← 📐 Rank, Elimination & LU
Subtopic 2 of 4

3.2Rank

The idea

The rank counts the independent rows (equivalently, columns) of a matrix: it equals the number of pivots after elimination.

Think of it like this: The number of truly different instructions in a list, ignoring repeats and sums of earlier ones.

Why it exists

Rank decides invertibility, the number of solutions, and the dimensions of all four subspaces.

How it works

  1. Eliminate and count non-zero rows (pivots).
  2. rank A = rank Aᵀ ≤ min(m, n).
  3. rank(AB) ≤ min(rank A, rank B) and rank(A + B) ≤ rank A + rank B.
  4. An outer product uvᵀ has rank 1.

Worked example: GATE 2017

P + Q has rows R1, R2 and R3 with R3 − R2 = R1. Only two rows are independent, so the rank is 2.

Now play with it

  • →Try the "Diagonal pair" in the rules lab: rank A = rank B = 1 but rank(AB) = 0.
  • →Look for the dependent row in the 3×3 rank-2 example before eliminating.
A (3×3), pivot columns shaded
123
456
789
Reduced row echelon form
10-1
012
000
rank r2
nullity = n − r3 − 2 = 1
dim column space2
dim row space2
dim null space1
dim left null space (m − r)1

The n = 3 columns split into r = 2 (gold: pivot columns) + 1 (green: free variables).

rank A1
rank B1
rank(A + B)2
rank(AB)0
✓ rank(A + B) = 2 ≤ rank A + rank B = 2
✓ rank(AB) = 0 ≤ min(rank A, rank B) = 1
✗ rank(AB) = rank A × rank B? No: the product rule is false in general
✓ det(AB) = det A · det B   ✗ det(A + B) = det A + det B? (usually not)
FactStatement
Row rank = column rankrank A = rank Aᵀ ≤ min(m, n)
Productrank(AB) ≤ min(rank A, rank B)
Sumrank(A + B) ≤ rank A + rank B
Outer productrank(uvᵀ) = 1 for non-zero u, v
Gram matrixrank(AᵀA) = rank A
Invertiblen × n matrix has rank n ⇔ det ≠ 0

Row operations never change the rank, so reduce to echelon form and count the non-zero rows.

Check yourself

rank(AB) is always

Remember: Rank = number of pivots; rank(AB) ≤ min; rank(A + B) ≤ sum.
Exam notes (the dense version)

The rank is the number of pivots = the number of independent rows = the number of independent columns (so rank A = rank Aᵀ ≤ min(m, n)). Rules: rank(AB) ≤ min(rank A, rank B), rank(A + B) ≤ rank A + rank B, rank(AᵀA) = rank A. A matrix uvᵀ (column times row) has rank 1. An n×n matrix is invertible exactly when rank = n.

Previous year questions: Rank, Elimination & LU (4)

PYQs are grouped by topic, so this is the full set for Rank, Elimination & LU; try the ones that match this subtopic. Open all 4 on the topic page.