Linear Algebra
4.99 min read

Left Invertibility and Injectivity

A matrix ARm×nA \in \mathbb{R}^{m \times n} is left invertible if there exists LRn×mL \in \mathbb{R}^{n \times m} such that LA=InLA = I_n. Composing LL on the left of AA gives the identity.

The connection to injectivity: AA is left invertible if and only if AA is injective. Proof: LA=ILA = I means for any x\mathbf{x} with Ax=0A\mathbf{x} = \mathbf{0}, we get x=Ix=(LA)x=L(Ax)=L0=0\mathbf{x} = I\mathbf{x} = (LA)\mathbf{x} = L(A\mathbf{x}) = L\mathbf{0} = \mathbf{0}. So the null space is trivial.

The left inverse "undoes" AA — if Ax=bA\mathbf{x} = \mathbf{b}, then x=Lb\mathbf{x} = L\mathbf{b}. This only works because AA is injective: each b\mathbf{b} has at most one pre-image, so the formula is unambiguous.

Formal View

Theorem 4.9
ARm×nA \in \mathbb{R}^{m \times n} is left invertible \Longleftrightarrow AA is injective \Longleftrightarrow Nullity(A)=0\text{Nullity}(A) = 0.

Why This Matters

Left invertibility is the algebraic certificate of injectivity.

  • Least squares: the pseudoinverse A+=(ATA)1ATA^+ = (A^T A)^{-1} A^T is a left inverse when AA has independent columns
  • Compressed sensing: measurement matrices must be injective to recover sparse signals uniquely

Quiz

Question 1

A 5×35 \times 3 matrix with rank 3 is left invertible.

Common Mistakes

  • Thinking a left inverse means the matrix is square — rectangular injective matrices have left inverses too.