Differential Equations · Undergraduate

The Matrix Exponential e^(At)

Quick answer

The system x′ = Ax with x(0) = x₀ has the solution x(t) = e^(At)x₀, where the matrix exponential is the power series I + At + (At)²/2! + ⋯. For a diagonal matrix, exponentiate each diagonal entry. For a nilpotent matrix the series stops after a few terms. If A = PDP⁻¹, then e^(At) = Pe^(Dt)P⁻¹. When A and B commute, e^(A+B) = e^A e^B, which handles matrices like 3I + N. A useful check: det e^A = e^(tr A).

What you'll learn

  • Define the matrix exponential by its power series
  • Compute e^(At) for diagonal and nilpotent matrices
  • Use diagonalization to compute e^(At)
  • Solve x′ = Ax with x(t) = e^(At)x(0)

The series

For a square matrix AA, define

eAt=I+At+(At)22!+(At)33!+⋯e^{At} = I + At + \frac{(At)^2}{2!} + \frac{(At)^3}{3!} + \cdots

The series always converges. Differentiating term by term gives ddteAt=AeAt\tfrac{d}{dt}e^{At} = Ae^{At}, so the system x′=Ax\mathbf{x}' = A\mathbf{x} with x(0)=x0\mathbf{x}(0) = \mathbf{x}_0 has the solution x(t)=eAtx0\mathbf{x}(t) = e^{At}\mathbf{x}_0.

MatrixeAte^{At}
diagonal (a00b)\begin{pmatrix} a & 0 \\ 0 & b \end{pmatrix}(eat00ebt)\begin{pmatrix} e^{at} & 0 \\ 0 & e^{bt} \end{pmatrix}
nilpotent N=(0100)N = \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}I+Nt=(1t01)I + Nt = \begin{pmatrix} 1 & t \\ 0 & 1 \end{pmatrix}
rotation (0−110)\begin{pmatrix} 0 & -1 \\ 1 & 0 \end{pmatrix}(cos⁡t−sin⁡tsin⁡tcos⁡t)\begin{pmatrix} \cos t & -\sin t \\ \sin t & \cos t \end{pmatrix}
A=PDP−1A = PDP^{-1}PeDtP−1Pe^{Dt}P^{-1}

Why diagonalization helps

If A=PDP−1A = PDP^{-1}, then Ak=PDkP−1A^k = PD^kP^{-1}, since the inner P−1PP^{-1}P factors cancel. Every term of the series has the form P(⋅)P−1P(\cdot)P^{-1}, so the whole sum is PeDtP−1Pe^{Dt}P^{-1}, and eDte^{Dt} only needs ordinary exponentials of the eigenvalues. This is the matrix form of the eigenvector method for linear systems.

Worked examples

Common mistakes

Practice problems

  1. For A=(1003)A = \begin{pmatrix} 1 & 0 \\ 0 & 3 \end{pmatrix}, find the top-left entry of eAe^{A}.

    Answer

    ee

    Full solution

    eAe^{A} is diagonal with entries e1e^1 and e3e^3.

  2. For the same AA, find the bottom-right entry of eAte^{At} at t=ln⁡2t = \ln 2.

    Answer

    88

    Full solution

    e3ln⁡2=23e^{3\ln 2} = 2^3.

  3. For N=(0100)N = \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}, find the top-right entry of eNte^{Nt} at t=5t = 5.

    Answer

    55

    Full solution

    eNt=I+Nte^{Nt} = I + Nt, whose top-right entry is tt.

  4. For A=(2102)A = \begin{pmatrix} 2 & 1 \\ 0 & 2 \end{pmatrix}, find the top-right entry of eAe^{A}.

    Answer

    e2e^2

    Full solution

    As in Example 2, the entry is te2tte^{2t} at t=1t = 1.

  5. For A=(0−110)A = \begin{pmatrix} 0 & -1 \\ 1 & 0 \end{pmatrix}, find the top-left entry of eAte^{At} at t=πt = \pi.

    Answer

    −1-1

    Full solution

    It is cos⁡π\cos\pi.

  6. Use det⁡eA=etr⁡A\det e^{A} = e^{\operatorname{tr} A} to find det⁡eA\det e^{A} for A=(1234)A = \begin{pmatrix} 1 & 2 \\ 3 & 4 \end{pmatrix}.

    Answer

    e5e^5

    Full solution

    The trace is 1+4=51 + 4 = 5.

  7. For N=(010001000)N = \begin{pmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{pmatrix}, find the top-right entry of eNte^{Nt} at t=2t = 2.

    Answer

    22

    Full solution

    N3=0N^3 = 0 and N2N^2 has a single 11 in the top-right corner, so eNt=I+Nt+N2t22e^{Nt} = I + Nt + \tfrac{N^2t^2}{2}. The entry is t22\tfrac{t^2}{2}.

  8. A student computes eNe^{N} for N=(0100)N = \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix} by exponentiating each entry. Find the correct bottom-left entry of eNe^{N}.

    Answer

    00

    Full solution

    eN=I+Ne^{N} = I + N, whose bottom-left entry is 00.

Frequently asked questions

What is the matrix exponential?

e^(At) = I + At + (At)²/2! + (At)³/3! + ⋯, the same series as for eˣ with a matrix in place of x.

How does it solve x′ = Ax?

The derivative of e^(At) is Ae^(At), and e^(A·0) = I, so x(t) = e^(At)x(0) satisfies both the equation and the initial condition.

How do I compute e^(At) for a diagonalizable A?

Write A = PDP⁻¹. Then e^(At) = Pe^(Dt)P⁻¹, where e^(Dt) has e^(λt) on its diagonal.

Is e^(A+B) = e^A e^B?

Only when A and B commute. In general the two sides differ.

What to learn next