Matrices
Open in the Padhaiko app5.1 Review of algebra of real and complex matrices
Matrix operations and special matrices
A matrix of order has rows and columns. Sum and scalar multiple are entrywise; the product exists when the columns of equal the rows of , with . Matrix multiplication is associative and distributive but not commutative. and . A square matrix has an inverse iff , with .
For a complex matrix, is the entrywise conjugate and is the conjugate transpose.
| Real matrix | Condition | Complex analogue | Condition |
|---|---|---|---|
| Symmetric | Hermitian | ||
| Skew-symmetric | Skew-Hermitian | ||
| Orthogonal | Unitary |
5.2 Rank of matrices and its application in systems of linear equations
Rank
The rank of a matrix is the order of its largest non-vanishing minor, equivalently the number of non-zero rows after reducing to row echelon form by elementary row operations (swapping rows, scaling a row, adding a multiple of one row to another). Row operations do not change rank. , and .
Consistency of (Rouché–Capelli)
Let be and the augmented matrix. Then:
| Condition | Conclusion |
|---|---|
| inconsistent, no solution | |
| consistent, a unique solution | |
| consistent, infinitely many solutions with free parameters | |
| Homogeneous | always consistent; non-trivial solutions exist iff (for square : iff ) |
5.2
Examine the consistency of , , for and , and solve it when it is consistent.
Row reduction of the augmented matrix: apply and :
using last.
Ranks: in both cases. For the last row reads , so and the system is inconsistent.
For : , so there are infinitely many solutions with free parameter. Put : and .
: no solution. : , , for any .
5.3 Vector space, linear dependence and independence
Vector space and linear independence
A vector space over is a set closed under vector addition and scalar multiplication, satisfying the usual axioms (commutativity and associativity of addition, a zero vector, additive inverses, distributivity, ). Examples are , the polynomials of degree at most , and matrices. A subspace is a subset that is itself a vector space (it must contain and be closed under both operations).
Vectors are linearly dependent if for some constants not all zero; otherwise they are linearly independent. Writing the vectors as rows of a matrix: they are independent iff the rank equals (for vectors in , iff the determinant is non-zero). A basis is a linearly independent set that spans , and the number of vectors in it is the dimension.
5.4 Eigenvalues: Cayley–Hamilton theorem and its applications
Eigenvalues
For a square matrix , a non-zero vector with is an eigenvector, and the eigenvalue. Non-trivial requires , the characteristic equation. For a matrix it is , where trace and sum of the principal minors. The sum of the eigenvalues is the trace and their product is . Triangular matrices have their diagonal entries as eigenvalues.
Cayley–Hamilton theorem
Every square matrix satisfies its own characteristic equation: if is the characteristic equation of then
Applications: (i) the inverse, since and multiplying by gives ; (ii) reducing higher powers of (or polynomials in ) to a combination of .
5.5 Eigenvectors, diagonalization of matrices
Eigenvectors and diagonalization
For each eigenvalue , the eigenvectors solve . Eigenvectors belonging to distinct eigenvalues are linearly independent. If an matrix has independent eigenvectors, put them as the columns of ; then
A matrix with distinct eigenvalues is always diagonalizable. A real symmetric matrix has real eigenvalues and mutually orthogonal eigenvectors for distinct eigenvalues, so with normalised eigenvectors is orthogonal () and .
5.5
For find the eigenvalues and eigenvectors, diagonalize , verify the Cayley–Hamilton theorem, and find and .
Eigenvalues: , so (trace , determinant ).
Eigenvectors: for , gives , so . For : , so .
Diagonalization: with , and ,
Cayley–Hamilton: . Hence and
The fifth power: with :
; , ; ; .
5.6 Reduction of quadratic forms into canonical forms (three variables only)
Quadratic form and canonical form
A quadratic form in is with a real symmetric matrix (off-diagonal entries are half the coefficients of the cross terms, so -coefficient goes to ). Using the orthogonal matrix of normalised eigenvectors and the substitution ,
the canonical form (sum of squares), where are the eigenvalues of . The rank is the number of non-zero , the index the number of positive ones, and the signature is .
| Eigenvalues of | Nature of |
|---|---|
| all | positive definite |
| all | negative definite |
| all , at least one | positive semi-definite |
| all , at least one | negative semi-definite |
| both signs occur | indefinite |
5.6
Reduce to canonical form and state its nature, rank, index and signature.
Matrix: . Here , and .
Characteristic equation: . Trying : , so and .
Canonical form: with ( the orthogonal matrix of normalised eigenvectors),
All eigenvalues are positive, so is positive definite, with rank , index and signature .
; positive definite; rank , index , signature .
Practice Exercises — Matrices
Verify the Cayley–Hamilton theorem for and use it to find .
Characteristic equation: (trace , determinant ).
Verification: and , so .
Inverse: , so
.
Find the rank of .
Row reduction: and give rows and .
Next: makes the third row zero, leaving two non-zero rows. The minor confirms that the rank is at least .
.
Test for linear dependence: (a) , , ; (b) , , .
(a) The determinant , so the vectors are dependent. Indeed , i.e. .
(b) , so the vectors are independent and form a basis of .
(a) dependent, ; (b) independent.
Find the eigenvalues and eigenvectors of .
Eigenvalues: expanding along the first row, , so .
Eigenvectors: : . : gives . : gives . These are mutually orthogonal, as expected for a symmetric matrix.
with eigenvectors , , .
Show that is Hermitian, find its eigenvalues, and an eigenvector for the larger one.
Hermitian test: .
Eigenvalues: trace and determinant , so and . Both are real, as the theory predicts.
Eigenvector for : gives , so . Check: row 1 of is and row 2 is .
Hermitian; ; eigenvector for .