ProofComplete

Proof of Matrix Exponential Properties

We prove the fundamental properties of eAte^{At}, establishing it as the solution operator for constant-coefficient linear systems.


Main Result

Theorem5.9Matrix exponential as solution

For any constant n×nn \times n matrix AA, Φ(t)=eAt=∑k=0∞(At)kk!\Phi(t) = e^{At} = \sum_{k=0}^\infty \frac{(At)^k}{k!} satisfies:

  1. The series converges for all t∈Rt \in \mathbb{R}.
  2. Φ(0)=I\Phi(0) = I and Φ′(t)=AΦ(t)\Phi'(t) = A\Phi(t).
  3. Φ(t+s)=Φ(t)Φ(s)\Phi(t+s) = \Phi(t)\Phi(s) for all t,st, s.
  4. Φ(t)\Phi(t) is invertible with Φ(t)−1=Φ(−t)\Phi(t)^{-1} = \Phi(-t).
  5. det⁡Φ(t)=etr(A)t\det\Phi(t) = e^{\text{tr}(A)t}.

Proof

Proof

Convergence (Property 1). For any matrix norm, ∥Aktk/k!∥≤∥A∥k∣t∣k/k!\|A^k t^k/k!\| \leq \|A\|^k |t|^k/k!. The series ∑∥A∥k∣t∣k/k!=e∥A∥∣t∣\sum \|A\|^k|t|^k/k! = e^{\|A\||t|} converges, so the matrix series converges absolutely.

Initial condition and ODE (Property 2). At t=0t = 0: eA⋅0=I+0+⋯=Ie^{A\cdot 0} = I + 0 + \cdots = I. For the derivative: term-by-term differentiation (justified by uniform convergence on compact sets) gives

ddteAt=∑k=1∞kAktk−1k!=A∑k=1∞Ak−1tk−1(k−1)!=A∑j=0∞(At)jj!=AeAt.\frac{d}{dt}e^{At} = \sum_{k=1}^\infty \frac{kA^k t^{k-1}}{k!} = A\sum_{k=1}^\infty \frac{A^{k-1}t^{k-1}}{(k-1)!} = A\sum_{j=0}^\infty \frac{(At)^j}{j!} = Ae^{At}.

Group property (Property 3). Since AA commutes with itself, the Cauchy product of the series for eAte^{At} and eAse^{As} yields eA(t+s)e^{A(t+s)}:

eAteAs=(∑j=0∞Ajtjj!)(∑k=0∞Akskk!)=∑n=0∞∑j+k=nAntjskj!k!=∑n=0∞An(t+s)nn!=eA(t+s).e^{At}e^{As} = \left(\sum_{j=0}^\infty\frac{A^jt^j}{j!}\right)\left(\sum_{k=0}^\infty\frac{A^ks^k}{k!}\right) = \sum_{n=0}^\infty\sum_{j+k=n}\frac{A^n t^j s^k}{j!k!} = \sum_{n=0}^\infty\frac{A^n(t+s)^n}{n!} = e^{A(t+s)}.

Invertibility (Property 4). From Property 3 with s=−ts = -t: eAte−At=eA⋅0=Ie^{At}e^{-At} = e^{A\cdot 0} = I.

Determinant (Property 5). Let ϕ(t)=det⁡(eAt)\phi(t) = \det(e^{At}). Then ϕ(0)=1\phi(0) = 1 and ϕ(t+s)=ϕ(t)ϕ(s)\phi(t+s) = \phi(t)\phi(s) (since det⁡(AB)=det⁡Adet⁡B\det(AB) = \det A\det B). Differentiating: ϕ′(0)=tr(A)\phi'(0) = \text{tr}(A) (by the formula for the derivative of the determinant). The only continuous solution to ϕ(t+s)=ϕ(t)ϕ(s)\phi(t+s) = \phi(t)\phi(s) with ϕ′(0)=tr(A)\phi'(0) = \text{tr}(A) is ϕ(t)=etr(A)t\phi(t) = e^{\text{tr}(A)t}. ■\blacksquare

■

Applications

ExampleDeterminant verification

For A=(1203)A = \begin{pmatrix}1 & 2\\0 & 3\end{pmatrix}: tr(A)=4\text{tr}(A) = 4, so det⁡(eAt)=e4t\det(e^{At}) = e^{4t}.

Direct: eAt=(ete3t−et0e3t)e^{At} = \begin{pmatrix}e^t & e^{3t}-e^t\\0 & e^{3t}\end{pmatrix}, det⁡=et⋅e3t=e4t\det = e^t \cdot e^{3t} = e^{4t}.

RemarkNon-commuting matrices

If AB≠BAAB \neq BA, then in general e(A+B)t≠eAteBte^{(A+B)t} \neq e^{At}e^{Bt}. The correct formula involves the Baker-Campbell-Hausdorff series: eAeB=eA+B+12[A,B]+⋯e^A e^B = e^{A+B+\frac{1}{2}[A,B]+\cdots} where [A,B]=AB−BA[A,B] = AB-BA.