A Vector That a Matrix Cannot Turn
Most vectors, when multiplied by a square matrix, change both direction and length. A small set — the eigenvectors — change only their length. Their direction is fixed by the matrix. They are the invariant directions of the transformation.
Finding the eigenvectors of a matrix is finding its hidden coordinate system — the directions along which the matrix's action becomes simple multiplication. Once you know these directions, problems involving the matrix's repeated action (powers of $A$, exponentials of $A$, differential equations) become straightforward.
What an Eigenvector Is
Given a square matrix $A$ of size $n \times n$, a nonzero vector $v$ is an eigenvector of $A$ if there exists a scalar $\lambda$ such that
$$A v = \lambda v.$$
The scalar $\lambda$ is the eigenvalue associated with $v$. The equation says: applying $A$ to $v$ produces the same vector $v$, scaled by $\lambda$.
The zero vector trivially satisfies this for any $\lambda$, so by convention the zero vector is not called an eigenvector.
Quick Facts:
Defining equation: $Av = \lambda v$ for a nonzero $v$.
Eigenvalue: the scalar $\lambda$. Can be zero, negative, or complex.
Eigenvector: the direction. Any nonzero scalar multiple of an eigenvector is also an eigenvector for the same eigenvalue.
Characteristic equation: $\det(A - \lambda I) = 0$. Its roots are the eigenvalues.
Number of eigenvalues: an $n \times n$ matrix has $n$ eigenvalues counted with multiplicity (over $\mathbb{C}$).
Grade introduced: undergraduate linear algebra; touched in CBSE Class 12 (matrices/determinants); CCSS-M HSN-VM.C.12 (work with vectors and matrices); NCERT Class 12 Chapter 3 — Matrices.
How to Find Eigenvectors
The procedure has two stages.
Stage 1 — Find the eigenvalues
Solve the characteristic equation:
$$\det(A - \lambda I) = 0.$$
This is a polynomial in $\lambda$ of degree $n$. Its $n$ roots are the eigenvalues.
Stage 2 — Find the eigenvector for each eigenvalue
For each $\lambda$, solve the homogeneous linear system:
$$(A - \lambda I) v = 0.$$
The non-trivial solutions form the eigenspace for $\lambda$. Any nonzero element of the eigenspace is an eigenvector.
Examples of Eigenvectors
Quick. Find an eigenvector of $A = \begin{pmatrix} 3 & 0 \ 0 & 2 \end{pmatrix}$.
Characteristic equation: $\det \begin{pmatrix} 3 - \lambda & 0 \ 0 & 2 - \lambda \end{pmatrix} = (3 - \lambda)(2 - \lambda) = 0$. Eigenvalues $\lambda = 3, 2$.
For $\lambda = 3$: $(A - 3I)v = \begin{pmatrix} 0 & 0 \ 0 & -1 \end{pmatrix} v = 0$ gives $v = \begin{pmatrix} 1 \ 0 \end{pmatrix}$.
For $\lambda = 2$: $v = \begin{pmatrix} 0 \ 1 \end{pmatrix}$.
Final answer: eigenpairs $(\lambda_1, v_1) = (3, (1, 0))$ and $(\lambda_2, v_2) = (2, (0, 1))$.
Standard (Wrong Path First — Where Intuition Breaks). Find the eigenvectors of $A = \begin{pmatrix} 4 & 1 \ 2 & 3 \end{pmatrix}$.
The wrong path. The rusher computes the trace ($4 + 3 = 7$) and the determinant ($12 - 2 = 10$) and assumes the eigenvalues are 4 and 3 — the diagonal entries.
The flaw: diagonal entries are eigenvalues only for diagonal (or upper-triangular) matrices. This matrix has off-diagonal entries, and the eigenvalues must be solved from the characteristic equation.
The rescue. Compute $\det(A - \lambda I) = (4 - \lambda)(3 - \lambda) - 2 = \lambda^2 - 7\lambda + 10 = (\lambda - 5)(\lambda - 2)$.
Eigenvalues: $\lambda = 5, 2$.
For $\lambda = 5$: $(A - 5I)v = \begin{pmatrix} -1 & 1 \ 2 & -2 \end{pmatrix} v = 0$. Equation $-v_1 + v_2 = 0$, so $v = (1, 1)$.
For $\lambda = 2$: $(A - 2I)v = \begin{pmatrix} 2 & 1 \ 2 & 1 \end{pmatrix} v = 0$. Equation $2v_1 + v_2 = 0$, so $v = (1, -2)$.
Final answer: $(\lambda, v) = (5, (1, 1))$ and $(2, (1, -2))$.
Stretch. Find the eigenvalues of the rotation matrix $R = \begin{pmatrix} 0 & -1 \ 1 & 0 \end{pmatrix}$ (90° rotation).
Characteristic equation: $\det \begin{pmatrix} -\lambda & -1 \ 1 & -\lambda \end{pmatrix} = \lambda^2 + 1 = 0$.
So $\lambda^2 = -1$, giving $\lambda = i$ and $\lambda = -i$.
Final answer: the eigenvalues are complex: $\lambda = \pm i$.
