Opening subject page...
Loading your content
Discover how two special matrices mirror the roles of 0 and 1 in everyday arithmetic—and why that matters for every matrix calculation you'll ever do.
Long before calculators could multiply matrices in a blink, mathematicians needed a systematic way to solve large systems of equations. Matrices—rectangular arrays of numbers—emerged gradually as a bookkeeping tool, and with them came two special players whose behavior mirrors what you already know about 0 (the additive identity) and 1 (the multiplicative identity) in the real numbers. Understanding their backstory will help you see why mathematicians defined them exactly the way they did.
In short, mathematicians recognized that the arithmetic you learned in middle school—adding zero changes nothing, multiplying by one changes nothing—needed exact counterparts in the world of matrices. The zero matrix and the identity matrix fill those roles perfectly, and that's the concept this lesson unpacks.
Before we dive into diagrams and equations, let's nail down four foundational ideas. If you've already mastered basic matrix addition and multiplication, these should feel natural—like giving names to behavior you've already seen.
Matrices are grids of numbers, so let's lay them out visually. The first diagram below shows how the zero matrix acts under addition; the second shows how the identity matrix acts under multiplication. Pay close attention to which entries combine and why the result always matches the original matrix A.
In the diagram above, every entry of A is paired with a corresponding 0 from the zero matrix. The sum for each position is simply the original entry, so the result is a copy of A. This is exactly how adding 0 to a real number works—just extended to every cell of the grid.
The identity matrix's 1s sit on the main diagonal (top-left to bottom-right), and every off-diagonal entry is 0. When you compute each dot product (row of A times column of I), the 1 "picks out" the entry you want, and the 0s annihilate everything else. The result? A carbon copy of A.
Let's formalize what the diagrams just showed us. These properties hold for matrices of compatible dimensions, just as their real-number counterparts hold for all real numbers.
In this equation, every entry of O is zero, so the (i, j)-entry of A + O is simply aij + 0 = aij. Notice that O must share the same dimensions as A; there is no single universal zero matrix—its size depends on context, just like there is only one real number 0, but many differently-shaped zero matrices.
The identity matrix Iₙ is the n × n square matrix whose (i, j)-entry equals 1 when i = j and 0 otherwise. This is often written using the Kronecker delta: δij. When A is square (m = n), the equation simplifies to A · I = I · A = A.
Just as a + (−a) = 0 in the reals, every matrix has an additive inverse (negate every entry), and their sum is the zero matrix. This makes the set of m × n matrices a mathematical structure called a vector space—but that's a topic for a future course.
This mirrors the real-number fact that a × 0 = 0. No matter what A contains, all those dot products reduce to sums of zeros.
One subtlety that trips students up is that the zero matrix and identity matrix come in different sizes. The table below catalogues some common examples and clarifies when each form appears.
| Matrix | Size | Appearance | Role |
|---|---|---|---|
| O2×2 | 2 × 2 | [ 0 0 ]
[ 0 0 ] | Additive identity for any 2 × 2 matrix |
| O2×3 | 2 × 3 | [ 0 0 0 ]
[ 0 0 0 ] | Additive identity for any 2 × 3 matrix |
| I2 | 2 × 2 | [ 1 0 ]
[ 0 1 ] | Multiplicative identity for 2 × 2 matrices |
| I3 | 3 × 3 | [ 1 0 0 ]
[ 0 1 0 ]
[ 0 0 1 ] | Multiplicative identity for 3 × 3 matrices |
| I4 | 4 × 4 | [ 1 0 0 0 ]
[ 0 1 0 0 ]
[ 0 0 1 0 ]
[ 0 0 0 1 ] | Used in 3-D graphics transformations |
A crucial point: the identity matrix is always square (n × n), because matrix multiplication A · I requires the number of columns in A to match the number of rows in I. The zero matrix, on the other hand, can be any rectangular shape, since matrix addition only requires matching dimensions. When a problem says "the identity matrix" without specifying size, the size is inferred from the matrix it's being multiplied with.
