0
$\begingroup$

This is a repost from MathStackExchange (https://math.stackexchange.com/q/4840786/) where however no solution has been found so far.

Given the tensor product of Hilbert spaces $\otimes_{i \in \mathcal{Z}} (\mathcal{H}_i, \psi_i)$ (here $\mathcal{Z}$ is the set of integer numbers, $\mathcal{H}_i = \mathcal{L}^2(\mathcal{R}, dx)$ and $\psi_i \in \mathcal{H}_i$), I would like to understand the following things:

  1. What is the domain for the sum of two self-adjoint operators $\hat{p}_1+\hat{p}_2$ acting on the Hilbert space $\mathcal{H}_1\otimes\mathcal{H}_2$, where each $\hat{p} = -i \frac{d}{dx}$ is the momentum operator?

  2. I would like also to understand what is the domain of existence for the operator defined as limit $\lim_{n \to \infty}\sum_{i = -n}^{n} \hat{p}_i$, and if it is still self-adjoint with dense domain. Any help is really appreciated.

P.S. Here I assume the Von Neumann definition of infinite tensor products of Hilbert spaces, but I am not sure if there are different definitions not equivalent to this one.

$\endgroup$
4
  • 1
    $\begingroup$ In which (physical) context does 2. appear? $\endgroup$ Commented Jan 10 at 8:29
  • $\begingroup$ @TobiasFünke For example in studying the macroscopic limit of a system $\endgroup$
    – MBlrd
    Commented Jan 10 at 20:02
  • $\begingroup$ OK. I don't have much experience with that, but isn't the "usual" procedure to calculate the quantities of interest first and then take the TDL? $\endgroup$ Commented Jan 10 at 20:12
  • $\begingroup$ @TobiasFünke if you have to work with states at thermodynamical equilibrium, you can impose that your state is KMS and work on that. I would like to have a better understanding on this macroscopic quantity to understand when it is well defined and when not $\endgroup$
    – MBlrd
    Commented Jan 10 at 20:16

1 Answer 1

-1
$\begingroup$

Trivially, all operators $$(A_k): \mathit H_k\to \ H_k$$ act on factor $k$ only and act as identity on the complement, so the embedding functor to its tensor form is

$$(A_k) \to 1_1 \otimes \dots\ 1_{k-1} \otimes A_k \otimes \ 1_{k+1}\dots $$

This is modelled for use of multidimensional integrals as the norm functional for each factor.

$$\int_{X_1,\dots, X_n} f_1(x_1) f_2(x_2) \dots \partial_{x_1} g_1(x_1) g_2(x_2) \dots d^nx =\int f_1(x) g'(x_1) dx_1 * \prod_{k=2}^n f_k(x_k)g_k(x) dx_2\dots dx_n $$

It follows that $$\partial_{x_k} f(x_1,\dots\, x_n)=0 \quad \text{iff the k^{th} factor is the constant function}$$ that is, by the philosophy of tensor products, the factor $$f_k(x_k)=1,$$ the product does not depend on $x_k$ or $f_k(x_k)=0$, the product belongs to the subspace without that variable.

The difficulty to construct infinite tensor products is the endless series of formation of equivalence classes yielding a transfinite limit of finite products or of infinite products yielding finite expectations.

For finite tensor products, work begins with the step, to show, that the Hilbert space of function over $\mathbb R^n$ has bases of products of a single basis in $\mathit L^2(\mathbb R)$ by their tensor product in all variables, that reduce to ordinary products under the multidimensional integral as the constituting bilinear form.

$\endgroup$
1
  • $\begingroup$ what are the f and g? $\endgroup$
    – MBlrd
    Commented Jan 10 at 20:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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