1
$\begingroup$

Let $(E,\|\cdot\|)$ be a separable Hilbert space on $\mathbb{R}$. We consider a sequence of strongly continuous semigroup $\{T^{(n)}\}_{n=1}^\infty$ on $E$ [of course, for $n \in \mathbb{N}$ and $t>0$, $T^{(n)}_t$ is a bounded linear operator].

For $n \in \mathbb{N}$, we write $(A_n,\text{Dom}(A_n))$ for the generator of $T^{(n)}=\{T^{(n)}_t\}_{t>0}$. That is, \begin{align} A_nf&:=\lim_{t\to 0}\frac{T^{(n)}_tf-f}{t},\quad \text{Dom}(A_n):=\left\{f \in E : \lim_{t\to 0}\frac{T^{(n)}_tf-f}{t} \text{exists in }E\right\}. \end{align}

Question.

Let $T=\{T_t\}_{t>0}$ be a strongly continuous contraction semigroup on $E$. Then, the corresponding generator $(A,\text{Dom}(A))$ is a densely defined closed linear operator on $E$.

We now assume the following conditions.

  1. $T$ is a symmetric semigroup on $E$. In other words, $(A,\text{Dom}(A))$ is self-adjoint on $E$.
  2. For every $f \in \text{Dom}(A)$ and $n \in \mathbb{N}$, we have $f \in \text{Dom}(A_n)$ and $\lim_{n \to \infty}\|A_nf-Af\|=0$.

Under these conditions, can we show that $\lim_{n \to \infty} \|T_t^{(n)} f-T_{t}f\|=0$ for any $t>0$ and $f \in E$ ?

Note that each $T^{(n)}_t$ is not necessarily contractive. In my impression, such a result generally does not hold. However, I have no counterexamples.

$\endgroup$
5
  • $\begingroup$ If you have $\|T_t^n\| \le Me^{\omega t}$ with the same $M, \omega$, this is a version of Trotter-Kato approximation theorem. Are you asking if this is true without the above stability assumption? $\endgroup$ Commented Apr 25, 2022 at 14:07
  • $\begingroup$ @GiorgioMetafune Thank you for your comment. Yes. I'm thinking of a situation where it's not immediately clear if the stability assumption holds. $\endgroup$
    – sharpe
    Commented Apr 25, 2022 at 14:11
  • 1
    $\begingroup$ Without the stability assumption there are counterexamples, but I do not know if they appear also in your specific situation. I know F. Kuhnemund, M. Wacker: The Lie Trotter formula dose not hold for arbitrary sums of generators, Semigroup Forum 60 (2000), 478-485. $\endgroup$ Commented Apr 25, 2022 at 14:35
  • $\begingroup$ @GiorgioMetafune It is not immediately clear for me whether the sequence of semigroups in this paper converge to a symmetric semigroup. Anyway, thank you very much for the information. $\endgroup$
    – sharpe
    Commented Apr 25, 2022 at 15:02
  • $\begingroup$ That I do not know; however it is the only counterexample I know. $\endgroup$ Commented Apr 25, 2022 at 15:05

0

You must log in to answer this question.

Browse other questions tagged .