Differential equations are where IB Mathematics: Applications and Interpretation HL turns calculus into modelling. A differential equation describes how fast something changes, and solving it tells you what happens over time: a cup of tea cooling, two populations rising and falling, a spring settling down. Calculus has 41 suggested teaching hours at AI HL (19 at AI SL), and differential equations are one of the main HL-only calculus topics.
AI HL treats the topic differently from AA HL. AA HL focuses on analytical techniques such as the integrating factor and homogeneous equations; AI HL focuses on modelling, numerical methods and systems of equations, with a GDC on every paper. If you are on the AA course, read our AA HL differential equations guide instead. If you want support with the AI version, our IB Maths AI HL tutoring covers each method below.
What the AI HL topic covers
- Setting up a differential equation from a description, for example "the rate of change is proportional to the amount present".
- Separation of variables for equations of the form dy/dx = f(x) g(y).
- Slope fields: sketching and interpreting them.
- Euler's method for first-order equations and for coupled systems.
- Coupled linear systems dx/dt = ax + by, dy/dt = cx + dy, solved with eigenvalues and eigenvectors.
- Phase portraits: equilibrium points, stability, saddle points and spirals.
- Second-order equations, rewritten as a coupled system and solved numerically.
The eigenvalue work builds directly on the matrices topic, so revise our AI HL matrices and Markov chains guide first if eigenvectors feel shaky.
Setting up and separating variables
Most modelling questions start with a sentence. "The temperature falls at a rate proportional to the difference between it and the room temperature" becomes dT/dt = −k(T − T_room), with k > 0. The minus sign says "falling"; the bracket says "difference".
To separate variables, move every y (or T) term to one side and every x (or t) term to the other, integrate both sides, and add one constant. Then use the given conditions to find the constants.
Worked example 1: a cooling drink
A drink at 90 °C is left in a room at 22 °C. Its temperature T after t minutes satisfies dT/dt = −k(T − 22). After 10 minutes it is at 60 °C. Find k, and find when the drink reaches 30 °C.
- Separate: ∫ 1/(T − 22) dT = ∫ −k dt.
- Integrate: ln|T − 22| = −kt + c.
- Rearrange: T − 22 = A e^(−kt), so T = 22 + A e^(−kt).
- Use T = 90 when t = 0: A = 68, so T = 22 + 68e^(−kt).
- Use T = 60 when t = 10: 38 = 68e^(−10k), so k = (1/10) ln(68/38) = 0.0582 (3 s.f.).
- Set T = 30: 8 = 68e^(−kt), so t = ln(68/8) ÷ k = 36.8 minutes.
Keep k unrounded in your GDC for step 6. Using 0.0582 gives 36.77, which still rounds to 36.8 here, but early rounding often costs the accuracy mark.
Slope fields
A slope field draws a short line segment at each grid point with gradient dy/dx. A solution curve follows the segments, never crossing them. Two quick checks help:
- Where dy/dx = 0 (an isocline of zero slope), solution curves have horizontal tangents.
- If a line has the same gradient everywhere it crosses, it may itself be a solution. For dy/dx = x − y, every point on y = x − 1 has slope x − (x − 1) = 1, which equals that line's gradient, so y = x − 1 is a solution curve.
Euler's method, including coupled systems
Euler's method steps forward using the current gradient: y_(n+1) = y_n + h × f(x_n, y_n), with x_(n+1) = x_n + h. For a coupled system, update both variables at once using the old values of each:
x_(n+1) = x_n + h × (dx/dt at step n), y_(n+1) = y_n + h × (dy/dt at step n).
Worked example 2: predators and prey
A model for a prey population x and a predator population y, with t in years, is:
dx/dt = 0.4x − 0.01xy, dy/dt = −0.3y + 0.005xy, with x = 100 and y = 20 when t = 0.
Use Euler's method with h = 0.1 to estimate x and y when t = 0.2.
| Step | t | x | y | dx/dt | dy/dt |
|---|---|---|---|---|---|
| 0 | 0 | 100 | 20 | 40 − 20 = 20 | −6 + 10 = 4 |
| 1 | 0.1 | 102 | 20.4 | 40.8 − 20.808 = 19.992 | −6.12 + 10.404 = 4.284 |
| 2 | 0.2 | 103.9992 | 20.8284 |
So x ≈ 104 and y ≈ 20.8 at t = 0.2.
The non-zero equilibrium is where both rates are zero: 0.4 − 0.01y = 0 gives y = 40, and −0.3 + 0.005x = 0 gives x = 60. So (60, 40) is the point where both populations would stay constant. Use the GDC's table or a spreadsheet-style program for longer runs, but write out the first step in full so the examiner can see your method.
Coupled linear systems and eigenvalues
For dx/dt = ax + by, dy/dt = cx + dy, write the matrix M = (a b; c d). If M has real, distinct eigenvalues λ1 and λ2 with eigenvectors p1 and p2, the solution is:
(x, y) = A e^(λ1 t) p1 + B e^(λ2 t) p2,
with A and B fixed by the initial values. The eigenvalues also tell you the long-term behaviour without solving anything:
| Eigenvalues | Equilibrium at the origin | Phase portrait |
|---|---|---|
| Both real and negative | Stable | Trajectories move into the origin |
| Both real and positive | Unstable | Trajectories move away from the origin |
| Real, opposite signs | Saddle point (unstable) | In along one eigenvector, out along the other |
| Complex, negative real part | Stable | Spirals inwards |
| Complex, positive real part | Unstable | Spirals outwards |
| Purely imaginary | Neither (a centre) | Closed loops (ellipses) around the origin |
Worked example 3: a saddle point
Solve dx/dt = x + 2y, dy/dt = 2x + y, with x = 3 and y = 1 at t = 0, and describe the phase portrait.
