed within a given model, appearing as a constant in the equation.
rate equation
The differential equation expressing how the dependent variable changes; the core product of the translation step.
classification
Identifying the type of differential equation — order, linearity, autonomy, separability — to determine available solution methods.
method selection
Choosing the most appropriate technique based on equation classification; a skill in its own right.
exact solution
A closed-form formula for the solution function, valid for all t in the domain.
numerical solution
An approximation computed by iterative methods such as Euler, improved Euler, or Runge-Kutta.
qualitative analysis
Analysis of system behavior — equilibria, stability, long-term trends — without finding a closed-form formula.
interpretation
Explaining what the mathematical result means for the real system in terms of variables, parameters, units, and context.
assumption
A simplification made in building the model that may not hold exactly in the real situation.
limitation
A constraint on the model's accuracy or applicability, arising from assumptions that break down in certain conditions.
sensitivity analysis
The study of how model outputs change when parameters change, using partial derivatives or numerical experimentation.
equilibrium
A constant solution of an autonomous equation, found by setting f(y) = 0 and solving.
stability
The behavior of solutions starting near an equilibrium; stable if they return, unstable if they depart.
long-term behavior
The behavior of the solution as t → ∞; determined by equilibria, stability, and eigenvalue analysis.
technology-assisted modeling
Using computational tools to extend analytical understanding; requires conceptual foundation to use correctly.
model communication
The clear written explanation of a model: context, symbols, equation, analysis, interpretation, limitations, and conclusion.
capstone problem
A complete applied modeling exercise requiring all steps of the process, with method selection left to the student.
basic reproduction number
In the SIR model, ℛ₀ = βN/γ; determines whether an epidemic grows (ℛ₀ > 1) or decays (ℛ₀ < 1).
Model Selection Practice
- A voltage across a capacitor satisfies a first-order linear equation driven by a sinusoidal source. What method would you use? Explain your classification.
- A population follows dP/dt = P(4 − P)(P − 1). What method is most appropriate for understanding long-term behavior? Identify the equilibria without solving.
- A spring-mass system is driven by F(t) = 3e^(−t)cos(2t). What form of particular solution would you guess? What method would you use?
- A system has two nonlinearly interacting chemical concentrations. Exact solution is unavailable. What approach would you take to understand long-term behavior?
- An initial value problem has a piecewise constant input that switches on at t = 5. Which transform method handles this most naturally?
- A function satisfies y'' + 4y = 0 with y(0) = 2, y'(0) = −1. What are the characteristic roots? What form does the general solution take?
- A system x' = Ax has eigenvalues −2 ± 3i with negative real part. Describe the qualitative behavior of solutions near the origin.
- A differential equation has no closed-form solution and you need values at t = 0, 0.1, 0.2, …, 1.0. What numerical method would you use? What step size issues would you consider?
Solution Method Review
- Separable. Solve dy/dx = x²/y² with y(0) = 2.
- Integrating factor. Solve y' + 3y = 6e^(−t) with y(0) = 1.
- Characteristic equation. Find the general solution of y'' − 5y' + 6y = 0.
- Repeated roots. Find the general solution of y'' − 4y' + 4y = 0.
- Complex roots. Find the general solution of y'' + 2y' + 5y = 0.
- Undetermined coefficients. Find a particular solution of y'' + y = 3cos(2t).
- Laplace transform. Solve y'' + 3y' + 2y = e^(−t), y(0) = 0, y'(0) = 1 using Laplace transforms.
- Euler's method. Apply Euler's method with h = 0.25 to y' = y − t, y(0) = 2, for t = 0, 0.25, 0.5, 0.75, 1.
- Phase line. Draw the phase line for dP/dt = P(2 − P) and classify each equilibrium.
- System eigenvalues. Find the eigenvalues of A = [−1, 2; 0, −3] and classify the equilibrium at the origin.
Interpretation Problems
- The solution to a population model is P(t) = 500/(1 + 4e^(−0.6t)). Identify the carrying capacity, growth rate, and initial population. Describe the long-term behavior and the shape of the solution curve.
- A damped oscillation is described by x(t) = e^(−0.5t)(3cos(2t) − sin(2t)). What is the damping coefficient? What is the natural frequency? Does the system oscillate? Does it eventually come to rest?
