Signals and Systems · State-space systems, causality, and stability
Stable eigenvalues can amplify the state before every mode decays
Problem
Consider the planar linear system $\dot x=Ax$ with \[ A=\begin{pmatrix}-1&4\\0&-2\end{pmatrix}. \] Every coordinate of $x$ is a fully visible state: there is no output map, no cancelled factor, and no hidden mode. 1. Compute the eigenvalues of $A$ and prove that $A$ is Hurwitz and diagonalizable. Compute the matrix exponential \[ e^{At}=\begin{pmatrix}e^{-t}&4(e^{-t}-e^{-2t})\\0&e^{-2t}\end{pmatrix}. \] 2. Starting from $x(0)=(0,1)^{\mathsf T}$, obtain \[ x_1(t)=4(e^{-t}-e^{-2t}),\qquad x_2(t)=e^{-2t}. \] Prove that $x_1$ attains a unique maximum at $t=\ln 2$ with value $1$. Compute $\lVert x(\ln 2)\rVert_2$ and show that \[ \lVert x(\ln 2)\rVert_2=\frac{\sqrt{17}}4>1=\lVert x(0)\rVert_2. \] 3. Prove that $x(t)\to 0$ as $t\to\infty$. Explain the transient amplification: $A$ is stable but not normal. Verify $AA^{\mathsf T}\ne A^{\mathsf T}A$ by direct multiplication. 4. Audit these claims: (i) strictly negative eigenvalues force $\lVert x(t)\rVert_2$ to be monotone decreasing, so the state can never exceed its initial Euclidean norm; (ii) the observed growth is a hidden or cancelled mode, or else $A$ is a defective Jordan block rather than diagonalizable; (iii) the exercise is a Hankel-rank proof that a larger realization is nonminimal. Do not introduce an unobservable or uncontrollable mode, and do not replace the packet by a minimal-realization argument.
Hint
Solve $\dot x_2=-2x_2$ first, then integrate $\dot x_1+x_1=4x_2(t)$ with the factor $e^{t}$.
Check your work
Work the problem yourself first. Then open it in Training to check your answer and read the full worked solution.
The answer check and full solution for this problem come with ProofAnvil Practice membership ($19 USD monthly). See membership. Or start with the free Signals and Systems sample problem: Try the free sample problem.
More Signals and Systems practice problems
- Throughout this problem, t∈R is measured in secondsSignal spaces and LTI systems
- Continuous-time time scaling x(t/2) is defined for every real t whenever x is defined on RSignal spaces and LTI systems
- Let u[n]=1 for n≥ 0 and u[n]=0 for n<0Signal spaces and LTI systems
- Continuous-time signals are real functions of t∈R with t in seconds, and discrete-time…Continuous and discrete convolution
- Continuous-time systems act on real functions of t∈R with t in seconds, and…Continuous and discrete convolution
- A rectangular voltage pulse of height 9.00 V and width τ=0.200 ms, centered at the…Fourier series and transforms
- Sampling is x[n]=x_c(nT_s) with f_s=1/T_s=8000 Hz and t in secondsLaplace and z transforms
- Use the bilateral Laplace transform with absolute-convergence ROCLaplace and z transforms
- Time t is in seconds and s is in s^(-1)Sampling, aliasing, and reconstruction