3
$\begingroup$

$\DeclareMathOperator\GL{GL}\DeclareMathOperator\SL{SL}\DeclareMathOperator\SU{SU}$I'm a PhD student in physics working in the broad area of photonic quantum computing. My current project looks at the equivalence of any two $n$-photon $m$-mode Fock states under linear optical evolution.

These two Fock states can be rewritten as symmetric and homogenous polynomials of degree $n$ in $m$ variables. The linear optical evolution implies a linear $\SU(m)$ i.e. special unitary matrices of dimension $m$, change of basis for these variables. So basically, given any two Fock states, I'm interested in calculating the polynomial invariants of representations of the compact Lie group $\SU(m)$.

I've been going through the book ‘Algorithms in invariant theory’ by Bernd Sturmfels but it only mentions the first fundamental theorem for $\GL(m,\mathbb{C})$ and $\SL(m,\mathbb{C})$ Lie groups. Does there then exist a relation between the polynomial invariants of $\SU(m)$ and its complexification $\SL(m,\mathbb{C})$?

The only related result I could find was in the article ‘Lifting smooth homotopies of orbit spaces’ by Gerald Schwarz. The proposition (5.8) in it mentions the following algebra isomorphism:

\begin{equation} \mathbb{R}[W]^{\rho (K)} \otimes \mathbb{C} \simeq \mathbb{C}[V]^{\sigma (G)}, \end{equation}

where $K$ is a compact Lie group and $G = K_{\mathbb{C}}$ is its complexification. Their representations are also defined as $\rho: K \rightarrow \GL(W)$ and $\sigma: G \rightarrow \GL(V)$, where W is a real vector space and $V = W_{\mathbb{C}}=W\otimes_{\mathbb{R}}\mathbb{C}$.

I was hoping to get an easier description or examples of this algebra isomorphism if possible or any resources that point to the same. Also, is it possible to calculate the non-polynomial invariants that are related to the $L^{2}$ norm of the Fock state polynomials since they are also invariants under the $\SU(m)$ action?

$\endgroup$
5
  • 3
    $\begingroup$ $\mathrm{SU}(m)$ is Zariski-dense in the complex group $\mathrm{SL}_n(\mathbf{C})$ and hence they have the same polynomial invariants for every (complex algebraic — equivalently holomorphic) finite-dimensional representation. [This is what Vladimir Dotsenko refers to as "Weyl unitary trick.] $\endgroup$
    – YCor
    Commented Oct 5, 2022 at 15:50
  • $\begingroup$ @‍VladimirDotsenko's reference to Weyl's unitary trick referenced by @YCor. $\endgroup$
    – LSpice
    Commented Oct 6, 2022 at 3:37
  • $\begingroup$ @YCor Thanks for your answer. Would you also know of any references or methods that talk about how to derive the 'non-polynomial' invariants of SU(n) then? For example, I'm interested in the L2 norm and the Bombeiri norm of the complex coefficients of my original symmetric and homogenous polynomials - which are also SU(n) invariant. These norms are polynomials in not just the coefficients but also their complex conjugates. I can't see how they can be expressed as linear combinations of the SL(n,C) invariants which are polynomials only in the coefficients but not their complex conjugates. $\endgroup$ Commented Oct 25, 2022 at 2:23
  • $\begingroup$ @DeepeshSingh yes you're right. The Weyl unitary trick only concerns the $\mathrm{SU}(m)$-invariants that ate polynomials in the coefficients. The Proposition 5.8 you're mentioning will not address polynomials also involving conjugates of coefficients. $\endgroup$
    – YCor
    Commented Oct 25, 2022 at 6:28
  • $\begingroup$ @YCor Thanks for your reply. Yes, Proposition 5.8 mentioned above doesn't provide us with extra 'non-polynomial' invariants. Would you however know of any references that comment on the same? I'm interested in knowing if the SL(n,C) invariants are sufficient or only necessary to show equivalence under SU(n) action. I've asked the new question here: mathoverflow.net/questions/432340/…. Thanks again! $\endgroup$ Commented Oct 25, 2022 at 9:02

1 Answer 1

3
$\begingroup$

The method/result you are looking for is commonly known under the name "unitary trick" (of Hurwitz and Weyl), - this keyword should bring you a great deal of accessible explanations of all possible levels (since your post does not give a clear idea of your level, it is better if you look through several different sources and choose a suitable one).

$\endgroup$
2
  • $\begingroup$ Thanks for your answer, Vladimir. Would you also know of any references or methods that talk about how to derive the 'non-polynomial' invariants of SU(n) then? For example, I'm interested in the L2 norm and the Bombeiri norm of the complex coefficients of my original symmetric and homogenous polynomials - which are also SU(n) invariant. These norms are polynomials in not just the coefficients but also their complex conjugates. I can't see how they can be expressed as linear combinations of the SL(n,C) invariants which are polynomials only in the coefficients but not their complex conjugates. $\endgroup$ Commented Oct 25, 2022 at 2:23
  • $\begingroup$ I've asked this question in more detail here: mathoverflow.net/questions/432340/… $\endgroup$ Commented Oct 25, 2022 at 2:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .