Linear Algebra Learning Studio · Module 03

Linear Systems and RREF

Learn augmented matrices, Gaussian elimination, reduced row-echelon form, pivots, free variables, rank, and solution classification.

Offline browser lab Interactive mathematics Student practice environment
Module 03 · Systems and elimination

Convert equations into structure, then classify every possible solution set

Gaussian elimination replaces a system with an equivalent system that is easier to read. Row operations preserve the solution set while exposing pivots, free variables, contradictions, rank, and the geometry of the solution.

Form augmented matricesUse legal row operationsRecognize REF and RREFClassify solution setsConnect pivots and rank
01

From equations to Ax=b

Separate coefficients, unknowns, and constants

Model conversion

Equation form

x + 2y − z = 3
2x − y + z = 3
−x + y + 2z = 2

Matrix form

12−1
2−11
−112
x
y
z
=
3
3
2

The augmented matrix appends b as a final column: [A|b]. Row operations act on entire equations, including the constant column.

02

Elementary row operations

Change representation without changing the solution set

Rᵢ↔Rⱼ

Swap rows

Reorder equations. This is useful when a better pivot appears lower in the matrix.

cRᵢ

Scale a row

Multiply an equation by a nonzero scalar. Never scale by zero.

Rᵢ+cRⱼ

Replace a row

Add a multiple of one row to another to eliminate an entry.

Column operations are not equation operations

Ordinary Gaussian elimination uses row operations because rows represent equations. Arbitrary column operations change the variables and therefore change the interpretation unless that change is explicitly tracked.

03

Pivots, free variables, and rank

Read the geometry from echelon structure

Pivot variables

A leading nonzero entry identifies a pivot column. The associated variable is determined by the equations once free variables are selected.

Free variables

A nonpivot variable can be assigned a parameter. Each free variable contributes one dimension to the null space.

rank(A) = number of pivot columns    |    nullity(A) = n − rank(A)
RREF evidenceConclusionGeometry
Pivot in every variable column; no contradictionUnique solutionA single point.
At least one free variable; no contradictionInfinitely many solutionsA line, plane, or higher-dimensional affine subspace.
Row [0 … 0 | c] with c≠0No solutionThe constraints do not intersect.
04

Interactive RREF laboratory

Reduce systems and inspect every row operation

Gaussian elimination and RREF analyzer

Select the number of equations and variables. The final column is the right-hand side b.

Complete row log
Augmented matrix [A | b]
05

Numerical perspective

Distinguish mathematical equivalence from stable computation

Exact classroom elimination often uses small integers and fractions. Numerical software instead chooses pivots to limit roundoff growth, commonly using partial pivoting. Solving Ax=b by elimination or a factorization is generally preferable to computing A⁻¹ explicitly and then multiplying by b.

Factorization viewpoint

Elimination can be organized as an LU factorization, A=LU (with row permutations when needed). One factorization can then solve many systems with the same A but different right-hand sides.

06

Review questions

Classify systems and interpret pivot structure

Knowledge check

Which operation is not an elementary row operation?

Knowledge check

What does a row [0 0 0 | −2] mean?

Knowledge check

A consistent system has 5 variables and rank 3. How many free variables are there?

Knowledge check

What is the relationship between RREF and the original system?

Knowledge check

When does Ax=b have a solution?