uniolympiads Ελληνικά

Trace and determinants

The identities behind most matrix problems: tr⁡(AB)=tr⁡(BA)\tr(AB) = \tr(BA), multiplicativity of the determinant, eigenvalues and block matrices.

What you should know by heart

The trace and the determinant are the two numbers that "summarise" a square matrix without depending on the basis. Over the complex numbers, if λ1,…,λn\lambda_1, \dots, \lambda_n are the eigenvalues of AA counted with multiplicity,

tr⁡A=λ1+⋯+λn,det⁡A=λ1⋯λn.\tr A = \lambda_1 + \dots + \lambda_n, \qquad \det A = \lambda_1 \cdots \lambda_n .

Both are coefficients of the characteristic polynomial: χA(λ)=λn−(tr⁡A)λn−1+⋯+(−1)ndet⁡A\chi_A(\lambda) = \lambda^n - (\tr A)\lambda^{n-1} + \dots + (-1)^n \det A.

Trace

  1. Linearity: tr⁡(αA+βB)=αtr⁡A+βtr⁡B\tr(\alpha A + \beta B) = \alpha \tr A + \beta \tr B.
  2. Swapping a product: tr⁡(AB)=tr⁡(BA)\tr(AB) = \tr(BA), even when AA is m×nm \times n and BB is n×mn \times m.
  3. Cyclicity: tr⁡(ABC)=tr⁡(BCA)=tr⁡(CAB)\tr(ABC) = \tr(BCA) = \tr(CAB). But not every permutation.
  4. Trace of powers: tr⁡(Ak)=λ1k+⋯+λnk\tr(A^k) = \lambda_1^k + \dots + \lambda_n^k.

Item (4) is the tool for nilpotent matrices. Over a field of characteristic 00, AA is nilpotent if and only if tr⁡(Ak)=0\tr(A^k) = 0 for k=1,…,nk = 1, \dots, n. By Newton's identities, the power sums determine all the elementary symmetric polynomials of the eigenvalues, hence χA\chi_A as well, so χA(λ)=λn\chi_A(\lambda) = \lambda^n.

Determinant

  1. Multiplicativity: det⁡(AB)=det⁡Adet⁡B\det(AB) = \det A \det B for square matrices.
  2. Transpose and scaling: det⁡A⊤=det⁡A\det A^{\top} = \det A and det⁡(cA)=cndet⁡A\det(cA) = c^n \det A.
  3. Block triangular: det⁡(PQ0S)=det⁡P⋅det⁡S\det\begin{pmatrix} P & Q \\ 0 & S \end{pmatrix} = \det P \cdot \det S, with P,SP, S square blocks.
  4. Sylvester's identity: det⁡(Im+AB)=det⁡(In+BA)\det(I_m + AB) = \det(I_n + BA).

Item (3), combined with multiplication by suitable triangular matrices, gives the Schur complement: if PP is invertible,

det⁡(PQRS)=det⁡P⋅det⁡(S−RP−1Q).\det\begin{pmatrix} P & Q \\ R & S \end{pmatrix} = \det P \cdot \det\left(S - R P^{-1} Q\right).

Typical moves in problems

  • The trace as an obstruction. If a matrix identity has the form "commutator = something", take the trace.
  • The determinant as a sign. To show det⁡(⋅)≥0\det(\cdot) \geq 0, write the matrix as ZZ‾Z \overline{Z} with ZZ complex. Then det⁡=det⁡Z⋅det⁡Z‾=∣det⁡Z∣2\det = \det Z \cdot \overline{\det Z} = |\det Z|^2.
  • The matrix that joins two determinants. When you compare determinants of different sizes, build a larger block matrix that contains both.

For practice, see the problems with the tools Trace, Determinants and Block matrices.

Further reading

  • Axler, Linear Algebra Done Right, 4th ed., §8D (Trace: A Connection Between Matrices and Operators)
  • Axler, Linear Algebra Done Right, 4th ed., §9C (Determinants)
  • Gelca & Andreescu, Putnam and Beyond, §2.3.2 (Determinants)
  • Gelca & Andreescu, Putnam and Beyond, §2.3.1 (Operations with Matrices)

Contest problems using these tools

No problems with these tools yet.