0
$\begingroup$

Let $\mathcal{A}$ be a non-reflexive Banach algebra. For the definition of Arens product, please refer to this link. Here we let $\square$ denote the first Arens product and $\diamond$ denote the second Arens product. Let $\pi: \mathcal{A} \rightarrow \mathcal{A}^{\ast\ast}$ be the canonical embedding. For each $a\in \mathcal{A}$, define:

$$ L_a: \mathcal{A}^{\ast\ast} \rightarrow \mathcal{A}^{\ast\ast}, \hspace{0.3cm} A\mapsto a\square A $$

and:

$$ R_a: \mathcal{A}^{\ast\ast} \rightarrow \mathcal{A}^{\ast\ast}, \hspace{0.3cm} A\mapsto A\diamond a $$

Then the mapping $L: a\mapsto L_a$ defines a representation of $\mathcal{A}$. My question is:

  1. When $\mathcal{A} \square \mathcal{A}^{\ast\ast}$ (resp. $\mathcal{A}^{\ast\ast} \diamond \mathcal{A}$) is dense with respect to the norm topology, or equivalently, $\big[ L(\mathcal{A})(\mathcal{A}^{\ast\ast}) \big]^{\perp}$ (resp. $\big[ R(\mathcal{A})(\mathcal{A}^{\ast\ast}) \big]^{\perp}$) is $0$?

  2. In the case when $\mathcal{A}$ has a bounded approximate unit (with respect to the norm topology) $(a_i)_{i\in I}$, let $E$ be a weak-$\ast$ cluster point of $\big\{ \pi(a_i) \big\}_{i\in I}$. Then when $E$ is THE unit of the Banach algebra $\big( \mathcal{A}^{\ast\ast}, \square \big)$, or $\big( \mathcal{A}^{\ast\ast}, \diamond \big)$?

Update: In the case when $\mathcal{A}$ has an approximate unit, one can show that $A\square E = E\diamond A = A$. Then define:

$$ Z\big( \mathcal{A}^{\ast\ast}, \square \big) = \big\{ A\in\mathcal{A}^{\ast\ast}: \forall\, B\in \mathcal{A}^{\ast\ast},\, A\square B = B\square A \big\} $$ In the case when $\mathcal{A}$ has an approximate unit, I proved that $\mathcal{A} \subseteq Z\big( \mathcal{A}^{\ast\ast}, \square \big)$ if and only if any cluster points of $\big\{ \pi(a_i) \big\}_{i\in I}$ commutes with all elements in $\mathcal{A}^{\ast\ast}$ with respect to the both Arens products. However, I wonder if there is a necessary condition on $\mathcal{A}$ that implies such $E$ is a unit and unique.

$\endgroup$
10
  • 1
    $\begingroup$ I recommend changing the title, since neither of your questions have anything to do with irreducibility: if D is a Banach algebra and M is a left Banach D-module, then knowing that D$\cdot$M is dense in M says nothing about whether M is irreducible or not. (Consider D being the complex numbers, acting by scalar multiplication.) $\endgroup$
    – Yemon Choi
    Commented Dec 1, 2023 at 23:05
  • $\begingroup$ Thanks for pointing out since I struggled with looking for a better title for a while. Here the "irreducible" does not mean a "simple module" but I want to make it similar to a irreducible representation of a $C^*$-algebra. For instance, given a $C^*$-algebra $A$ and a $\ast$-representation $\pi: A\rightarrow B(H)$, $\pi$ is irreducible if $\pi(A)(H)$ is dense in $H$ (like my first question), or equivalently $\pi(A)'$ is trivial (like my second question). Please feel free to edit the title if you find a better word. $\endgroup$ Commented Dec 2, 2023 at 2:19
  • 2
    $\begingroup$ That condition defines a nondegenerate representation, not an irreducible one. And it's not equivalent to the commutant being trivial... $\endgroup$ Commented Dec 2, 2023 at 8:09
  • $\begingroup$ @MatthewDaws has already pointed out your error concerning terminology, but also please reread my initial comment and note that ${\mathbb C}$ is a perfectly good ${\rm C}^\ast$-algebra and the homomorphism which sends $\lambda$ to $\lambda I_2$ is a perfectly good star-representation on a Hilbert space which satisfies your density condition but is clearly not an irreducible representation, in any sense. $\endgroup$
    – Yemon Choi
    Commented Dec 2, 2023 at 12:33
  • 1
    $\begingroup$ Regarding your questions, are you aware of Arens regularity and some of the standard examples of Banach algebras that are not Arens regular? $\endgroup$
    – Yemon Choi
    Commented Dec 2, 2023 at 12:34

0

You must log in to answer this question.