AP PRECALCULUS — UNIT 4B · FUNCTIONS INVOLVING PARAMETERS, VECTORS, AND MATRICES
4.14 — Matrices Modeling Contexts
Notes — Real-World Use of Matrix Multiplication
💡 Learning Objectives (4.14.A)
By the end of this lesson you will be able to:
- Build a matrix model for a real-world context (transitions, networks, encoding)
- Use matrix multiplication to evolve a state from one time period to the next
- Interpret entries and columns of a transition matrix in context
- Recognize when a matrix model has a stable long-term distribution
1. Transition Matrices
A TRANSITION MATRIX describes how a system moves from one state to another over a fixed time interval. Its (i, j) entry is the proportion (or rate) of state j that transitions to state i in one period. So each COLUMN of a transition matrix sums to 1 if probabilities are conserved.
📘 Example — Migration between cities
A = [[0.9, 0.15], [0.1, 0.85]].
- Column 1: 90% of city 1's population stays, 10% moves to city 2
- Column 2: 15% of city 2's population moves to city 1, 85% stays
- Each column sums to 1.0 ✓
2. Computing the State at the Next Time
If x_n is the state vector at time n and A is the transition matrix, then:
x_(n+1) = A · x_n
This forward-step rule lets you predict future states one step at a time, or via a power: x_n = A^n · x_0.
3. Worked Example — Population Migration
📘 Example — Predict 3 years out
A = [[0.95, 0.10], [0.05, 0.90]], starting populations x_0 = [[8000], [12000]].
- x_1 = A·x_0 = [[0.95·8000 + 0.10·12000], [0.05·8000 + 0.90·12000]] = [[8800], [11200]]
- x_2 = A·x_1 = [[0.95·8800 + 0.10·11200], [0.05·8800 + 0.90·11200]] = [[9480], [10520]]
- x_3 = A·x_2 ≈ [[10058], [9942]]
- The populations are slowly equalizing.
4. Other Modeling Contexts
- Markov chains: weather (sunny → cloudy → rainy probabilities), customer loyalty, web page rankings, biological allele frequencies
- Networks / graphs: the (i, j) entry counts the direct connections from node j to node i; A^n counts paths of length n
- Inventory / production: ‘How many units of each product can we make from x kilograms of raw material?’ leads to a matrix equation
- Encoding / cryptography: messages multiplied by an invertible matrix; receiver multiplies by the inverse to decode
5. Steady-State (Equilibrium) Distributions
A STEADY-STATE vector x* satisfies A · x* = x* — applying A doesn't change the distribution. If A is a transition matrix and we start AT x*, we stay there forever.
Solving for the steady-state: rearrange A x* = x* into (A − I) x* = 0 and solve. For most transition matrices in AP precalculus problems, there is exactly one steady-state distribution (up to scaling).
📘 Example — Find a steady state
A = [[0.9, 0.2], [0.1, 0.8]]. Find x* with x*₁ + x*₂ = 1.
- Equations: 0.9 x*₁ + 0.2 x*₂ = x*₁ ⇒ −0.1 x*₁ + 0.2 x*₂ = 0 ⇒ x*₂ = 0.5 x*₁
- Combined with x*₁ + x*₂ = 1: x*₁ + 0.5 x*₁ = 1 ⇒ x*₁ = 2/3, x*₂ = 1/3
- Steady state: 2/3 of population in city 1, 1/3 in city 2.
6. Interpreting Entries
On AP-style problems, you'll be asked what individual matrix entries mean in context. For a transition matrix A, the entry a_(ij) typically represents:
- The proportion of group j that becomes group i in one time step
- Or, in network contexts, the number of direct connections from node j to node i
- Or, in production, the amount of input j needed to make one unit of output i
7. Long-Term Predictions
For long-term predictions x_n = A^n x_0, three behaviors are common:
- Steady-state convergence: x_n → x* (a fixed distribution)
- Cyclic behavior: x_n cycles among a few values
- Unbounded growth or decay: x_n grows or shrinks without limit (less common in well-behaved models)
8. Summary
- Transition matrix: column j tells where state j moves to (entries in [0, 1], summing to 1)
- Forward step: x_(n+1) = A x_n; long-run: x_n = A^n x_0
- Steady state x* satisfies Ax* = x* — solve as a linear system
- Matrices model migrations, networks, inventories, encoding, and more
- Interpret matrix entries in context with units (proportions, counts, dollars per unit, etc.)