- The phase plane for a two-dimensional system shows a stable spiral around the equilibrium (2, 1). Describe the long-term behavior of all nearby trajectories. What does a stable spiral mean physically for a mechanical system?
- A model predicts C(t) = 100e^(−0.12t) mg/L for drug concentration. What is the half-life? When does concentration fall to 5 mg/L? Is this model appropriate for the first 30 minutes after administration?
- The sensitivity of an equilibrium population to growth rate r is ∂P*/∂r = −150. If r is overestimated by 0.01, how does the predicted equilibrium change? What does this say about the importance of accurately estimating r?
Technology Problems
- Use a graphing utility to plot slope fields for y' = y(2 − y) on 0 ≤ t ≤ 4, 0 ≤ y ≤ 3. Sketch solution curves for y(0) = 0.1, y(0) = 1, and y(0) = 2.5.
- Use a numerical ODE solver to solve y' + y = sin(t), y(0) = 0, for 0 ≤ t ≤ 20. Plot the solution and describe its long-term behavior. Compare with the exact solution.
- Use numerical simulation to solve the SIR system from Capstone Problem 5 with β = 0.0003, γ = 0.1, N = 1000, S(0) = 990, I(0) = 10, R(0) = 0, for t = 0 to t = 100 days. Plot all three populations on the same axes.
- Use a CAS or numerical tool to find the Laplace transform of f(t) = t²e^(−3t). Verify using the transform table.
- Use a spreadsheet or Python to apply the improved Euler method to y' = y − t² + 1, y(0) = 0.5, with h = 0.1, for t = 0 to t = 2. Compare with the exact solution y(t) = (t+1)² − 0.5e^t.
- Use a graphing tool to plot the phase portrait for x' = x − xy, y' = xy − y (Lotka-Volterra predator-prey). Describe the shape of the orbits near the nontrivial equilibrium.
Error Analysis
Identify the error in each of the following modeling or solution steps and explain how to correct it.
Problem 1
A student models a growing population with harvesting by writing: dP/dt = H − rP
What is wrong with this equation? What should it be?
Problem 2
A student solves y' + 2y = 4 and writes the general solution as y = 2.
What has the student confused? What is the correct general solution?
Problem 3
A student classifies the equilibrium y* = 3 of y' = (3−y)² as stable, because solutions approach it from the left.
What is wrong with this classification?
Problem 4
For y'' − 4y' + 4y = 0, characteristic equation gives (r−2)² = 0, so the student writes y = Ce^(2t).
What step was omitted? What is the correct general solution?
Problem 5
A mixing problem: 100-L tank, clean water in at 5 L/min, mixed solution out at 5 L/min. Student writes dA/dt = −5A.
What error did the student make? What is the correct equation?
Problem 6
A student uses Euler's method with h = 2 for a rapidly oscillating equation and reports the results as accurate.
What problem might arise? What should the student have done?
Problem 7
A student interprets T(t) = 20 + 60e^(−0.1t) by saying: "The temperature approaches 0 as t → ∞."
What error has the student made? What is the correct limiting behavior?
Chapter 14 Checkpoint
Chapter 14 Checkpoint
Complete without assistance — assesses the full modeling process, method selection, interpretation, sensitivity, and communication
Section 1 — The Modeling Process
- State the ten steps of the differential equation modeling process in order.
- Define variable, parameter, and rate equation. Give an example of each from any model in this chapter.
- Explain the difference between an exact solution and a qualitative analysis. When is each most appropriate?
Section 2 — Method Selection
- For each equation, classify it and name the primary solution method:
(a) y' − 5y = e^(2t) (b) y'' + 9y = 0 (c) y' = y(3 − y) (d) y'' + 4y' + 13y = sin(t)
- A system has coefficient matrix with eigenvalues −2 ± 3i. What is the equilibrium classification? Describe the long-term behavior of nearby solutions.
Section 3 — Interpretation
- The solution to a model is
P(t) = 100 − 80e^(−0.5t). What is P(0)? What is the limit as t → ∞? What differential equation does this function satisfy? Describe the qualitative behavior.