The complex eigenvalues are expected — a 90° rotation has no real invariant direction (every real vector is rotated, none is merely scaled). The eigenvectors live in the complex plane.
Where Eigenvectors Show Up — From Google to Quantum Mechanics
Eigenvectors are not a school exercise. They are the engine behind some of the most important algorithms and physical theories of the past century.
Google PageRank. The web is represented as a giant link matrix. The principal eigenvector of that matrix is the PageRank vector — the "importance" of every page. Larry Page and Sergey Brin won a Turing-level prize for this insight.
Principal Component Analysis (PCA). In data science, PCA finds the directions of greatest variance in a dataset — the eigenvectors of the data's covariance matrix. The first principal component is the eigenvector with the largest eigenvalue.
Quantum mechanics. The eigenvectors of an operator are the system's eigenstates. The eigenvalues are the measurable values (energy levels, spin values, position eigenstates).
Structural engineering. Buildings have natural vibration frequencies — the eigenvalues of the stiffness matrix. The eigenvectors are the mode shapes. Earthquake-resistant design ensures the building's natural frequencies do not match common earthquake frequencies.
Image compression. Singular value decomposition (SVD) uses eigenvectors of $A^T A$ to compress images.
Markov chains. The stationary distribution of a Markov chain is the principal eigenvector of the transition matrix.
The destination, in every direction: any time a transformation gets applied repeatedly, eigenvectors are the directions that survive.
The Common Errors That Cost Marks
1. Reading diagonal entries as eigenvalues for non-diagonal matrices.
Where it slips in: Matrix with non-zero off-diagonal entries — student reports diagonal entries as eigenvalues.
Don't do this: Skip the characteristic equation.
The correct way: Solve $\det(A - \lambda I) = 0$ unless the matrix is upper or lower triangular.
2. Returning the zero vector as an eigenvector.
Where it slips in: Solving $(A - \lambda I)v = 0$ — student writes $v = 0$.
Don't do this: Treat $v = 0$ as a valid solution.
The correct way: The zero vector is excluded by definition. Find a nonzero solution to $(A - \lambda I)v = 0$ — this requires the matrix $A - \lambda I$ to be singular, which is why $\lambda$ satisfies the characteristic equation in the first place.
3. Confusing eigenvalues and eigenvectors.
Where it slips in: Asked for the eigenvector, student reports the eigenvalue.
Don't do this: Mix up the scalar with the vector.
The correct way: $\lambda$ is the scalar (eigenvalue). $v$ is the vector (eigenvector). They come in pairs.
4. Missing complex eigenvalues.
Where it slips in: Rotation matrices and other matrices whose characteristic polynomial has no real roots — student declares "no eigenvalues."
Don't do this: Stop when no real solutions exist.
The correct way: Eigenvalues live in the complex plane. A real matrix can have complex eigenvalues (they come in conjugate pairs).
The real-world version. In November 1940, the Tacoma Narrows Bridge collapsed when its torsional eigenmode — one of the eigenvectors of its stiffness matrix — was excited by a sustained 64-km/h wind.
The bridge's lowest-frequency torsional eigenvalue was just below the dominant frequency of vortex shedding at that wind speed. Resonance amplified the eigenmode until the structure failed. Modern earthquake-resistant design now starts by computing every building's eigenvalues before construction.
The Mathematicians Who Discovered Eigenvectors
Augustin-Louis Cauchy (1789–1857, France) introduced the term characteristic value in 1829, proving foundational theorems about symmetric matrices.
David Hilbert (1862–1943, Germany) coined the term eigenwert (German: "characteristic value") in his work on integral equations around 1900 — the source of the modern English word.
John von Neumann (1903–1957, Hungary/USA) developed the theory of eigenvectors in infinite-dimensional Hilbert spaces, providing the mathematical framework for quantum mechanics.
Conclusion
An eigenvector of a square matrix is a nonzero vector that the matrix scales without rotating.
The scaling factor is the eigenvalue $\lambda$. The defining equation is $Av = \lambda v$.
Find eigenvalues by solving $\det(A - \lambda I) = 0$; find eigenvectors by solving $(A - \lambda I)v = 0$ for each $\lambda$.
The single most common mistake is treating diagonal entries as eigenvalues for non-diagonal matrices — always go through the characteristic equation.
Eigenvectors power PageRank, PCA, quantum mechanics, and structural engineering.
A Practical Next Step — Three Problems
Find the eigenvalues and eigenvectors of $\begin{pmatrix} 5 & 4 \ 1 & 2 \end{pmatrix}$.
Show that $\begin{pmatrix} 1 \ -1 \end{pmatrix}$ is an eigenvector of $\begin{pmatrix} 2 & 1 \ 1 & 2 \end{pmatrix}$ and find the eigenvalue.
Find the eigenvalues of the rotation matrix $\begin{pmatrix} \cos\theta & -\sin\theta \ \sin\theta & \cos\theta \end{pmatrix}$.
Want a live Bhanzu trainer to walk through more eigenvector problems? Book a free demo class — online globally.
Was this article helpful?
Your feedback helps us write better content