Let $A: D(A) \subset H \rightarrow H$ generate a strongly continuous semigroup $T(t)$ on a Hilbert space $H$ and $B\in \mathcal{B}(H)$. Consider the two control systems: $$(1)\; x'(t)=Ax(t)+ Bu(t) \qquad \text{ and } \qquad (2)\; x'(t)=R(\lambda_0,A)x(t)+Bu(t),$$ where $\lambda_0 \in \rho(A): \Re \lambda_0 \ge\omega>\omega_0(T)$ (the type of $T(t)$).


H. O. Fattorini, Some remarks on complete controllability, Siam J. Control, 4 (1966), pp. 686-694.

it is proved that the approximate controllability of $(1)$ and of $(2)$ are equivalent. See for instance Proposition 2.3.

My question: is there any relation between exact null controllability of $(1)$ and of $(2)$? If so, any reference that consider this topic would be helpful.


1 Answer 1


No. Many PDE systems, e.g. the heat and wave equation, allow exact null controllability for controls restricted to a proper subset of the physical domain. On the other hand, resolvent operators are smoothing, so if the initial condition for (2) has a singularity outside the controlled region, there is no way for the control to remove it.

  • $\begingroup$ Thank you. Could you please clarify the last claim, say for heat equation on $L^2(\Omega)$ e.g., why (2) is not exactly null controllable ? $\endgroup$
    – S. Euler
    Feb 28, 2021 at 17:29
  • $\begingroup$ If the initial condition has a singularity in the uncontrolled region, it will persist in time, because $R(\lambda_0,A)x$ is smoother than $x$ and $Bu$ is zero at the location of the singularity. $\endgroup$ Feb 28, 2021 at 19:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.