1.1110 min read

Applications: Measurement, Laws, and Models

Linear systems appear across science and engineering whenever we must find unknown quantities subject to constraints. Here we explore three concrete examples.

Chemical balancing: To balance $\text{CH}_4 + \text{O}_2 \to \text{CO}_2 + \text{H}_2\text{O}$ , we require the number of atoms of each element to be equal on both sides. Each element gives a linear equation in the unknown molecule counts.

Model fitting: Given data points, finding the coefficients of a polynomial or linear model that fits them best is a linear system problem. For example, fitting a line to $(x_i, y_i)$ pairs means solving $ax_i + b = y_i$ for $a$ and $b$ .

CT scan reconstruction: Each X-ray beam through the body gives one linear equation relating the densities of tissues along its path. With thousands of beams, the system of equations can reconstruct a 2D slice.

Formal View

Example 1.11 — Chemical Balancing

To balance

x_1 \text{CH}_4 + x_2 \text{O}_2 \to x_3 \text{CO}_2 + x_4 \text{H}_2\text{O}

, conservation of atoms gives: \begin{align*} \text{C:}\quad & x_1 = x_3 \\ \text{H:}\quad & 4x_1 = 2x_4 \\ \text{O:}\quad & 2x_2 = 2x_3 + x_4 \end{align*} Setting

x_1 = 1

and solving gives

(1, 2, 1, 2)

\text{CH}_4 + 2\text{O}_2 \to \text{CO}_2 + 2\text{H}_2\text{O}

Why This Matters

Linear systems are the mathematical language of constraints — they arise everywhere something must balance or sum to a target.

Medical imaging (CT, MRI) reconstructs internal structure by solving large linear systems
Structural analysis computes forces in beams and trusses via equilibrium equations
Economics finds equilibrium prices by solving supply-demand linear systems
Machine learning fits linear models by solving or minimizing systems of linear equations

Learning Resources

Balancing chemical equations

Khan Academy

Balancing equations using linear algebra.

12 min

Quiz

Question 1

Fitting a line $y = ax + b$ to three data points is always a linear system.

Common Mistakes

Assuming applications are always "nice" — real systems often have no solution or infinitely many (fitting exactly vs. overdetermined).