Differential equations are the mathematics of systems defined by how they change. Instead of stating a quantity directly, a differential equation states a rule linking the quantity to one or more of its derivatives.
This makes differential equations ideal for physics, engineering, biology, economics and any field where the present rate of change helps determine the future state. Motion, heat, populations, circuits, chemical reactions, epidemics and control systems all become differential-equation problems once their dynamics are written mathematically.
Series route: Mathematics Learning Hub → How Mathematics Works → Differential Equations.
1. What a differential equation is
A differential equation relates an unknown function to its derivatives. A simple example is dy/dt = ky, which says the rate of change of y is proportional to y itself.
The equation does not give one number. It describes an entire family of possible trajectories.
2. Why rate laws are natural
Many real laws are local. Newton’s second law relates force to acceleration. Radioactive decay relates current amount to decay rate. Heat flow relates temperature differences to rate of transfer.
These laws describe what the system is doing now. Differential equations propagate those local rules through time to produce global behaviour.
3. Order tracks how many derivatives appear
The order of a differential equation is the highest derivative appearing in it. A first-order equation involves a first derivative. A second-order equation may involve acceleration or curvature through a second derivative.
Higher order usually means the system needs more initial information to determine a unique trajectory.
4. Initial conditions select one path
The equation dy/dt = ky has infinitely many solutions of the form y = Ce^{kt}. The initial condition y(0)=y₀ determines C and selects the particular trajectory y=y₀e^{kt}.
This reveals a fundamental architecture: law + state = evolution.
5. Boundary conditions constrain solutions at edges
Some problems specify information at spatial or temporal boundaries rather than at one initial time. A beam may be fixed at both ends. A temperature may be specified on the boundary of a region.
Boundary-value problems often behave differently from initial-value problems because the solution must satisfy conditions across a region rather than simply evolve from a starting state.
6. Separable equations isolate variables
If a first-order equation can be written in the form dy/dx = g(x)h(y), then the variables may be separated and integrated under suitable conditions.
This method works because the equation can be rearranged into one side depending only on y and the other only on x.
7. Exponential growth and decay are the canonical example
The equation dy/dt = ky says change is proportional to current amount. Positive k produces exponential growth; negative k produces exponential decay.
The same structure appears in populations, compound growth, radioactive decay and simple continuous-time models of adoption or depletion.
8. First-order linear equations have a general structure
An equation of the form y′ + p(x)y = q(x) can be solved using an integrating factor. The method transforms the left-hand side into the derivative of a product.
The key move is representation: rewrite the differential relationship so ordinary integration becomes possible.
9. Second-order equations model inertia and oscillation
Mechanical systems often involve acceleration, producing second-order equations. A mass-spring system, for example, can be modelled by m x″ + c x′ + kx = F(t).
Mass stores inertia, damping removes energy, stiffness restores displacement, and forcing injects external input. One equation compresses the entire mechanism.
10. Homogeneous and forced behaviour are different
The homogeneous equation describes the system’s natural modes. A forcing term describes external input.
Solutions often decompose into natural response plus forced response. This distinction is central in circuits, vibrations and control engineering.
11. Characteristic equations convert differential problems into algebra
For constant-coefficient linear differential equations, assuming exponential trial solutions converts differentiation into multiplication by powers of a parameter.
The resulting characteristic polynomial can reveal growth, decay and oscillation through its roots.
12. Complex roots encode oscillation
Complex conjugate characteristic roots produce sine and cosine behaviour multiplied by exponential envelopes. This is why complex numbers naturally enter real physical oscillations.
The complex representation compresses rotational behaviour that later reappears as real oscillation.
13. Systems of differential equations handle interacting states
Real systems usually have multiple variables. Predator and prey populations interact. Electrical currents and voltages interact. Position and velocity evolve together.
A system can often be written compactly as x′ = Ax or x′ = f(x,t).
14. Linear algebra reveals system modes
For x′=Ax, eigenvalues and eigenvectors of A often determine the basic modes of evolution. Eigenvalues with negative real parts tend to produce decay; positive real parts produce growth; imaginary parts produce oscillation.
This is a major bridge between differential equations and linear algebra.
15. Phase space shows the system instead of one variable
A dynamical system can be visualised in phase space, where each point represents a complete state. A trajectory then shows how the whole state evolves over time.
This perspective is especially powerful when explicit formulas are unavailable.
16. Equilibria are states where change stops
An equilibrium satisfies f(x)=0 in an autonomous system x′=f(x). At such a state, the system remains stationary unless disturbed.
The next question is stability: if the system is nudged slightly, does it return, remain nearby or move away?
17. Stability is more important than exact formulas in many systems
Engineers often care less about a closed-form solution than whether disturbances decay. Ecologists may care whether populations remain bounded. Control systems must keep outputs near desired states.