Let's verify both identity properties with a concrete 3 × 3 matrix. We'll start with addition and then move to multiplication.
A = [ 4 −1 6 ]
[ 0 3 −5 ]
[ 2 8 1 ]A + O₃ₓ₃ = [ 4+0 −1+0 6+0 ] = [ 4 −1 6 ]
[ 0+0 3+0 −5+0 ] = [ 0 3 −5 ]
[ 2+0 8+0 1+0 ] = [ 2 8 1 ]A · I₃ = [ 4 −1 6 ]
[ 0 3 −5 ]
[ 2 8 1 ]The analogy between real-number arithmetic and matrix arithmetic is surprisingly tight, but there are important differences. The table below puts the parallels—and the divergences—side by side.
| Property | Real Numbers | Matrices |
|---|---|---|
| Additive identity | a + 0 = a | A + O = A |
| Multiplicative identity | a × 1 = a | A · I = A |
| Additive inverse | a + (−a) = 0 | A + (−A) = O |
| Multiplicative inverse | a × (1/a) = 1 (a ≠ 0) | A · A⁻¹ = I (if A is invertible) |
| Multiplication by zero | a × 0 = 0 | A · O = O |
| Commutativity of × | Always: ab = ba | Not guaranteed: AB ≠ BA in general |
| Zero-product property | ab = 0 ⟹ a = 0 or b = 0 | Fails: AB = O is possible with A ≠ O and B ≠ O |
The last two rows highlight where the analogy breaks down. Matrix multiplication is not commutative: the order matters. And unlike real numbers, two nonzero matrices can multiply to give the zero matrix. Keep these exceptions in mind as you move forward, even though the identity properties themselves transfer perfectly.
You might wonder: "Why make such a big deal about two simple matrices?" The answer is that the zero and identity matrices are the foundation for nearly everything else in linear algebra. Here's a preview of where these concepts lead.
| Concept You Know | Where It Leads |
|---|---|
| A · I = A | Matrix inverses: Finding A−1 such that A · A−1 = I. This is how you "divide" by a matrix to solve systems of equations. |
| A + O = A | Vector spaces: The zero matrix is the zero vector of the space of m × n matrices, a key object in abstract algebra and functional analysis. |
| det(I) = 1 | Determinants: The identity matrix has determinant 1, which connects to volume scaling, invertibility tests, and eigenvalue theory. |
| I has 1s on diagonal | Eigenvalues: Finding eigenvalues involves solving det(A − λI) = 0, where λI is a scalar multiple of the identity. |
In a college-level linear algebra course, you'll see that the identity matrix appears in nearly every major formula—from the characteristic equation to change-of-basis transformations. For now, the essential insight is this: I and O aren't just curiosities. They are the structural pillars on which all of matrix algebra is built, exactly as 0 and 1 are the pillars of ordinary arithmetic.
[ 5 −3 ] / [ 2 8 ] (a 2 × 2 matrix). Compute B + O2×2 and verify it equals B.[ −4 7 ] / [ 1 −9 ]. Show each dot-product calculation.[ 10 ] / [ −5 ] / [ 3 ]. To "apply no transformation," the engine multiplies p by I3. Verify that I3 · p = p, then explain why a programmer might ever intentionally use the identity matrix (instead of simply skipping the multiplication).The zero matrix O and the identity matrix I are the matrix equivalents of the numbers 0 and 1 in real-number arithmetic. Adding the zero matrix to any matrix A of matching dimensions leaves A unchanged (A + O = A), and multiplying any square matrix A by the appropriately sized identity matrix also leaves it unchanged (A · I = I · A = A). These properties—the additive identity and multiplicative identity—are not just algebraic curiosities; they are the structural backbone of matrix algebra, enabling concepts like matrix inverses, determinants, and eigenvalues.
Key nuances to remember: the zero matrix can be any rectangular shape, while the identity matrix is always square. Matrix multiplication is not commutative in general, and the zero-product property fails for matrices. Despite these differences, the identity properties transfer perfectly from the reals, giving you a familiar anchor as you navigate the more complex landscape of matrix operations.