Mathematical modeling with ODEs

Mathematical modeling with ODEs is the systematic process of translating a physical, biological, or engineering system into a differential equation. Once you have the equation, you can analyze it (find equilibria, study stability) and solve it (analytically or numerically) to predict the system’s behavior.

The general workflow:

1. Choose dependent and independent variables, set a frame of reference.
2. Choose convenient units of measurement.
3. Identify the underlying principle governing the system: Newton’s laws, conservation, rate laws, etc.
4. Express that principle as an equation in your chosen variables.
5. Solve by integrating both sides (or applying a solution method).
6. Apply side conditions (initial values, boundary conditions) to pin down constants.

The modeling step (3–4) is often the hardest. The mathematics (5–6) is mechanical once the equation exists.

Worked example: falling object

Model a falling object’s position over time.

Step 1: Variables. Let $h(t)$ = height above ground at time $t$. Independent variable: $t$. Dependent: $h$.

Step 2: Units. SI: meters, seconds.

Step 3: Principle. Newton’s second law: $F=ma$. The only force (ignoring air resistance) is gravity, $F=-mg$ (negative because we measure height upward). The acceleration is $a=d^{2}h/d{t}^2$.

Step 4: Equation:

$m\frac{d^{2}h}{d{t}^2}=-mg\Rightarrow \frac{d^{2}h}{d{t}^2}=-g$

Step 5: Integrate twice.

$\frac{dh}{dt}=-gt+c_1,h(t)=-\frac{1}{2}g{t}^2+c_{1}t+c_2$

Step 6: Apply initial conditions. Suppose $h(0)=1$ (m), $v(0)=0$ (released from rest). Then $c_1=0$, $c_2=1$:

$h(t)=-\frac{1}{2}g{t}^2+1$

The object falls. At time $t=\sqrt{2/g}\approx 0.45$ s, $h=0$ (reaches the ground).

Common modeling principles

Each domain has its standard principles:

• Mechanics: $F=ma$ (Newton). For damped oscillators: $m\ddot{x}+\gamma \dot{x}+kx=F(t)$.
• Heat transfer: Newton’s law of cooling, $\frac{dT}{dt}=-k(T-T_{\text{env}})$.
• Population dynamics: $\frac{dp}{dt}$ = (birth rate − death rate) × $p$. See Malthusian model and Logistic model.
• Chemical kinetics: rate of reaction proportional to reactant concentrations. First-order: $\frac{d[A]}{dt}=-k[A]$.
• Electric circuits: Kirchhoff’s voltage law, $L\ddot{q}+R\dot{q}+q/C=V(t)$.
• Compound interest (continuous): $\frac{dM}{dt}=rM$, exponential growth.
• Radioactive decay: $\frac{dN}{dt}=-\lambda N$, exponential decay.

Choosing the right model

Modeling involves trade-offs:

• Simpler models are easier to solve and interpret but capture less detail. The simplest population model (Malthusian model) ignores environmental limits.
• More complex models capture more reality but are harder to solve. The Logistic model adds carrying capacity.
• Even more complex models add competition, predation, age structure, spatial variation — eventually requiring PDEs or stochastic equations.

Start simple. Add complexity only when the simple model fails to match observations.

What a model gives you

Once you have an ODE describing the system:

• Predict the future by solving the ODE.
• Find equilibria by setting derivatives to zero — these are the long-term states.
• Analyze stability — see Stability of autonomous systems — to know whether equilibria are physical or sensitive.
• Compute response to inputs via Laplace transform, Convolution integral, etc.
• Identify parameters by fitting model output to data.

Limitations

ODE models are deterministic — same initial conditions always give same future. Real systems often involve randomness (use stochastic differential equations) or spatial variation (use PDEs). They also assume continuity — population sizes, for example, are integers but ODEs treat them as real numbers, which is a fine approximation only when populations are large.

For real engineering practice, ODE models are the starting point — useful for initial analysis, then refined or replaced when accuracy demands it.