Qualitative analysis can therefore be more valuable than symbolic solving.
18. Nonlinear equations change the landscape
Nonlinear differential equations can produce multiple equilibria, bifurcations, limit cycles and chaos. Superposition no longer applies.
The failure of linearity means local behaviour and global behaviour can diverge dramatically.
19. Chaos is deterministic but sensitive
A deterministic system can still become practically unpredictable when nearby initial conditions diverge rapidly. This is sensitive dependence on initial conditions.
Chaos does not mean randomness. It means deterministic rules can amplify tiny uncertainty until long-range prediction becomes unreliable.
20. Partial differential equations add multiple independent variables
When a quantity depends on several variables such as space and time, partial differential equations arise. Temperature T(x,t), wave displacement u(x,t) and fluid velocity fields are examples.
PDEs describe how local spatial structure and temporal change interact.
21. Diffusion smooths gradients
The heat equation models diffusion: regions of high concentration or temperature tend to spread into neighbouring lower regions.
The equation formalises a local flow law and predicts global smoothing over time.
22. Wave equations preserve propagation
Wave equations model disturbances that propagate through space while oscillating. Sound, strings and electromagnetic waves all generate PDE structures of this kind under different physical assumptions.
23. Numerical methods are essential
Most realistic differential equations do not have convenient closed-form solutions. Numerical methods approximate trajectories at discrete steps.
Euler’s method uses the current slope to step forward. More advanced Runge–Kutta and multistep methods improve accuracy and stability.
24. Step size creates an accuracy–cost trade-off
Smaller steps often improve local accuracy but increase computational work. Some equations are stiff, meaning naive step sizes can create unstable numerical behaviour even when the true solution is well behaved.
Numerical differential equations therefore require error control, not just repeated arithmetic.
25. A worked mechanism: cooling
Newton’s law of cooling can be modelled as dT/dt = −k(T−Tₐ), where Tₐ is ambient temperature.
- The driving quantity is the temperature difference T−Tₐ.
- The negative sign says the difference shrinks.
- The solution is T(t)=Tₐ+(T₀−Tₐ)e^{−kt}.
- The initial condition T(0)=T₀ selects the trajectory.
- As t grows, the exponential term decays and T approaches Tₐ.
The equation translates a local rate law into a global temperature history.
26. Common failure modes
- Equation without meaning: solving symbolically without identifying what each variable represents.
- Initial-condition loss: forgetting the information needed to determine constants.
- Units drift: ignoring dimensional consistency in rate laws.
- Linear-method overreach: using superposition in nonlinear systems.
- Exact-solution bias: treating numerical or qualitative analysis as inferior when no closed form exists.
- Stability blindness: computing a trajectory without asking whether small disturbances grow or decay.
27. Differential equations and calculus
Calculus defines derivatives and integrals; differential equations use them as constraints on unknown functions. Solving often requires integration, differentiation and limit reasoning.
28. Differential equations and linear algebra
Systems of linear differential equations become matrix problems. Eigenstructure reveals natural modes and stability. Numerical solvers also rely heavily on linear algebra at each step.
29. Differential equations and probability
Stochastic differential equations add random forcing to continuous dynamics. Probability distributions themselves may evolve according to differential equations such as Fokker–Planck equations.
30. Differential equations and modelling
A differential equation is only as useful as its modelling assumptions. Which rates matter? Which variables are omitted? Are parameters constant? Is the process really continuous?
Good modelling separates mathematical correctness from model adequacy.
31. Differential equations as a machine
A useful machine model is:
State Variables → Rate Law → Differential Equation → Initial/Boundary Conditions → Analytical/Numerical Solution → Stability and Model Check → Interpretation.
32. What mastery looks like
- translate mechanisms into rate laws;
- distinguish initial-value and boundary-value problems;
- recognise separable, linear and common second-order forms;
- interpret equilibria and stability;
- use numerical methods with error awareness;
- connect matrix eigenstructure to system modes;
- separate deterministic chaos from randomness;
- return every solution to its physical or conceptual meaning.
33. Conclusion
Differential equations work by turning local laws of change into global trajectories. Initial conditions identify the starting state. Boundary conditions constrain allowable solutions. Linear algebra reveals system modes. Numerical methods approximate systems too complicated for symbolic solutions. Stability analysis determines whether disturbances disappear or grow.
Calculus measures change. Differential equations make change into a law of motion. That is why they sit at the heart of mathematical modelling.
How Mathematics Works | Batch 03
- How Mathematics Works | Discrete Mathematics
- How Mathematics Works | Linear Algebra
- Differential Equations — this article
- How Mathematics Works | Real Analysis
Return to the Mathematics Learning Hub or the How Mathematics Works root.