Eigenvalues and eigenvectors describe the special directions of a linear transformation. When a matrix transforms space, most vectors change both length and direction, but an eigenvector stays on its original line. It may stretch, shrink, or reverse direction, and the number that tells how much is the eigenvalue.
This idea matters in physics, engineering, computer graphics, data science, and differential equations because it reveals the natural directions of a system.
For a square matrix A, an eigenvector v and eigenvalue λ satisfy A v = λ v, where v is not the zero vector. To find eigenvalues, solve the characteristic equation det(A - λI) = 0, then substitute each λ into (A - λI)v = 0 to find its eigenvectors. Geometrically, eigenvectors are invariant directions under the transformation, while eigenvalues are scale factors along those directions.
In a 2 by 2 example, A = [[3, 1], [0, 2]] has characteristic equation (3 - λ)(2 - λ) = 0, so λ = 3 and λ = 2.
Understanding Math: Eigenvalues and Eigenvectors
A useful way to think about this topic is to compare a transformation with a machine that acts on every arrow in a plane or space. Most arrows leave the machine tilted in a new direction. The special arrows give a simpler report about what the machine does.
They show the separate growth rates built into the transformation. In three dimensions, these directions can be visualised as important axes of the action. A negative scale factor means an arrow ends up on the opposite side of the origin.
A zero scale factor means every vector on that special line is sent to the zero vector. This tells you that the transformation loses information and cannot be reversed.
Finding these directions involves two linked steps. First, students search for the scale factors that make a certain modified matrix singular. Singular means it has no inverse and its rows or columns are dependent.
That condition matters because it guarantees there is a nonzero direction left over to solve for. After each scale factor is found, substitute it back into the modified matrix and solve the resulting system of linear equations. The free variables describe all eigenvectors belonging to that scale factor.
Any nonzero multiple of one eigenvector works too. This is why an eigenvector is really a direction rather than one unique arrow.
Sometimes a matrix has repeated eigenvalues. A repeated value does not automatically provide enough independent eigenvectors. For example, a shear transformation can have one repeated eigenvalue but only one invariant line.
Such a matrix cannot be diagonalised. Diagonalisation means rewriting the transformation in a coordinate system made from eigenvectors, where the matrix has entries only along its main diagonal. In that coordinate system, each coordinate is simply scaled on its own.
This makes repeated applications much easier to calculate. A matrix raised to a large whole-number power is difficult in its original form, but diagonal form turns the task into raising separate scale factors to that power.
Students meet this idea whenever a system changes over time. In a vibrating bridge model, eigenvectors describe independent vibration patterns, while eigenvalues help determine how those patterns grow, fade, or oscillate in a related time model. In image compression and data analysis, principal directions identify the directions with the greatest spread in a data set.
In quantum physics, measured values are linked to eigenvalues of operators. When learning the method, keep track of the difference between a value and a vector. Check every proposed vector by applying the original matrix, not just the modified one.
Remember that the zero vector is never allowed as an eigenvector, since it gives no direction. Also watch for arithmetic signs in determinants, since one small sign error can change every result.
Key Facts
- Eigenvalue equation: A v = λ v, where v ≠ 0.
- Characteristic equation: det(A - λI) = 0.
- For a 2 by 2 matrix A = [[a, b], [c, d]], det(A - λI) = (a - λ)(d - λ) - bc.
- If λ > 1, the eigenvector is stretched; if 0 < λ < 1, it is shrunk; if λ < 0, it is reversed and scaled.
- For A = [[3, 1], [0, 2]], the eigenvalues are λ = 3 and λ = 2.
- Eigenvectors for different eigenvalues are linearly independent.
Vocabulary
- Eigenvalue
- An eigenvalue is the scalar λ that tells how much an eigenvector is scaled by a linear transformation.
- Eigenvector
- An eigenvector is a nonzero vector that stays on the same line after a matrix transformation.
- Characteristic equation
- The characteristic equation det(A - λI) = 0 is the equation used to find the eigenvalues of a square matrix.
- Linear transformation
- A linear transformation is a rule that maps vectors to vectors while preserving vector addition and scalar multiplication.
- Eigenspace
- An eigenspace is the set of all eigenvectors for one eigenvalue, together with the zero vector.
Common Mistakes to Avoid
- Using the zero vector as an eigenvector is wrong because A0 = λ0 is true for every λ and gives no special direction.
- Solving det(A) - λI = 0 is wrong because the determinant must be taken after subtracting λI, so the correct form is det(A - λI) = 0.
- Assuming every vector is an eigenvector is wrong because most transformed vectors change direction and do not remain on the same line.
- Forgetting to find eigenvectors after eigenvalues is incomplete because λ only gives the scale factor, while the eigenvector gives the actual invariant direction.
Practice Questions
- 1 Find the eigenvalues of A = [[4, 0], [0, 7]]. Then state one eigenvector for each eigenvalue.
- 2 For A = [[2, 1], [0, 3]], solve det(A - λI) = 0 and find an eigenvector for each eigenvalue.
- 3 A transformation sends one vector to twice itself and sends another vector to a vector pointing in a different direction. Explain which vector is an eigenvector and why.