Let $\mathcal{H}$ be a separable infinite-dimensional Hilbert space and consider the bounded linear operators from $\mathcal{H}$ to $\mathcal{H}$: $\mathfrak{B}(\mathcal{H})$. Recall the definition of the Hilbert-Schmidt norm of an operator $\mathscr{A}$:

$$ \lVert \mathscr{A} \rVert_{HS}^2 = \textrm{Tr}(A^*A) = \sum_i \lVert A e_i \rVert^2 $$

where $(e_i)_{i=1}^\infty$ is a complete orthonormal system for $\mathcal{H}$. An operator with finite HS-norm is said to be Hilbert-Schmidt and we denote the set of all Hilbert-Schmidt operators from $\mathcal{H}$ to $\mathcal{H}$ by $\mathfrak{B}_2(\mathcal{H})$.

The Bochner mean of a random variable defined on a Banach space $\mathcal{X}$ (if it exists) is given by a limit of simple functions, in the sense that choosing some approximating sequence of integrable simple functions $s_n$ satisfying

$$ \lim_{n \to \infty} \int_\Omega \lVert X-s_n \rVert_\mathcal{X} \, \mathrm{d}P = 0 $$

we define

$$ E(X) = \int_\Omega X \, \mathrm{d}P = \lim_{n \to \infty} \int_\Omega s_n \, \mathrm{d}P $$

The Pettis mean of a random variable defined on a Banach space $\mathcal{X}$ (if it exists) is defined to satisfy:

$$ \varphi(E(X)) = E(\varphi(X)) \quad \forall \varphi \in \mathcal{X}^* $$


Let $(\Omega, \mathbb{F}, P)$ be a probability space and consider some random operator $\mathscr{X}: \Omega \to \mathfrak{B}(\mathcal{H})$. My question is about the interaction between evaluating a random operator and finding the expectation of the random operator.

For which of the expectations defined above (if any) does it hold that:

$$ E(\mathscr{X}(h)) = E(\mathscr{X})(h) \quad \forall h \in \mathcal{H} $$

Does it maybe hold if we assume that $\mathscr{X}$ takes values in $\mathfrak{B}_2(\mathcal{H})$ (since this space is also separable and a Hilbert space so the Bochner and Pettis integrals coincide)?


Your Answer

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

Browse other questions tagged or ask your own question.