- The equilibrium of a model is y* = a/b where a, b > 0. Compute ∂y*/∂a and ∂y*/∂b. Interpret each sensitivity in words. If a is overestimated by 10% and b is underestimated by 10%, in what direction does the predicted equilibrium shift?
Section 4 — Numerical Comparison
- Apply Euler's method with h = 0.5 to
y' = 2 − y, y(0) = 0, for t = 0, 0.5, 1.0, 1.5, 2.0. Find the exact solution and compare. What is the error at t = 2?
Section 5 — Communication
- Write a brief model communication (five to eight sentences) for the following: A cup of coffee cools in a room at 20°C. At t = 0 the temperature is 90°C, and after 10 minutes it is 70°C. You model the cooling with Newton's law. Include the equation, the solution, the determination of the cooling constant from data, and the interpretation.
Section 6 — Assumptions and Limitations
- Name three assumptions of a constant-coefficient second-order mechanical vibration model. For each assumption, describe one real situation where it would break down.
Capstone Project Guide
A capstone project for this course asks you to choose a real applied situation, build a differential equation model from scratch, carry out a complete analysis, and write a clear report.
Project Requirements
- 1Choose a situation. Any area involving quantities that change over time with a describable rate of change: biology, pharmacology, ecology, engineering, economics, physics, environmental science, epidemiology.
- 2Define variables and parameters with units. List everything before writing the equation.
- 3Build the differential equation. Translate the verbal description into mathematical language. Explain the equation in words.
- 4Classify and choose a method. Explain your classification and your choice.
- 5Solve or analyze. Carry out the complete solution or qualitative analysis.
- 6Check. Verify by substitution or comparison with numerical output.
- 7Interpret. Explain equilibria, stability, long-term behavior, and parameter influence.
- 8Sensitivity analysis. Choose one or two key parameters and analyze how the equilibrium or solution changes when they change.
- 9Limitations. Identify at least two assumptions and discuss how they limit accuracy.
- 10Communicate. Write a clear report following the structure in Section 14.12. A checklist rubric appears in Appendix R.
Possible Project Topics
Suggested Starting Points
- Medication dosing: how do repeated doses build toward a steady-state concentration?
- Epidemic spread: how does vaccination affect ℛ₀ and the epidemic's peak?
- Structural vibration: how does damping affect beam response to a periodic force?
- Logistic growth with seasonal harvesting: what harvest rates are sustainable?
- RC or RLC circuits: how does the circuit respond to a step voltage or sinusoidal source?
- Chemical reaction kinetics: how do reactant concentrations evolve in a first-order reaction?
- Simple energy balance climate model: how does a feedback parameter affect temperature over time?
- Predator-prey dynamics: under what conditions do populations coexist near the coexistence equilibrium?
Bridge Note — Advanced Applied Mathematics
Where This Foundation Leads
Partial differential equations take the single independent variable of this course and replace it with two or more, bringing heat flow, wave motion, diffusion, electrostatics, and fluid flow within reach. Numerical methods courses formalize the Euler and Runge-Kutta ideas into rigorous error analysis and finite element methods. Dynamical systems courses extend Chapter 12 into chaos theory, strange attractors, and complex system behavior. Control systems use the Laplace framework from Chapter 10 to design feedback mechanisms. Mathematical biology, epidemiology, and environmental science extend the population and mixing models of Chapters 5, 9, and 14 into high-dimensional systems with spatial structure and stochasticity. Financial mathematics applies PDEs to option pricing and interest rate modeling. Data science uses these ideas in neural ODEs, state-space models, and parameter estimation from time-series data.
The mathematics of change is the mathematics of the real world. You have learned its language. Now go use it.
MGU Library Connection
- Calculus II and Calculus III — extension of integration and multivariable methods underlying ODE techniques
- Numerical Methods — formal treatment of ODE solvers, stability analysis, finite difference methods, and PDEs
- Probability and Statistics — connections to stochastic differential equations and parameter estimation
- Data Science Foundations — neural ODEs, state-space models, and dynamic system inference from data
- Physics, Engineering, Biology, Economics, and Environmental Science pages — discipline-specific applications of every technique in this book
- Appendix R of this textbook — Capstone Project Rubric
- Appendix S of this textbook — Advanced Applied Mathematics Readiness Check