- Matrix M = (1 2; 2 1). Characteristic equation: (1 − λ)² − 4 = 0, so 1 − λ = ±2, giving λ = 3 or λ = −1.
- For λ = 3: −2x + 2y = 0, so an eigenvector is (1, 1). For λ = −1: 2x + 2y = 0, so an eigenvector is (1, −1).
- General solution: (x, y) = A e^(3t) (1, 1) + B e^(−t) (1, −1).
- At t = 0: A + B = 3 and A − B = 1, so A = 2 and B = 1.
- Solution: x = 2e^(3t) + e^(−t), y = 2e^(3t) − e^(−t).
Check by differentiating: dx/dt = 6e^(3t) − e^(−t), and x + 2y = 2e^(3t) + e^(−t) + 4e^(3t) − 2e^(−t) = 6e^(3t) − e^(−t). It matches.
One eigenvalue is positive and one negative, so the origin is a saddle point. Trajectories come in along the line y = −x (the λ = −1 direction) and leave along y = x (the λ = 3 direction). For this starting point, the e^(3t) terms dominate, so in the long run x and y both grow and y/x approaches 1. To draw arrows correctly, test one point: at (1, 0), dx/dt = 1 and dy/dt = 2, so the trajectory there moves up and to the right.
Second-order equations as coupled systems
A second-order equation such as d²x/dt² + 0.5 dx/dt + 4x = 0 (a damped spring) is handled by setting y = dx/dt. That gives the coupled system dx/dt = y, dy/dt = −4x − 0.5y, which Euler's method can step through.
Worked example 4: a damped spring
With x = 1 and dx/dt = 0 at t = 0, use Euler's method with h = 0.1 to estimate x at t = 0.2.
- Step 1: x1 = 1 + 0.1 × 0 = 1; y1 = 0 + 0.1 × (−4 × 1 − 0.5 × 0) = −0.4.
- Step 2: x2 = 1 + 0.1 × (−0.4) = 0.96; y2 = −0.4 + 0.1 × (−4 × 1 − 0.5 × (−0.4)) = −0.4 − 0.38 = −0.78.
So x ≈ 0.96 at t = 0.2, and the spring is moving back towards its rest position with velocity about −0.78.
Common mistakes
- Forgetting the constant when separating variables, or adding one to each side and never combining them.
- Rounding k or an eigenvalue too early. Store values in the GDC.
- Updating y with the new x in a coupled Euler step. Each new value must come from the old pair.
- Eigenvectors the wrong way round in the general solution. Pair each e^(λt) with its own eigenvector.
- Arrows missing or reversed on a phase portrait. Test one point to find the direction.
- Calling a saddle point stable. Only trajectories exactly on the incoming eigenvector reach it.
- Forgetting to interpret. AI questions often end with "what does this mean for the populations?" A sentence in context earns that mark.
Exam technique: Papers 1, 2 and 3
Every AI HL paper requires a GDC: Papers 1 and 2 are 120 minutes and 110 marks each, and Paper 3 is 60 minutes and 55 marks with two extended problem-solving questions. Use the GDC to solve the characteristic equation, run Euler iterations and check a solution by graphing, but show the set-up by hand. Our GDC skills guide explains how to build a quick iteration.
Differential equations suit Paper 3's style: a model is introduced, solved numerically, compared with an exact method, then interpreted. Our HL Paper 3 guide shows how to manage those long questions. Logistic growth is a natural partner topic, covered in our post on logistic models and linearising data.
Course change note: in the new courses (first exams May 2029), Euler's method leaves AA HL. Students sitting exams in 2027 and 2028 take the current course, in which Euler's method is part of both AA HL and AI HL.
How Maths Bodhi helps
Our tutors teach AI HL differential equations from the model outwards: first turning words into an equation, then choosing between separation, Euler's method and eigenvalues, then explaining what the answer means. Sessions include GDC practice on the student's own calculator and written feedback on set-up and interpretation. See how our IB maths home tutors work with DP students.
Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online. You can book a free demo class.
Frequently asked questions
- Is Euler's method in IB Maths AI HL?
- Yes. AI HL uses Euler's method for first-order differential equations, for coupled systems and for second-order equations rewritten as coupled systems.
- How do you solve coupled differential equations in AI HL?
- Write the system as a matrix, find its eigenvalues and eigenvectors, and when the eigenvalues are real and distinct write the solution as A e^(λ1 t) p1 + B e^(λ2 t) p2, using the initial values to find A and B.
- What does a phase portrait show?
- It shows how x and y change together over time, with arrows on the trajectories; the eigenvalues tell you whether the equilibrium is stable, unstable, a saddle point or a spiral.
- How is a second-order differential equation handled in AI HL?
- Set y = dx/dt to turn it into two first-order equations, then use Euler's method to step x and y forward together.
- How is AI HL different from AA HL for differential equations?
- AA HL focuses on analytical methods such as the integrating factor and homogeneous equations, while AI HL focuses on modelling, numerical methods, coupled systems and phase portraits with a GDC.
- Does Euler's method leave the IB maths syllabus in 2029?
- It leaves AA HL in the new courses first examined in May 2029, but it is not on the published list of content removed from AI HL.
